Calibration model for Moisture and IPA

Anonymous's picture

I am trying to develop calibration models for  prediction (at-line mode) of moisture content and Residual solvent content (in this case isopropyl Alcohol) during fluid bed drying. Please let us know which wavelengths should be used in the calibration models for moisture and isopropyl alcohol. i am using Brimrose NIR with wavelength range of 900 to 2300nm. If any one is already worked on this kind of work, please share with us procedure of how to prepare calibration samples. I am thinking of two ways. one with offline sample preparationa and the second option is collecting real time samples from the batch for preparation of calibration model. Please suggest which is the better option. 
Note: I am talking about two different products one with Moisture and second product with Isopropyl alcohol.

tricour's picture

Hi Tiru,
I am very concerned to help you about this application, so feel free to describe exactly your request and i will transfer your coordinates and need to our technical specialist at Brimrose. So he could come to you asap to provide support.
Kind regards
Thomas Ricour
BRIMROSE EMEA MD for spectrometer
[email protected]

lois_weyer's picture

I can say definitely that if you try to take samples out of the batch and analyze them offline, you will get deceptive calibrations.  The IPA and/or water will evaporate partially when you take them out no matter how careful you are.  Also, the temperature at which you take the readings will affect the spectra.  I suggest that you use covered, offline samples, heating them to the process temperature if possible. Add-in water or IPA, tumble well, analyze before taking the NIR spectra and perhaps analyze them again afterwards.  I have used a constant-humidity room adjusted to a series of humidities to do this for a moisture calibration, so that the samples didn't change between analysis and NIR scans.

anikalam's picture



We did in line monitoring of moisture with NIR during fluid bed granulation. Although we used the full wavelength range, moisture has very specifics bands around 1450 and 1900 nm. I am not sure about the IPA. 


Collecting real time samples would be far better approach for calibration development in my opinion. Difference in sample preparation technique would affect the physico-chemical properties of the samples, that in turn would affect the NIR signal. We did LOD on the real time samples and developed the calibration based on that. Hope this helps. Good luck.