|
Diffuse reflectance measurements are a mainstay of Near-Infrared Spectroscopy (NIRS). One potential difficulty has always been the shifting of the baseline and potentially non-uniform light scattering due to particle size differences between samples.
The physics of diffusely reflected light have been a topic of research for over one hundred years. Numerous papers have been published on the subject (1-4). For many quantitative applications of NIR spectroscopy to the analysis of opaque solids, care has been taken to make the particle sizes of samples uniform and mostly reproducible This has often led to milling, sieving, and grinding of larger NIR samples.
The "pesky" complications caused by non-uniform of different particle-sized samples do, however, follow the rules of physics. If peak or baseline shifting occur, then it follows that these phenomena can be used to measure their direct cause. In other words, the particle size can be determined from the "problems" that we often have to work around.
METHODOLOGY:
A Luminar 2000 AOTF/NIR Spectrometer, equipped wit ha fiber-optic reflectance probe, was used for powder measurements. NIR spectra were obtained of various materials of similar chemical composition, but differing in particle size. These spectra were then compared.
A second derivative transformation was performed to enhance any differences among the spectra. Earlier work done in NIRS has shown that derivative transformations could lessen particle size problems.
RESULTS AND DISCUSSION:
Figure 1 shows two distinct spectra, for essentially, the same chemical substance: table sugar. One spectrum is of the granular type, while the other is of the finely powdered confectioner's type.
Figure 1. NIR spectra of granular and confectioner's sugars after second derivative.
Even ignoring the water absorption peaks at about 1440 and 1920 nm, there are both peak size and wavelength differences between the two sample spectra that can only have been caused by particle size differences. It was seen in prior work (5) that these absorption differences follow a predictable pattern.
It was demonstrated that the absorption at any single wavelength is proportional to the reciprocal of the (mean) particle size of the samples. This was shown to be true for both organic and inorganic materials. The baseline shift seen in the absorbance spectra of the previous two sugars is only mostly cancelled by the second derivative function, demonstrating that all optical effects cannot be easily removed.
Some work has been done using normalizing functions and a few commercial software packages are available for this purpose.
Several representative pharmaceutical materials were sized and their mean particle sizes were determined via a reference method (laser scattering) (5). The absorbance spectra of three representative sieve sizes of aspirin, USP are shown in figure 2. Granules of 40, 100, and 200 mesh cover the gamut from granular to "micronized" (so-called because the particle size is reported in microns or 10-6 meters). The major features of the materials are identical, but the absorbance values increase with particle size.
Figure 2.
NIR diffuse reflectance spectra of 40, 100, and 200 mesh aspirin, USP.
The fact that the differences are greater at higher wavelengths is confirmed with Figure 3. In this the reciprocal of the particle size is plotted as a function or absorbance at four different wavelengths.
Figure 3.
1/x vs Absorbance for aspirin spectra at four different wavelengths.
Similar results are obtained for inorganic and organic salts as well as neutral compounds. They need not even have conventional NIR chromophores to work, as the scattering is a physical , not a chemical phenomenon.
It may be seen that NIRS may be used for the rapid determination of the partiv\cle size of solid materials. While each product within a company may require its own calibration, NIR provides an inexpensive and easily implemented means of making quantitative particle size determinations.
Click here to go back to the main Publications page.
|
