| Author | Message | ||
     Solomon Abebe (Chesa) |
Hi there, This is my first time to use NIR discussion. I hope am well come. I am new to NIR field and i think i like it. At the moment am using a transflectance probe in parallel with ATR-FTIR probe to measure solution concentration during protein crystallisation process. i found out that The NIR prediction during pure (homogenous) solution is pretty good but once crystallisation occurs the particles interfere and in the probe and pls model prediction is erratic. I guess the particles cause none-linearity. To improve the prediction i collected data on slurry solution and built a pls calibration model to predict the crystalliastion process. The problem exists with a bit of improvement. The data i used were absorption spectra. I tried the transmission data since i may get a trend opposite to that of the absorbance. Unfortunately, the prediction happen to be almost identical. My question how can i get similar prediction? The other problem is ATR-FTIR has been proven capable of predicting solution concentration in number of Papers. But i have a problem of particles incrustation on the probe which gives high absorbance during low solution concentration. How could i avoid that? There was a suggestion of heating the probe to avoid incrustation; would it make a problem since, IR itself a heat radiation? Your suggestions are much appreciated. Regards Solomon | ||
| Bruce H. Campbell (Campclan) |
Solomon, From your description I believe you calibrated only on the homogeneous system. If that is indeed correct, the growth of the particles will, among other things, cause an effective decrease in the path length. This would explain why the calibration doesn't hold in use when particles form. When you used transmittance instead of absorbance, the regression coefficients would be of opposite sign, compared to the use of absorbances, thereby "correcting" for the use of transmittance. I would be more leary of the non-linearity of transmittance in this case, leading to possible incorrect analyses. Heat effect from the NIR beam is quite low incomparison to heating the sample from an independent source. In any case, the amount of heat from the NIR beam should be near constant for all your measurements, provided the beam doesn't stay on the sample continuously. If the growth of crystals on the optics is very slow, you may not have a problem with incrustation; this would have to be evaluated for your system. Bruce | ||
| hlmark |
Bruce - I agree with the first part of your assessment, but not the second. If the calibration was done using only the homogeneous, clear solutions, then the model would not include corrections for the effects of scattering. This seems to be a premier case of not including all the variability of the samples in the calibration model. On the second part, concerning possible heating effects, I have to disagree. What you say is true for dispersive NIR instruments, the vast majority of which disperse the radiation before the sample. In this case, only the very small amount of energy in the narrow bandwidth selected from the beam will fall on the sample, and have virtually no ability to heat it perceptibly. With an FTIR instrument, on the other hand, the beam is broad-band and therefore even after passing through the interferometer, the beam has as much energy as it did when leaving the lamp. This broad-band radiation indeed has enough energy to heat the sample; indeed, upon putting your finger at the focus of the beam from an FTNIR insturment, you will feel the heat almost immediately. Cases are known where samples have been burnt by FTNIR beams. Howard \o/ /_\ | ||
| Lois Weyer (Lois_Weyer) |
If I understand the question about heating the probe correctly, you are asking whether or not one can heat a probe without affecting the spectra. I have had an experience in which a salt was precipitating out on my probe because the temperature of the probe was colder than that of the solution because part of the probe stuck out into colder air. In that case, we were able to heat-wrap the probe casing and prevent precipitation. The heat did not affect the spectra because it was just making up for the loss in temperature. I don't know how high you want to heat your probe, though. Also, it might break if you heat it too high - check the specs. | ||
| Solomon Abebe (Chesa) |
Thank you guys for discussing my problem. As Bruce said, if the Calibration model is been built up on the homogenous solution, i agree that the prediction will be wrong. But i have built a calibration model up on slurry solution which are the condition could occur during crystallisation. I found the prediction to be good until incrustation start to build on the surface of the probe. As soon as the incrustation occurs the reading will be the absorption due to the particle incrusted instead of the solution concentration that is decreasing during re-crystallisation. I reckon if i have a technique that the probe only could measure the liquid i may have a good prediction of solution concentration. I have thought of fine filter around the probe but the problem is the filter will be blocked by the particles as well. What method would you think i could use to detect on the solution? regards Solomon | ||
| hlmark |
Solomon - If you can't use heat to keep the material from precipitating on the probe, is it possible to remove the encrustation physically? I was thinking about the possibility of either stirring the solution in the vicinity of the probe very vigorously, or pumping the solution over the probe very vigorously, so as to physically wash away the encrustation as soon as it might form. Another possiblity would be some sort of mechanical wiper mechanism that would wipe away the encrustation. Usually, these things are easier to deal with if you can remove them as soon as they happen, while they are still small. Howard \o/ /_\ | ||
| Solomon Abebe (Chesa) |
Thanks Howard, it is not possible to remove incrustation physically because the probe is in a reactor of 500 ml and indeed it is close to an impeller with the 200 rpm. As u said i have tried to increase to 300 rpm but not more than that i should increase further and see. The wiper is interesting but seams hard work. Solomon | ||
| Bruce H. Campbell (Campclan) |
Solomon, If you are getting incrustation on the probe, I would think you are getting incrustation on all parts of the reactor. If so, you have what I think is a larger problem, phasing. Is the incrustation material the same as the particles that form? And if the incrustation doesn't occur on the other parts of the reactor, it may be possible to devise some surface treatment of the optics to prevent incrustation without preventing the transmission of the beam. The latter route may require recalibration. Bruce | ||
| Solomon Abebe (Chesa) |
Bruce, I see incrustation on the impeller even on the glass reactor it self after completing the experiment. What kind of surface treatment do you suggest? Solomon | ||
| Bruce H. Campbell (Campclan) |
Solomon, You may have three "phases" in your reaactor. One phase is the homogeneous solution, another the particles that form surrounded by the homogeneous solution, and the third the incrustation. The two latter may have definite differences, which would not be unusual in my experience. If indeed this is the case, you would have to decide if you really want to have incrustation. I have no experience removing the causes of incrustation, but a thought you may want to investigate would be to put a thin layer of an optically transparent material on the probe optics. A possible material may be a fluorocarbon. Others reading this may have experience and can suggest whether or not this approach would work. Bruce | ||
| hlmark |
It'll probably have to be trial-and-error. Fluorocarbon ("Teflon") will have good chemical properties, but might be subject to rapid wear in an environment containing particles that are being vigorously stirred. One important point here is that different kinds of fluorocarbons have different optical properties. The common "Teflon" is poly-tetrafluoroethylene (TFE), and is the highly scattering white type that we're all used to. But poly-fluoroethylenepropylene (FEP) is only very slightly hazy, almost clear, in fact, and may be better optically. Howard \o/ /_\ | ||
| Solomon Abebe (Chesa) |
What about their temperature resistivity the materials you suggest? I am working temp. up to 95 deg.C and what kind of addhesive are we going to apply? Will the addhesive be capable of handeling the environement? Solomon | ||
| Solomon Abebe (Chesa) |
Actually if these materials are really good, do not creep, no crystal can form on them.... It is better to make a filter around the probe out of the material you suggest to get only the solution instead of sticking them on the surface of the probe. What do you think? Solomon | ||
| hlmark |
Fluorocarbon is generally good up to 200 degrees Celcius. But using a porous filter will allow the solution to contact the probe, and if the incrustation is due to precipitation in situ, then it won't solve the problem because the precipitate will still form on the probe. "Trial and error" means just that: you're going to have to try different things until you find one (or some combination) that works. \o/ /_\ | ||
| Solomon Abebe (Chesa) |
Hi guys, we are thinking to purchase an NIR probe for solution and solute concentration measurement during crystallisation process. There are three kind of probes suggested a transmission probe with variable gap, a transflectance probe with variable gap and a reflectance probe. Which one do you think is appropriate to the tasks? Are there transmission probe with variable gap? The NIR instrument we have is an interferometer ABB Bomem with InGaAs detector. Kind regards Solomon |