| Author | Message | ||
     Hu Xu (Xuhu) |
NIR heating | ||
| Hu Xu (Xuhu) |
Hi, everybody I am a research student in UK. My research project is about thermoset curing by high intensity near infrared heating. But I know little about infrared before. So I have to teach myself the essential knowledge. One of the literatures said ¡°most polymers are semi-transparent in the NIR-range. In contrast, most organic materials have very strong absorption in the short to mid range infrared region¡±. I try to find the detail about polymer or organic material¡¯s optical property in infrared region. But from my research little book and paper works on it. May be I have missed or it¡¯s too simple to say. It has annoyed me for more than a month. Please give me an answer or recommend a few books. Thank you for your help. Hu Xu | ||
| Bruce H. Campbell (Campclan) |
Hu Xu, You could read a number of books, journals, etc, but there are Cothup-type charts that give molar absorptivities. You could then calculate the amount of photons absorbed at the various wavelengths, given the type of polymer you have. For example, if your polymer contains a CH2 group, that molar absorptivity is 0.1 around 1700 to 1780 nm. There are other places that group absorbs, such as around 2300 to 2400 nm with molar absorptivity of 0.25. You might have better luck doing such types of calculations for the amount of heat or energy absorbed. Bruce | ||
| Hu Xu (Xuhu) |
Bruce. Thanks for your answer. What I am interesting is that how to make qualitative analysis of the IR optical property of organic material, for example how to explain "most polymers are semi-transparent in the NIR-range. In contrast, most organic materials have very strong absorption in the short to mid range infrared region". Thanks. Hu | ||
| DJDahm |
In the NIR range, the overtones and combinations of the mid-IR bands occur. Each time you move up to the next higher overtone, the absortion generally decreases by a factor of between 10 and 100. So the NIR range seems semi-transparent. | ||
| Hu Xu (Xuhu) |
DJDahm, Would you please recommend a book or paper about what you said? Thanks. | ||
| DJDahm |
What? My word isn't good enough for you? Actually I don't have a good reference off the top of my head. Maybe someone else has one to offer. One of the disadvantaes of having lived 60 winters is that you don't remember where you learned things. Most of the books on vibrational spectroscopy dwell on the selection rules and at what wavelengths the absorption occurs rather than the intensity of the absorption. You might try the theory chapters in the handbooks, and see if they have references. I think Bruce's point would be mine; you really need to look at the data to know what's going on. Many compilations of IR spectra will have polymer spectra, usually of cast films. These will be thin, and the thickness is usually not given. We published a few NIR spectra of olefinic polymers in JNIRS last year. I believe that they were about 0.25 mm thick. Typically a polymer film in transmission in the mid-IR range will have a background absorption of between 10 and 20%, most of which is really scatter loss. Then there will be absorption bands that are very intense. But there will be large gaps where most of the light is transmitted. Generally, the more complex the structure is, the more absorption bands will be present. As you go down in wavelength, the bands will become less intense, but there will be may more of them, sometimes covering almost the whole NIR range. Around 2000 nm in the Near IR, a 0.25 mm film might absorb 50 to 80% of the light in the at the absorption maxima. At 1200 nm, maybe 20% will be abssorbed at the maxima. If you are lucky, you might find spectra of polymers that have been scanned with an FTIR instrument all the way through the IR and Near-IR range. | ||
| DJDahm |
Incidently, a good starter book on Infrared in general, not the spectroscopy of it is: "Infrared Radiation" by Ivan Simon (1966). | ||
| hlmark |
There are spectra of a number of polymers in Jerry Workman's new "Handbook of Organic Compounds" by Academinc Press. Unfortunately, the handbook is VERY expensive. Howard | ||
| Hu Xu (Xuhu) |
DJDahm, It's very nice of you to tell me the detail I want to know. I really appreciate your help. As you said, few books or papers about the intensity of absorption is published. But that's key to my project because I need to calculate how deep the IR can propagate before the radiation is toltally absorbed by semi-transparent material. So both IR spectroscopy of polymer and radiation heat transfer knowledge are needed. Would you please give some advices on this subject? Do you know someone or some groups have experience or is working on this subject? Thanks. Xu Hu | ||
| hlmark |
Xuhu - the classic books about diffuse reflection are the one by Paul Kubelka (strong on Kubelka-Munk theory!), and one by Harry Hecht. They were both published many years ago and are probably out of print, although you should be able to find them in university libraries - or try AMAZON.COM. There was also one published by what used to be CRC (the Chemical Rubber Company, which I think doesn't exist under that name any more) - I wasn't too impressed with the CRC book, but the others are pretty good, if you can find them. Good luck. Howard | ||
| hlmark |
Xuhu - I almost forgot - another classic book on the subject is by Kortum. Originally written in German, but there was an English translation available, too. I'm sure it is also out of print. In fact, I may have confused Kortum's book for Paul Kubleka - Kubelka wirte the classic article dealing with diffuse reflection, but I think maybe not a book: look for Kortum instead. Howard | ||
| Art Springsteen |
Just to confirm Howard Mark's response- Kortum is out of print. You might try contacting the publisher, Elsevier. Art Springsteen | ||
| DJdahm |
Xu Hu, I'm afraid you have "tapped the well" as far as my knowledge in this area goes. The two books that that Howard mentioned are out of print, but they can be found at any university that has a decent science library (especially a University that actually existed in the 1960's). Both have the English title: Reflectance Spectroscopy Wendlandt and Hecht (Originally in English) The first 90 pages are indeed "classic" stuff. Kortum (originally in German, but with an English Translation) very mathematical, but very good. As far as your questions go, I think the US military would have to be considered the world's leader in the technology you are interested in. I haven't seen their stuff written, but I imagine most of it has been declassified by now. As far as curing thermoset resins with IR goes, I think you would want to record the IR spectra while you were curing them. You will then know where they absorb. Keep in mind that it is not bad if the stuff that you are curing absorbs the radiation. That adds to the heat. However you want to find a wavelength that goes through any container that you may need to hold the resins in. It seems to me in principle a straightforward problem of wavelength selection, but I don't know your application exactly. If you want to cure the resins, the Near IR may not be a good choice, because the resin will not aborb the heat as effectively. Finally, the "Handbook of Vibrational Spectroscopy, Volume I, Edited by Chalmbers and Griffiths" is either out, or should be any day now, from Wiley. Don Dahm | ||
| hlmark |
Xu Hu - I agree with most of Don's comments, and have a couple more to add: 1) The question of using NIR as a heating source will depend on physical factors as well as chemical and spectroscopic ones. One major advantage of NIR in this respect is that a container that won't absorb NIR can be made from quartz (or even glass, in a pinch, if you can work at suitable short wavelengths), or made from any material and use a quartz window to introduce the NIR radiation. If the sample is thick enough, it will absorb all the radiation, even for the weak NIR bands. A container can even be made from Teflon, which also doesn't absorb NIR. There is also another non-absorbing plastic, although I don't know the name of it, but chemically it's a polymeric ether that has effectively all -C-O-C- bonds and no -CH, so it doesn't absorb NIR either. It's available from an outfit called Fresnel Optics. I have no idea what it's other properties are, though. 2) The Hankbook of Vibrational Spectroscopy by Griffiths and Chalmers is not yet published. I happen to know because I'm an associate editor, and we still have to write the prefaces. In any case, it does not include compendia of spectra, as the Handbook of Organic Compounds by Jerry Workman does, and that one is already published although as I said previously, the price is very high. Another compendium of NIR spectra is the old one from Sadtler (now Biorad) which has spectra of many pure organic materials but I don't think it includes polymers. Your best bet on all these counts is, a Don said, a good University Library. Howard | ||
| Shihying Chang |
>Xu, >By following your description in previous mails, I >think you may want to try the direction of Radiation >Chemistry and the topics related to radiation >modification of polymers. Only very few researchers >dedicated in that area. Prof. emeritus Joseph >Silverman over U. Maryland and Prof. Malcom Dole >(another Univ.)are among them. Dr. Dole has a very >complete book about the radiation chemistry and >dosimetry dealing with the depth profile of radiation >penatration. They were member of the Manhattan >project and a lot of the project materials may be >useful to your research. >Good luck. > >Shihying Chang | ||
| Hu Xu (Xuhu) |
Cool! I feel very happy to read all this intercourse. I think all the people that give me help will also be happy. As a Chinese proverb says: HELPING OTHERS IS THE ORIGIN OF HAPPINESS. In fact why I use NIR is base on its semi-transparent property to polymer because the polymer is on metal substrate which can reflect most of the NIR radiation that penetrate the polymer. So the polymer is possible to heat up and cure evenly. I will try to find the books you have recommended. That's not easy because I have check the online catalog of British library and get nothing. Maybe you can tell me the exact name. I have so many new knowledge to assimilate. I have another question to who familiar with Infrared lamp. What kind of IR or NIR lamp should be used as high intersity source and where to buy in UK? Thanks!!! Xuhu | ||
| Art Springsteen |
Re: Source for curing. If you decide to work primarily in the NIR, the best and most efficient source would simply be a big ol' tungsten/halogen lamp. Get them from Osram or Phillips or any other lamp supplier. Art Springsteen | ||
| andres rozlosnik (Andres) |
Hi Any of you have experience in measuring contrast between the slag and Pure steel in the near infrared band? Thanks Andres | ||
| Art Springsteen |
Andres, The steel should essentially be spectrally flat in the NIR while you should be able to pick up some of the silicate bands in the slag. Most metals, or at least all I've measured, have little or no NIR structure. Most minerals do. Slag is primarily silicaeous, so... Art S. | ||
| hlmark |
Andres - I think I want to make a few comments in addition to what Art said. When I first saw your message, my first thought was to wonder if you were talking about steel that was in, or had just come out, of a blast furnace. More generally, though, that led to thoughts that what you might expect to measure will depend on the conditions prevailing at the time of the measurement. If my first thought was correct, then for both the steel and the slag you would expect any spectral features from the sample to be overwhelmed by blackbody background radiation. In this case, you would see the same radiation from both the slag and the steel, except for any differences in temperature. If the samples are cool, then Art's description of the slag would be correct. The measurement of the steel however, would again depend on conditions. For all metals, both theory and experience tell us that the metal will have no absorbance. There is one big assumption made here, however: that the metal is pure, and that you are measuring the surface of the metal itself. Steel, on the other hand, will oxidize (especially if hot) and the oxide can undergo further reactions, for example with atmospheric moisture. Therefore you can get absorbance bands from the oxide (which we would expect to be weak) and also absorbance bands from -OH, which may be strong, depending on how many have formed. This brings you into a whole area of surface chemistry, large areas of which are complicated and not well understood, but if you can get some rust and measure it, that might give you some idea of what to expect. Howard | ||
| Art Springsteen |
Andres, Howard is absolutely correct. At furnace or even at any advanced temperature, the emission will swamp any spectral features. I made the assumption that you were looking at ambient temp. If you are, the silicates in the slag will be far different than the iron/steel/iron oxides. If you ARE dealing with hot stuff, NIR is not the way to go...Some sort of emission spectroscopy will work better. Art S. | ||
| Solomon |
Hello every body, i am trying to build NIR calibration model which can predict the particle size and solution concentration at the same time during crystallisation process. I found it difficult to build a calibration matrix in order to build this model. The information data i got for particle size is on raw spectra but for solution concentration these raw spectra have to be treated to reduce the partcles effect on the spectra by applying such as 2nd order derivatives. So now i have to kinds of data the raw and the 2nd order drivatives data i have to use one of the data to build the model according to my knowledge, if you have any suggestion, could you please forward me? | ||
| Nuno Matos (Nmatos) |
If I understood your problem why not using two independent models? It will be better for models maintenance. Other point, instead of the 2nd derivative try using MSC (I personally prefer MSC over SNV). Did I answer your doubts? Nuno | ||
| hlmark |
Solomon - Some other thoughts: you really do need two calibrations. For the concentration calibration you will probably want to eliminate the effect of particle size, which is what you want to measure with the other calibration. Thus, getting a good calibration for particle size may help you with the concentration calibration since you will be able to do a better job of eliminating the effect of it from the data. A complementary thought is that if you use MSC (Multiplicative Scatter Correction) to eliminate particle size from the data so you can calibrare for concentration, the coefficients of the MSC for each reading may well be what you could use to estimate the particle size for that reading! In looking at this question, I perused the whole thread, and have a suggestion for anyone trying to get hold of books that are out-of-print, as were discussed in the answers to Xuhu's request: both amazon.com and barnesandnoble.com have a used/out-of-print section where bookstores with all sorts of books list their holdings. I've recently been able to pick up a copy of Kortum that way. But you may hve to keep your eye on them for a while, until the book you want appears. Howard \o/ /_\ | ||
| Solomon |
Thanks Howard, as you said two calibration models could be robust. But I wanted to use one calibration model for both. As you suggested using MSC coefficients, but how can I verify the correlation of the coefficients to the particle size? | ||
| Solomon |
Thanks Nuno, i will check the out come using MSC but 2nd order derivativs also elimnate the base line shift due to particles. I know i can build to models but i want to build one model for both. | ||
| Nuno Matos (Nmatos) |
The thing I don't like in 2nd order derivative is it's high sensibility leading to falses unmatches. For the model duality you can try the PLS2 algorithm. Nevertheless using PLS1 is prefereble. Most software packages allow to include in one calibration file two calibrations (pe GRAMS). What software you're using? Nuno | ||
| hlmark |
Solomon - either we're misinterpreting what you mean when you say you want to use one model for both parameters, or you're not understanding the way these things work. You can create the two models from a single (correctly set up) data set, but you would still need two models. A calibration model, after all, is simply the numerical expression of the relationship between the property you are calibrating for, and the spectral data you are using to perform the analysis. So there will be one relationship for each different property (or constituent). Here's an analogy: let's ignore the fact that we're using multivariate maathematics. Imagine you were back in school and the instructor gave you a solution containing a mixture of copper sulfate (which passes blue light and absorbs yellow) and potassium dichromate (which passes yellow light and aborbs blue) (And further, let's assume that copper dichromate is soluble). So in order to analyze for the copper, you'd have to use yellow light, and to analyze for the dichromate, you'd have to use blue light. So even though each analysis can be done with a single wavelength, since they use different wavelengths, you would have to independently find the relationship between the concentration of each material and the wavelength it absorbs. This can be done with a single data set for the calibrations, and in the end, both measurements could be made from one spectrum after you've developed the relationships (models) but you would still need two different models for the two materials. Essentially the same thing applies here. Howard \o/ /_\ | ||
| Gabi Levin |
Hi, At last I see someone picking the relevant point, I was busy so I didn't get around to it. If Solomon is familiar with Chemometric software, he can see that it is possible to feed into the same "spreadsheet" a set of spectra, and a set of variables for which he wants to create a calibration. By running consecutive "regressions" each for one variable at a time, he will obtain a set of "calibrations, each for a variable, but they will all use the same set of spectra. I suspect that this is what Solomon means by saying one calibration - one set of spectra and he does not need to collect separate sets of spectra. The next question is - how do you want to do the predictions for unkonwns - if off line -then it is simple - you feed the unknown spectrum to the software, go to the Predict module, choose the correct "calibration" and click "predict" to get the value of the specific variable. You need to manually do so for all the variables. If you have a "Predcit" software that can accept a spectrum, and run predictions from several models, (still sequentially, but without need for operator intervention) for several variables, then you can get, in one click several predictions, for several variables. If you run online, you need a real time software, that stores several models, collects a spectrum, and predicts, (sequentially, but automatically without operator's intervention) all the required variables. From "Operational" point of view it is like having "ONE MODEL" although in practice, you have several models, created sepaartely for each variable, but from a single set of spectra. I hope this helped a little bit. Gabi Levin Brimrose |