https://www.euro-online.org/enog/inoc2007/Papers/mac-slots.html https://www.euro-online.org/enog/inoc2007/Papers/m https://www.euro-online.org/enog/inoc2007/Papers/mac-slots.html

NIR quantification of soda-lime glass content in glass/wood mixtures

janskvaril's picture
Forums: 

Dear Sir or Madam,
I'm able to make a NIR calibration of a soda-lime glass content in glass/wood mixtures. Is there anyone how can help me to explain the phenomena behind? Why the soda-lime glass is detectable by NIR?
Thank you
With kind regards
Jan

joseacayu's picture

Hi Jan,
NIR spectra reflects the physicochemical properties of matter, from the vibrations of their bonds, particularly C-H. Thus, soda-lime glass content provides its characteristic 'fingerptint' to the spectra you have recorded. I don't know if it helps to clarify your doubt.
Regards,
José A. Cayuela

Andy McLauchlin's picture

Hi Jan,
Did you record a NIR spectrum of the glass? 
How does it compare with the spectrum of the wood?
Does your calibration work whatever wood you use?
Can you be sure that you are modelling the glass content and not a decrease in the wood content?
Regards,
Andy

dwhopkins's picture

Hi Jan,
I suspect that the wood samples are supplying a diffusely scattering background that is reasonably consistent, and I'll guess that that are mainly monitoring the Si-OH band that is unique and strong in the glass samples.  I have some questions, since you didn't provide any details.  You need to be sure you have validated your calibration sufficiently before you use it for routine measurements. 
Did you use reflectance or transmission measurements?  How are the samples prepared?  Are the glass components fragments of a uniform particle size?  What was the range of glass contents that you used?   How many samples did you use for the training set?  How many independent samples have you used for validation?  What method did you use for calibration, and what pre-treatments did you use for your spectral measurements?  If you used PLS or PCR for your calibration, can you show us the regression coefficient plot, so we can evaluate better why your calibration is measuring the glass content.
Best wishes,
David Hopkins