| Author | Message | ||
     ali |
What is the difference between NIR transmittance and NIR reflectance regarding the analysis of meat proteins? Are they interchangable? | ||
| Tony Davies (Td) |
Dear Ali, Yes, either way is possible. If you use reflectance then you need to measure in the �Norris� region 1100-2500 nm. If you use transmittance then you need to switch to the �Herschel� region, 780-1100 nm. This allows you to use longer pathlengths (10-15mm) and is my preferred method because it samples a larger volume of material. You do need an instrument that is optimised to the region. Best wishes, Tony | ||
| yonll |
Hello, For reflectance spectra of (chicken) meats, please look at my recent publications in Meat science, 58, 395-401 (2001) and 63, 301-307 (2003). Best regards, Yong | ||
| Nagarajan (Nagarajan) |
Hello, I am working on diffuse reflectance of powders in the mid IR region. Can anybody suggest me how to do the sample preparation with & without substrate in order to use them in Integrating sphere. | ||
| ali |
Dear Tony, I am going to replace FT-NIR transmittance spectroscopy with the traditional kjeldal method for determination of protein content of meat in our lab. I intend to use PLS for calibration/prediction processes. Here are several questions before me that need your help: 1) What is the problem with using a wider range than the so-called �Herschel� region. Won't that produce more selectivity considering the noise filtering capability of PLS approach? 2) Regarding the sample preparation. After grinding and blending the meat, how should I feed the semi-solid material to the FT-IR instrument? How should I make a homogenized/thin film of meat? 3) In comparison with NIR reflectance, are the absorbance profiles of the two methods comparable? ( I have not worked with NIR reflectance at all) 4) In choosing calibration samples, should I look for a wide calibration or a narrow population? 5) Have there been any previous study on the determination protein content of meat by FT-IR transmittance? Thanks in advance for your information and help. Ali Shamsaie Department of Food Chemistry Faculty of Pharmacy Tehran University of Medical Sciences Tehran, Iran P.O. Box: 14155-6451 emial: [email protected] | ||
| Michel Coene (Michel) |
Dear Ali, Can you tell us what instrument you work with? If you will be working with a traditional FT-IR converted to NIR you might have quite a sampling problem. These machines usually take quite a small volume of sample and if you want the measurement to be representative of a large chunk, you will need to blend it extremely fine. Traditional transmission meat analysers usually have a big fat lamp and large filters to send light through several cm3 of meat. The problem with the wavelength range is exactly that the higher wavelengths get absorbed faster and will not penetrate completely through a thick sample. You can only use them in reflection because the pathlength there is shorter. Finally you use FT-IR and FT-NIR in the same question. Please make a clear separation of both methods. It might look like just changing a beamsplitter, but it is a whole different ballgame!! | ||
| Art Springsteen |
Dear nagarajan, This is a somewhat complex question to answer. It depends on a number of things. A) what kind of sphere are you using? (Up or downlooking, sidelooking?). b) What are the powders? Typically, one mixes the powders with something like KBr to dilute and minimize Rehstrahlen bands but you might be able to measure them neat. Secondly, you could compress either the KBR/powder mixes into pellets and measure the pellets but one must be careful in that the reflectance is very density depenedent. Mid-IR is a bit more of a problem than working in the NIR, since powder cells with transparent windows aren't as readily available, thus it's a bit more difficult to use the sphere to measure. Regards, Art Springsteen Avian Technologies LLC www.aviantechnologies.com | ||
| Tony Davies (Td) |
Dear Ali, I assume you are hoping to replace the Kjeldahl method with FT-NIR transmittance. Answers to your questions: 1) In transmittance applications you have to decide on a path length. If you want to use a moderately large path length then you have to limit absorptions by the use of shorter wavelengths (i.e. Herschel region). If you use large path lengths and longer wavelengths then absorbance will be large and the detector will have a very small signal and you will get a very noisy spectrum. Putting this sort of data into PLS is not a good idea. 2) Sample homogenisation is very important in meat analysis. You probably need to use one that was designed for meat. These will produce a very consistent paste. It is still going to be very difficult to produce a representative and consistent thin film, which is why I recommend the use of short wavelengths and thick samples. Most of my meat experience comes from work on the development of the Foss �MeatSpec�. The homogenised sample was put into a plastic bag, which was squeeze to a known thickness in a sample holder. This was analysed using Herschel region transmission spectroscopy. 3) Similar but not the same; because you have to change wavelength regions you are not looking at the same thing. 4) The calibration samples should span the population range that you want to analyse. If this is a very wide range you might have to split the calibration into �high� and �low� in order to achieve the required analytical performance. 5) I have done a search on the NIRP database and did not find any reference to meat and FT-NIR. This is not a bad thing but you should ask your instrument supplier for advice. Best wishes, Tony | ||
| ali |
Dear Michel, Currently we have no IR instrument. Is there a major difference between FT-IR and FT-NIR? I have not any direct experience with any type of IR instruments but think FT-NIR is equivalent to FT-IR in the NIR region! and all FT-based IR instruments use beamsplitters. Please correct me if I am wrong. Thanks. ali | ||
| Michel Coene (Michel) |
Ali, The difference between FTNIR and FTIR is one of physics. Usually one instrument can be used for both, by changing beamsplitter(NaCl/glass), lamp (Nernst glower/Halogen lamp) and detector. The big difference is pathlength. FTIR requires extremely short pathlengths, and you face problems such as the fact that you can not use glass in the optical system. I don't know if you feel like squeezing that meat between two large NaCl crystals... Mid-Ir (FT_IR) gives more information and is easier to calibrate, but sampling is far from obvious. In the NIR region you can use glass cells and lenses, and not worry about CO2 in the air. The price you pay is a more complicated calibration procedure (and that is where PLS comes in). If you haven't bought an instrument yet and you only want to use it for analyzing meat, this should strongly influence your choice. If "routine" measurements are good enough, e.g. because you want to make statistics of the cattle population in your country, then you should look at custom made analysers like the meatspec. With AOTF or Diode Array technology, you can even build systems that measure fat and protein while the meat passes by on a conveyor belt. If you want to study the spectroscopy itself (or get more accurate results), then you should look at FT systems, AOTF (Brimrose), scanning (NIRSystems)and possibly high end diode arrays. Be carefull however to consider sampling. As I said before, the bigger the sample volume actually "touched" by the light, the less you have to worry about your sample not being representative. An FT-IR system will typically have a sampling volume of a few square mm diameter and only microns in depth!!! Theory and litterature will show you all kinds of beautifull results but if you have to analyse 1000+ samples then you better have an easy sampling procedure or you will soon start to see all kinds of "monday-morning" errors appear (aka "presentation errors" in the official lingo). | ||
| ali |
Dear Michel, Thanks for your helpful informations. Whare can I find more detailed information on the subject in the net? | ||
| Michel Coene (Michel) |
Unfortunately there are not many full articles appearing on the net. Scientific magazines want to sell paper copies (or internet subscriptions) and manufacturers are afraid the competition will steal their good ideas. Try this site and the things mentioned above. A good place to ask for advice is the Danish Meat Research Institute, as they have quite a reputation in the field. Look in the sites of the manufacturers and do searches with Google etc. Do not forget you will not only have to choose hardware but also a chemometric package (unless it is included with the hardware). The Unscrambler (Camo) is very userfriendly in the sense it does not require you to exactly understand all the mathematics behind but gives you a "feel" of your data instead. This is however partially compensated because it integrates extremely badly with spectrometer-control software. I need to go through seven different file-formats to make one calibration on one component. Grams integrates much better but will maybe feel slightly more "technical". Pirouette used to be big in the US, but I don't see much happening recently (am I wrong?). Matlab will give unlimited possibilities but requires a solid background, and programming experience will come in handy. There are others, and some spectrometer suppliers simply write their own. | ||
| Andreas Hannisdal |
Hi, I'm analysing heavy petroleum products and need some support. Due to the very strong absorptivity of these samples I have to use my diffuse-reflectance probe mounted onto a gold background ensuring an apparent path length of only 2*0.16 mm. So to the case. How can I expect my spectra to differ from a typical transmittance experiment? The heavy crude oils contain some scattering components (inorganic particles, wax and asphaltene aggregates). The probe end design is a detector positioned in the center of 7 light sources. Andreas | ||
| Bruce H. Campbell (Campclan) |
Andreas, You may try to warm the samples. This may enable you to increase the path length. I did that with some petroleum samples and it helped. Bruce | ||
| David W. Hopkins (Dhopkins) |
Andreas, Even at a cell physical path of 0.16mm, you may be measuring these turbid samples more in reflectance than transmission, especially at longer wavelengths. When the absorbance readings are above 0.8 or 1.0, the reading is dominated by reflectance. At lower absorbances, the light may actually be penetrating to the gold background and back. Therefore, at lower apparent absorbances your spectra will look most like true transmission scans, and at higher absorbances, the bands will appear to be somewhat reduced from what you would expect from a pure transmission measurement. I think you might have less difficulties with cell flow, blockage and cleaning if you just used pure reflectance measurements on a cell without a gold backing and a more convenient and longer path. Is that an option? Have you tried it? Best regards, Dave | ||
| Andreas Hannisdal |
Bruce and David, Thanks for the answers. I agree that the pure reflectance option is a possibility. However, very short path length is necessary (less than 1 mm for these samples) so a transflectance probe with this short path length makes it hard to get the sample (viscous) in there. I think the best option is a transmission or reflectance cell with optical cables connected to it.(Works fine for online measurements as well) Thanks for the comments about the differences in spectra! Andreas | ||
| David W. Hopkins (Dhopkins) |
Andreas, I liked Bruces's suggestion to operate at a higher temperature, to dissolve some of the higher melting point solids. However, even at higher temperatures, it is likely that the samples will be very turbid. The point is, in Reflection measurements, the sample properties determine the depth of penetration and the total optical path in the material (although we do not know the total path). Therefore the geometry of the cell can be opened up to enable handling viscous liquids easily. It is necessary to select a cell with a depth sufficient to be infinitely thick for all samples and all wavelengths, and from what you say, a cell with a depth of 0.5 or 1.0 cm would probably work well. Then you avoid problems with transflection or transmission cells with very narrow gaps. I think you understood this, but I was puzzled that you continued to talk about the necessity of a short path. Regards, Dave |