The optical performances of components exposed to high power laser beams during long periods are generally decreasing with time. We analyze surface modifications that occur after some hundreds of hours at a fluence of 0.5 kW/cm<sup>2</sup> to 10 kW/cm<sup>2</sup>. The results obtained with two dedicated test benches over thousands of hours are discussed. Exposition of tens of components has been achieved in the multi-component bench to simulate real optical systems and to acquire statistical confidence in the results. The single component bench allows continuous measurements of the surface temperature and of the scattered light. These measurements show how these beamprints can lead to the surface destruction. We investigate the effects of the beam fluence, the laser wavelength and the chamber pressure. Experiments at two wavelengths have shown that the deposition rate and the surface absorption increase with decreasing wavelength. The efficiency of oxygen in reducing the contamination speed has been investigated using absorption mappings. Laser cleaned surface absorptions are comparable to initial measurements, showing that damage did not occur. ESCA analysis of the beamprints showed that a few nanometers carbon layer has been implanted.
A system and method have been developed at CEA to retrieve phase information from multiple intensity measurements along a laser beam. The device has been patented. Commonly used devices for beam measurement provide phase and intensity information separately or with a rather poor resolution whereas the MIROMA method provides both at the same time, allowing direct use of the results in numerical models. Usual phase retrieval algorithms use two intensity measurements, typically the image plane and the focal plane (Gerschberg-Saxton algorithm) related by a Fourier transform, or the image plane and a lightly defocus plane (D.L. Misell).
The principal drawback of such iterative algorithms is their inability to provide unambiguous convergence in all situations. The algorithms can stagnate on bad solutions and the error between measured and calculated intensities remains unacceptable.
If three planes rather than two are used, the data redundancy created confers to the method good convergence capability and noise immunity. It provides an excellent agreement between intensity determined from the retrieved phase data set in the image plane and intensity measurements in any diffraction plane.
The method employed for MIROMA is inspired from GS algorithm, replacing Fourier transforms by a beam-propagating kernel with gradient search accelerating techniques and special care for phase branch cuts. A fast one dimensional algorithm provides an initial guess for the iterative algorithm.
Applications of the algorithm on synthetic data find out the best reconstruction planes that have to be chosen. Robustness and sensibility are evaluated. Results on collimated and distorted laser beams are presented.
The use of high power dye laser beams during long periods is difficult because the performances of the optical components are decreasing with time. To understand this phenomenon, a set-up for research of experimental conditions that can minimize this dramatic trend has been built. After being irradiated by a laser beam of some kW/cm<SUP>2</SUP>, the optical components are locally covered by droplets. The size of the droplets depends on experimental conditions. They can be eliminated by cleaning. The high absorption of the deposits (200 ppm to 1000 ppm) leads us to search experimental conditions to limit the layer deposition speed. XPS measurements prove that the contamination is merely made up of organic compounds and hydrogenated carbon. The experiments show that a clean room environment under a controlled airflow is not sufficient to assure a very slow deposition. Vacuum from 10<SUP>-2</SUP> to 10<SUP>-6</SUP> mbar is even worse than room conditions and only the presence of oxygen can limit or eliminate these deposits.
We show how the anisotropy of scattering enables to point out the origin of
microroughness in optical coatings with no ambiguity. Experimental results prove that
residual roughness due to material microstructure is negligible with coatings produced by
Ion Assisted or Ion Plating deposition. In these conditions, and provided that the substrate
is measured before coating, prediction of scattering from a stack of any design requires
the knowledge of only two cut-off frequencies that describe the action of materials at
interfaces. These parameters can be determined with single layers and then can be used
for more complex optical systems.
Since isotropy degree variation of scattering does not depend on the design of the
coating, it enables to show that the differences that can occur between calculation and
measurements are due to thickness or index errors on the layers. Therefore the substrate
roughness plays a primordial role and we compare, using numerical calculation, the
roughness spectra that can be obtained with mechanical or optical measurements.