Abstract

Reflection gratings for use in vacuum ultraviolet (VUV) spectrometers are characterised by having a sinusoidal groove profile and a small step height (h~40-200 nm). The theory of the diffraction efficiency, as a function of wavelength, for a grating with a step height in this range is well established [1]. With a specified sinusoidal groove profile, the diffraction efficiency curve has a pronounced null at one VUV wavelength. The specified usable wavelength range of the spectrometer starts above this wavelength. Recent use of a commercial spectrometer [2] gave results which indicated that the wavelength region containing this zero response, and the associated surrounding region of very low diffraction efficiency, in fact yielded well resolved spectral content [3]. This raised the need to know the actual step height and its variation across the grating. Optical surface profiling has been used to quantify the step height of the grating in areas of 45 μm × 51 μm. The instrument used to measure the grating profile was a Bruker-AXS NT-9800 optical surface profiler in phase-stepping interferometry (PSI) mode. The spatial period of the grating is ~30 pixels wide so the step profile is well sampled in the imaged surface profile. Nevertheless, this is an exacting sample to measure by this technique and a series of potential artefacts that can affect the accuracy of the measurement have been identified and eliminated. These experimental details will be described. An example of the optical profile obtained, by best practice, for a region 8 μm × 8 μm is shown in fig. 1. Analysis of the data set containing ~22,000 sinusoidal cycles shows a near Gaussian distribution of step heights centred at a height of (56.25+/−0.03) nm. The full width at half maximum of the distribution is 13 nm. Critical appraisal of whether there are further sources of error that need to be found and eliminated in the calibration of the measurement is ongoing but at this time the step height measurement is believed to be accurate. However, there is no uncertainty about the height variation that has been measured. This height variation means the grating will have a small, non-zero diffraction efficiency at and around the expected null diffraction wavelength for the grating wherever this null actually occurs. The null wavelength is directly proportional to the step height. Thus, this measurement does establish that VUV spectrometers may respond in spectral regions where state-of-the-art claims say they will not.

© 2015 IEEE