Near-normal (10° angle) incidence

For several years NPL has provided measurements and transfer standards of regular reflectance at near-normal (10°) incidence in the thermal infrared spectral range of 2.5 µm to 55.6 µm (4000 cm^-1 to 180 cm^-1). NPL has subsequently achieved UKAS Accreditation (Calibration No. 0478SI) for this service. NPL uses an absolute reflectance technique (using a Strong-type or "V-W" reflectometer - Fig 1) to calibrate "in-house" reference standard mirrors. This is a difficult and time consuming process as the results are very sensitive to any alignment errors in the reflectometer. Stable samples with well known properties and a long calibration history, as well as a good understanding of the measurement system go a long way toward assuring that valid results are obtained each time.

Figure 1: 'V-W' reflectometer in the sample ('W') position.

Due to the difficulties of the absolute technique it is impractical to use this as a day-to-day calibration service. Therefore, for the calibration service NPL measures samples relative to the "in-house" reference mirrors using strict substitution and an adapted reflectometer. For most samples this enables NPL to provide a quick and inexpensive service with only a small increase in uncertainty over that of the "in-house" reference mirrors themselves. Measurement by substitution of the sample with a reference mirror in the same plane ensures that most sources of error (e.g. alignment, linearity) are significantly reduced and in the case where sample and reference are close to being identical these errors are virtually eliminated. With this technique the largest source of error is usually the uncertainty of calibration of the reference standard.

Figure 2: 'V-only' aperture plate and sample holder.

In order to achieve this measurement the usual "V-W" part of a Strong-type or "V-W" reflectometer has been replaced with a "V-only" attachment (Fig 2). A sequence of scans is taken, typically sample (S1), reference (RF), dark / zero offset (D1) and sample again (S2). Relative reflectance is then calculated from: [(S1+S2)-(2xD1)]/[2(RF-D1)]

The result is then multiplied by the absolute reflectance value of the reference mirror to give the absolute reflectance of the sample.

Larger angles (18° - 85°)

In 1999 NPL made generally available a service allowing calibrated measurements at angles of incidence of between 18° and 85°. Measurements are made at the angle of interest, separately in s- and p-polarisations, relative to an "in-house" calibrated NPL reference mirror. A mutually reinforcing pair of accurately set polarisers is used to control the polarisation.

Figure 3: Harrick-type angular reflectometer.

In this case the "V-W" attachment is replaced by that of a Harrick-type angular reflectometer (Fig 3). A sequence of scans similar to that used for the near-normal reflectance is carried out to obtain a relative value. As with the near-normal case, the relative value is multiplied by the absolute reflectance of the reference standard at the same angle and polarisation to obtain an absolute calibration. In order to obtain an absolute calibration of the NPL "in-house" reference standard mirrors the mirrors are calibrated at near-normal angle in the usual way, to give an absolute calibration. A combination of literature and fitted optical constants, n(ν) and k(ν), are then used to provide reference values at larger angles^[1].

Ordal et al.^[2] are considered to provide the most reliable data for the real and imaginary components of refractive index, n(ν) and k(ν), of the highest purity aluminium. The NPL reference mirrors were produced under normal "good practice" conditions and must contain slight traces of impurities, which will have the effect of slightly reducing k(ν). However, the n(ν) values of Ordal et al. can be validly applied to the real NPL mirrors and the near-normal absolute calibration is used to calculate special values of k(ν). The n(ν) and special k(ν) values are then used with the Fresnel Equations to give s- and p-polarised spectral reflectance for the NPL reference mirrors for any angle of incidence. This technique is insensitive to errors in the n(ν) data of the reference mirror, due to the fitted k(ν) values forcing the reflectance to be correct at 10° angle. The form of the Fresnel Equations then give reflectances at other angles which are insensitive to errors in n(ν) due to the compensating changes produced by the special k(ν).

NPL has been able to further exploit this technique and offer a reference standard mirror with tables of n(ν) and k(ν) values valid for the individual standard^[3]. This enables the end-users to calculate the reference reflectance values for any angle they choose (up to 85°). This service is only available for non-overcoated aluminium mirrors, which NPL supplies.

Artefacts

NPL offers a range of stocked artefacts for use as regular reflectance reference standards:

high values of reflectance

• Aluminium on a glass substrate.
• Aluminium on a glass substrate with a silicon oxide overcoat. (Tougher surface, but with an absorption feature near 8 µm).

spectrally varying, mid to low values of reflectance

• Uncoated float glass. (Reflectance ~ 0-30 %).
• Uncoated BK7 glass. (Reflectance ~ 0-40 %).

In addition NPL will calibrate any suitable sample submitted for measurement.

Reference

1. F.J.J. Clarke, S, P.I.E., 2776, 184 - 194 (1996)
2. M.A. Ordal, R.J. Bell, R.W. Alexander, L.A. Newquist and M.R. Querry, Applied Optics, 27, 1203-1209 (1988)
3. F.J.J. Clarke, J.R. Birch, C.J. Chunnilall, M.P. Smart, "FTIR - Measurement Standards and Accuracy". In course of publication: special issue of Vibrational Specroscopy - proceedings of International Conference on Advanced Vibrational Specroscopy.