| Author | Message | ||
     Anne Orcutt (Anne) |
We currently are using a filter NIR with the standard NIR wavelenth range and many of our calibrations use wavelengths in the upper half of the range. We are shopping for a new scanning NIR instrument. One vender is showing us an instrument that due to the detecter, it only uses the lower half of the NIR range. Is this acceptable? What impact wiil this have? Since many of our current calibrations use higher wavelengths, will we lose sensitivity or accuracy or robustness? | ||
| W. Fred McClure (Mcclure) |
Anne, You did not tell us what your application (product, material, property) you are working on. However, if you are developing new applications, you are shooting yourself in the foot if you do not have access to the full NIR spectral range 780-2500 nm. I cannot discuss specifics here but if you want to contact me directly ([email protected]) I can discuss my NIR experiences with you. | ||
| Andrew Hind |
Anne, I concur with Fred's comments that you should not limit yourself with respect to spectral range. Scanning instruments are available which cover the range from 175 - 3300 nm. If you are involved in research/method development, this additional wavelegnth range is likely to be important. Please contact me if you would like to discuss the matter further ([email protected]). | ||
| Andrew Hind |
Anne, I agree with Fred's comments regarding wavelength range. Scanning instruments are available whcich cover the range from 175 - 3300nm. This increased range is likely to be significant if you are involved in any type of research or method development work. Please contact me if you would like to discuss the matter further ([email protected]). | ||
| Peter Tillmann (Tillmann) |
Dear Anne, comming from a filter instrument you will most probably develop new calibrations anyhow. As Fred said the answer is dependend on your task, i.e. product, measurement set-up, ... For us (feed stuffs in a laboratory set-up) migrating from 1100-2500 nm to lower wavelengths results in much cheaper instruments with no loss of precision as preliminary tests revealed. As was shown in Cordoba eventually calibration transfer across instrument types (different manufactures, different wavelengths range, ...) is promising. But most of the time recalibration on the new instrument will be easier. Yours Peter | ||
| Peter Tillmann (Tillmann) |
Dear Anne, comming from a filter instrument you will most probably develop new calibrations anyhow. As Fred said the answer is dependend on your task, i.e. product, measurement set-up, ... For us (feed stuffs in a laboratory set-up) migrating from 1100-2500 nm to lower wavelengths results in much cheaper instruments with no loss of precision as preliminary tests revealed. As was shown in Cordoba eventually calibration transfer across instrument types (different manufactures, different wavelengths range, ...) is promising. But most of the time recalibration on the new instrument will be easier. Yours Peter | ||
| Tony Davies (Td) |
Dear Anne, A quick �off-the-cuff� response (which I am not good at). It does depend on your applications. You should be able to discover what other people are using and your potential vendor should also have some information. I think that many filter-based calibrations used higher wavelengths but in scanning instruments there are several reasons why shorter wavelengths will work as well if not better. The information is repeated many times but with the filter instrument there may have been only one choice of wavelength (the repeats were not included). Generally, 700 -1800 nm covers almost all applications. If you are considering 700 � 1100 nm then more caution would be advised There will be some calibrations which cannot be replaced with Herschel wavelengths. When you go to a scanning instrument, you will have the benefit (?) of advance chemometrics such as PLS and you will stop worrying quite so much (But definitely not completely) about specific wavelengths. There are many other important specifications when considering which instrument to purchase. The first is NOISE not wavelengths. Good luck! Tony | ||
| Gabi Levin |
Dar Anne, There is very little that I can add that wasn't said already. Not knowing your application, compositions, etc. renders any advice I can give limited in use. The only generalized statement I can make is that if you have a spectrometer that scans at least the range 1100 to 2300nm than you have the capability to expand beyond what you are doing today. Not having at least this much of a range limits your capability to develop new applications, some that possibly you are not even considering today because of the limited nature of filter instruments to begin with. If you wish to consult with me on a more specific basis, and discuss on line or lab implementation, I will be glad to do so, but because of commecrial apsects I can not do it here. My e-mail is [email protected] Thanks. Gabi Levin | ||
| Anne Orcutt (Anne) |
Application: Moisture and Fat in confections. We have shown that that on several manufacturers' scanning instruments, we can transfer the calibration coeffiencts from our filter instrument, do a bias adjustment and continue with MRL calibration. Our goal is to use MRL until we collect enough data to switch to PLS. The instruments we tested are similar in noise, precision, etc. | ||
| Devils advocate |
Are you sure you need a scanning instrument? It seems like a lot of money for a relatively simple application. Are you unhappy with the precision of your current machine, or just looking to replace a piece of industrial archeology? | ||
| hlmark |
Anne - "Devil's advocate" makes a good point. Filter instruments were among the first technologies to be used for spectroscopic analysis, way back in the dark ages "BC" (Before Computers) and are still used as the simple, inexpensive solution for routine analysis. I know where there are some available "pre-owned" as they say, which I will tell you about off-line, since that has commercial implications. Howard \o/ /_\ | ||
| DJDahm |
I would like to amplify a little (and perhaps give a slightly different take) on some of the previous responses. If I were being argumentative (which I would never do), I would say that you are not doing NIR spectroscopy unless you have a scanning instrument, you are just using wavelengths in the NIR region to do an analysis. As a consultant, I don't work with any customers that do not have (or are in the process of buying) a scanning instrument and a good chemometrics package. There is a need to choose between FT versus a grating instrument. There is a bottom end wavelength at which an FTIR instrument can operate reliably. That wavelength has been moving down with time. Then there is the detector selection. If we take the visible/NIR cutoff as where the optimum detector choice canges, it is about at 1100 nm. In the wavelength range below 1100, there is limited vibrational information available, so many scanning istruments have that as the lower end of the wavelength range. If you scan only below 1100, many of us would say you are not really doing NIR Spectroscopy, but rather visible. For data collected in the reflection mode, the data above 1800 nm or so is of limited utility because absorptions tend to be so strong as to make quantitation in the presence of surface relection less reliable. So, in my experience, 1100 to 1800 nm winds up being the critical range for many analysis, and I reccomend to my clients that they make sure that the instrument they buy operates reliably in this range. However, the adjoining wavelengths can add versitility that at times can be very important for specific aplications. Now when you are considering a particular analysis, it is frequently folly to pay for versitility. Some vendors have optimized low cost instruments (which in this case might be called analyzers) for particular applications. In my opinion, the weak link in what we as a NIR community (instrument manufaturers, professors, consultants, etc) provide is an effective way of selecting a low cost analyzer for a new application on the basis of results we have obtained on a scanning instrument. | ||
| Tony Davies (Td) |
Don, I wish you would not do it! What? Make controversial statements that I have to respond to!! 1) Filter instruments not spectroscopy? I thought that Beer-Lambert was the basis for spectroscopy. It was defined for single measurements using monochromatic radiation. Nothing about scanning! What is your definition of spectroscopy? Yes, I know filters do not produce monochromatic radiation but if we want to be argumentative scanning instruments don�t either. (Especially true for FT instruments!). 2) �Bottom-end� wavelength could be a bit confusing for FT spectroscopists who think in wavenumbers. I assume you mean small wavelengths i.e. high wavenumbers? 3) NIR spectroscopy? Please do not add to the confusion. The NIR range was defined by an IUPAC convention. It is 780-2,500nm or 12,800-4,000 wavenumbers. Spectroscopic measurements inside this range are NIR spectroscopy. Measurements outside are visible or midIR. Defining the operation according to what absorptions (we think) we are measuring becomes very confusing. 4) The 700-1100nm range There are a large number of scanning instruments in use with the 700-1100nm range. They operate in the transmission mode but most of the applications are diffuse transmission rather than clear liquids. The advantage is that absorptions are less intense so that path lengths can be longer. I would argue that there is more information (in the number of different vibrations) but it may be getting very overlapped in this part of the NIR range. 5) Data above 1800? You may be correct but I think this is where filter instruments score. They used many wavelengths above 1800nm. If they had not worked, would we have the multitude of scanning instruments that are currently available? 6) Translating research to low cost instruments In the �80s it was a case of translating research to the available (filter) instruments. For quite some time, there were less than 25 scanning instruments worldwide! The problem really came by telling people that PLS was the answer to all their problems but they would need a scanning instrument. So the problem is that everyone assumes that scanning instruments are �better� and I have not been asked for many years to suggest a low cost alternative. To be honest, it is some time since I have persuaded a potential customer that it might be a good idea to employ a consultant to help them decide which instrument system they should buy! But may be I give too much free advice! Best wishes, Tony | ||
| David Russell (Russell) |
I appreciate TD's well written reply to DJDahm. As a person who started in this business extracting data from scanning lab instruments to determine the optimum filter selection I would probably have been offended if had read the original note before the reply. And as far as I'm concerned, I was doing both Spectroscopy and Chemometrics even though perhaps at a novice level at that time. What many people tend to forget is that there are many applications where interferences that could be "detected" by a scanning instrument are not an issue, and many applications that can only be captured by a filter instrument due to limitations in fiber optics and other components in the sample path that are available for scanning instruments. As a person who spends a significant portion of their time supporting filter instrument applications I consider the "art" involved just as important as that which I use to develop scanning applications in the NIR and other wavelength regions. | ||
| djdahm |
Tony, Well said. And David said: "And as far as I'm concerned, I was doing both Spectroscopy and Chemometrics even though perhaps at a novice level at that time." Of course, he was. And I suppose that Bruce Campbell could say that he played golf, even if Tiger Woods was in the room. In summary, my position is: A scanning instrument with full NIR cabability is the instrument of choice when one is serious about being in the NIR Spectroscopy game (Fred's suggestion). Of the NIR range, the 1100-1800 nm range is the one that most often has the information necessary to attack a "typical" problem. When folks are thinking about an instrument to perform a particular analysis, low cost alternatives (which may have the cabability of doing the analysis just as well) are often overlooked. I wish I was better at being able, for a particular problem, to go from looking at spectroscopic information obtained on a full range scanning instruemnt to selecting a good, low cost option. Tiger Woods is a better golfer than Bruce. Now Tony, have I said anything controversial this time. Don Dahm PS Oh, I forgot. More people should hire Tony to help them select an instrument. | ||
| Tony Davies (Td) |
Don, Nothing controversial; until the last line! We must not use this forum for advertising. However,it is a real problem, maybe my approach is wrong. I agree with everything you said in the last message but the idea that Bruce is not the undiscovered world's best golfer is an assumption! Does this have any bearing on NIR spectroscopy? Just to go back to the original question I am not sure that Anne wants to "get serious with NIR spectroscopy" She needs analytical results. Moisture and fat in confections does not need a scanning spectrometer. If she wants to add other uses then that is another question. Best wishes to all, Tony |