Problems with commercially available equipment

Several makes and designs of accessory for the measurement of absolute regular reflectance are available to fit many of the infrared grating spectrophotometers in current use. The common principle of these devices is shown in the two parts of Figure 1. During the sample reading R_s(λ) the device mirrors M1, M2, M3, produce the same attenuations as during the reference reading R_r(λ), but there are two additional reflections from the sample so that the sample reflectance, r(λ), is given by:

r(λ)={ Rs(λ) / Rr(λ) }^½

where λ represents a wavelength within the spectral region scanned.

Figure 1. Schematic configurations to show the principle of "V/W" type reflectometers,
with a sample reading on the left and a reference reading on the right.

Infrared grating spectrophotometers usually have three or four gratings, which are automatically deployed in sequence to cover the scanned thermal infrared spectral range of about 2.5 µm to 55 µm (4000 - 180 cm^-1). For some years users or would-be users of infrared absolute regular reflectance accessories have been conscious of problems with their function, which may be revealed by discontinuities in measured values across grating changes, or by indicated values which are known from other evidence to be erroneous. The equivalent problem in ultraviolet/visible region instruments is neither so apparent nor usually so serious, due to the absence of grating changes and to the much larger area of detector used, which make those instruments much less sensitive to optical misalignment than infrared ones. Systematic errors are revealed at grating changes because there are sudden and large changes of the state of polarisation, changes of spatial distribution of blaze efficiency across the grating apertures and often also changes of the solid angle of collection.

Fourier transform spectrometers also give erroneous reflectance values due to their being very sensitive to slight imperfections of optical beam substitution between the sample and reference scans.

Improvements at NPL to remove systematic errors

An absolute regular reflectance accessory, made by SPECAC to fit in the sample compartment of a Perkin-Elmer 580B spectrophotometer, was modified at NPL to give a more nearly ideal optical configuration as well as to permit fine adjustments during alignment. Figure 2 shows the configurations used for the sample and reference scans respectively, and a small platform carrying the sample aperture A and mirror M2 can be moved reproducibly from its position for the sample scan so that this mirror is then in position M2' for the reference scan.

In normal use of the spectrophotometer, the source image is formed at S. When the regular reflectance accessory is in place the radiation from the source optics is deflected by the plane mirror M1 to form a source image at mirror M2 or M2', and the concave mirror M3 is needed to form a further source image at S via the folding mirror M4. Due to unequal conjugates M2-M3 and M3-M4-S, the source image at S is now demagnified, although the solid angle of image formation is correspondingly increased (see Figure 2). Before the spectrophotometer was specially modified (see below), the demagnified source image at S failed to overfill the full height of the virtual slit image at the same position. The result of this was that any slight misalignments of the spectrophotometer optics, when combined with an inevitably less than perfect optical substitution between sample and reference scans, caused differences of vignetting between sample and reference scans somewhere in the wide spectral range covered (2.5 µm to 55 µm).

Figure 2(a): Actual optical configuration for a sample scan.

Figure 2(b): Actual optical configuration for a reference scan.

To prevent these effects, the optical components which are moved, that is A and M2, each needed to be set for tilt and azimuth on their platform correct to better than 0.5 minutes of arc. Even with careful handling, it was not possible to maintain such alignment for more than two days of active use without a critical realignment procedure. Other limitations of the technique were that sample mirrors had to be mounted against the aperture plate parallel to better than 0.5 minutes of arc, and their departure from flatness had to be no more than this. Even with all these conditions satisfied, valid results could still be obtained only after meticulous realignment of the spectrophotometer optics, particularly of the tilt adjustments of the four gratings.

These problems have been eased at NPL by building two additional limiting stops into the spectrophotometer optics. The first stop is conjugate with the grating in use, and preserves a constant solid angle of collection from the sample. The second stop is the more important: it reduces the entrance slit height used, so that the demagnified source image at S now properly overfills the virtual slit image there. One benefit of these improvements is that the movable components A and M2 need be aligned correct only to 4 minutes of arc. Another benefit is that distorted samples, departing from optical flatness by several minutes of arc, can now be measured reliably.

Calibration service

For calibration service work a relative reflectance measurement technique is used, where the sample is measured relative to an absolutely calibrated NPL reference standard mirror by strict substitution. This eases the problems of misalignment. So long as the sample can be substituted for the reference mirror without change of tilt or location, systematic errors caused by imperfect optical alignment are nullified.

References

1. CLARKE, F. J. J. Metrology and standards at NPL for the infrared region. Advances in standards and methodology in spectrophotometry, Elsevier, 1987.
2. CLARKE, F.J.J. Infrared regular reflectance standards from NPL. Proc. Soc. Photo-Opt. Instrum. Eng. 1996, 2776, 184-195.