| Author | Message | ||
     Eric |
Hi all, we have argued the filter for a few days. Any comments are welcomed! We will install a long wavelength pass filter on our FT-NIR, which blocks the short wavelengh light. Only the light wavelength above 2000 nm can pass through the filter. The questions are: 1)Shoud the absorbance be same after the filter installing? since A=I/Io. Filter blocks part of light, the Io decreases. I also decreases. therefore, absorbance (A) should be same. Is it correct? 2)Since the absorbance is same, we can not increase the pathlength, because A=xlc. concentration (c) and absorbance (A) are constant, so we cannot use a longer pathlength. right? 3)The filter blocks some light. we can change to a more powerful light lamp. But, how much can we increase? Is there any formula? 4)If a more powerful light lamp is applied, how is the temperature effect? anyone has experience? 5)What does the filter exactly do? only blocks some light? in other words, what can be improved? Thank you!!! Eric | ||
| Tony Davies (Td) |
Eric, I agree with your 1) but what do you want to do? Best wishes, Tony | ||
| Eric |
Hi Tony, A guy suggested us to install a filter on our FT-NIR if we want to measure low concentrations, but I do not know what can be improved by the filter. Can we use a longer pathlength, a more powerful lamp? others? Eric | ||
| hlmark |
Eric - that's a lot of complicated questions! I'll give it a try: 1)Shoud the absorbance be same after the filter installing? since A=I/Io. Filter blocks part of light, the Io decreases. I also decreases. therefore, absorbance (A) should be same. Is it correct? A1) Theoretically, yes. In practice, there are possible interactions with other parts of the instrument operation, especially in an FTIR. First of all, in those regions where the filter absorbs completely, if you try to calculate reflectance/transmittance, you may run into "divide-by-zero" problems, since Io will be zero, as well as I. Even if noise causes no data point to be exactly zero, the observed noise on the transmittance spectrum in those regions will be enormous. This may or may not be a problem when you come to display the spectrum. A more insidious problem that the FTIR instrument may be subject to is that when you cut out the high-frequency part of the spectrum you will be changing the nature of the interferogram. This is especially important around the centerburst, because most FTIRs use the centerburst to calculate phase corrections for the entire spectrum. If the centerburst characteristics are not what the software expects, it may not be able to do that correction properly. This may not only affect the computed abosrbance, it may affect the entire abilty of the system to come up with a meaningful spectrum. It will depend on how much of the spectrum is removed, and how the rest of the system responds to that. The key word here is SYSTEM: an instrument is more than a collection of parts, those parts are designed to work together in a certain way, and arbitrarily changing one of them can have major consequences on how the parts interact. 2)Since the absorbance is same, we can not increase the pathlength, because A=xlc. concentration (c) and absorbance (A) are constant, so we cannot use a longer pathlength. right? 2A) If you can still get a good spectrum (after the considerations of question #1) then yes, the absorbance of the sample in those regions of the spectrum where your filter lets radiation through will be the same. As for changing the pathlength, you may be able increase it if the absorbance in regions of interest is low. There is an optimum (fairly broad, but still an optimum) transmittance/reflectance value at about 33%; if the absorbance is too low, it is beneficial to increase the pathlength so that the transmittance achieves that value (and vice versa if the absorbance is too high). 3)The filter blocks some light. we can change to a more powerful light lamp. But, how much can we increase? Is there any formula? 3A) You can increase the source until you overload either the detector or the A/D converter, whichever happens first. You'll be able to tell because the spectrum will become distorted. 4)If a more powerful light lamp is applied, how is the temperature effect? anyone has experience? 4A) YES, you can expect the sample temperature to increase due to having more radiation falling on it. This will be partially compensated by the reduction in energy achieved by eliminating part of the spectrum. The degree of increased heating will depend on the characteristics of both the source and the sample. People have been known to damage samples by too much heat, even in unmodified FTIR instruments. 5)What does the filter exactly do? only blocks some light? in other words, what can be improved? 5A) That will depend on the design of the filter. I recall your bringing this question up before but I don't recall the details - you should have used the same thread so that we could check back on the earlier discussion. The bottom line is that if your instrument is already optimized, there probably isn't too much improvement you can make unless you're willing to undertake a major re-engineering effort. Howard \o/ /_\ | ||
| Bruce H. Campbell (Campclan) |
Previous discussions are archived. To go to the archives, click topics under Discussion at the top left. Then follow the path to the one you want. You can then do a search if you are having trouble finding the appropriate section. Bruce | ||
| hlmark |
Bruce - how do you manage to be 11:16 AM when it's only 6:30 here now??? Maybe that's why you have the energy to do searches. But when it's 5:00 AM and I can barely keep my eyes open, I need the info spoon-fed to me, I can't go traipsing off into the archives!! \o/ /_\ | ||
| Eric |
Hi Howard, Lots of thanks first! Now, I installed the filter on our FT-NIR. The absorbance of a solution in the region 2000--2500 nm of the spectrum is almost same as without filter. Since the absorbance of the sample is not low in this region, I cannot increase the pathlength. Perhaps, what I can do now is to increase the source. Does that mean I increase the energy available from the source? How about the detection limit then? Thanks again! Hi, Bruce, thank you for your rapid response. Eric | ||
| Tony Davies (Td) |
Hi Howard, Your message was timed at 11:30; times are GMT. I found the original question from Eric by searching on: long+pathlength (on page 2 of the output - March 22). But I cannot find the suggestion to use a cutout filter. It must be on another thread. "Filter" gives too many hits to be useful. Hi Eric, I think your best course would be to improve your S/N and you can do this easily with your FT-NIR. You just need to average a LOT of scans. Tony | ||
| Eric |
Hi Howard and Tony, We can get the previous discussion here: Bruce list/I need help/filter? detectro? more powerful lamp Eric | ||
| Edward Stark |
Hi I think that the addition of the filter is intended to increase the dynamic range of the FT by reducing the spectral range. This is discussed in Chapter 8 of Griffiths and de Haseth "Fourier Transform Infrared Spectrometry" Wiley 1986. A specific reference is on page 285. Ed Stark | ||
| hlmark |
Eric - Well, it seems you lucked out with the filter, that it didn't effect the instrument operation. As I said earlier, you can increase the source until something overloads. To some extent you can reduce that by turning down the gain in the signal channel; that may not be possible, though, depending on the design of your instrument. Or if it does, then at some point you will overload the detector preamplifier. Increasing the source energy will directly increase the S/N. But as Tony said, scanning and coadding the data many times is a much easier way to improve the S/N. On the other hand, you're fighting the square-root-of-N problem when you do that: you have to coadd four times as many scans in order to double the S/N. Another way to reduce noise is to get a quieter detector, but that would probably mean one that is cryogenically cooled. What type of detector is in the instrument now? Everything helps, but eventually you will run into some limit on each one. \o/ /_\ | ||
| Eric |
Hi, The detector in our FT-NIR is deuterated tryglycine sulfate (DTGS), which is not cryogenically cooled. I have problem to find suitable lamp now. the instrument is Spectrum One NTS (PerkinElmer). Does anyone know any suitable lamps? Eric | ||
| hlmark |
Eric - yes, I had guessed as much. DTGS is pretty standard for FTIR, especially for the mid-infrared. Come to think of it, you may not need to use a cryogenic detector, if you're replacing DTGS: you might be as well off using a lead sulfide (PbS) detector. Those are much more sensitive than DTGS in the NIR and will save you much hassle in not having to deal with the cryogenics. But I suggest you work with a good electro-optical engineer to select, optimize and install the PbS detector, since you will almost certainly need to replace the detector power supply and preamplifier, as well. Which also makes me guess that your whole system is set up for mid-infrared operation,, meaning that the source is a globar or ther mid-IR source of some sort. In that case, to replace the source is easy: the standard NIR source is a tungsten-halogen lamp. These run at much higher temperature than globars, and temperature is the main consideration in how much optical energy a lamp puts out. You could even try a tungsten-halogen automobile headlamp. BTW: there is another rather insidious situation you may run into: If you replace the source and the detector and are successful in reducing the noise level appreciably, you may reduce the noise below the level of the least-significant-bit (LSB) of your A/D converter. If this happens, you will no longer get any benefit from coadding your data, since there will be no random component in the output of the A/D converter. Coadding works only on the random part of the data and if you eliminate that your noise will be that of the LSB. Replacing the A/D converter is not really an option unless you want to go into the instrument business: replacing the A/D converter would involve major software changes as well as hardware changes - you would essentially be building a different instrument. It can eb done, but it's pretty far afield from making NIR measurements! \o/ /_\ | ||
| Eric |
Hi Howard, The system of our current instrument is as fellowing: source: tungsten halogen source Detector: DTGS (for transmission) & InGaAs (for reflectance) Beamsplitter: multi layer CaF Eric | ||
| hlmark |
Hmmm .... you might check the specs on the two detectors, to see if the InGaAs is quieter than the DTGS. If so, and if you're making transmission measurements, then maybe you can use the InGaAs instead of the DTGS for that. At least then you could use the same electronics, and you'd only have to change the instrument mechanically. Otherwise, you're back to finding a different (probably cryogenic) detector. Of course, some detectors, even InGaAs, can get improved performance with only modest cooling, say to about -40 deg C. Again, check out your detector specs to see what can be done with that. Then you'll "only" have the problem of preventing condensation. Since you're already using a high-temperature source, there's not too much more you can improve that. A higher-wattage source might help, but only if you can get the extra energy into the optical system. This may be difficult since the current optics are designed for the current source, and extra filament length or whatever a different lamp might have that gives the extra energy is likely to be wasted. Making it work can be tricky, if possible at all. Howard \o/ /_\ |