Meas. Sci. Technol. 9 (1998) 256–260. Printed in the UK PII: S0957-0233(98)85267-3 Measurement of standard aluminium mirrors, reflectance versus light polarization J Campos, J Fontecha, A Pons, P Corredera and A Corr ´ ons Instituto de F´ ısica Aplicada (CSIC), Serrano 144, 28006 Madrid, Spain Received 19 June 1997, in final form 3 October 1997, accepted for publication 27 October 1997 Abstract. The dependence of the reflectance of standard aluminium mirrors on the polarization status of the incident light has been calculated and measured for an angle of incidence close to normal (8 ◦ ), so that the systematic uncertainty introduced by not controlling the light polarization status in spectrophotometric measurements of specular reflectance can be known and then eliminated or minimized. 1. Introduction Aluminium mirrors are mainly used to calibrate reflecto- meters such as those based on spectrophotometers and to determine some correction terms associated with the mea- surements of total reflectance (specular plus diffuse) with integrating sphere attachments. The uncertainty associated with these aluminium mirrors depends on the characteristics of the incident beam (the angle of incidence, spectral dis- tribution, spatial distribution, polarization and so on) since the value of the reflectance depends on those parameters. However, the polarization dependence of aluminium mir- rors at ‘normal incidence’ is not considered in the special- ized literature, perhaps because it is assumed to be small compared with other uncertainty sources in the measure- ment. We remind the reader that standard reflectance for ‘normal incidence’ is measured at 8 ◦ incidence by interna- tional convention. Elemental calculations point out that the difference between the reflectance values for a P polarized beam and an S polarized beam at an incidence angle as small as 8 ◦ is of the order of the measurement uncertainty of reflectometers of normal quality. So the polarization status of the light should be considered as a source of uncertainty when accurate measurements are needed and the degree of polarization of the incident light is not known. This is exactly the case of reflectometers based on spectrophotometers or grating monochromators, for which the degree of polarization of the light changes with wavelength and is very high for some wavelengths. This paper describes the determination of an uncertainty value due to polarization for two high-quality aluminium mirrors made in our institute. Aluminium is the most widely used metal for conventional reflectance standards in the UV, visible (VIS) and near-IR (NIR) spectral ranges, due to its high reflectance value, stability and ease of deposition. The structure of an aluminium mirror is formed by a more or less thick glass with a face polished to optical quality, over which a film of aluminium is deposited. In this case the film is deposited by high-vacuum evaporation of high-purity (99.99%) aluminium filaments. The aluminium film oxidizes immediately in contact with air [1–3], so the structure is completed by an oxide layer on top of the aluminium. The oxidation process, as described in the literature, depends greatly on the evaporation conditions, although it is fast at the beginning and slower later. Generally this process is completed within several weeks, with 40 ˚ A being a typical long-term layer thickness for samples kept under normal laboratory conditions. This natural process contributes to protecting the aluminium surface from possible later chemical attacks. From an optical point of view, this layer is considered non-absorbing in the UV, VIS and NIR spectral ranges. Consequently, the polarization dependence of the reflectance is due to the metal film and the oxide layer as well, both of which will have to be considered in the calculations. There is great experience in the fabrication of this type of high-quality mirrors in our institute, mainly for space applications [4]. 2. Theoretical calculations The whole structure of a high-vacuum evaporated aluminium mirror after oxidation can be seen in figure 1. In the case under study, the nominal thickness of the aluminium film is 1200 ˚ A for mirror 1 and 1500 ˚ A for mirror 2. These values are standard for evaporated mirrors made in the facilities of our institute. It can be accepted in both cases that the aluminium film is thick enough to be considered infinite, since, for an aluminium thickness 0957-0233/98/020256+05$19.50 c  1998 IOP Publishing Ltd