AQ-00275-000, Rev. 3
24
4. Here it is assumed that the geometric scattering properties of the sample and the reference stan-
dard are identical. Therefore, the efficiency factor for radiation reflected from each sample will be
the same. If the geometric scattering properties of the sample differ greatly from those of the refer-
ence, significant systematic errors may be introduced into the measurements. Therefore, the geo-
metric scattering properties of the sample and the reference should be matched as closely as
possible.
Reflectance Measurement of Non-Ideal Samples
Two simple ideal geometric reflectance distributions may be described regarding reflectance spec-
troscopy: perfect specular reflection, and perfect diffuse reflection. Specular reflection is simply
the reflection of light according to Fresnel's law, without scattering or diffusion. Diffuse reflection
is reflection according to Lambert's law, and is therefore described as "Lambertian." For the pur-
poses of integrating sphere reflectometry, the Spectralon standards provided with this accessory
serve as an excellent approximation to the ideal of perfect Lambertian reflection. A clean, flat,
first-surface mirror provides an excellent approximation to the ideal of perfect specular reflection.
If your sample closely approximates either of the ideal reflectance characteristics, it is not difficult
to select an appropriate reference standard. Many samples, however, do not conform to this
requirement. Samples which are neither nearly specular nor Lambertian should be measured rela-
tive to a standard with similar geometrical scattering properties. If calibrated standards that meet
this criterion are not available - as is often the case - another approach must be taken.
The first such alternate approach is simply to measure the sample as if it were either nearly Lam-
bertian or nearly specular, and accept that such a measurement involves a higher-than ordinary
degree of systematic uncertainty. Such an approach is more than adequate for many applications,
especially those where the emphasis lies with repeatability, rather than absolute accuracy of mea-
surement.
A more ambitious approach involves modeling the geometric reflectance distribution of the sample
in question as an ideal "mixed" distribution, composed of two components, one of which is per-
fectly specular, and the other, perfectly Lambertian. The sample is measured in both total hemi-
spherical (specular included) and diffuse (specular excluded) geometries, and the relative
magnitude of the two components is estimated from the results. Systematic errors involved in mea-
suring a partly-specular sample against a diffuse reference, or a partly-diffuse sample against a
specular reference, can then be corrected in accordance with this model. An example of such an
approach is presented below.
A first surface mirror is measured with the instrument in the standard hemispherical configuration,
first using a specular reference standard (M), then using a diffuse reference standard (M'). The
geometric correction factor,
γ
, is calculated as follows:
Next, the test sample is measured, using a diffuse reference, in both the standard hemispherical
('specular included') and diffuse ('specular excluded') configurations. The results are given as R
SPIN
γ
= M/M'
Eq. 7