| Author | Message | ||
     Erik Skibsted |
Hi I am about to start deveoping a quantitative NIR method to predict the amount of coating sprayed on a tablet. I tried something similar a year ago, just using the reflectance spectra and it worked quite good. But now we would like to develop a method for our production line and the coating for the tablet consists some metallic compounds, I believe it is TiO2 and meaby also some Fe-compounds. The coating is pearly white and from some sources I have heard that NIR is not working for coating determination when there is TiO2 present in the coating, I guess it reflects the light as specular reflectance with little or no information? From my previous experiments I learned that you can make an "indirect" calibration i.e. you calibrate on the NIR signal of the core tablet and as you put more and more coating on the tablet you measure less and less of the core tablet. Meaby if the TiO2 (in the concentrations we are going to use in the coating formulation) are not too high then it should be possible to "see" the core tablet?? Secondly I thought about the varying penetration depth of the NIR light in the different regions and meaby it is better to concentrate on the third overtone region where the penetration depht is higher then the first overtone region? Any suggestion and practical experiences are warmly appriciated - I'll buy you a beer at NIR2005 in New Zealand if you have some good suggestions/comments :o) Regards Erik Skibsted University of Amsterdam, Novo Nordisk | ||
| hlmark |
Erik - your analysis of the problem seems reasonable. However, you're missing one key point: if the TiO2 is indeed interfering because of the effects of optical scatter, then you have to realize that optical scatter becomes greater as the wavelengths get shorter. Higher-order overtones occur at shorter and shorter wavelengths as the order increases, which in turn means that the scattering due to the TiO2 will increase and the penetration depth will therefore decrease. One alternetive you might consider is trying to measure the spectrum by transmission through the sample. The energy will be low, but at least whatever you do manange to measure will contain a signature of the tablet. Howard \o/ /_\ | ||
| Erik Skibsted |
Thanks Howard Well I have the option of measuring in transmission and I definitely try that out as well. In the film coating composition there are other components, and meaby it it possible to calibrate quantitatively to some of these components ? Well I am looking forward to attack the problem, and I'll share my experience here on the board | ||
| Tony Davies (Td) |
Erik, I think that transmission is a good idea but remember that it will be diffuse transmission. Many years ago I did a study on coated and un-coated tablets and got very poor results from the coated tablets working in reflection (we didn't have transmission capabilities in those days!). I just couldn't see the analyte. My explanation was that the energy that did get into the tablet was lost by internal reflection and could not escape from the tablet but it may have just been due to the difficulties already identified by yourself and Howard. If I was correct then the same thing could happen in transmission. I will be interested to know how you get on - Good Luck! Tony | ||
| Gabi |
Hi Eric, Your points are well understood - The questions I have are: 1. Are your coatings very "glossy"? Because when you say "pearly" I tend to think of a shiny surface. 2. Even so, in a correct set-up, the illumination and the collection angles should eliminate any specular reflectance. If the tablet is not flat, and the illuminated area includes the whole tablet, you may be getting specular reflectance from some of the surfaces. In that case you should check the set-up. 3. I don't think it is the TiO2 itself, because I had evry good success with TiO2 in the coating, as well as with non flat glossy ones, because I usually illuminate only the center part, avoiding the possibility of specular reflectance. 4. About the depth of penetration, Howard is right. 5. The issue of using information from the tablet, if you penetrate beyond the coating, and you manage to collect some of it, then if you use PLS1 you are using tablet information whether you like it or not. Look at the loading weights of your regression and see if you identify tablet material features in the loading weights. You can collect spectra of the tablet after removing the coating and use it to study the spectral features that come from the tablet. 6. About doing transmission - I am usually against it because we are "diluting" the spectral information of the thin coating with "tons" of spectral information from the tablet, thus reducing the sensitivity to small changes in the coating weight. The thicker the tablet, the worst off we are because we loose more signal, and dilute to a larger degree. If you wish to send me some spectra, I might be able to undesrand few things that I can't without seeing spectra. Particluarly I would like to look at the uncoated and coated tablet spectra if you care to send. My e-mail is [email protected] and I will be glad to share some information with you. Thanks, Gabi |