OSA's Digital Library

Optics Express

Optics Express

  • Editor: C. Martijn de Sterke
  • Vol. 18, Iss. 16 — Aug. 2, 2010
  • pp: 16580–16586
« Show journal navigation

Multi-objective genetic algorithm applied to spectroscopic ellipsometry of organic-inorganic hybrid planar waveguides

Vasco R. Fernandes, Carlos M. S. Vicente, Naoya Wada, Paulo S. André, and Rute A. S. Ferreira  »View Author Affiliations


Optics Express, Vol. 18, Issue 16, pp. 16580-16586 (2010)
http://dx.doi.org/10.1364/OE.18.016580


View Full Text Article

Acrobat PDF (2547 KB)





Browse Journals / Lookup Meetings

Browse by Journal and Year


   


Lookup Conference Papers

Close Browse Journals / Lookup Meetings

Article Tools

Share
Citations

Abstract

The applicably of multi-objective optimization to ellipsometric data analysis is presented and a method to handle complex ellipsometric problems such as multi sample or multi angle analysis using multi-objective optimization is described. The performance of a multi-objective genetic algorithm (MOGA) is tested against a single objective common genetic algorithm (CGA). The procedure is applied to the characterization (refractive index and thickness) of planar waveguides intended for the production of optical components prepared sol-gel derived organic-inorganic hybrids, so-called di-ureasils, modified with zirconium tetrapropoxide, Zr(OPrn)4 deposited on silica on silicon substrates. The results show that for the same initial conditions, MOGA performs better than the CGA, showing a higher success rate in the task of finding the best final solution.

© 2010 OSA

1. Introduction

The usage of enhanced multi-objective algorithm is helpful in spectroscopic ellipsometry data analysis because it enables the combination of measurements from multiple samples, complex multilayer structures, different incidence angles and/or the usage of external reflectance (or transmittance) data. This can enhance the information content about a set of samples and thus be beneficial to reduce ambiguity, improving the results confidence. Usually, this type of analysis is accomplished using the conventional optimization techniques with one objective function, which suffers from constraints associated with local optimization.

In this context, spectroscopic ellipsometry multi-sample analysis will be used to calculate the refractive index and thickness values of sol-gel derived planar waveguides of methacrylic acid (McOH) modified zirconium tetrapropoxide, Zr(OPrn)4, classed as di-ureasil-zirconium oxo-cluster hybrid (di-ureasils-Zr-OMc). Di-ureasils-Zr-OMc hybrids have been used in the last years as integrated optics substrates for Y-power splitters, optical filters, and Fabry-Perot cavities [8

8. R. A. S. Ferreira, C. M. S. Vicente, V. Fernandes, A. G. Macedo, E. Pecoraro, R. Nogueira, P. S. André, P. V. S. Marques, and L. D. Carlos, “Organic-inorganic hybrids for the new generation of optical networks,” in Proc. of International Conference on Transparent Optical Networks (ICTON 2009)(IEEE, S. Miguel (Portugal), July. 2009), pp. Tu.B4.2–1.

10

10. D. C. Oliveira, A. G. Macedo, N. J. O. Silva, C. Molina, R. A. Sá, . Ferreira, P. S. Andre, K. Dahmouche, V. de Zea Bermudez, Y. Messaddeq, S. J. L. Ribeiro, and L. D. Carlos, “Photopatternable di-ureasil-zirconium oxocluster organic-inorganic hybrids as cost effective integrated optical substrates,” Chem. Mater. 20(11), 3696–3705 (2008). [CrossRef]

]. In this work we proposed an approach based on a multi-objective genetic algorithm (MOGA) simultaneously with a single objective local optimization algorithm applied to spectroscopic ellipsometry, with the subjacent advantage of improving the uniqueness and confidence in the final solutions and reducing the computation time require to solve the problem. As far as we know, this approach to solve the complex problems associated with spectroscopic ellipsometry is a novelty.

2. Experimental

2.1 Sample preparation

The reagents O, O’-Bis(2-aminopropyl) polypropylene glycol-block-polyethylene glycolblock-polypropylene glycol (Fluka), commercially known as Jeffamine-ED600®, average molecular weight 600 g.mol−1, 3-isocyanatepropyltriethoxysilane (ICPTES) (Aldrich 95%), ethyl alcohol absolute P.A. (Carlo Erba), tetrahydrofuran P.A. (stabilized - Riedel-de Haën), HCl (ACS Reagent 37% - Sigma-Aldrich), zirconium tetrapropoxide, Zr(OPrn)4 (Aldrich, 70 wt. % in 1-propanol), methacrylic acid (McOH, CH2 = C(CH3)COOH, Aldrich, 99%) and butanol (Sigma, 1-butanol, ≥ 99%) were used as received. The synthesis of the di-ureasil host, termed as d-U(600), contains 8.5 (OCH2CH2) polymer chains with both ends grafted to a siliceous network by means of urea linkages was recently published [11

11. E. Pecoraro, S. García-Revilla, R. A. S. Ferreira, R. Balda, L. D. Carlos, and J. Fernández, “Real time random laser properties of Rhodamine-doped di-ureasil hybrids,” Opt. Express 18(7), 7470–7478 (2010). [CrossRef] [PubMed]

]. The di-ureasil d-U(600) was doped with 40% mol of Zr(OPrn)4 with the Zr(OPrn)4:McOH molar ratio of 1:1. The hybrids were processed as thin films deposited on oxidized silicon wafers (with silica thickness of 1 μm). To enable the wafers to be spun, they were held by suction on a chuck, placed on the axis of the spin coater (MIKASA, 1H-DX2) accelerated at 500 rpm for 15 s and 1000 rpm for 30 s. The films were then dried at 50 °C for 12 h, for complete solvent removal. In order to evaluate the performance of both types of optimization by comparing the proximity to the final solution and the appraised reduction of the computation time, we present a comparison between single and multi-objective optimization genetic algorithm considering two films, termed as dUZ40-1 and dUZ40-2, whose ellipsometry data was simultaneously optimized.

2.2. Experimental techniques

The Scanning Electron Microscopy (SEM) images were obtained with a field emission type microscope (JEOL-JSM6335F) operating at 3.0 kV. To reveal the substrate and the film cross section, the samples were cleaved and cleaned with an air flux. The films were fixed on the SEM sample holder with a conductive adhesive. A good image contrast was achieved without the need of any coating process.

The Spectroscopic ellipsometry measurements were made using an AutoSE ellipsometer (HORIBA Scientific) with a total of 250 points in the wavelength interval 440-850 nm, an incidence angle of 70°, an acquisition time of 22 ms per point and an average of 10 measurements per point.

3. Ellipsometry fundamentals

A theoretical model of the sample structure must be employed to calculate Is and Ic, leaving the parameters of interest (thickness and refractive index) as variables that are obtained, when the calculated ellipsometric data match the experimental one.

The goodness of the fit is generally evaluated with the (unbiased) estimator, the mean squared error (MSE) given by
MSEIcIs=12NM1j=1N{(IcjmodIcjexp)2+(IsjmodIsjexp)2}
(1)
where N is the number of experimental data, M is the number of the model variable parameters, and the subscripts exp and mod stands for the measured and simulated data, respectively.

The optical constants for the film were calculated using the Lorentz model, which expresses the relative complex dielectric constant as a function of the frequency ω (eV), described by
ε=ε+(εsε)ω02(ω02ω2)+iΓω
(2)
where ε, εs, ω0 (eV), Γ (eV) are the high frequency relative dielectric constant, the static relative dielectric constant, the oscillator resonant frequency and the damping factor, respectively.

This model does not take into account all the resonant transitions, being limited in terms of application range. However, for a restrict wavelength range (as in the present case) this model is adequate to describe the relative complex dielectric function of the films.

4. Multi-objective optimization

Two of the commonest approaches with multi-objective optimization problems are to combine the individual objective functions into a single composite function or to determine a set of Pareto optimal solutions. For the first approach the weights assigned to each normalized objective function, converts the problem into a single objective one [13

13. A. Konak, D. W. Coit, and A. E. Smith, “Multi-objective optimization using genetic algorithms: A tutorial,” Reliab. Eng. Syst. Saf. 91(9), 992–1007 (2006). [CrossRef]

]. In the later approach a set of solutions that are non-dominated or superior with respect to each other, is determined [13

13. A. Konak, D. W. Coit, and A. E. Smith, “Multi-objective optimization using genetic algorithms: A tutorial,” Reliab. Eng. Syst. Saf. 91(9), 992–1007 (2006). [CrossRef]

]. For a given set of functions zi with i = 1,...,k, a point x is said to dominate another point y (x > y) only if zi(x)≤ zi(y) for i = 1,...,k and zj(x) < zj(y) for at least one objective function j [13

13. A. Konak, D. W. Coit, and A. E. Smith, “Multi-objective optimization using genetic algorithms: A tutorial,” Reliab. Eng. Syst. Saf. 91(9), 992–1007 (2006). [CrossRef]

]. The solutions which are non-dominated by any other belong to the so called Pareto set. The corresponding objective function values in the objective space are termed as Pareto front.

To describe the application of multi-objective optimization in ellipsometric data analysis, we considered the combination of measurements for two films. Using conventional optimization for the data analysis, the MSE value for each film is combined in the one objective function simply by summation or weighted summation. With multiple objective optimization for each film k an objective function (MSEIcIs,k)is built. The same principle is valid for films that are measured at different positions or when measurements are performed at different angles of incidence. Another possibility is the application in cases where external data is available, assigning one objective function to ellipsometric data and the other, for example, to reflectance data.

For complex problems MOGA extends the advantages of GA because it takes into consideration the individual fitness of each function, finding multiple non-dominated solutions. Therefore, the quality and uniqueness of MOGA solutions are, in principle, superior to those provided by single objective optimization. It is important to note that evolutionary algorithms that use the non domination sorting (NDS) are not suited for problems with more than four objective functions. This is due to the fact that a major number of initially generated solutions will be actually non-dominated, which makes difficult to guide the search. These problems are tackled with many-objective optimization algorithms [13

13. A. Konak, D. W. Coit, and A. E. Smith, “Multi-objective optimization using genetic algorithms: A tutorial,” Reliab. Eng. Syst. Saf. 91(9), 992–1007 (2006). [CrossRef]

].

Among the various MOGA’s, one of the most popular, due to the good overall efficiency and easiness to implement, is a NSGA II (Non Dominated Sort Genetic Algorithm) based algorithm [14

14. K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitist multiobjective genetic algorithm: NSGA-II,” IEEE Trans. Evol. Comput. 6(2), 182–197 (2002). [CrossRef]

]. Originally this algorithm is bi-objective but it can readily be extended for more functions.

It is useful to implement a method for constraint handling, whenever it is possible. We have implemented a procedure in which every new individual, generated by simulated binary crossover (SBX) [15

15. K. Deb and M. Goyal, “A Combined Genetic Adaptive Search (GeneAS) for Engineering Design,” Comput. Sci. Inform.. 26, 30–45 (1996).

] or polynomial mutation [15

15. K. Deb and M. Goyal, “A Combined Genetic Adaptive Search (GeneAS) for Engineering Design,” Comput. Sci. Inform.. 26, 30–45 (1996).

], is checked to verify if all chromosomes are in the feasible region. An infeasible chromosome is replaced with one resultant of an arithmetic crossover of the two parents. Common methods for GA’s constraints handling are based on penalty functions, which takes in account the feasibility of the solutions in the fitness values of the individuals. Our strategy simply relies on the working principle of genetic algorithms, if the repaired solution has poor fitness it will be naturally rejected.

Seeking the comparison between single and multi-objective optimization, we implemented a single objective genetic algorithm, termed as common genetic algorithm (CGA), which uses the same crossover and mutation operator than the MOGA, and the traditional tournament selection. Nor NDS nor crowding distance comparison is used. Therefore both algorithms are identical and the same building blocks are used, enabling a fair comparison of the results. The control parameter of SBX is 20 and polynomial mutation is 20 both for MOGA and CGA and the number of individuals generated by crossover, for both algorithms, is equal to the population size. The implemented single objective LO method was the Interior-Point algorithm with Hessian calculated by Broyden–Fletcher–Goldfarb–Shanno quasi-Newton approximation. The initial point for the LO method is the best solution found by the GA. The minimization algorithms were implemented in MATLAB®.

5. Example of the multi-objective optimization testing

Figure 1(a)
Fig. 1 (a) SEM photo of the cross-section and (b) structure model of the dUZ40 planar waveguides.
shows an SEM image of a dUZ40 film, revealing that the di-ureasil layer prepared by spin-coating has a uniform crack-pinhole-free dense microstructure with an average thickness of 0.835 ± 0.028 μm. Based on this layered structure the model structure scheme for the dUZ40 planar waveguides in Fig. 1(b) will be considered in the ellipsometric data analysis. The surface roughness was previously studied by atomic force microscopy, revealing low root-mean-square roughness values (<1 nm) [10

10. D. C. Oliveira, A. G. Macedo, N. J. O. Silva, C. Molina, R. A. Sá, . Ferreira, P. S. Andre, K. Dahmouche, V. de Zea Bermudez, Y. Messaddeq, S. J. L. Ribeiro, and L. D. Carlos, “Photopatternable di-ureasil-zirconium oxocluster organic-inorganic hybrids as cost effective integrated optical substrates,” Chem. Mater. 20(11), 3696–3705 (2008). [CrossRef]

]. Therefore, no surface roughness layer was considered in the structure model.

The final solution determined with MOGA/LO and CGA/LO yielding to 1.0220 ± 0.0009 µm and 1.0219 ± 0.0009 µm for WSiO2 and 0.7948 ± 0.0006 µm and 0.7924 ± 0.0012 µm for WUZ40, respectively, for the two planar waveguides (Table 1

Table 1. Individual (thickness for the SiO2, WSiO2 and dUZ40, WdUZ40 layers) and shared (high frequency relative dielectric constant, ε, static relative dielectric constant, εs, oscillator resonant frequency, ω0, and damping factor, Γ) parameter values determined using MOGA/LO and CGA/LO algorithms for dUZ40 films.

table-icon
View This Table
). These values are in good agreement with the thickness values commercially available for the buffer layer of SiO2 and with the organic-inorganic layer thickness estimated by SEM images (Fig. 1(a)). The attained refractive index values for the planar waveguides were 1.5111 ± 0.0010 and 1.5055 ± 0.0007 for a wavelength of 535 nm and 632.8 nm, respectively. These results are in good agreement with those previously reported for analogous di-ureasil planar waveguides [9

9. C. Molina, R. A. Sá, L. D. Ferreira, R. R. Carlos, S. J. L. Gonçalves, Y. Ribeiro, P. J. Messaddeq, O. Moreira, A. P. Soppera, P. S. V. Leite, Marques, and V. de Zea Bermudez, “Planar and UV written channel optical waveguides prepared with siloxane-poly(oxyethylene)-zirconia organic-inorganic hybrids. Structure and optical properties,” J. Mater. Chem. 15(35-36), 3937–3945 (2005). [CrossRef]

,17

17. C. M. S. Vicente, E. Pecoraro, R. A. S. Ferreira, P. S. André, R. Nogueira, Y. Messaddeq, S. J. L. Ribeiro, and L. D. Carlos, “Waveguides and gratings fabrication in zirconium-based organic/inorganic hybrids,” J. Sol-Gel Sci. Technol. 48(1-2), 80–85 (2008). [CrossRef]

]. The final ε value is smaller than 1, which results from the fact that the used model does not take into account all the resonant transitions. Therefore, this value cannot be taken as the permittivity value at high frequencies but only a fitting parameter for the analyzed wavelength interval.

The relevant question to optimize the minimization procedure is to decide the right moment to switch from GA to LO. To improve the GA search, it is necessary to properly dimension the population size, the initial range and the number of generations.

To quantify the MOGA and CGA algorithms performance Fig. 3(a)
Fig. 3 (a) Euclidean distance to the final solution (50 runs and population of 400 elements) as function of the generation’s number for CGA and MOGA algorithms (population size of 400 individuals). The lines are visual guides. (b) Convergence rate for 50 iterations as function of the population size for the two considered algorithms.
shows the Euclidean distance from the solution obtained after the application of GA to the one achieved after the LO optimization, as function of the generation number. In this analyze it was considered a population of 400 elements, being the final value the average of 50 runs, with the standard deviations represented as errors bars. Since GA's are stochastic optimizers, there is a natural randomness associated with the presented results, thus they should be taken as indicative of a trend. The Euclidean distance to the final solution is clearly dependent on the number of generations, being also clear that the proposed MOGA algorithm produces results with a distance to the final solution smaller that the CGA and with low dispersion around the average value of the 50 runs. It is also noticeable that the MOGA Euclidean distance displays a decay behavior with the number of generations and, consequently, presents an asymptotic behavior as the number of generations unlimitedly increases. Thus, we can conclude that it is not worthwhile the unlimited increase in the number of generations to improving the method’s accuracy.

The difference in performance of both algorithms could be explained regarding the non dominated solutions. The MOGA solutions that are simultaneously better for both samples are preferred to evolve, while the CGA ones with higher probability to evolve correspond to the better fitness for the sum of all solutions (independently of each individual fitness).

6. Conclusions

We have introduced for the first time the use of hybrid multi-objective optimization in spectroscopic ellipsometry data analysis. The procedure was validated, using planar waveguides of sol-gel derived organic-inorganic hybrids on oxidized silicon wafers. For the implementation of the hybrid optimization a single objective common genetic algorithm (CGA) was compared to a multi-objective genetic algorithm (MOGA), showing a higher success rate in the task of finding the best final solution for MOGA. Furthermore, we demonstrated that the hybrid MOGA is potentially faster than the CGA and also that its efficiency can be improved when the right moment to switch methods is properly chosen.

Acknowledgments

Funding was provided by Fundação para a Ciência e a Tecnologia, FEDER (PTDC/CTM/72093/2006, SFRH/BD/41943/2007) and COST Action MP0702. The authors also thank E. Pecoraro from Instituto de Telecomunicações, University of Aveiro, for help in the hybrids’ synthesis.

References and links

1.

P. Drude, “Über die Gesetze der Reflexion und Brechung des Lichtes an der Grenze absorbierender Kristalle,” Annalen der Physik 32, 584–625 (1887).

2.

A. Rothen, “The Ellipsometer, an Apparatus to Measure Thicknesses of Thin Surface Films,” Rev. Sci. Instrum. 16(2), 26–30 (1945). [CrossRef]

3.

M. Land, J. J. Sidorowich, and R. K. Belew, “Using Genetic Algorithms with Local Search for Thin Film Metrology,” in Proceedings of the Seventh International Conference on Genetic Algorithms, T. Bäck, ed., (Morgan Kaufmann Publishers, Inc, 1997), pp. 537–544.

4.

O. Polgár, M. Fried, T. Lohner, and I. Bársony, “Comparison of algorithms used for evaluation of ellipsometric measurements - Random search, genetic algorithms, simulated annealing and hill climbing graph-searches,” Surf. Sci. 457(1-2), 157–177 (2000). [CrossRef]

5.

A. Kudla, “Application of the genetic algorithms in spectroscopic ellipsometry,” Thin Solid Films 455–456, 804–808 (2004). [CrossRef]

6.

O. Polgár, P. Petrik, T. Lohner, and M. Fried, “Evaluation strategies for multi-layer, multi-material ellipsometric measurements,” Appl. Surf. Sci. 253(1), 57–64 (2006). [CrossRef]

7.

B. Neto, A. L. J. Teixeira, N. Wada, and P. S. André, “Efficient use of hybrid Genetic Algorithms in the gain optimization of distributed Raman amplifiers,” Opt. Express 15(26), 17520–17528 (2007). [CrossRef] [PubMed]

8.

R. A. S. Ferreira, C. M. S. Vicente, V. Fernandes, A. G. Macedo, E. Pecoraro, R. Nogueira, P. S. André, P. V. S. Marques, and L. D. Carlos, “Organic-inorganic hybrids for the new generation of optical networks,” in Proc. of International Conference on Transparent Optical Networks (ICTON 2009)(IEEE, S. Miguel (Portugal), July. 2009), pp. Tu.B4.2–1.

9.

C. Molina, R. A. Sá, L. D. Ferreira, R. R. Carlos, S. J. L. Gonçalves, Y. Ribeiro, P. J. Messaddeq, O. Moreira, A. P. Soppera, P. S. V. Leite, Marques, and V. de Zea Bermudez, “Planar and UV written channel optical waveguides prepared with siloxane-poly(oxyethylene)-zirconia organic-inorganic hybrids. Structure and optical properties,” J. Mater. Chem. 15(35-36), 3937–3945 (2005). [CrossRef]

10.

D. C. Oliveira, A. G. Macedo, N. J. O. Silva, C. Molina, R. A. Sá, . Ferreira, P. S. Andre, K. Dahmouche, V. de Zea Bermudez, Y. Messaddeq, S. J. L. Ribeiro, and L. D. Carlos, “Photopatternable di-ureasil-zirconium oxocluster organic-inorganic hybrids as cost effective integrated optical substrates,” Chem. Mater. 20(11), 3696–3705 (2008). [CrossRef]

11.

E. Pecoraro, S. García-Revilla, R. A. S. Ferreira, R. Balda, L. D. Carlos, and J. Fernández, “Real time random laser properties of Rhodamine-doped di-ureasil hybrids,” Opt. Express 18(7), 7470–7478 (2010). [CrossRef] [PubMed]

12.

O. Acher, E. Bigan, and B. Drévillon, “Improvements of phase-modulated ellipsometry,” Rev. Sci. Instrum. 60(1), 65–77 (1989). [CrossRef]

13.

A. Konak, D. W. Coit, and A. E. Smith, “Multi-objective optimization using genetic algorithms: A tutorial,” Reliab. Eng. Syst. Saf. 91(9), 992–1007 (2006). [CrossRef]

14.

K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitist multiobjective genetic algorithm: NSGA-II,” IEEE Trans. Evol. Comput. 6(2), 182–197 (2002). [CrossRef]

15.

K. Deb and M. Goyal, “A Combined Genetic Adaptive Search (GeneAS) for Engineering Design,” Comput. Sci. Inform.. 26, 30–45 (1996).

16.

E. D. Palik, Handbook of Optical Constants of Solids (Academic Press, New York, 1998).

17.

C. M. S. Vicente, E. Pecoraro, R. A. S. Ferreira, P. S. André, R. Nogueira, Y. Messaddeq, S. J. L. Ribeiro, and L. D. Carlos, “Waveguides and gratings fabrication in zirconium-based organic/inorganic hybrids,” J. Sol-Gel Sci. Technol. 48(1-2), 80–85 (2008). [CrossRef]

OCIS Codes
(120.2130) Instrumentation, measurement, and metrology : Ellipsometry and polarimetry
(310.6860) Thin films : Thin films, optical properties

ToC Category:
Instrumentation, Measurement, and Metrology

History
Original Manuscript: June 9, 2010
Revised Manuscript: July 3, 2010
Manuscript Accepted: July 5, 2010
Published: July 22, 2010

Citation
Vasco R. Fernandes, Carlos M. S. Vicente, Naoya Wada, Paulo S. André, and Rute A. S. Ferreira, "Multi-objective genetic algorithm applied to spectroscopic ellipsometry of organic-inorganic hybrid planar waveguides," Opt. Express 18, 16580-16586 (2010)
http://www.opticsinfobase.org/oe/abstract.cfm?URI=oe-18-16-16580


Sort:  Author  |  Year  |  Journal  |  Reset  

References

  1. P. Drude, “Über die Gesetze der Reflexion und Brechung des Lichtes an der Grenze absorbierender Kristalle,” Annalen der Physik 32, 584–625 (1887).
  2. A. Rothen, “The Ellipsometer, an Apparatus to Measure Thicknesses of Thin Surface Films,” Rev. Sci. Instrum. 16(2), 26–30 (1945). [CrossRef]
  3. M. Land, J. J. Sidorowich, and R. K. Belew, “Using Genetic Algorithms with Local Search for Thin Film Metrology,” in Proceedings of the Seventh International Conference on Genetic Algorithms, T. Bäck, ed., (Morgan Kaufmann Publishers, Inc, 1997), pp. 537–544.
  4. O. Polgár, M. Fried, T. Lohner, and I. Bársony, “Comparison of algorithms used for evaluation of ellipsometric measurements - Random search, genetic algorithms, simulated annealing and hill climbing graph-searches,” Surf. Sci. 457(1-2), 157–177 (2000). [CrossRef]
  5. A. Kudla, “Application of the genetic algorithms in spectroscopic ellipsometry,” Thin Solid Films 455–456, 804–808 (2004). [CrossRef]
  6. O. Polgár, P. Petrik, T. Lohner, and M. Fried, “Evaluation strategies for multi-layer, multi-material ellipsometric measurements,” Appl. Surf. Sci. 253(1), 57–64 (2006). [CrossRef]
  7. B. Neto, A. L. J. Teixeira, N. Wada, and P. S. André, “Efficient use of hybrid Genetic Algorithms in the gain optimization of distributed Raman amplifiers,” Opt. Express 15(26), 17520–17528 (2007). [CrossRef] [PubMed]
  8. R. A. S. Ferreira, C. M. S. Vicente, V. Fernandes, A. G. Macedo, E. Pecoraro, R. Nogueira, P. S. André, P. V. S. Marques, and L. D. Carlos, “Organic-inorganic hybrids for the new generation of optical networks,” in Proc. of International Conference on Transparent Optical Networks (ICTON 2009) (IEEE, S. Miguel (Portugal), July. 2009), pp. Tu.B4.2–1.
  9. C. Molina, R. A. Sá, L. D. Ferreira, R. R. Carlos, S. J. L. Gonçalves, Y. Ribeiro, P. J. Messaddeq, O. Moreira, A. P. Soppera, P. S. V. Leite, Marques, and V. de Zea Bermudez, “Planar and UV written channel optical waveguides prepared with siloxane-poly(oxyethylene)-zirconia organic-inorganic hybrids. Structure and optical properties,” J. Mater. Chem. 15(35-36), 3937–3945 (2005). [CrossRef]
  10. D. C. Oliveira, A. G. Macedo, N. J. O. Silva, C. Molina, R. A. Sá, . Ferreira, P. S. Andre, K. Dahmouche, V. de Zea Bermudez, Y. Messaddeq, S. J. L. Ribeiro, and L. D. Carlos, “Photopatternable di-ureasil-zirconium oxocluster organic-inorganic hybrids as cost effective integrated optical substrates,” Chem. Mater. 20(11), 3696–3705 (2008). [CrossRef]
  11. E. Pecoraro, S. García-Revilla, R. A. S. Ferreira, R. Balda, L. D. Carlos, and J. Fernández, “Real time random laser properties of Rhodamine-doped di-ureasil hybrids,” Opt. Express 18(7), 7470–7478 (2010). [CrossRef] [PubMed]
  12. O. Acher, E. Bigan, and B. Drévillon, “Improvements of phase-modulated ellipsometry,” Rev. Sci. Instrum. 60(1), 65–77 (1989). [CrossRef]
  13. A. Konak, D. W. Coit, and A. E. Smith, “Multi-objective optimization using genetic algorithms: A tutorial,” Reliab. Eng. Syst. Saf. 91(9), 992–1007 (2006). [CrossRef]
  14. K. Deb, A. Pratap, S. Agarwal, and T. Meyarivan, “A fast and elitist multiobjective genetic algorithm: NSGA-II,” IEEE Trans. Evol. Comput. 6(2), 182–197 (2002). [CrossRef]
  15. K. Deb and M. Goyal, “A Combined Genetic Adaptive Search (GeneAS) for Engineering Design,” Comput. Sci. Inform.. 26, 30–45 (1996).
  16. E. D. Palik, Handbook of Optical Constants of Solids (Academic Press, New York, 1998).
  17. C. M. S. Vicente, E. Pecoraro, R. A. S. Ferreira, P. S. André, R. Nogueira, Y. Messaddeq, S. J. L. Ribeiro, and L. D. Carlos, “Waveguides and gratings fabrication in zirconium-based organic/inorganic hybrids,” J. Sol-Gel Sci. Technol. 48(1-2), 80–85 (2008). [CrossRef]

Cited By

Alert me when this paper is cited

OSA is able to provide readers links to articles that cite this paper by participating in CrossRef's Cited-By Linking service. CrossRef includes content from more than 3000 publishers and societies. In addition to listing OSA journal articles that cite this paper, citing articles from other participating publishers will also be listed.

Figures

Fig. 1 Fig. 2 Fig. 3
 

« Previous Article  |  Next Article »

OSA is a member of CrossRef.

CrossCheck Deposited