From raw to corrected spectra Log Out | Topics | Search
Moderators | Register | Edit Profile

NIR Discussion Forum » Bruce Campbell's List » Equipment » From raw to corrected spectra « Previous Next »

Author Message
Top of pagePrevious messageNext messageBottom of page Link to this message

Benoit Igne (benoit)
Junior Member
Username: benoit

Post Number: 10
Registered: 11-2006
Posted on Thursday, May 24, 2007 - 4:47 pm:   

Dear Howard,

Thank you for those additional details, I now have to read the literature to perfectly understand the concept.

Thanks again,

Benoit
Top of pagePrevious messageNext messageBottom of page Link to this message

Howard Mark (hlmark)
Senior Member
Username: hlmark

Post Number: 98
Registered: 9-2001
Posted on Thursday, May 24, 2007 - 11:59 am:   

Benoit - in your original message you said you were doing transmission spectroscopy. If you're measuring by reflectance, then the same principles hold, but the implementation is slightly different: the "100% sample" should be something that is highly reflecting; packed Teflon powder is commonly used for that. Cups prepacked with powdered Teflon are available from many of the instrument vendors, each of which uses a cup that will fit their own instrument, of course! Similarly, the "0%" sample would be a cup packed with a black material, such as powdered carbon; lampblack is often used for that.

For the common NIR reflectance instruments, you should talk to the manufacturer of your instrument, and see what they recommend; many of them have some of these capabilities built-in in the instrument.

\o/
/_\
Top of pagePrevious messageNext messageBottom of page Link to this message

Benoit Igne (benoit)
Junior Member
Username: benoit

Post Number: 9
Registered: 11-2006
Posted on Thursday, May 24, 2007 - 8:36 am:   

Dear Howard,

Thank you for you answer. I will try that right away. The two books you mentioned are in my university's library so I will go check them out.

I just have one question about the "100% sample". You said that it has to transmit the same amount of light as a sample with no absorbance. Thus, in the case of a whole grain analysis, an empty cell scan will be a good "100% sample", right?

Thank you again,

Benoit
Top of pagePrevious messageNext messageBottom of page Link to this message

Howard Mark (hlmark)
Senior Member
Username: hlmark

Post Number: 97
Registered: 9-2001
Posted on Thursday, May 24, 2007 - 4:58 am:   

Benoit - I forgot to mention the classic book in this field: "Spectrochemical Analysis", by James Ingle and Stanley Crouch, published by Prentice-Hall. It's an advanced book and may be a little hard reading for a beginner, but it has everything you need to know. It's also somewhat expensive.

If you go to Amazon.com and do a search for "Instrumental Analysis" you'll find some books at a more elementary level, including the one "Undergraduate Instrumental Analysis" by James Robinson and Eileen Skelly-Frame. It covers many techniques besides spectroscopy, but should contain the basic information you're looking for. There are many other fine books listed there, also.

Howard

\o/
/_\
Top of pagePrevious messageNext messageBottom of page Link to this message

venkatarman (venkynir)
Senior Member
Username: venkynir

Post Number: 41
Registered: 3-2004
Posted on Wednesday, May 23, 2007 - 11:26 pm:   

Thanks for Harward mark for brief description. I have worked on NIR spectral collection for pattern recognition of the selective polymers in on-line. During execution of Pattern recognition technique (chemometrics) , I have seen there is drift in pattern and after few hours , the system start hanging . I implemented the method suggested by Prof.Harward quite long time before .The results are very good and no hanging of the system . It is also possible to eliminate conveyor spectra data .Good Suggestion to Benoit By Harward mark.
Top of pagePrevious messageNext messageBottom of page Link to this message

Howard Mark (hlmark)
Senior Member
Username: hlmark

Post Number: 96
Registered: 9-2001
Posted on Wednesday, May 23, 2007 - 7:25 pm:   

Benoit - the way you treat the data depends on how you want to report it: as transmission or as absorbance. Doing it for transmission is easier; also, once you've done it for transmission, you can convert to absorbance simply by applying the formula A = log (1/T) (where "log" means common log, that is to the base 10).

The whole idea behind using black and white, or any other references, is to be able to convert a measurement to something resembling an "absolute" scale, or at least, one that is relatiely independent of the instrument variations. Since transmission can go from 0 (if the sample transmits nothing) to 1 (if the sample transmits all the light impinging on it), the idea is to have something avialable that can be put in the instrument instead of a sample, that can provide known amounts of transmittance, that the sample readings can be referred to. The easiest things to use normally is a dark sample (so you can determine the instrument response in the absence of any light coming through the sample) and then a "100% sample" (or a "white" sample). The "100% sample" doesn't have to actually transmit 100% of the light, but what it does have to do is to transmit the same amount of light as a sample with no absorbance. Typically, the solvent that the actual sample is present in, is used for the "100% sample", and the instrument response for the "100% sample" is used to determine the instrument response when the sample has no absorbance due to the analyte.

Then, knowing the dark and 100% responses, the following equation can be applied:

T(sample) = (sample - dark) / (100% - dark)

where "sample", "100%" and "dark" represent the instrument responses for the three materials, and T(sample) means the actual transmission of the sample.

Ideally, this will remove the effect of instrument drift and other variations that can creep into the readings. At the very least, therefore, the "dark" and "white" standards, like any standards, have to be stable materials, whose transmission characteristics can be counted on to not change over time.

This is the situation in a nutshell. There are several books about instrumental analysis that contain the information you're looking for. You could look in the chapter about UV-Vis analysis in any of these books, if there's no chapter on IR or NIR; these considerations are independent of the wavelength range, except for what actually constitutes suitable standards.

Howard

\o/
/_\
Top of pagePrevious messageNext messageBottom of page Link to this message

Benoit Igne (benoit)
Junior Member
Username: benoit

Post Number: 8
Registered: 11-2006
Posted on Wednesday, May 23, 2007 - 5:36 pm:   

Dear NIR community,

I am doing some work on a NIR transmittance instrument that takes spectra on demand. So, there is no scan of empty cell, black and white references and no treatment of the spectrum.

I would like to be able to correct this spectrum from environmental changes by measuring references but I do not really know how to do that!

Can I just scan an empty cell and then subtract the spectra of the empty cell to sample's spectrum or I need to collect white and/or black references and do something with them?

From some work I did on IR, I know that we collect a background spectrum that is then subtracted to the sample's spectrum but I do not know how this subtraction is done! Regarding the white and black reference, I know that it is necessary for reflectance instruments but I do not know how the reflectance spectrum is corrected with those scans and I do not know if it also applies to transmittance instruments (I doubt!!!)

I wasn't really able to get any good answer from the literature (or I did not look at the good place), so I hope you will be able to advise me and/or indicate some literature where I can find the solution to my question!

Thank you for your inputs!

Benoit

Add Your Message Here
Posting is currently disabled in this topic. Contact your discussion moderator for more information.