Radiation emitted by a spectrophotometer’s light source is by nature unpolarized; however, this light will inevitably become partially polarized by a spectrometer’s optical components. The primary causes of this polarization are due to:
(1) The reflection type diffraction grating used in the instrument’s monochromator. This is because of the very narrow ruling distance of the lines that make up the grating, radiation with an electric vector parallel to the ruling is reflected preferentially over the perpendicular electric component.
(2) Each of the spectrometer's mirrors has the potential to contribute additional polarization. The degree to which each mirror polarizes light is a function of the angle of the incident light. For aluminum mirrors at near normal angles (less than 10 degrees from the perpendicular to the mirror surface) the polarization contribution is zero; however, as the angle of incidence increases so does the polarization. Polarization is total at incident angles close to 56 degrees from normal (Brewster’s angle).
(3) Narrow monochromator slits introduce additional polarization. To achieve the minimum spectral bandwidth of 0.05 nm on the Lambda 900 spectrometer, a geometrical bandwidth of 18 micrometers is needed.
(4) The photomultiplier tube detector for the UV-VIS range exhibits slightly different sensitivity for different directions of polarization.
As seen in Figure 1 below, the measure of polarization is generally non-zero for any given spectrometer. Most spectrophotometers have their radiation partly polarized parallel to the slit. This effect is especially strong at the wavelengths for Wood's anomalies, which reside at around 500 nm and 2000 nm for the Lambda 900. This figure clearly shows the characteristic polarization profile for each of the respective gratings used in the UV-Visible and NIR spectral ranges. The instrumental polarization increases as a function of increasing wavelength in a similar fashion for both grating regions; thereby, reinforcing the fact that the diffraction grating is by far the most dominant element contributing to the inherent polarization of the spectrophotometer’s light.
 |
| Figure 1 |
One possible way to control the effects of polarization would be to take into account the degree of native polarization for a given instrument; however, this can never be applied in practice because of the number of dependencies involved. Another possibility would be to use deliberately polarized radiation; however, the most practical procedure is the removal of existing instrumental polarization, i.e., to depolarize the instrument beam. The best choice for this is an optical device, called a depolarizer, which is mounted in the instrument’s light beam and generates pseudo-depolarized radiation by scrambling; however, absolute depolarization is not easy to perform across the entire wavelength range of the instrument.
The depolarizer used in the Lambda 950/1050 is of the Hanle type, which consists of two wedges of differing optical material fastened together. The first wedge is made of double refracting natural quartz material, while the second wedge, manufactured from silica, is used to correct the direction of the beam. In order to achieve maximal depolarization the angle of rotation of the depolarizer must be individually adjusted for the spectrometer in which it is installed. This is accomplished by rotating the depolarizer around the instrument’s beam axis in a trial-and-error fashion until maximal depolarization is obtained.
The spectra in Figure 2 is from a NIST mirror measured at 30 degrees on a URA. The red spectrum was measured with a common beam depolarizer in place. The black spectrum was measured with the depolarizer removed. In the previous figure we saw the dramatic difference in the polarization state of the Lambda 950/1050 at the detector/grating change. This is a result of the fact that the change occurs at the low wavelength end of the NIR grating where the polarization is highly negative and the high wavelength end of the UV-Vis grating where the polarization is highly positive. These opposite polarization states of the two gratings are incorporated as artifacts into the spectrum of a polarizing sample when there is no depolarization of the sample beam.
 |
| Figure 2 |
The next two figures display the results of the effects from lack of proper depolarization as a function of reflection angle on URA measurements.
As can be seen in Figure 3, there are no apparent polarization effects at angles below 15 degrees. The first effects are noted in the spectrum at 20 degrees as a minimal step at the detector/grating change point. The step continues to increase in size with increasing angle at 30 degrees.
 |
| Figure 3 |
As can be seen in Figure 4, the step continues to increase in size as the angle approaches Brewster's Angle (56 degrees) for aluminum where the step would be maximal.
 |
| Figure 4 |
