Sciencemadness Discussion Board

Can Digital Photography be used as the receptor array for FT-IR?

LSD25 - 7-2-2008 at 14:07

I have been putting some thought into this,

Now there are very small Near Infra-Red, Mid-Infra-Red & Far-Infrared filtered Digital camera lenses as well as the usual visible wavelength lenses.

I have also been looking up the use of diffraction gratings and how they split various light up into it's wavelengths, and more especially the overlap between the various sub-divisions of light when passed through a grating.

http://scitoys.com/scitoys/scitoys/light/cd_spectroscope/spe...

http://www.jobinyvon.com/SiteResources/Data/Templates/1divis...

http://www.rsc.org/ej/JA/2000/b001066i/b001066i-f2.gif

Now, I may well be wrong, but it seems that if one mounted 4 (or more) lenses on a rotary servo (Vis, NIR, MIR & FIR), it should be possible to record the amount of each wavelength passing through the grating. It might even be possible to use a high shutter speed and adapt a cheap(ish) digital camera to take onboard the images viewed through all 4 of the rotating lenses. Now if this was passed through to a PC it might even be possible to write some basic code (or to modify the code on an existing shareware package) to analyse the images, extract the peaks and identify the wavelengths.

However, it might well be a lot easier to manually plot the various points on the images with the computer, identify the wavelengths and then allow the computer to extrapolate from this to provide an overall graph of the data taken from the images.

Just a thought, given the extraordinary prices charged for this machinery it might well be that a useable variant may be able to be made at home (it may lack a lot of things, but that is life). Anyone with ideas as to whether or not this will work are more than welcome to have their say, as I said, this is only an idea - I am far from convinced that it will work.

niggaknow - 7-2-2008 at 17:16

Everything you always wanted to know about IR, but were afraid to ask.

http://www.4shared.com/dir/2077108/ab6c3d06/sharing.html

http://www.geocities.com/antiquesci/PE337/PE337-1.htm

[Edited on 7-2-2008 by niggaknow]

niggaknow - 7-2-2008 at 17:23

See this article.

Attachment: JCE1992p0077-monochromators.pdf (1.5MB)
This file has been downloaded 606 times

niggaknow - 7-2-2008 at 17:37

FT-IR specs on the market (with Michelson interferometers) are different from the dispersive type (gratings and prisms).

Attachment: JCE1979p0681-FTIR.pdf (2.5MB)
This file has been downloaded 735 times

not_important - 7-2-2008 at 17:44

OK, first off the CCD in digital camera only responds to about 1100 nm http://www.andor.com/learn/digital_cameras/?docID=315 so you're not going to even get all the NIR. Here's a short list of detectors used for IR work, you can use it as a starting point for further research http://www.brukeroptics.com/accessories/Notes/FTIR/detectors...

Second FT-IR is Fourier transform IR, which does not use dispersive methods but rather is a multiplex spectrometer. You can build dispersive IR gear, that's what I learned on, but calling it FT-IR is wrong.

[Edited on 8-2-2008 by not_important]

LSD25 - 11-2-2008 at 03:22

OK,

Finally some feedback (Excuse my not replying earlier, I had to track down the password, etc.).

I would like to now more, I simply (obviously erroneously) believed that by using a single light source/pulse and a diffraction grating that all such variants were described as variants of FT-IR spectroscopy.

As to the limits on the digital cameras, that is one of the things I feared - but did not know. Thank you.

Do you happen to know if there is any way one could use the IR photodiode-array from a CD/DVD player and the various parts therein (although I believe a different light source would be needed) in order to come closer?

Would it be possible to come up with a way to build something approximating those extremely expensive array's?

The major difficulty I have with those units is the price - no question - but it would also be nice to come up with something...

For instance, is it possible to use Near IR spectroscopy to identify various functional groups? Or is it possible to use some of these ideas with N/IR microscopy (I have a wierd little paper where they use photoshop to increase the contrast between different areas of digital photography - could this be useful?).

I have done a few searches on NIR spectroscopy & here are some of the more interesting results:

http://www.thermo.com/com/cda/technology/detail/0,2165,13191...

http://www.spectroscopynow.com/coi/cda/detail.cda?id=1881&am...

http://tinyurl.com/2d4zh5

In any event, it may be possible for members here to build UV-Vis/NIR spectorophotoscopes and build up a database of various compounds. Although it appears that these are nowhere near as well defined as IR spectroscopy, this would have to be set-off against the cost and availability of the necessary equipment. In any event, if such a thing were possible it would represent a large leap forward when compared to the extremely basic analytical options available to most members now.

From the link you posted, a digital camera may well be an option as a replacement for expensive arrays in this respect, while a light source will still be needed, this is a step forward.

I'd seriously like this to be worked through to a conclusion, it would be nice to know with some precision what is in the tubes after working up a reaction. It may not be unequivocal, but it would provide a sound starting point for working out what is happening.

not_important - 11-2-2008 at 07:01

Near IR is mostly used for (C,O,N)-H and C=O bond analysis, meaning you're not looking at carbon-carbon and maybe not carbon-oxygen single bond as I recall.

The cutoff of CCD arrays is such that your are looking at 3rd and 4th harmonics, which are rather weak and broad; I don't know how useful they are without the lower overtones and fundamental.

The photodiodes for CD/DVD application are in a array of eight diodes that really would effectively be only six in dispersive applications. They are also optimized for wavelengths right on the edge of visible down into the red proper http://www.datasheetcatalog.com/datasheets_pdf/T/Z/A/1/TZA10...

UV/Vis gives different sorts of information, useful when you know what is in there and want to know how much, but less so when you're trying to find out exactly got brewed up.

For near-UV/Vis/very-near IR people have used the imaging arrays of scanners as detectors in dispersive arrays, the one dimensional stream of data is just fine for looking at a spectrum. You get 1500 to 4000 points of data, which is a high enough resolution to be real useful; a 2D sensor would hand you much more data which you would the need to fold down into a 1D stream.

Using a laser, some narrow band pass and block optical filters, and a CCD array you might be able to do Raman spectroscopy, which can give similar information to that from standard IR spectroscopy.

FT spectroscopy does not use dispersive elements, prisms or gratings, but basically collects a blob of frequencies with a single detector and throws computers at the data to get a spectrum. The detectors can be expensive because of the difficulty of getting rapid, sensitive detection across that range of frequencies; some include integral thermoelectric coolers.


http://www.chemistry.adelaide.edu.au/external/soc-rel/conten...

vulture - 11-2-2008 at 13:23

Most CCDs are in fact sensitive to IR light, try pointing a remote at a digital camera. However, most cameras are also equipped with a hotmirror to block as much IR light as possible (it leads to moire and blooming IIRC). So you might want to check astrophotography sites as they usually do IR conversions to get that end of the night sky spectrum.

not_important - 11-2-2008 at 15:13

Quote:

Most CCDs are in fact sensitive to IR light, try pointing a remote at a digital camera. However, most cameras are also equipped with a hotmirror to block as much IR light as possible (it leads to moire and blooming IIRC). ....


This is true, but as shown in the link I gave, this is the very near IR. Most CCDs themselves poop out around 1000 nm, silicon CCDs can't go above 1,1 micron. See http://www.wrotniak.net/photo/infrared/ for IR and digital cameras.

The Fuji S3 Pro UVIR cameras were designed to support UV and IR photography. Spectral response is 350 to 1000 nm.



Even NIR is considered as 0,8 to 2,5 microns, the 1,1 limit only takes to higher overtones of the bands in that region.

LSD25 - 12-2-2008 at 01:16

Righto,

Here are a couple of links to amateur websites (actually pages) where they have modified cameras for astrophotography by removing the IR filters and they give details of how to acheive this.

http://www.vub.ac.be/STER/www.astro/AstroCCD/AstroQC.htm

http://www.ayton.id.au/gary/Science/Astronomy/Ast_PhotoDigit...

Here is an abstract of an article on the limits and details of astrophotography:

http://adsabs.harvard.edu/abs/1976S&W....15....4B

Thanks Vulture...

Not Important,

given the discussion in the google book I linked to above (read as much of that as possible - it is truly interesting), is there any chance of using these modified cameras in the manner suggested? (I had a look at the Fuji & also the Nikon, but the prices are horrendous & if amateur photogaphers are modifying cameras to get similar shots, I'd like to imagine the members here were able to do the same).

Here for instance is a page (including spectra) examining the perfomance of several cameras and modified variants of them in the relevant range:

http://www.pbase.com/terrylovejoy/image/29755088

NB This site also gives details on how to modify the cameras in question as well as great links to other sites.

[Edited on 12-2-2008 by LSD25]

not_important - 12-2-2008 at 02:59

I'm not sure what you mean by " using these modified cameras in the manner suggested".

They will work with a dispersive element for visible light spectroscopy, more properly for the near UV down to aboy 350 nm on up to 900 to 1000 nm. (note that human vision goes up to the high 600s to mid 700s, depending on the person)

This is useful for atomic emission spectra, and absorption spectra. With a detector with a large enough number of elements the UV end can be extended downwards using a phosphor to convert the short UV to visible or near UV that the detector will sense.

For absorption use, traditional uv-vis model, you don't even need a lot of elements as the bands generally are rather wide. Even the cheap "webcam" units will give enough resolution for that. However:
Quote:
The uv-vis spectra have broad features that are of limited use for sample identification but are very useful for quantitative measurements. The concentration of an analyte in solution can be determined by measuring the absorbance at some wavelength and applying the Beer-Lambert Law.


For emission spectra - atomic spectroscopy, atomic-absorption, and atomic fluorescence, you'll want more resolution, which means more elements in the detector and better dispersive optics. That's why I suggested a flatbed scanner detector as these generally have 2000 to 4000 elements, which very roughly corresponds to a 4 to 16 megapixel camera. However because they are linear sensors you only have to deal with a few 10s of kb of data per scan, instead of the megabytes of a digital camera.

Note the 'atomic' in these types of spectoscopy. You are dealing with electrons around isolated atoms, not molecular structures (except for some high temperature stable simple compounds). For organic molecular structures you need near-to-mid Iror Raman equipment. The first can't use the CCDs sensitive range, the second can but also needs narrow line lasers and optical filters. The sensor is not the most expensive element. These types of spectroscopy usually benefit from FT methods, see the Fellgett advantage, and use a single sensor element.

One other place digital cameras might be useful is with chromatography, where the extended pickup range and contrast enhancement might help visualise spots. Only simple software qould be needed, IrfanView would likely do the job.

It is important to note that the astronomers often refer to the the very near IR range as simply 'IR'. The atmospheric absorption bands start punching holes in the spectrum at about 800 nm with strong, wide bands around 4 and 5-8 microns. This limits astronomical viewing to windows such as those at ~870, 1100, and 1500 nm. The astronomers IR includes some of the emission lines of hydrogen, iron, and helium. Not all of these can be detected by silicon CCDs, as they lie above 1 micron in the NIR proper.

http://www.globalwarmingart.com/wiki/Image:Atmospheric_Absor...

http://www.atnf.csiro.au/people/mdahlem/pop/multi/ir.html

in a fashion related
http://pagesperso-orange.fr/redlum.xohp/laser/gratingOSA.htm...

LSD25 - 12-2-2008 at 03:29

Ok, Long, long way above my head at present - I will be doing some more reading and then I'll modify this post.

In case I forget, thanks for your help.

[EDIT No.1]

I decided to go right to the start of what you had written above and look up this:

http://en.wikipedia.org/wiki/Charge-coupled_device

I'll now start from there in an attempt to get up to speed.

[Edited on 12-2-2008 by LSD25]

[EDIT No.2]

I looked this up:

http://en.wikipedia.org/wiki/Image_scanner#Applications_Prog...

I must point out that the use of infrared cleaning as is mentioned at the bottom of the page - or the use of infrared light to find those parts of a scan which absorb/filter IR - suggests that not only do scanners have the capacity to decode IR signals (or probably NIR), but they can be used to pick up absorbance. Is this correct? or am I on the wrong track?

[Edited on 12-2-2008 by LSD25]

not_important - 12-2-2008 at 06:56

IR cleaning -

check the link to the full article; they use a NIR source to get a NIR image. This technique is being applied to colour film, which is somewhat opaque to NIR no matter what the visible colour is. Scratches and dust have a different brightness than the relatively flat toned IR background, so the IR imge can be enhanced to create and adjustment mask.

The scanner CCD isn't giving a special "I'm NIR" signal out, except in the context of "there is NIR illumination being supplied"

The scanner CCD has similar properties to the camera CCDs, the same 350 to 1000 nm range.

LSD25 - 12-2-2008 at 12:34

Yeah, I thought as much - but as NIR offers probably the best chance of this being developed for home use and also offers the use of glass containers for samples, this is probably the best we could hope for.

NIR spectroscopy is gaining some serious attention according to the literature:

Handbook of NIR Spectroscopy (Googlebook): http://tinyurl.com/ywtbg2

A brochure:

http://las.perkinelmer.com/Content/RelatedMaterials/Brochure...

A review article dealing with NIR spectroscopy:

http://www.odin.life.ku.dk/News_letters/Q2_Q3_2005/news001.p...

This is good as it gives a basic overview of the strengths & weaknesses of NIR, as well as giving some detail of the procedures used.

not_important - 12-2-2008 at 15:07

I fear that it is of limit use in the way that you seek, as it really only gives information on hydrogen to (C, N, O, S) bonds, and very little on the carbon skeleton. Besides that:
Quote:
In conclusion, NIR absorption bands are typically broad, overlapping and 10–100 times weaker than their corresponding fundamental mid-IR absorption bands. These characteristics severely restrict sensitivity in the classical spectroscopic sense and call for chemometric data processing to relate spectral information to sample properties
...
pharmaceutical industry and regulatory bodies have been slow to adopt the NIR technique, most probably since it lacks the ability of mid-IR to identify samples by mere inspection of spectra and involves calibration by sophisticated mathematical techniques


For identification you need to have the sample libraries built up, which sort of defeats the purpose of "know with some precision what is in the tubes after working up a reaction" unless you aleady can make the various products in pure form and are optimising the reaction or trying alternative methods.

The main advantages, being able to measure samples in containers or tablets, is of little use to the amateur lab, where it is easy to obtain a drop of the materials for analysis and generally it is not intentionally being mixed with other materials besides what's in the reaction mix.

The other problem with what you propose is that the Si CCDs only sense up to 1000 nm, maybe 1100 nm in some types with cryocooling. The real information in NIR is in the 1500 to 3000 nm range, and you data from most of the NIR range to untangle the signals into something useful.

LSD25 - 13-2-2008 at 06:06

Yeah, I get that it is nowhere near as useful as the dedicated MIR range... Nonetheless it may provide a useful tool for a home lab and that is all that I was considering it as. I am willing to do the testing and develop the process with solvents and pure reagents to start with in order to develop the procedure, as well as being willing to separate the component fractions of the mixture in order to try and determine which one is the one I wish to keep.

Even if this works out to be of little utility, it is an interesting exercise and given the current price of electronics - it is hardly likely to break the bank. If it also provides a useful starting point for further development, that would be a plus - but the entire topic came from wild speculation, so I am suprised that it has any utility at all.

not_important - 13-2-2008 at 20:25

It's just not the NIR is less useful for identification of unknowns, but that the range that silicon CCDs respond to has only very limited usefulness. I just don't think you'll find it useful for product identification, especially as there is little to no data on the carbon skeleton - n-propyl, iso-propyl, allyl, and tert-butyl alcohol would all look rather similar, you might not even see carbonyl oxygens, ethers, and substituted amines and amides.

First, note that there can be confusion as to what is meant by the term NIR

Astronomers typically use:
* near: (0.7-1) to 5 µm
* mid: 5 to (25-40) µm
* long: (25-40) to (200-350) µm

Based on the types of detectors that are useful:

* Near infrared (NIR): from 0.7 to 1.0 micrometers, from the approximate end of the response of the human eye to that of silicon photodiodes
* Short-wave infrared (SWIR): 1.0 to 3 micrometers (from the cut off of silicon to that of the MWIR atmospheric window. InGaAs covers to about 1.8 micrometers; the less sensitive lead salts cover this region
* Mid-wave infrared (MWIR): 3 to 5 micrometers (defined by the atmospheric window and covered by InSb and HgCdTe and partially by PbSe
* Long-wave infrared (LWIR): 8 to 12, or 7 to 14 micrometers: the atmospheric window, covered by HgCdTe and microbolometers
* Very-long wave infrared (VLWIR): 12 to about 30 micrometers, covered by doped silicon, microbolometers, and pyroelectric detectors.

Yet another scheme:

* Near-infrared : 0.75-1.4 µm
* Short-wavelength infrared 1.4-3 µm
* Mid-wavelength infrared also called intermediate infrared (IIR): 3-8 µm. Used to track jet aircraft by their engine plumes.
* Long-wavelength infrared : 8–15 µm, thermal imaging
* Far infrared (FIR): 15-1,000 µm

Second, the information in this range is pretty piss poor. Look at the chart on this web page - the frequency range you are trying to use has bands that are 3rd and 4th overtone. The intensity is 1/1000 to one ten-thousandth as strong as the fundamental in the NIR, generally used for IR spectroscopy. Path lengths are on the order of 10 cm, besides being weak the bands are broad, fuzzy, and over-lapping.

http://www.spectroscopynow.com/coi/cda/detail.cda?id=1881&am...

The main use of the range covered by Si CCDs seems to be measuring the ration of reflected light at 500 and 800-900 nm in order to determine the amount of vegetative coverage, and combined with 350-450 nm estimating soil moisture. All of that is based on the chlorophyll absorption spectrum.

Aromatic Thai Rice Identification by Near-infrared Reflectance - starting at 1300 mn and longer
http://www.tropentag.de/2005/abstracts/full/158.pdf

Mini-NIR spectrophotometer for measuring water content
http://newjournal.kcsnet.or.kr/main/j_search/j_download.htm?...


Attached is a chart of NIR bands. Remember that 3rd overtone is a bit like identifying distance shapes in heavy fog - that blob can be a person, shrub, or mailbox.

Attachment: nir-chart_grid_rev-3.pdf (668kB)
This file has been downloaded 2525 times

LSD25 - 14-2-2008 at 01:01

Thanks for that, my view is that as this is THE current research topic in Spectroscopy (as demonstrated by the sheer number of articles - because industry wants to be able to spend less money for better data in process management), I suspect that this area of spectroscopy will become more and more useful in the near future.

In fact, there are a large number of scholarly articles being written about the more accessible ranges of NIR and how to make better use of the data provided thereby. While this is not at a breakthrough stage, I strongly suspect (given the apparent volume of time and money thrown at the problem) that somebody will make a fundamental discovery to make better use of this data (I'll attach some of the more interesting ones soon).

I'd dearly like to be in a position to take advantage of such a breakthrough - I know I ain't smart enough or knowledgable enough to solve these problems, although I enjoy trying to take advantage of the hard work of others (standing on the shoulders of giants, so to speak). The best part of this idea is that unlike most labs, I have access to a variety of experienced computer programmers, so if the breakthrough is software based (as I strongly suspect it will be), the playing field will be evened out far more than it is with the oh-so-dear MIR arrays, etc.

not_important - 14-2-2008 at 02:25

It is a current hot topic, but mostly for QC and similar monitoring applications, not for general research. Best when you know what you should have, or shouldn't have, and you want to know how much of it is there, how pure it is. In these applications you already know what you have and what might show up that you need to control for, you can dispense with much of the information that is needed to determine structure and focus on a few wavelengths that distinguish A for B.

Because the wavelengths used can be transmitted through fiber optics, it's particularly useful for industrial process control and monitoring applications. The NIR band has the advantage of being less sensitive to stray thermal radiation. There is also a much wider range of optics materials to chose from as well; sapphire with its transmission out to 5 um works well and its hardness and temperature and solvent resistance are quite useful in industrial settings. But I doubt your personal applications include taking the IR spectrum of abrasive slurries being pumped over the viewing port.

Part of the reason it is hot topic is there's a lot more demand for industrial gear than for research lab instruments. More sales == bigger profits.

I don't understand why you keep referring to detector arrays, as in your last sentence above. Simple dispersive scanning and FT spectrometers can use a single element detector, arrays are used for dispersive instruments that give multichannel output and for imaging devices.

As I'm not in the business of chemical product manufacturing, I'd much rather have a FT-IR that covers 2 to 20 um than a half dozen of the 1 to 4 um process control ones that simply don't see much of the useful information.

LSD25 - 15-2-2008 at 00:45

Thanks mate, yeah I know the limits of the proposed equipment and trust me, I'd much prefer a proper instrument (in fact, I'd also like to have the $'s to buy the same without qualms). Unfortunately this is not likely to be the case in the foreseeable future. I'll try and get 'something' going to start with and then we'll see what works, what doesn't and whether further experimentation is worthwhile.

I'd like to try and set something up later which utilises the components of the DVD/CD type setups, this seems a lot more involved and I would prefer to get some runs on the board (actual hands on experience) before I even attempt such a thing.

Now, I do not understand (guess you are probably not all that suprised?) how the units could work with only one detector? My problem is in understanding how a single detector differentiates between the various wavelengths?

But yes, I too would honestly and sincerely prefer a true FT-IR setup that could acheive everything my little heart desires, I just don't see how to get there.

not_important - 15-2-2008 at 02:58

Scanning dispersive spectrometers use a prism or diffraction grating to get the spectrum, and allow a narrow range of frequencies to react the detector at any one time. The 'scanning' is the sweeping across the frequency/wavelength range to measure all the spectrum.

FT models split the 'white' light into two paths, slightly wiggle the length of one path, and recombine the paths to get a complex interference pattern which the computer then uses FFT processing to convert into the frequency spectrum.


In both cases you want the detector to have as flat a response to all wavelengths as it can, although differential models reduce the need. Most dispersive IR spectrometers were/are scanning, specialised ones use multiple detectors to avoid scanning by measuring a few wavelengths of interest; these generally are process control or specific monitoring applications.

Spectrometers for UV/visible light, usually called spectrophotometers, are mostly dispersive in nature, earlier ones and simple models are similar to scanning IR spectrometers in that they measured one wavelength at any given instant. More recently CCDs and other linear arrays have been used as detectors to allow sampling of the entire spectrum at once. However for many applications the single wavelength measurement wasn't a problem, because the task was measuring the absorption at some important wavelength to track the concentration of a particular chemical.

Spectroscopes also use film to record the full spectrum at once, and can give very high resolution; the newest ones use linear detectors. These are used for light emitted from excited atoms, while spectrophotometers are used for measuring light absorbed by molecules.

For a generally useful IR spectrometer, even for NIR, you'll be using a single detector. There's some applications for IR below 1 um, but they are fairly limited, QC related to water content and similar, or some biological research stuff.



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

http://elchem.kaist.ac.kr/vt/chem-ed/spec/vib/ir-instr.htm

http://en.wikipedia.org/wiki/Ultraviolet-visible_spectroscop...

LSD25 - 16-2-2008 at 18:12

OK, sorry but the diagrams I have seen to date didn't make that sufficiently clear, thus the failure to comprehend.

So, if I am correct - either the difraction grating (and therefore also the difracted spectrum) moves or the detector moves? This would, I assume require a fair degree of calibration insofar as the location of the various wavelengths within the spectra, as the only way to differentiate between the various wavelengths (or most, I would assume that by using the three photoreceptor diodes in a CD unit would differentiate between some of them) would be the location of the same within the spectrum?

So, working on the assumption (which well may be misplaced) that this is indeed possible, what you are in effect saying is that a functional and functioning IR spectroscope is within the reach of the home hobbyist?

I mean, it is technically interesting - although not exactly difficult - to imagine a road culminating in a 'shuttle' which moves from one side of the spectra to the other, measuring, collecting and sending to a PC data on the intensity of IR radiation. Similarly, it is hardly difficult to imagine that such an apparatus could be calibrated via the comparison of the spectra obtained thereby with a range of pure, known compounds against the known spectra of those compounds. Lastly, if this were indeed feasible, the sending of radiation-intensity data to a PC would enable the automatic plotting of the IR spectra obtained by this method (once a value for the IR-wavelength is assigned to the position of each component of the spectra where it intersects the shuttle track).

The only real problem with this is whether or not you were serious when you stated that the receptor array used in DVD/CD players descends into the actual IR range... I suppose one could also look about and try and find similar instruments which contain diodes/transistors which are capable of working in that range. The existence of these receptors at a decent price, appears to be the decisive factor in the design of a home-model. If the hypothesis that one could assign a wavelength value to the positions of the shuttle holds true, then this would obviate the need for the unit to be able to identify as well as quantify the wavelengths within the spectra.

PS I apologise for the late reply, I was working on converting a temperature controlled ceramic hotplate to a speed controlled, stirred, heat-controlled ceramic hotplate. You may have guessed, I like gadgets.:D

microcosmicus - 16-2-2008 at 19:45

Quote:

I mean, it is technically interesting - although not exactly difficult - to imagine a road culminating in a 'shuttle' which moves from one side of the spectra to the other, measuring, collecting and sending to a PC data on the intensity of IR radiation.


Another possibility is to use circular as opposed to linear motion My thought would be to make a
carefully balanced mirror wheel mounted between centres (like a gyroscope) and turn off the power
to the wheel while measuring. This way, even if you didn't know the exact angular speed, you could
be reasonable confident that it was constant during the scan. Using your spectral lines of known
substances as reference (likely several spectral lines from one substance would do) you could then
calibrate the spectrum. To scan both at once, perhaps an arrangement whereby the reference spectrum
goes through the top of the slit, the spectrum to be measured through the bottom --- measuring the
calibration simultaneously with the real McCoy rules out issues of conditions changing between
the calibration and the measurement.

not_important - 16-2-2008 at 21:26

for dispersive instruments the grating can move, the slit can move, or a movable mirror sweeps the spectrum across a fixed mirror. These are rotary movements.

Dual beam models do the same thing, but they also send a beam through the sample and another through a similar path without a sample. The beams share the detector, a rotating mirror sends one than the other beam to the detector/monochromator. This allows absorption from water vapour and CO2 to be compensated for, as well as other wavelength related variations in response. If the beam selector also gives some dark time, those periods can be used to determine the background signal level.

Even single beam instruments chop the beam, this gives an AC signal to process, getting away from problems related to DC drift in the detector and electronics; it's easier to build a low noise AC amp that a DC one.

Moving the detector is rare, some types have a physical structure that would generate false signal from vibration; as the signal levels are generally low moving the detector gives the possibility of electrical noise pickup.

As the detectors in the early machines were fairly slow to respond to changes in intensity, the sweep rates weren't high; it could take a minute or more to get a spectrum. Newer detectors are faster, but given that you are looking at a very narrow band of wavelengths, and IR sources are not very intense, scanning instruments are generally not real fast.

FT-IR moves a mirror along the light path, parallel with the beam. Often the mirror is mounted on a transducer which can be similar to a speaker cone, or a slab piezoelectric material. FT-IRs are usually self-calibrating, they include a gas laser (laser diodes tend to be too broadband and noisy) that functions as a reference frequency.

A simple array like that in CD players really isn't of any help here, you need to look at hundreds to thousands of wavelengths. The most use you'd would get of it is in monitoring light at right angles to the spread of the spectrum; this usually represents misaligned or poor quality optics.

Any silicon photodiode, be it a single diode or transistor or a CCD, likely is dropping down to zero response at about 1 um, and just about all do so by 1100 nm. Anything you build using Si diodes will be limited to about 1000 nm and shorter wavelengths, in that range you'll mostly find the very weak 3rd overtones for H-O and H-N bonds.

I think that you would get more useful information building a Raman spectroscope working in the visible and/or very near IR; it could give you information close to what conventional MIR gives.


By the way, this equipment used a 2D CDD, kept at -50 to -100 C

http://www.avaloninst.com/content/raman_information/Echelle....

LSD25 - 20-2-2008 at 01:34

Sorry, I was somewhat overoccupied for the last few days,

Anyhow, I was looking at the dedicated mid-IR photodiodes, correct me if I am wrong but it appears that the PbSe variety are useful for detecting these wavelengths and do not require anything below room temperature in order to operate properly?

http://sales.hamamatsu.com/assets/applications/SSD/irdetecto...

Here is where they are used in Oz:

http://www.capgo.com/Resources/Temperature/NonContact/NonCon...

This is where my thought process is leading me at the present time anyhow...

As to the detector moving, I was visualising moving the detector a 10th of a mm, then stopping for a period and taking in the spectra at that point. Quite frankly, if it takes 10 hours for this to completely log the spectra - so fucking be it.

So, what do you think?

not_important - 20-2-2008 at 02:52

PbSe is good from ~1 um out to 4 um at room temperature, out to 5.5 to 7 um with cooling to -50 C down to LN2. Cooling also boosts the sensitivity by an order of magnitude.

Some characteristic curves
http://www.lasercomponents.de/uk/fileadmin/user_upload/home/...

It is easier to move a mirror or grating, usually done as a rotation, than to move the detector, which will have to follow an arc that traces the focal plane.

And trust me on this, it's not just measuring while moving, but changes in resistance and noise pickup caused by moving even after stopping, that make waving a hi-Z detector about less than desirable. Remember, you're working with a detector with a resistance of 100K to several meg, a bias voltage of several 10s to several hundred volts, and an amplifier front end with an input Z of 10s to a couple hunder megs.

For PbSe, a chopping frequency around 1 KHz works well. Use a PLL to extract the actual frequency from the amplified signal, and drive a synchronous demodulator with the PLL followed by a low pass filter to recover the 'DC' signal.

LSD25 - 20-2-2008 at 06:34

OK, I'm currently looking for details on the feasibility of moving the mirror/grating, the trouble is the horrid lack of comprehension on my part.

The reason for trying to avoid cooling is simply that to do so will really make this a lot harder for me to do and probably for a whole lot of others to even contemplate trying.

http://www.cappels.org/dproj/syncdet/syncdet.html

To separate the wheat from the chaff so to speak, or to isolate the signal from the noise. You suggest this with a PL Loop? Here is a google-book on the theory, so at least we are on the same page:

http://tinyurl.com/35kamv

Now for mine, I suspect that this one is getting closer to where we want to go (they even provide sample code):

http://www.edn.com/archives/1998/061898/13di.htm

And here is an article pointing us to where to find this circuitry essentially ready-made:

http://www.radio-electronics.com/info/receivers/synchdet/syn...

So, as my motto is 'why buy when I can make and why make when I can modify...', can I yank the ready made circuitry out of something else in order to make this project as simple as is humanly possible?

not_important - 20-2-2008 at 08:01

PLL - like this:
http://www.uoguelph.ca/~antoon/gadgets/pll/pll.html

http://www.fairchildsemi.com/pf/MM/MM74HC4046.html

http://yyg.stanford.edu/projects/rangefinder/LM565.pdf



sync demod

http://www.analog.com/en/prod/0,,773_862_AD633%2C00.html

http://www.analog.com/en/prod/0,,773_862_ADL5391%2C00.html


For gain blocks, ordinary op amp should do nicely. You could also use switched capacitor filter chips for a a signal bandpass and the output lowpass, although ordinary R-C filtering will likely be enough for the LP filter.

http://para.maxim-ic.com/cache/en/results/4431.html

This is all audio frequency stuff, there are also chips use for RF that will not work at low enough frequencies for your application. There may be something out there that already has the combination of functions on it, but the IC building blocks make it pretty easy to do.

Wow, I wasn't aware what bad shape the US dollar was in. When I was working with this stuff a decade ago, these were all jelly bean parts, the budgetary pricing at the hundreds level was a little over a dollar on down.

LSD25 - 21-2-2008 at 14:08

Do you perhaps know of some way to source PbSe photodiodes that would be more effective than what I have found so far? I mean, and this extends way beyond the continuing argument over pro or contra drug chemistry, this could be of serious moment to virtually everyone on this site if only it can be made to work.

It is just that what appeared to be attainable the other day, is really not that easy on further research. It is potentially not so much that the photodiodes aren't available in the suggested equipment, it is more that there is very little data on how to know which equipment contains the same.

PS I am not looking for specific suppliers, it is more that I need a suggestion as to where they can be found more generally.

microcosmicus - 21-2-2008 at 16:30

While it is directed towards infrared thermometry, this page contains lot of useful
data about detector sensitivities, absorption of materials, instrument design and
the like which should be just as useful in building a spectrometer:

http://www.omega.com/literature/transactions/volume1/thermom...

As for where to find PbSe devices, maybe have a look at flame detectors and
thermometers because I have seen them mentioned in that context. I
even came across mention of one in an alcohol detector.

[Edited on 21-2-2008 by microcosmicus]

not_important - 21-2-2008 at 17:16

For NIR purposes flame detectors are indeed the most likely source.

Contactless/remote temperature sensors usually use micro thermocouple or bolometer elements, and include a filter that cuts off wavelengths short than some value, generally 4.5 to 5.5 um. This would almost work in traditional scanning IR instruments, although the short wavelength cutoff is a problem.

Contactless sensors intended for metal working or ceramics applications might be another source for NIR sensors, above 400 C emissivity at wavelengths shorter than 3 um becomes significant so the detectors used will be sensitive to shorter wavelengths than the more general purpose ones.

Low cost sensors for flame detection and temperature measurement often are based on lithium tantalate. These are robust but lower sensitivity, an order of magnitude or two less sensitive that the sensors typically used in IR-type analytical instruments. Because they are directly sensing the emitter, flames or hot material, the lower sensitivity in not much of a problem. In dispersive spectrometers only a small fraction of the IR emitter's light is hitting the sensor, just a narrow band of wavelengths, and that gets reduced further by the absorption of the sample being analysed. This can drop the signal right down into the noise level,


This book looks like it might be a useful reference on this general topic
http://www.blackwellpublishing.com/book.asp?ref=1405121033

LSD25 - 22-2-2008 at 16:11

So you mean something like these:

http://www.accutherm.com.au/html/test_products_international...

This is not to say that I have used or would use this supplier, I just want to know about the range of products offered and whether any of the same are useful.

The range of the infrared dealt with by these instruments is from 7µm-14µm on average. I assume the cheaper instruments of this type which are capable of measuring the same range of temperatures would also use this range of wavelengths?

The ones that utilise a laser to emit IR radiation and then measure the return, are they of interest or not?

Finally, is it possible to remove the filter (I assume there is a filter which cuts off the emmissions below 7µm? or is this likely to be the minimum wavelength of the sensor?).

PS This would be the far-IR would it?

Is there any absorbence data in this range that could be utilised?

What is the approximate range of the wavelengths used in IRDA? What about the wavelengths used by those IR tracking devices &/or the IR testing devices?

Lastly, what about the IR thermometers with the massive range - the ones that can measure from 30C-1700C, wouldn't they utilise the lower wavelengths?

Would it be possible to use NIR & FIR to generate a reasonable range of absorbence, the difficulties involved in finding a low-cost MIR sensor are beginning to look prohibitive...

not_important - 22-2-2008 at 20:50

Quote:
Originally posted by LSD25
So you mean something like these:

http://www.accutherm.com.au/html/test_products_international...

This is not to say that I have used or would use this supplier, I just want to know about the range of products offered and whether any of the same are useful.

The range of the infrared dealt with by these instruments is from 7µm-14µm on average. I assume the cheaper instruments of this type which are capable of measuring the same range of temperatures would also use this range of wavelengths?


Yes, the range is used in part because the blackbody radiation curves for the temperature range is always a slope increasing as the wavelength decreases, and higher temperatures are brighter than lower ones so a simple 'brightness' reading can be used as a measure of temperature.

Another reason for that range is that it is basically the transmission range of silicon, which as a hard, fairly chemically inert, and reasonably temperature tolerant material makes good windows for the detectors.
Quote:
The ones that utilise a laser to emit IR radiation and then measure the return, are they of interest or not?


The laser is just for targeting, as with gunsights.
Quote:
Finally, is it possible to remove the filter (I assume there is a filter which cuts off the emmissions below 7µm? or is this likely to be the minimum wavelength of the sensor?).


Some of the sensors look like the filter is part of the casing, on the inside surface of the cap. Those might be difficult as the sensors are small and you'd be working close to the actual sensor. Also some sensor materials are sensitive to water and/or air, if you didn't know what the sensor was you'd be taking a gamble.

Some sensor materials have response curves not much wider than the filter, others are much wider. The thermopile type have their range mainly determined by the 'blackness' of the hot junctions for a given wavelength. Some of the others become transparent at various IR wavelengths. Again, you'd need to know the sensor type to say.

I noticed that some suppliers of these sorts of sensors offered a range of window materials, with overlapping ranges but overall covering noticeably more than 7-15 um.
Quote:


PS This would be the far-IR would it?

Is there any absorbence data in this range that could be utilised?


Long wavelength IR or IR-C might be better, because far IR often means beyond 15 um. In astronomy it falls within the midrange IR

This is roughly wave numbers of 1400 to 700. In that range are many of the carbon-carbon bond bands, C-F, Cl-C-H, C-SO3 salts, sulfamides, sulfates organic and inorganic, some C-H, and -O- . This range is sometimes called the "fingerprint region", because the bands are affected by other parts of the molecule, giving patterns that at somewhat distinctive of a certain compound.

But you're still missing important parts, you want to go to at least 4 um if not even shorter. This range contains the generic bands for classes of bonds; 2 to 4 um is O-H, N-H, C-H, and similar, 4 to 5 um is CC and CN triple bonds plus H-C where the carbon is also double bonded, 5 to 7 is C=C, C=O, C=N, and certain C-C single bonds

charts of bands

http://www.kayelaby.npl.co.uk/chemistry/3_8/3_8_5a.html

http://www.kayelaby.npl.co.uk/chemistry/3_8/3_8_5b.html
Quote:

What is the approximate range of the wavelengths used in IRDA? What about the wavelengths used by those IR tracking devices &/or the IR testing devices?


IRDA : 850 to 900 nm

trackers - depends on which generation. 2 to 5 um for hot objects, 300 to 1600 C, including the 4,2 um CO2 emission band in jet exhaust. Newer ones also use 8-13 um, cooler targets and in an atmospheric window. Remember the sky is rather dark in the IR; Rayleigh scattering goes down as the wavelength increase, and space is cold looking like -50 to -150 C.

IR testers? Likely the 7-15 um thermal range, never really looked into them.
Quote:
Lastly, what about the IR thermometers with the massive range - the ones that can measure from 30C-1700C, wouldn't they utilise the lower wavelengths?

They might be "two colour" devices, looking at two different bands to handle the shifting of the blackbody curves.
Quote:
Would it be possible to use NIR & FIR to generate a reasonable range of absorbence, the difficulties involved in finding a low-cost MIR sensor are beginning to look prohibitive...


The problem is that cheap sensors means high volume production, which also means many devices sold and eventually junked.

Optics can be expensive as well, especially for the wider bandwidth used by conventional investigative (lab) IR. While visible light spans less than an octave, lower cost UV-Vis a range of 3:1 and better UV-Vis a half decade, NIR a 3:1 range, full analytic IR goes from 3 to 15 um - 5:1 range - or better. And many materials exhibit absorption bands somewhere in that range. Mirrors can show appreciable changes in reflectivity, air itself gets in the way to a degree.

--------------------------------------------------------------------------

An alternative that you may want to do some serious reading on is Raman spectroscopy. It yields information quite similar to conventional IR spectroscopy, but rather than bands at fixed wavelengths it uses re-emitted photons that have had their frequencies shifted by small amounts related to the particular local molecular structure. And it can work with aqueous solutions.

Postage-stamp overview - shine an intense beam of monochromatic light at or through a substance, capture some of the light coming off at 80 degrees to the beam. Block the wavelength of the original beam, run the rest through a spectroscope and look at the spectrum.

The shifts are such that with yellow to Really Near IR light the shifted frequencies stay within the range silicon diodes sense, meaning scanner and camera CCDs would function as detectors. As the scattered light is very weak, the smaller the area of the detector the less background ionizing radiation adds to the noise, so scanner CCDs might be better.

The holographic blocking filter might be expensive, the gratings for dispersing the light are also not real cheap. But the rest of the optics should be conventional, glass lenses and so on. CD (640 nm) or DVD (780 nm) lasers might do as the light source, as might the lasers in laser printers (diodes @ 655 or 790 nm). The scattered light is longer wavelength than the laser.

The volume phase holographic gratings favoured for these applications are expensive, but also are possibly within the reach of amateur production.

LSD25 - 23-2-2008 at 02:43

Like this one:

http://www.umich.edu/~morgroup/virtual/labeled/virtual.html

Here is another basic introduction:

http://carbon.cudenver.edu/public/chemistry/classes/chem4538...

The first one is probably the most interesting, it describes in detail how it works and describes the use of a, what appears to be, fairly normal IR camera. My question with this is, how and why would we need a holographic grating? Couldn't we use a normal grating, like the one in DVD/CD systems? This coupled with a laser from a DVD/CD and then using a Si/CCD from a scanner would appear to be the easiest, lowest tech, option for home use.

The only question is, would it work? From what I understand the spectra would not be anywhere near as well organised as the one in the link, although if we used a laser as the excitor, then the monochromatic beam should exclude any need for a filter?

Yes, I agree this might well be the best option for the home chemist.

Scratch that about the holographic grating, they are hardly expensive:

http://www.edmundoptics.com/onlinecatalog/displayproduct.cfm...

Also, I assume that by carefully selecting the incident laser light, the Raliegh scattering could surely be excluded - given that surely some lasers are used more than others and surely there is limited wavelength filters available to exclude them?

I don't know for sure, but that is what I am now going to look for.

[Edited on 23-2-2008 by LSD25]

I just found something which makes me wonder if the makers of the infrared thermometers with the attached IR laser device which measures the reflected light haven't already dealt with this problem, simply because this book:

http://tinyurl.com/3c79y7

Suggests that they must be filtering out the wavelength coinciding with the IR radiation they illuminate the article with.

Also, would it be correct to suggest that if we used a narrow wavelength of light - like a true red LED and then a true red filter, we would substantially remove anything in the red range? If for instance we did this again with a blue LED and a blue filter we would have something to compare the red LED spectra with, that which was ommitted by the red & blue filters could then be worked out by reference to the other spectra?

I don't know, but by filtering out the wider wavelengths, the entirety of the Raliegh scatter from each colour ELD should then be excluded? That which is not Raliegh scatter from each should then be present in the spectra from the other colour of LED should it not? Just a wild theory anyway...:o

PS Have a good look at all the links on that first one with the CCD camera, it goes right into the necessary design features and is fucking brilliant.

[Edited on 23-2-2008 by LSD25]

not_important - 23-2-2008 at 06:32

The gratings in CD/DVD systems are specialised for beam generation needed for that application, they are not GP diffraction gratings.

Those hologratings are replica gratings and likely don't work well enough for this application, although they are good for conventional spectroscopes. See

http://www.edmundoptics.com/onlinecatalog/DisplayProduct.cfm...

http://www.edmundoptics.com/onlinecatalog/displayproduct.cfm...

for the types used for NIR Raman work.

LEDs are not that narrow band, nor are the intense enough. The scattered light is very weak, you need a intense light source to get useful amounts of scattered light. Before lasers this would be done with an array of mercury vapour discharge lamps enveloping the sample.

Because the shifted lines are so weak, you need very good grating to avoid having the excitation wavelength light scattered by the grating from contaminating the spectrum, swamping it out. That's also why a narrow band blocking filter is used, 10 to 20 nm wide with less than 5 to 10 nm going from OD 0,3 to 4,0 (50% to 0,01% transmission), to block unshifted laser light from the analyser.

A similar narrow bandpass filter is used to clean up the laser light before it illuminates the sample. Many lasers have small side bands, which would confuse the resulting spectrum.

The umich people are using a packaged laser, possibly with a cleaning filter as they label a filter holder in the path from laser to sample. That's also a fairly powerful laser,
Quote:
Laser:

The laser is the source of the light used to induce the Raman effect. In these experiments, the laser is a Neodynium Yag laser (often abbreviated as "Nd:YAG") which emits light at a wavelength of 532 nm. The intensity of the light can often be around 0.5 Watts.

TECHNICALLY SPEAKING: This is a flash-lamp pumped, mode-locked, doubled Nd:YAG which weighs more than most of the graduate students in the lab and works sometimes only when it wants to. Plans are to upgrade to a diode-pumped 2 Watt laser sometime in the future.
or class 3B to 4 - eye protection, no shiny objects in the lab including watches and jewelry. The LSO would write you up if you forgot those rules on class III and up lasers, and tell you stories of people hearing a "crack" as the beam fried their eye.

Note that they use a holographic notch filter and holographic grating, which may be a VPH one similar to the second set in the Edmonds links I gave.


The filters used in contactless temperature measurement are very broad band. For devices with IR illumination, they are using the illumination to determine the emissivity of the surface in the IR, I suspect the equipment just turns the laser on to calibrate, then off to measure. Alternating those states would give a signal whose high/low values are laser+thermal and just thermal, allowing continuous calibration and monitoring.

Ah, LN2 cooling on their CCD, and they are working in the visible range so it is just for noise reduction meaning really low light levels.


notch filter example attached

Attachment: 1050.pdf (275kB)
This file has been downloaded 6618 times

LSD25 - 23-2-2008 at 14:08

Hang on a minute,

Don't most CCD camera's come straight out of the box with IR filters already installed?

I am unsure what this would give in the scattered light side (the stuff we want), but you were saying that this would be a shorter wavelength than the incident light?

So, thereotically at least, if we could match a simple laser type device to the bottom end of the IR filter on CCD cameras to start with, this should remove the Raleigh scatter and the reflected incident light as well, leaving only the shorter wavelength ('Stokes shifted'?) light to be recorded (what wavelengths would we be talking about?).

Would this work sufficiently to allow us to escape having to purchase a notch filter? It is just that that would probably be the single most expensive part of the entire system. This may require, given the crude nature of the available laser devices, the building of an adjustable monochromator (tuneable mirrors, a grating, etc. same as would be used in an IR diffraction device), but that looks to be feasible (just). This would allow for greater choice in light sources, which could then be narrowed down to the appropriate wavelength while the matching of this incident light to the bottom end of the existing IR filter would thereotically mean that the built-in filter would remove the problems?

not_important - 23-2-2008 at 21:15

Quote:
Originally posted by LSD25
Hang on a minute,

Don't most CCD camera's come straight out of the box with IR filters already installed?

I am unsure what this would give in the scattered light side (the stuff we want), but you were saying that this would be a shorter wavelength than the incident light?

So, thereotically at least, if we could match a simple laser type device to the bottom end of the IR filter on CCD cameras to start with, this should remove the Raleigh scatter and the reflected incident light as well, leaving only the shorter wavelength ('Stokes shifted'?) light to be recorded (what wavelengths would we be talking about?).


No, normally the Stokes scattered light, longer wavelength than the source, is used as it is much more intense than the anti-Stokes at ordinary temperatures.

Yes, digital cameras do generally have IR filters on them, but those are not sharp cutoff filters and useless for this application, ignoring them "facing the wrong way".
Quote:
Would this work sufficiently to allow us to escape having to purchase a notch filter? It is just that that would probably be the single most expensive part of the entire system. This may require, given the crude nature of the available laser devices, the building of an adjustable monochromator (tuneable mirrors, a grating, etc. same as would be used in an IR diffraction device), but that looks to be feasible (just). This would allow for greater choice in light sources, which could then be narrowed down to the appropriate wavelength while the matching of this incident light to the bottom end of the existing IR filter would thereotically mean that the built-in filter would remove the problems?


You can use a monochromator to clean up the laser before illuminating the sample. If you use real good gratings you can use multiple monochromators to extract the spectrum information. But a notch filter makes it easier; the Stokes scatter is 10^6 to 10^8 times less in intensity that the Raleigh scatter.

Note that most instruments that use multiple monochromaters seem to be scanning ones, with a single detector and moving parts.

ScatteredLightChromators&RamanLevels.png - 33kB

LSD25 - 25-2-2008 at 04:37

I am trying to cope with the start of uni at present and I am having time-issues as it is (I still have over 10 projects ongoing). I would just like to know have you had any success finding a cheaper, useable variant of the notch filter or alternatively, the PbSe photodiode?

I am going to have to order one of these in and I have to think very carefully which is going to be the easiest and most likely outcome. At a pinch, a few overtimes and I could probably afford both in time.

I think for the moment I will try and make a working monochromator, or a couple of the same - so that if I manage to get either (especially the PbSe diode) I have a good headstart on the scanning part of the proposed equipment.

If anyone reading this has any ideas to contribute, especially on where to find a PbSe photodiode or a useful notch filter in Oz, please feel free...

{EDIT}

I was just reading up on the theory of the bolometers(?), the idea is that the sensor includes an IR absorbent material which, when it absorbs IR radiation emits a miniscule amount of heat which is read by a secondary part of the sensor and which can then determine what and how much radiation was absorbed depending upon the nature of the absorbent material and the wavelength of light is was irradiated with.

I was thinking to myself, hmmm, how to use this to build a sensor for IR spectroscopy... Perhaps if I used an aldehydic material as the sensor, it would pick up and absorb the light... Nah, wouldn't work, BUT

What if the heat sensor was mounted immediately adjacent to the sample to be irradiated and a tuneable monochromator was used to vary the wavelength of light the sample was irradiated with? Heat sensors may be somewhat easier to design and build, the sample will absorb IR radiation and will emit heat. The nature of the monochromator could then allow us to determine (probably by extensive calibration with known compounds) what wavelength of light is emitted at what mirror rotation position(s)?

Of course this will be slower than any other approach, although I suspect that the heat increase would be miniscule and could probably be factored in if a heatsink and fan were used on the casing.

The thing is, that by measuring the heat emission caused by absorbance of IR radiation, we are actually measuring absorbence, not (as in most IR spectroscopy) the level of the remaining IR radiation passing through the sample AFTER absorbence by the sample.

[Edited on 25-2-2008 by LSD25]

The design of microbolometer type spectroscopy should, if it is feasible, probably use low-cost thermistors with a measurable change in resistance with a 0.1C increase in temperature (I think this is feasible). This would be mounted series with another thermistor backing IR transparent material (dunno, how about NaCl/KBr?). The difference in the resistance would be the heat increase of the first less noise. If another was mounted as well, with IR opaque (ie. full absorbence) material, the percentage difference between the 1st & second will be able to be worked out by expressing it as a percentage of the difference in resistance between the second (no increase) and third (maximum increase).

The DOD type of microbolometers use Si as the heat sensor. This is possibly as cheap as it is going to get.

From what I understand, microbolometers of this type are not normally cooled (that is why they are so bloody big at present). The change in resistance would allow us to get a clear, electrical signal, which when plotted with the known position of the tuning mirrors (and the wavelength(s) attributed thereto) should allow for a relatively simple plotting program.

Please tell me whether or not this will work?

[Edited on 25-2-2008 by LSD25]

LSD25 - 27-2-2008 at 01:39

Double-posting, well sorta unavoidable at present...

I have been looking into this further - apparently the new Si gap thermosensors (digital output apparently - they are also quite cheap) can pickup a 0.05C change in temperature - they are also virtually linear in gain so it is quite possible that by reference to a small heat increase / time, it should be possible to operate this fairly well without a major heat increase being necessary. In order to allow the overall speed of the unit to be improved, a peltier type junction-board with thermostat could probably be used to rapidly cool and maintain the sensor at a set temperature before and after irradiation of the sample.

The only question I can find nothing on is whether or not the concept underlying microbolometers, that when an IR absorber is irradiated with IR it gives off heat (conversion/conservation of energy - I have heard of that somewhere?) can be applied succesfully to organic/non-organic chemicals which absorb in discrete wavelengths and not in others. I cannot see why it would not work, I would just like to know for sure.

not_important - 27-2-2008 at 02:51

In my day microbolometers used extremely find gold powder, chemically produced, as thei absorbing surface. There's a very black form of nickel that might do.

These ICs - MAX6633, MAX6634, MAX6635 : http://www.maxim-ic.com/quick_view2.cfm/qv_pk/3074
will resolve 0,0625 degree C

The problem with many of these sensors is that they are too large, their thermal time constant is fairly long. This means the scan takes a long time to complete, or that you will have very low resolution. If you stretch the time to complete a scan out too far, drift gets to be a real problem; even the IR source drifts some.

I think you'd have problems trying to use a feedback loop with a TEC. For maximum sensitivity you want the sensor to heat up quickly, meaning it has low thermal mass and is thermally isolated. To cool it down you want it to have a good thermal path to the TEC, which has fairly large thermal mass.

I think you run into problems using the temperature rise of the sample. First off you'll need a very small sample to make it responsive; a test tube full isn't going to heat up much in the monochromator beam. Remember that you're tossing out most of the radiation at any given time, and broadband IR emitters aren't terribly intense.

Second is the thermal capacity of the sample. Different materials will heat up a different delta for a given amount of absorption. This can make it difficult to determine structure based just on the spectrum, and even matching against known spectrum could be harmed by a relatively minor amount of a second substance that affect the heat capacity of the sample.

I think you'll find it's easier to detect small changes in intensity than it is to detect small changes in temperature.

LSD25 - 27-2-2008 at 05:17

I was actually just thinking of using a silicon thermistor - these really are as small as one is likely to see. They are fairly efficient at picking up changes in temperature and the responses are fairly quick. The cost of these is absolutely negligable, especially when viewed alongside trying to detect changes in intensity.

As to the ambient heating of the sample, I was considering using a blank sample of either KBr or NaCl as a zero, presumably the heat increase in this would be a good reference to that of the sample? The rise in temperature appears to cause a straight-line increase/decrease in resistance, as such this should be able to be worked out.

As to the problem you foresee with the whole concept, that is the main thing I am worried about. It goes to the very heart of the idea - is the heat increase of an organic/other substance when exposed to a narrow wavelength of IR radiation a measurable and useable indicator of that compounds absorbence at that wavelength?

I 'hope' that it will prove to be so, but I am far from convinced... However, I suspect that it might well be, given that the emission of heat caused by the absorption of IR radiation must be in some way connected to the amount of absorption, I just hope that this is measurable. I have looked and cannot find any data, one way or the other, which would enable me to answer this question properly in relation to organics, not withstanding the fact that this is what appears to be the conceptual background to the use of microbolometers themselves (albeit, from what I have found so far, all non-organic).

However, if the half-baked theory holds good (and worse have), then the mere adulteration of the sample with other compounds will obviously lead to the corruption of the absorbence data, as it does with any other type of spectroscopy. The spectrum, based upon the heat emission / absorbence should really not be all that different, within the boundaries of measurement, from that obtained by detecting the intensity of the remaining radiation.

BUT, this has yet to be tried and I cannot even find a patent on the idea... Nonetheless, I cannot see why it would not work (which worries me, as people tend to patent the damndest things), if absorption of IR radiation = emission of heat (and both NASA & the US DOD appear to be of that opinion), and leaving aside the difficulties in actually collecting the data, then this would seem to be a fucking good bet for a purely speculative patent application.

LSD25 - 1-3-2008 at 06:09

Double-posting, well nobody to blame but myself:(

Anyhow, I was thinking - I can buy an entire (admittedly fairly clunky) Michelson interferometer online for under $200, so this set me to thinking (Oh god NO, I hear you say?), there are also good topics on making the same:

http://www.colorado.edu/physics/phys5430/phys5430_sp01/PDF%2...

If a microbolometer works by (1) absorbing IR with the IR absorbent coating; and (2) measuring the change in resistance of the internal resistor as a result of the increase in heat transferred from the IR absorbent part, is it possible to use one (or a number of these - say 20-30 silicon resistors coated with an IR absorbent - at the right ranges of course) of these as the sensor for actual FT-IR?

It seems to me that the benefit of using it this way would be that the increase in heat with each absorption of IR radiation would be limited by the fact that the whole thing would be over and done with more rapidly.

not_important - 1-3-2008 at 09:44

Quote:
Originally posted by LSD25...
As to the problem you foresee with the whole concept, that is the main thing I am worried about. It goes to the very heart of the idea - is the heat increase of an organic/other substance when exposed to a narrow wavelength of IR radiation a measurable and useable indicator of that compounds absorbence at that wavelength?
...


The absorbed IR will show up as heat, however:

1) the change in temperature is proportional to the intensity of the radiation and the heat capacity of the substance. The heat capacity of the substance depends on a number of things, including the degrees of freedom of the molecule - directly related to the complexity of it. So differing compounds will have different heat capacities, and thus differing temperature increases for a given amount of some wavelength absorbed. So unlike a transmission IR where the intensity of say the C=O band is roughly the same for most compounds, the temperature based one will give radically different amounts of change for differing compounds. In effect the spectrum is flattened out for the more complex molecules.

2) there's not a lot of energy at a given wavelength for most IR sources, making for a very small temperature rise.

3) for solutions, mulls, or other forms used when working with solids, the heat capacity gets hit over the head by the other substance, even if it is transparent for the wavelengths being measured.

Putting those together I think it would be difficult to determine what an unknown is from looking at its structure.

I'm not sure I understand what you're trying to do with the multiple pickups - care to amplify on that?

LSD25 - 1-3-2008 at 11:25

I was looking at a site earlier tonight where it showed a java-applet of FT-IR and what is the actual end-result.

I was thinking to myself, that if it were possible to place several radiation-heat receivers in a small area, then the changes in the fringes would be more apparent more quickly - thus less radiation & less heat.

I must admit though, I am still hopelessly befuddled about how one converts that signal to anything, even the one that would be received with a proper sensor.

not_important - 2-3-2008 at 00:45

Not sure why several receptors would do that, focusing all the IR onto a single detector should do the task.

The FT part of FT-IR is close to FM if you've not the maths background. But there are plenty of code libraries around. Get to understand the basic concepts needed, the several types of transforms, windowing, and so on, and you could cobble together a functional FT system.

As FTs are used in NMR and some newer MS, it's a concept you likely want to become a little more familiar with.

LSD25 - 2-3-2008 at 05:12

Righto,

I'll make that my main effort for the next week or so...

That is the joy of this, I ain't seeking to build a better mousetrap - just a cheaper one - the information exists, it is just a matter of assimilating enough of it to work out what is necessary and how shit works.

The main problem I have is working out how to visualise 'seeing' something when the sensor is only sensing the presence of an IR signal, whether by direct absorbence or via heat absorbence. This is somewhat easier to get my head around when dealing with a single wavelength obviously the difference in the signal intensity when it goes through a sample is the absorbence level of the signal and the wavelength is able to be determined by reference ot the position of the mirrors in a monochromator...

OK, now from the little I have read so far, the fringes resulting from running light through the interferometer are the radiation from the output separated into discrete fringes, from which one can derive the entirety of the spectra of both the source of the light, but also that of the source passed through the sample - which effectively filters out part of the spectra.

That is obvious, the problem I have is not so much the FT, but more how does a single-point type sensor collect suffifcient data from which to collect enough of the fringes so as to allow the FT to be run?

I mean, once I comprehend this - the FT part could be worked out, without even being required to understand the equation - simply by virtue of the fact that it is mathematics and that is what computers do best. The same (or a virtually similar) program could then be used in whatever spectroscopy was utilised to collect the signal.

not_important - 2-3-2008 at 07:35

If you took the simple Michelson interferometer design, and have the IR source be an idealised one at infinity so the photons can be thought of as flat wavefronts - parallel with or perpendicular to the end mirrors, the the fringes exist in time, or in space along the axis of radiation traveled, but not as the conventional bullseye pattern. Think of it with a single wavelength, nice straight wavefronts marching along, reflecting off the beam splitter and mirrors. When they arrive at the detector plane, it's rows of peaks and troughs parallel to that plane; the entire detector surface brightens and dims simultaneously.

The real world isn't that pretty, but close enough. So far as the detector goes, it's not seeing a bullseye fringe pattern but just a modulation of the intensity of the light, modulation tied to the rate of movement of the moving mirror.

The difficult part of an IR interferometer is the beamsplitter, it has to work across a decade of wavelengths. There have been IR systems that use purely reflective optics, interdigitating the fix and movable mirrors, as a means of getting good performance across a wide range of wavelengths.

LSD25 - 2-3-2008 at 07:57

Thanks,

Here is a good introduction to the principles:

http://mmrc.caltech.edu/FTIR/FTIRintro.pdf

I also found this which is probably the most simple version of the theory I have yet found:

http://www.laserfocusworld.com/display_article/25338/12/none...

So, looking at the problem from that viewpoint and taking part of what you said to its logical conclusion - the sensor only really takes in the intensity of the light at the point at which the sensor operates and that this when viewed in conjunction with the position of the moveable mirror gives the basic information which is deduced by the computer?

Ok, so when the system is first built, one just takes a spectrum of the light source, probably one of a sort which has detailed spectral information available - using this to work out what is what & where it is, then various samples of known compounds, etc.?

LSD25 - 29-4-2008 at 01:16

Ok,

Back to this for a specific question - if one had a Class 2, Laser Pointer, 0.95mW output, wavelength 630-680nm, would it be feasible to use a grating (as linked to above), a filter (is it possible to just cutoff from 630nm on up?), a lens and a CCD device?

If the Class II device is too light-duty, is there a way to modify the same in order to increase the output?

Alternatively, could the weak raman spectra be intensified by using a photo-editing package as used by High School students in fluorescence microscopy (I think I cited that idea here), IIRC they used the photo-editing package to increase contrast and thereby intensify the weak image to something useable.

not_important - 29-4-2008 at 07:01

If the shifted (Raman) light is too weak, it is too weak - down in the noise and not recoverable. You're never looking at these with your eyes, it's always electronics and data massaging before it's put on a display for you to see. The problem is the huge difference in intensity of the laser and simple scattered light from it vs the Raman light; defects in the grating and lens and dust in the air can scatter so much of the laser frequency that the portion that lands elsewhere is still many orders of magnitude greater than the Raman light, more than the resolution range of the sensor.

Post processing would be useful to remove 'hard' background data errors such as adjusting cells with higher or lower sensitivity than average, or mapping out stuck cells. Some linear (scanner) CCDs have support for some oft hat processing built into them, scan a uniform surface and then load the resulting intensities into a small RAM on the chip; the values are used by the chip to tweak the apparent sensitivity of each pixel.

You can't boost the output from a laser diode very much, at least for any useful amount of time. Better to get a higher powered diode; that's why CD/DVD drives were talked about because their diodes are higher power.

You can get sharp cutoff filters, but they will be expensive. If you have good quality gratings you can do the task using them, but there is strong emphasis on 'good quality'

LSD25 - 29-4-2008 at 11:14

I was actually looking at the Edmund catalogue, the cutoff filters are pretty dear, but if they are likely to work (same for the grating - the more lines the better, right?) then it will work out comparatively cheap anyhow. I was actually reading up on how they take multiple images in astronomy in order to cancel out the noise.

So, if I got one of the higher end diffraction gratings from the catalogue, the rayleigh scattered light would be recorded, but it could then be discounted?

PS What about if the sample was put in a colloidal silver solution? I was reading that this would improve the return significantly? If so, colloidal silver kits are available (although a touch expensive). This book (http://tinyurl.com/4a3gno) does, from page 129, detail the preparation of colloidal silver from readily available silver nitrate (using either trisodium citrate or borohydride) for Surface Enhanced Raman. Would this be of any practical utility for what I am thinking of?

Vogelzang - 8-5-2008 at 10:57

How about buying one of these?
http://cgi.ebay.com/PERKIN-ELMER-237B-Grating-Infrared-IR-Sp...


Or you could buy one of these new for less than $10K.
http://www.bucksci.com/m500.htm

Vogelzang - 8-5-2008 at 11:03

There's lot of knowledgable people over here
http://www.diyaudio.com/forums/forumdisplay.php?s=&forum...
that can help you with tube circuits.