OptiReOpt DLL consists of two major parts, namely on-line characterization and on-line reoptimization parts. On-line characterization on the basis of in situ spectral photometric measurements is used for the precise determination of deposited layer thicknesses. It is also possible to use spectral photometric data for determining termination moment of the current layer deposition.

Layer thicknesses determined by the on-line characterization DLL functions are required for the on-line reoptimization procedure. When thicknesses of already deposited layers are known and it appears that some layers have significant thickness errors, the reoptimization of remaining coating layers can be used to improve resulting coating performance.

Computational speed of OptiReOpt is outstandingly high. This is achieved due to the use of native OptiLayer DLLs, that are optimized for the utilization of all features of the modern hardware, including processors architecture and ability to perform parallel computations at multiprocessor systems or systems with processors supporting hyperthreading. Typically characterization step requires from fractions of seconds to several seconds while reoptimization step requires even smaller time. This time depends on the number of layers in the coating, and the number of measured spectral points.

The main goal of this section is to outline application potentialities of OptiReOpt as well as the most important ideas related to the on-line characterization and reoptimization. The detailed description of OptiReOpt DLL functions is provided in the next sections.

On-line characterization

The main goal of the on-line characterization is the determination of thicknesses of already deposited layers. On-line characterization functions of OptiReOpt DLL can be also used for monitoring deposition processes by generating a command for the termination of current layer deposition.

The on-line characterization of thicknesses of already deposited layers requires in situ measurements of spectral transmittance and/or spectral reflectance just after the end of deposition of each coating layer. If OptiReOpt DLL is used for the monitoring purposes, then in situ spectral scans of coating transmittance and/or reflectance should be done during layer depositions in small time intervals that can provide a sufficient accuracy of layer thickness monitoring. Basic requirements to the in situ measurement data and major ideas connected with the on-line analysis are discussed in Section "In situ measurement data".

There are two computational modes of the OptiReOpt characterization routines. These modes are referred to as SEQUENTIAL and TRIANGLE. In the case of SEQUENTIAL mode only thickness of the last deposited layer is determined by the characterization routine. In the case of TRIANGLE mode thicknesses of all already deposited layers are determined at each characterization step. Operational time of OptiReOpt DLL in SEQUENTIAL mode is faster than in TRIANGLE mode. At the same time, in general, TRIANGLE mode provides more reliable thicknesses of all already deposited layers. It is advisable to use SEQUENTIAL mode only in the case of very stable deposition processes and high-accuracy in situ measurement data.

On-line reoptimization

The main goal of the on-line reoptimization is the compensation for errors in thicknesses of already deposited layers. On-line reoptimization is done on the basis of the on-line characterization results. Thus high accuracy and reliability of these results is essential for the success of the on-line reoptimization.

The OptiReOpt reoptimization routine corrects theoretical thicknesses of those design layers that haven't been deposited yet. This correction is done with respect to the theoretical targets used for coating design so as to provide the best possible fit of theoretical targets by spectral characteristics of reoptimized design. If for some reason theoretical design targets are not available, then the correction is done with respect to the spectral transmittance/reflectance calculated for the initial theoretical design.

The OptiReOpt reoptimization routine is extremely efficient from the computational point of view and does not require interrupting the deposition process in order to wait for the corrected theoretical design.

In situ measurement data

High accuracy of in situ measurement data is a crucial issue for the reliability of on-line characterization and success of on-line reoptimization. Because all spectral photometric devices exhibit calibration drifts in time, all possible precautions must be done in order to avoid or at least minimize such drifts of measurement data. It is advisable to make an instrument calibration just before each new measurement scan.

Spectral region of supplied measurement data should be limited to the region with the most reliable transmittance/reflectance data. Short wavelength and long wavelength boundary regions of the nominal range of spectral photometric monitoring device should be specifically examined in this respect. Because of high computational efficiency of OptiReOpt DLL there are no special requirements on limiting the total number of data points supplied for the characterization analysis. At the same time 1-2 nm step of spectral measurements in visible range is usually quite sufficient for providing enough detailed input spectral information.

There are practical situations in which some measurement drifts are inevitable in spite of all possible precautions. In such situations OptiReOpt can apply internal algorithms for automatic compensation of systematic errors in measurement data sets. OptiReOpt distinguishes between two types of such errors: shift of a data set as a whole for some value (the corresponding compensation procedure is referred to as data shift correction) and errors connected with the inaccuracy of scale calibration (compensation by scale adjustment).

Material refractive indices in vacuum and air

OptiReOpt DLL is intended for the support of manufacture of arbitrary multilayer optical coatings. In the case of two-component optical coatings high and low index materials are typically referred to as H and L materials.

For some manufacturing processes layer materials refractive indices exhibit variations after the coating exposition to air environment. The main source of such variations is layer porosity. But refractive index variations due to other factors, like small chemical or structural changes, are also possible. One should always keep in mind that on-line monitoring and on-line characterization are done in vacuum while manufactured optical coating is intended for use in quite different environment. For this reason reliable information about layer material refractive indices in vacuum and air (or other environment where the manufactured coating is used) is required for the successful on-line characterization and reoptimization.

It is highly recommended to perform special pre-investigation of layer material properties in vacuum and ambient environment. Using literature data on material refractive indices is dangerous because actual layer material indices are dependent on the type and parameters of the deposition process in use.

It is necessary to investigate not only layer material refractive indices but also layers bulk inhomogeneity. Even dense layers produced by energetic deposition processes often feature bulk inhomogeneity (usually bulk inhomogeneity of a positive type). Ignoring layers inhomogeneity may lead to the accumulation of errors in the course of the on-line characterization and to the loss of accuracy of thickness determination with the growing number of deposited layers.

We recommend using OptiChar and OptiRE programs of the OptiLayer software family for the detailed investigation of layer material properties in vacuum and air.

Material correction factors

Not only layer refractive indices but also layer thicknesses may exhibit variations after the exposition to air. The ratio of layer thicknesses in air and vacuum is called correction factor. Correction factors are different for different layer materials and are dependent on the type of deposition process and many other factors. We recommend performing preliminary comparative investigation of single layer parameters in vacuum and air in order to check whether layer thicknesses are stable after the exposition to air or non-unit correction factors should be introduced for the some layer materials. OptiChar characterization software is especially well suited for such investigation.

Correction factors can be also used in the case of indirect spectral photometric monitoring when photometric measurements are done for a witness sample while manufactured samples are located in other positions inside the chamber as compared to this sample. In this case correction factors represent the ratios of layer thicknesses of manufactured samples to the corresponding layer thicknesses of witness sample.

If indirect optical monitoring is used and layer thicknesses also exhibit variations after the exposition to air, then correction factors should take into account both factors.

Coating designs and coating characteristics in vacuum and air

Because on-line characterization is done in vacuum while manufactured coating is used in different environment (referred to as air) one should always distinguish between coating design and coating characteristics in vacuum and air. In situ measurements and the on-line characterization procedure provide information about actual coating spectral characteristics and actual coating design in vacuum. On-line reoptimization is done for the coating design in air and with respect to air target spectral characteristics. It is possible that the type of these characteristics as well as their spectral range and incidence angle are different from those used for monitoring purposes.

The correspondence between coating designs and coating characteristics in vacuum and air is established basing on layer material indices in vacuum and air and material correction factors setting the ratios of air and vacuum layer thicknesses.

Design correction criteria

On-line reoptimization is based on the results of on-line characterization, which is quite complicated problem, especially in the case of coatings with many layers. From a mathematical point of view the on-line characterization problem belongs to the so-called inverse recognition problems and has all specific features typical for these problems. One of these features is possible instability of inverse problem solution which exhibits itself in growing solution inaccuracy. OptiLayer software mathematical routines incorporate all latest achievements of the theory of inverse problems in order to minimize solution instability. Nevertheless this potential threat always exists because of the specific nature of the on-line characterization problem.

Accuracy of the on-line characterization is severely affected by errors in measurement data. Thus all possible efforts should be applied to providing the maximum accuracy of input photometric data. But even with high accuracy data the solution instability may grow dramatically if design corrections are performed too many times. Thus the decision on whether to perform the design correction or not should be done very carefully basing on several specific criteria.

First of all it is not advisable to perform the design correction if registered thickness errors are small enough. In this instance the threshold error level should be specified so that all thickness errors not exceeding this level are ignored. It is impossible to indicate the universal threshold error level suitable for all designs. We recommend using the analysis options of OptiLayer in order to find an acceptable error level for your specific design.

It is possible that the OptiReOpt reoptimization procedure provides only an insignificant improvement of the merit function value. In such situations it is also not advisable to perform a design correction because this correction may raise the instability of subsequent on-line characterization and reoptimization procedures. Thus the merit function variation threshold level should be specified so that the current theoretical design is not corrected if the reoptimization procedure provides the merit function variation below this level.