In recent years, demand for an optical element having a high degree of freedom shape is increased. High-precision aspherical shape is required for the X-ray focusing mirror etc. For the head-mounted display etc., optical element of the free-form surface is used. For such an optical device fabrication, measurement technology is essential. We have developed a high- precision 3D nanoprofiler. By nanoprofiler, the normal vector information of the sample surface is obtained on the basis of the linearity of light. Normal vector information is differential value of the shape, it is possible to determine the shape by integrating. Repeatability of sub-nanometer has been achieved by nanoprofiler. To pursue the accuracy of shapes, systematic error is analyzed. The systematic errors are figure error of sample and assembly errors of the device. This method utilizes the information of the ideal shape of the sample, and the measurement point coordinates and normal vectors are calculated. However, measured figure is not the ideal shape by the effect of systematic errors. Therefore, the measurement point coordinate and the normal vector is calculated again by feeding back the measured figure. Correction of errors have been attempted by figure re-derivation. It was confirmed theoretically effectiveness by simulation. This approach also applies to the experiment, it was confirmed the possibility of about 4 nm PV figure correction in the employed sample.
Aspherical optical elements with high accuracy are important in several fields such as third-generation synchrotron
radiation and extreme-ultraviolet lithography. Then the demand of measurement method for aspherical or free-form surface
with nanometer resolution is rising. Our purpose is to develop a non-contact profiler to measure free-form surfaces directly
with repeatability of figure error of less than 1 nm PV. To achieve this purpose we have developed three-dimensional
Nanoprofiler which traces normal vectors of sample surface. The measurement principle is based on the straightness of
LASER light and the accuracy of a rotational goniometer. This machine consists of four rotational stages, one translational
stage and optical head which has the quadrant photodiode (QPD) and LASER head at optically equal position. In this
measurement method, we conform the incident light beam to reflect the beam by controlling five stages and determine the
normal vectors and the coordinates of the surface from signal of goniometers, translational stage and QPD. We can obtain
three-dimensional figure from the normal vectors and the coordinates by a reconstruction algorithm. To evaluate
performance of this machine we measure a concave aspherical mirror ten times. From ten results we calculate measurement
repeatability, and we evaluate measurement uncertainty to compare the result with that measured by an interferometer. In
consequence, the repeatability of measurement was 2.90 nm (σ) and the difference between the two profiles was ±20 nm.
We conclude that the two profiles was correspondent considering systematic errors of each machine.
High accuracy optical elements are applied in various fields. For example, ultraprecise aspherical mirrors are necessary for
developing third-generation synchrotron radiation and XFEL (X-ray Free Electron LASER) sources. In order to make such high
accuracy optical elements, it is necessary to realize the measurement of aspherical mirrors with high accuracy. But there has been
no measurement method which simultaneously achieves these demands yet. So, we develop the nanoprofiler that can directly
measure the any surfaces figures with high accuracy. The nanoprofiler gets the normal vector and the coordinate of a
measurement point with using LASER and the QPD (Quadrant Photo Diode) as a detector. And, from the normal vectors and
their coordinates, the three-dimensional figure is calculated. In order to measure the figure, the nanoprofiler controls its five motion axis numerically to make the reflected light enter to the QPD’s center. The control is based on the sample's design formula.
We measured a concave spherical mirror with a radius of curvature of 400 mm by the deflection method which calculates
the figure error from QPD’s output, and compared the results with those using a Fizeau interferometer. The profile was
consistent within the range of system error. The deflection method can’t neglect the error caused from the QPD’s spatial
irregularity of sensitivity. In order to improve it, we have contrived the zero method which moves the QPD by the
piezoelectric motion stage and calculates the figure error from the displacement.