The effect of sample packing of a powder

0,2 0.15 0.1 0.05 _ j -0.05 -0.1 -0.15 CM CD CM (N W avelength (nm )

Figure 33 Sample packing effects upon allopurinol powder

The example show packing effects caused by shaking same the sampling vial between measurements of the allopurinol powdered pure drug. Spectra between 1100 and 2500 nm

The effect of shaking the powdered drug sample between each measurement can be seen in Figure 33. There are differences in the spectra which are not apparent when repeatedly measuring the powder and sampling vial held in the same position. By removing the sampling vial from the instrument and shaking

the contents, different areas of the powder are being examined and different portions of the powder sample surface are brough intpo contact with the light. This has led to greater light scattering.

0.06 0.04 0.02 - • -0.02 -0.04 -0.06 -0.08 -0.1 -0.12 o 0 0 CM CD <o CM CM CM CM CN W a v e le n g th (n m )

Figure 34 Sample packing effects - Second derivatives

The example shows packing effects caused by shaking the sampling vial between

measurements of the allopurinol powdered pure drug. Second derivative spectra are shown between 1100 and 2500 nm

A full Spectral shift of this nature is caused by the re-orientation of the way the powder particles are packed and is due to the non-regular shape of such

particles and changes in bulk density. Irregularly dispersed particles reflect light at different angles and produce a different response on the detector.

This is a consequence of sampling that will have to be considered when comparing results within all other experiments. The effect produced does not

affect the actual peak position in terms of wavelength but does have a bearing on the intensities and relative intensities of all peaks.

To take this effect into account when constructing spectra to be stored in the database, all samples should be shaken in between measurements and sufficient measurements should be made to take account of these effects. Physical effects such as this can be reduced by applying the mathematical second derivative to the spectra. A description of derivatives is contained within section 1.10.1. A second derivative plot of the spectra is contained within Figure 34. The effects evident in the spectra are no-longer noticeable as the peaks at a particular wavelength now appear to the same order of magnitude and there is no variation in the baseline. The same principle can be applied to tablets. Measurements of tablets rotated between measurements and sampled on both sides produced variation between the recorded spectra (Figure 36).

0.7 -r 0.6 - 0.5 - 0.4 - u 0.3 0.2 - 8 ^ ? •«— t - CM 0 0 § § Wavelength (nm)

Figure 35 Twelve spectra from one allopurinol lOOmg tablet BP, held in one position Spectra produced between 1100 and 2500 nm

Areas of variation include the curvature of the tablet surface, the packing within the tablet and surface effects such as scratching, scoring and embossing.

These effects have occurred because the portion of the tablet directly in contact with the incident ray has varied. This spectral variation is not present when sampling the same portion of one tablet repeatedly (Figure 35).

0.8 0.7 0.6 0.5 0.4 0.3 0.2 00 <D i h- h- do CD Wavelength (nm)

Figure 36 Aspirin 300 mg caplet, 10 rotations on each side of the caplet. Spectra between 1100 and 2500 nm

The effect of varying tablet position can be seen by measurements taken of one aspirin tablet (caplet). The tablet is capsule like in shape and contains a curved surface with scoring on one side of the caplet. The variation in spectra produced (Figure 36) is of the same type to that noted when shaking powdered samples between measurements. A spectral shift is noted but the magnitude of each individual spectrum is the same - the baseline has shifted, not the peak heights or peak wavelength positions. The detector iris was positioned in the middle of the sampling platform with the iris diameter being less than the

diameter of the tablets being sampled. Stray light as a result of the tablet not covering the iris was not thought to have contributed to the light scatter. As with the powdered drugs the spectral shift effect was reduced by using second derivative plots of the spectra (Figure 37). Experiments in this thesis will attempt to build the maximum amount of variation possible for each sample into the spectral libraries.

Measurements to produce pure drug spectra will therefore continue to

incorporate the shaking of sample vials between each measurement and tablets will be rotated between measurements and sampled on both sides of the tablet.

0.06 0.04 0.02 -0.02 -0.04 -0.06 -0.08 (M (O 0 0 O ■<- CO (Û ro CO s i _ CNI 0 0 lO CM T - CM CO CM CM CM CM CM (D 00 CD CO 0 > CD Wavelength (nm)

Figure 37 Aspirin 300 mg caplet, 10 rotations on each side of the caplet. Second derivative spectra between 1100 and 2500 nm

It would appear that where differences in chemical characteristics are important, for example when seeking to identify samples, second derivative spectral