| Author | Message | |||
     hlmark (Unregistered Guest) Unregistered guest |
Fabien - I don't know, but there are limits to how many restrictions you can put on yourself and still accomplish anything. If you want a solvent that has no NIR absorbance bands, then it cannot have any -CH, which for an organic solvent means (essentially by definition) no hydrogen. But there has to be SOMETHING attached to the carbon atoms; and in practice that means a halogen (except for CS2, which is still something you might want to consider). There are some other non-hydrogen-containing molecules, such as COCl2, but that's phosgene which is a VERY poisonous gas used in chemical warfare; you certanly want to stay away from all those. So you're pretty much left with the various perhalogenated hydrocarbons. If any of those get into the upper atmosphere I suppose there is always a possibility of it decomposing into species that will react with ozone. But as long as you're only using lab-scale quantities, and you're careful not to release anything unnecessarily or in excessive amounts, I don't think you have to worry about the effect on the atmosphere. The alternative is to paralyze yourself from doing any sort of science. \o/ /_\ | |||
| Fabien NEKELSON (C156787) New member Username: C156787 Post Number: 1 Registered: 3-2006 |
Thanks a lot for all your reply. The use of CCl4 in the lab is quite restricted due to its high toxicity. I am checking for Freon 113 but I was wondering if this chemical can react with the Ozone layer? Fabien | |||
| hlmark (Unregistered Guest) Unregistered guest |
Just looked up Freon-113, it's trichlorotrifluoroethane; the boiling point is 118 deg F, which equals 47.7 deg C (somewhat less than I thought, but handleable in the lab) \o/ /_\ | |||
| hlmark (Unregistered Guest) Unregistered guest |
Gabi - Looks like we both found the same band, that's plaguing Fabien. CCl4 is, indeed, an ideal solvent for NIR, except that I don't think any company will allow it's use any more - too dangerous. Some of the other perhalogenated molecules you list are also NIR-band-free, as you say, but most of them boil at such low temperatures that they are more useful as refrigerants than as solvents. There's only one or two that have reasonable boiling points. One is Freon-113 (don't recall the formula offhand), that boils at about 70+ deg C. Fabien could try that, although I don't know how good a solvent it is. Howard \o/ /_\ | |||
| Gabi Levin (Unregistered Guest) Unregistered guest |
Hi HOward, I found spectrum for chloroform (CHCl3, so we all talk about same thing) in the Atlas of NIR spectra, 10mm pathlength, Perkin Elmer 4000A (an old animal but was doing a good job in it's days) Peaks are: (I did not try to make accurate determination from the graph) ~1120 - weak ( W) ~1402 - M (Medium) ~1515 - very weak (VW) ~1675 - very strong (VS) ~1840 - M ~ 2020 - M ~2090 - M So, the chloroform can not be considered a "clean" NIR solvent. The best choice will be Carbon tetra chloride, CCl4 - this one has no C=H bonds and can be considered a true clean solvent. The atlas of NIR spectra shows the spectra for CCl4 to be almost perfectly "flat" at zero absorption (small humps at 1680 and 1860 with 50! mm pathlength) Another good option, if acceptable from chemical aspects is Carbon di-sulfide, CS2 - the atlas shows perfect flat up to 2200nm with 10mm path. Ethylene tetrachloro C=C with 4 Cl atoms is also a possibility, and tetra chloro - difluoro ethane, C-C with 4 chlor and 2 fluor atoms, since they are all free from any C-H bonds. I hope this will help a bit Gabi Levin Brimrose Corp. | |||
| hlmark (Unregistered Guest) Unregistered guest |
Fabien - According to the spectra in volume 2 of the Handbook of Organic Compounds, J. Workman, Academic Press (2001) chloroform has some moderately strong bands at 1673.8 nm, 1709.5 nm and a weaker one at 1758.6 nm. Thus those bands are integral to the solvent. You could try using a different solvent, as you suggested, keeping in mind to use one that does not have bands interfering with your analysis. Alternatively, you could try doing spectral subtraction, digitally subtracting the spectrum of the chloroform from the spectrum of your sample. The accuracy of this procedure will depend on many factors, including the stability of your instrument, the concentration of the analyte, the linearity of the measurement system, and others. There are also mathematical methods to orthogonalize the spectrum of the chloroform to the sample, allowing you to calculate the optimum weighting factor to use to perform the subtraction. Howard \o/ /_\ | |||
| Fabien NEKELSON (Unregistered Guest) Unregistered guest |
I am using spectroscopic grade chloroform as NIR solvent to measure the absorption spectrum of my compound in the range 900-2000 nm. I am particularly interested in the intra charge absorption band which should appear around 1700 nm. My problem is that in this region of the spectrum some weak absorption peaks are present even after scanning the baseline with the solution without the analyte. I have purified chloroform by distillation and filtration over alumina but still I got those peaks. I was wondering if you have any idea how to avoid those aborption coming from the solvent. Should I change to other organic solvent like DCM, toluene or THF? Best regards, Fabien | |||
| Paolo Cucuzziello |
I know that CCl4 has been used for NIR spectroscopy but I can't use it in my lab for enviromental/health issues. Any other suggestion? Would taking my solvent as a blank give me reliable measurements ? | |||
| David W. Hopkins (Dhopkins) |
Paolo, It is not clear what your application is, but I think it is possible to offer some general suggestions. If you are doing transmission spectroscopy of clear or nearly clear solutions, using the solvent as a blank may give good results. I have found that water-based solvents still may give problems, and the most stable blank is an air blank, even for highly scattering liquids. There are often problems with temperature in water systems. If you are doing reflection measurements, then reflecting standards such as the Spectralon (R) 99% reflecting standard gives good results. I have heard that the 80% reflecting standard is also used. This would, however, add more noise to the measurement than a reference with lower absorbance. It would also allow you to have more nearly balanced reference and sample signals, which may be an advantage. It is desirable to use a repeatable and stable reference for the reference measurements, as the variability of the measurement is included in the noise of the sample reading. Hope this helps. Best regards, Dave | |||
| hlmark |
Paolo - if you're doing clear transmission measurements then NIR is no different than any similar measurements in any other spectral region: the ideal reference material is the sample without any analyte; this will match the sample characteristics (temperature sensitivity, solvent absorbance, etc.) as well or better than anything else you might use. Air is also a good reference, as David indicates, and will certainly not show any absorbance of consequence, but has the disadvantage that, if you're measuring clear liquid samples in a cuvette or other similar type of liquid cell, you cannot match the number of air-cell interfaces. On the other hand, to measure one air reference and one sample-without-analyte reference is easy enough, and you can then compare the effect of using each as your reference reading. As for your particular question: there are a couple of Freons that are completely halogenated (so that they have no absorbance in the NIR) but still have high enough boiling points to keep as a liquid and handle using ordinary equipment. I think one of them is Freon-113. I don't know what it's solvent characteristics or associated health concerns might be, though. But you can check for characteristics of Freons, to try to find other suitable ones. Howard \o/ /_\ | |||
| Paolo Cucuzziello |
dear all, I am using NIR to do quantitative measurements of tertiary amines in a polyolic matrix. Amines are 1 or 2 pbw in the blend. I need to go as deep as +/- 0.15 % in my analysis that means I would like to have results like : amine A = 1.0 +/- 0.15 The problem I have now is that the polyol matrix is really killing the signals coming from the amines. I guess this is due to some effects like - low signals from the 3amines ( only C-H and very low C-C ) - strong hydrogen bonding ( broad peaks ) - low conc in the blend itself So , I would like to try solvents to get rid of the hydrogen bonding problem. The problem is that extra dilution will then play against me even more than before. Any suggestion is very much appreciated I know I am probably too much MIR-minded but I am looking for a peak to look at and say : this is my amine. Is there any chance to do that ? Rgds Paolo | |||
| Paolo cucuzziello |
I am working with cuvettes in Transmission mode. | |||
| David W. Hopkins (Dhopkins) |
Paolo, I think you should be able to find bands specific to the tertiary amines, but they may not be easy to demonstrate. Can you measure the pure material to at least locate the bands and make an attempt to assign them? There is a very fine chapter that helps interpret organic spectra, by Lois Weyer and Su-Chin Lo, "Spectra�Structure Correlations in the Near-infrared", in the 5-volume set edited by John M. Chalmers and Peter R. Griffiths, Handbook of Vibrational Spectroscopy, John Wiley & Sons, Ltd, Chichester, U.K., 2002, ISBN 0471988472. Lois or Peter might be able to help you obtain a copy. Is it possible to work at high temperature, to try to minimize the formation of H bonds? Or, can you put some material in liq Nitrogen to try to sharpen the bands by low temperature? The fracturing and scattering that would occur would serve to amplify the band peaks by increasing the optical path of the light within the sample. This may not be a good procedure for routine measurements, but at least it might serve to help you understand the spectroscopy. Have you tried derivatives to resolve the bands? Second or even fourth derivatives sharpen up broad bands and make it easier for our visual inspection of the bands. Unfortunately, fourth derivatives are more difficult to interpret, because of the multiplicity of peaks introduced for each spectral band. You have an interesting problem, and I would like to hear more about your progress. I do not know what you mean by concentration in pbw, but if you mean ppb by weight, you do indeed have a challenging problem. ppm would be a better possibility, I think. Best regards, Dave | |||
| Paolo Cucuzziello |
Thanks David , pbw stands for part by weight and 1 pbw stands for 1 gram over 100 grams. I have read the excellent work done by Lois and Su-Chin and I took spectra of the pure amines. To start up my work I took the spectra of triethylamine : in that spectra it is easy to pick up C-H stretchings between 5700 - 5800 1/cm and combinational bands between 4000 - 4500. I am completly missing the combination band around 8000 for the C-H str + C-N str. Unfortunately this latest point was the one I was focusing on because the matrix' spectra is almost flat in that area. Combinational bands around 4300 could be a chance but I have seen that even very fine haziness can strongly change this area making any attempt to work there very tricky.Might be this sensitivity is an artifact coming from the instrument or from the handling of the samples. I am currently trying to dig a bit more on this area. I'll try to freeze my samples and see what comes out. Rgds Paolo | |||
| dhopkins |
Paolo, I have not seen that abbreviation for percent by weight, thanks for the clarification. At least this sounds feasible. Based on Lois' spectrum for butylamine, I would expect some nice bands at the combination region about 5000 cm-1 and overtone at 6600 cm-1, but I don't see why you would expect anything at 8000 cm-1? Derivatives can get rid of the baseline offset problems caused by haziness, but I think you would want to know why it occurs. A little bit of water would probably be a bad thing. I'd be interested in seeing the scans of the pure amine and a representative 1 pbw mixture. Could you send send me electronic versions at [email protected], or even attach them for all who might want to observe your spectra and offer opinions? (See Documentation/ Formatting/ Attachment on left side bar for instructions.) Best regards, Dave | |||
| Paolo Cucuzziello |
Dave , butylamine is not a tertiary amine and I can understand that Lois found those peak . 3ary amines don't have any N-H bond and this makes them ghost chemicals in the NIR. In the paper from Weyer and Lo it is stated ( pag 9 ) that a peak near 7937- 7874 cm-1 may be assigned to a combination of CH and CN stretching modes. Here follow and example of what I mean for ghost chemicals :
best regards Paolo | |||
| Paolo Cucuzziello |
Dave I had to post my file as a txt but you should be able to read it with excel at least. columns have been changed a bit but where you see 1.1 means that the figures below that columns are the trasmittance values for a blend with 1.1 pbw N,N dimethylcyclohexylamine Rgds Paolo | |||
| Gabi Levin |
Hi everyone - Paolo is right - tertiary amines don't have any N-H bonds. So, I would not look for them where the N-H bonds should be . However, the compound Paolo is talking about does have several C-H bonds, in the methyl groups, and also in the cyclo part, at least I would assume so. These C-H bonds, although similar to the C-H bonds in the polyolefins (I assume polyolefins) they are not identical. I would expect the peaks to be shifted due to the presence of the amine group, the cyclic structure, and this could give rise to calibrations. However, don't expect to base the calibration on a few "preffered" wavelegths. Such calibrations will use a good number of PC's and will reuqire full spectrum. The major concern, is the low concnetration, roughly 1% by weight. This is a low range for creating a calibration on the basis of relatively small spectral differences, however, I was able in the past to achieve decent applications where I had situation with very similar compounds - monomer and dimer. The key to successful calibration will be the accuracy of the reference method. With a range from 0.5 to 1.5 % - the reference method should be as good as 0.02 or 0.03 to provide for a successful calibration. I hope this will help paolo in focusing on a useful approach. Many thanks, Gabi Levin | |||
| David W. Hopkins (Dhopkins) |
Paolo, Of course, you are right about the tertiary amine structure, I was overlooking that important fact. However, your spectra are very impressive, and it seems to me that there are good features at 5700 and 5790 cm-1 that show clear differences between the amine and the matrix. If you take the difference 1-2 (after converting to Abs), you see some interesting difference peaks at 6660 and 7020 that may be due to the amine and measureable at the 1 pbw level. I see some striking water bands in the matrix, and the water will contribute to a calibration. This may be a problem, unless you ignore the water band and include enough samples so the relationship between water content and amine content is eliminated. I think you may be up against the noise limit of the instrument, so you should consider averaging 9 or more scans of each sample. Do you have access to a monochromator-based instrument? You might like to try some representative samples and see what a lower photometric noise level would accomplish. I second Gabi's comment about reliability of the lab reference method, that will be key to your success, but I think you should be successful! The other key will be a good experimental design, with a "proper" number of calibration and test samples. Best regards, Dave | |||
| Gabi Levin |
Hi David, Thnaks for your comment. Looks like my guess about spectral differences of the C-H is correct, 5700 (1754nm) and 5790 (1725nm)is a correct location. The presence of water will complicate, but unless its concnetration is linked to the concnetration of the amine it is possible to calibrate for both. The way I will approach this problem would be plenty of process samples, to include most possible varaibilities of water, amine, matrix, and minute contaminants that may be present. Plenty I mean over a 100. Then I would use PLS1 with full spectrum. Then I would predict 30 to 40 new samples, and do my SEP on these. I would look at the loading weights (I hope his software provides it) of each PC by itself. I would look at the PCs that explain high portion of the Y (I hope he has this information provided) and watc for high loadings in the region where the amine shows differences from the matrix. If this repeats itself for a few PCs, his calibration is based on true spectral information and should yield good calibrations, granted that the reference method is good enough. I could use the spectrum in Excel format, along with the Y values to try the Unscrambler and see how it goes. Could Paolo or David e-mail them? [email protected] Thanks a lot | |||
| David W. Hopkins (Dhopkins) |
Gabi, Anyone can download the ascii text file from Paolo's posting of 8/25 10:36. Just click on the name of the attached file, in the message as printed in the string of messages at the nirpublications website. This is slick. He could have attached a graph, but by attaching the spectra, we can use any software to observe and manipulate the scans ourselves. Best regards, Dave | |||
| Riccardo De Cesare |
Ciao Paolo, looking at the spectra you posted it seems that you are working with a polyether polyol but I might be wrong. If this was the case then I would focus on the first overtone of the CH bands around 8200 cm-1. This to avoid the "noise" coming from water. I would also ( if possible ) increase the optic path lenght. I think you would never see that amine peak you are looking for ( not at 1 pbw level ) and you really have to be carefull not to build your calibration on what I call a "dilution effect". It can happen to find very good correlation even when a chemical is not absorbing at all. For example you could find a correlation for the amount of CCl4 in a blend of chemicals just because you are diluting the rest. This can always be an option but you should know already quite a lot of your blend to do so. Fortunately C-H stretch in cyclic alkanes ( N,N DMCHA )is shifted from that of linear alkanes and this is what I suggest to use if you don't have other cyclic alkanes in your blend. I don't know if this can be used for your analysis but I hope so. Ciao Riccardo |
polyol + amine