Second-order optical nonlinearities of azo chromophores covalently attached to a sol-gel matrix

.

.

!

ELSEVIER

20 October1995

CHEMICAL PHYSICS LETTERS Chemical PhysicsLetters245 (1995)36-40

Second-order optical nonlinearities of azo chromophores covalently attached to a sol-gel matrix Didier Riehl ", Frederic Chaput b, Yves L~vy a, Jean-Pierre Boilot u, Francois Kajzar c, Pierre-Alain Chollet c a Institut d'Optique Th~orique et Appliqu~e, URA CNRS 14, B~timent 503, B.P. 147, 91403 Orsay Cedex, France b Groupe de Clu'mie du Solide, Laboratoire de Physique de la Mati~re Condens~e, URA CNRS 1254 D, Ecole Polytechnique, 91128 Palaiseau, France c CEA (LETI - Technologies Avanc~es), DEIN/SPE - Centre d'Etudes de Saclay, 91191 Gifsur Yvette Cedex, France

Received31 May 1995;in final form21 July 1995

Abstract

Optically nonlinear disperse red 1 (DR1) was covalently bound into a silica gel network by the coupling of DR1 and 3-isocyanatopropyltriethoxysilane. Organic-inorganic hybrid materials present a large resonant second-order susceptibility (d33 55 pm/V for a fundamental wavelength of 1.06 I~m) with an excellent room-temperature stability. =

1. I n t r o d u c t i o n

During the last decade the second-order nonlinear optical (NLO) properties of a large variety of poled polymeric materials were investigated, especially for such potential applications as fast waveguide electro-optic modulation and frequency-doubling [1,2]. Concerning device applications, the major problem is the poor thermal and orientational NLO stability of these materials. The noncentrosymmetric alignment of the NLO chromophores resulting from a classical electric field poling process progressively relaxes to a random configuration. Several synthesis routes were investigated to stabilize the poling-induced nonlinearity in the NLO polymers [2]. The randomization of the aligned molecules was partially prevented by using crosslinkable polymer systems and high glass transition temperature polymer systems such as polyimides [3]. The basic idea is to slow down the molecular orientational motion which leads to the randomization.

In the last few years, current work has also focused on NLO sol-gel systems [4-12]. Sol-gel technology provides an attractive route to the preparation at low temperature of a rigid amorphous three-dimensional inorganic network exhibiting good optical properties. The silica sol-gel polymerization consisting of the formation of siloxane bridges I -Si-O-SiI

I

was generally initiated by adding water to an alcoholic solution of silicon alkoxide (such as tetraethoxysilane or TEOS). The preparation of numerous molecule doped gels and their optical applications were proposed, such as for information recording materials and dye-laser materials [13-15]. In fact, like other guest-host systems, doped NLO gels suffer from phase separation and thermal relaxation problems. To overcome these problems, active NLO molecules were covalently bound into the gel

0009-2614/95/$09.50 © 1995 ElsevierScienceB.V. All fights reserved 0009-2614(95)00990-6

SSDI

I

R. Riehl et al. / Chemical Physics Letters 245 (1995) 36-40

network by using functionalized alkoxide precursors F-R'-Si(OEt) 3 where F was a chemical function such as an amino or isocyanate group and R' an alkyl spacer. As expected, the thermal relaxation decreased and long-term stability of the nonlinearity at 100°C was observed for a sol-gel material in which chromophores were covalently locked in both ends [10,11]. The values of the nonlinearity reported for highly active chromophores in sol-gel materials were surprisingly low although covalent grafting was to lead to high chromophore loading densities. This suggests that the problem of phase separation was not entirely solved in sol-gel materials, because the covalent grafting on the gel precursor was either incomplete or partially destroyed during the acidic hydrolysis step of the sol-gel process. However, at a fundamental wavelength of 1064 nm, a second-order nonlinear coefficient, d33, of 157 p m / V was reported for azo active chromophores (disperse red one) dispersed in a silica film [9]. The proximity of the resonance, responsible for the strong enhancement of the NLO coefficient, makes the system inefficient for practical integrated frequency doubling applications due to the strong absorbance (A = 2) at the second harmonic wavelength of )t = 532 nm. Recently, we reported the sol-gel preparation of a new organic-inorganic composite material in which the optically nonlinear disperse red 1 (DR1) was covalently bound into the silica network by the

O~~--~

4'

~

~U~

O~N4CI'I~)3"Sl(Olgt)a

37

coupling of DR1 and 3-isocyanatopropyltriethoxysilane (ICPTEOS) [16,17] (Scheme 1). In this Letter, we present the optical properties of sol-gel films which exhibit a large and stable second-order nonlinear coefficient, d33, of 55 p m / V .

2. Sample preparation The coating solution was obtained from copolymerization of the dye attached alkoxysilane (DR1UPTEOS) with tetraethoxysilane (TEOS). The hydrolysis was performed under acidic conditions with acetone as a common solvent. The alkoxysilane : water (pH = 1) : acetone initial molar ratios were 1:10:6. The atomic molar silicon concentration in alkoxysilane was defined as: Sialkoxysilane = SiTEOS + SiDR1UPTEOs with SiDR1UPTEOs = 0.1 SiTEOS (sample noted DR1 : Si = 1 : 11) or 0.2 Sia~os (sample noted DR1 : Si = 1 : 6). After hydrolysis for one hour at room temperature, pyridine (Py) was added to the solution (the Sialkoxysilane/Py molar ratio was 0.3). Then the asprepared sol was passed through a 0.45 ixm filter before deposition. Thin films of azo-oxide gels (0.5 to 1 ~m of thickness) were spin-coated onto 1 mm thick microscope slides. The angular speed of the spinner varied from 1000 to 4000 rpm. The sample plates were dried at room temperature for a few minutes until the surface of the film became tack-free. The spin coated azo oxide xerogel was smooth and uniform with a bright red color. The chromophore load density was approximately 43 wt% for the sample DR1 :Si = 1:6.

~ 60-70"12

3. Linear optical properties 3.1. Absorption spectra SolventJ Si(Ogt). ÷ IlaO

o

~

~

~/

"c~u,

I

I

Scheme 1. The different steps leading to the organic/inorganic composite material.

Visible absorption spectra were recorded using an E G & G optical multichannel analyzer and a Xe arc lamp as light source. In order to monitor the temperature-induced changes, the samples were put on a temperature-regulated heating stage. Fig. 1 shows the absorbance spectra of a 0.55 Ixm thick DR1 (DR1 :Si = 1:11) (a) before heating, (b) after five

38

R. Riehl et al. / Chemical Physics Letters 245 (1995) 36-40

1•47

v~

,.2-1

Table 1 Refractive indices measured ( + 10 -3) from the prism coupler method at 633, 830, 1314 nm wavelength and from interference polychromatic spectra at 793 and 920 nm

1

\ I

l

400

450

I

I

500 550 ~, ( nrn )

n° ne ns I

I

600

650

Fig. 1. Absorbance spectra of a 0.55 p,m thick DRI (DRI :Si = 1 : I 1) sample. (a) Without heating ( ); (b) after five minutes at 80°C (" • "); (c) after 85 min at 160"C (- - -).

minutes at 80°C and (c) after 85 min at 160°C. The first spectrum is strongly asymmetric and noticeably differs from the usual curves obtained with DR1 in various environments (solution, organic polymer [18,19], gel [20]). In particular, an additional peak at 425 nm appears, probably corresponding to a partial aggregation o f azobenzene units [21]. Heating the sample tends to restore the usual absorption band shape and to remove the 425 nm absorption peak. This change is irreversible, suggesting a modification of the chemical environment o f the dye by removing residual solvent trapped in the pores of the matrix, and possibly an extent of the reticulation by siloxane bridges.

3.2. Thickness and refractive index determination

A=633 nm

A=793 nm

A=830 nm

A=920 nm

A=1314 nm

1.685 1.659 ! .507

1.616 -

1.613 1.596 1.503

1.605 -

1.584 1.569 1.497

The thickness h was equal to 1015 nm (+4 nm). n° and ne are the TE and TM indices and n s is the substrate index.

The thickness was found to be equal to 1015 nm. We note that the ordinary refractive index n ° is higher than the extraordinary one n e (negative birefringence) leading to the conclusion that the molecules are mostly lying in the film plane• In addition, the refractive indices were determined in the absorption band by using a Kramers-Kronig analysis of the visible absorption spectrum. Fig. 2 displays the thus calculated TE index dispersion for a DR1 (DR1 : Si = 1 : 6) sample. We also performed thickness and refractive index measurements at 0.633 Ixm to analyze the changes induced by heating and poling of the sample. After heating at a temperature of 150°C for 13 h, the thickness was 911 nm and the n ° and n e indices

1.74q 1.72-- I

1.7o-] ~ 1.68

The thickness of the films, and refractive indices outside the absorption band, were determined by measurements of the effective indices of the guided modes in the films using the prism coupling method [22], performed at three wavelengths (0.633, 0.83 and 1.32 ~m), using the commercial Metricon PC 2000 prism coupler. Another method using polychromatic light interferometric measurements allowed the determination o f the refractive index of the thin film over a large range o f wavelengths. That technique [23], used in reflection or transmission, provides the refractive indices as a function of the wavelength and the thickness o f the sample. The results obtained with these two methods are summarized in Table 1.

~ 1.66 1.64"~ 1 . 6 2 1.601•58'

0.50

'

'

'

I

. . . .

0.75

!

. . . .

I

. . . .

1.00 1.25 wavelength (Ixm)

I

1.50

Fig. 2. Ordinary refractive index dispersion for a DRI (DRI: Si = 1:6) sample. ( × ) Prism coupling measurements, ( O ) white light interference fringes measurements (in transmission), ( , ) resulting from the Kramers-Kronig analysis of the absorption spectrum; ( - - - ) fitting on a one-resonance Sellmeier equation.

R. Riehl et al. / Chemical Physics Letters 245 (1995) 36-40

were respectively 1.690 and 1.675. This shows a significant shrinkage of the thin film after heating corresponding to the formation of siloxane bridges. However, the average index is only slightly increased due to the removal of the solvent trapped in pores. Finally, a few days after poling the sample at 160°C, the thickness was 889 nm and the indices were 1.687 and 1.714. The negative birefringence observed before and after heating turns into a positive one after poling, showing the expected orientation of the chromophores perpendicular to the layer plane.

3.3. Nonlinear optical properties Corona poling was used to induce the noncentrosymmetry needed for macroscopic secondorder nonlinearities. The films, spin-cast on glass substrates, were put on a temperature controlled copper plate and a positive high voltage was applied to a needle placed in front of the sample. The poling procedure was optimized using in situ SHG experiments at 1.064 p~m fundamental wavelength. The source was a Q-switched Nd : YAG laser, with a pulse duration of 100 ns and a repetition rate of 1 It-It. The best results were obtained with a corona discharge voltage of + 5.6 kV, with the needle placed 12.5 mm from the sample. It was not possible to apply stronger high voltages for a long time (several hours) because of a progressive abrasion of the film. A 0.5 txm thick film was completely removed after 4 h at + 10 kV applied voltage. At temperatures less than 120°C, only weak SH signals were obtained. Fig. 3 shows the evolution of SH intensity as function of time for three different conditions: (a) the film poled at 120°C, (b) at 160°C, (c) the sample heated at 160°C for 17 h before starting the poling process. In the first two conditions a slow increase in the harmonic intensity was observed, followed by a fast rise, reaching a plateau after a few minutes. The same experiment was performed on the sample heated for 17 h and the most important part of the signal was obtained in the first two minutes. A possible explanation consists of the presence of ions in the residual solution trapped in the matrix, contributing to screen the electric field. As supposed above, the thermal curing tends to remove this residual solution

0

39

I I I 100 150 200 time (minutes)

50

I 250

I 300

Fig. 3. Evolution of SHG signals on 0.55 Ixm thick DRI (DRI : Si = 1 : I 1) sample. (a) Poling at 120°(7; (b) poling at 160°C; (c) poling at 160°C (the sample was previously stored at 160°C during 17 h).

and reduce the ionic mobility, and consequently increase the poling field. After obtaining the maximum of the signal, the temperature was progressively decreased to room temperature with the field kept on to prevent thermal diffusion of molecular orientations. Surprisingly, as shown in Fig. 4, the SH intensity began to decrease when the temperature fell below 120°C, and at ambient temperature the quadratic susceptibility d33 decreased to half its initial value. We have no definitive explanation about that process which may be due to a molecular disorientation, resulting from

I

1

.

[

0

~

T

-

1200

i\

"~ 0 . 6 -

T=160°C -

,.

poll? off

heatin

0.2T = Tamb 0.0-

I

0

20

I

I

40 60 time ( minutes )

I

80

Fig. 4. Evolution of SHG signal on a 0.55 p~m thick DRI (DRI:Si = 1:11) sample, during cooling from 160°C to room temperature.

40

R. Riehl et a L / Chemical Physics Letters 245 (1995) 36-40

steric constraints related to the shrinkage of the matrix. Concerning the stability of the poling-induced X ~2~,no significant decrease was observed at ambient temperature within a week. Total relaxation of X ~2) needs heating at 160°C for one hour. It is important to notice that if the sample is poled again at T >/ 120°C, the maximum is reached again in only a few minutes. To evaluate the quadratic susceptibility at 1064 nm the SH intensities were recorded versus incidence angle after the poling sequence, and the curves obtained were fitted to the theoretical SHG model [24,25], using an a-quartz crystal (d u -- 0.46 p m / V ) as reference. Our best result was obtained with a 0.9 i~m thick DR1 film (DR1 : Si = 1 : 6). The refractive indices, determined by using the methods described above, are n ( w ) = 1.59 and n ( 2 t o ) = 1.69. Fitting the experimental data with the theoretical model in the p - p polarization configuration and assuming that d33 = 3d13, gives the resonance-enhanced value of d33 = 55 p m / V , 3 days after poling. The analysis of the dispersion of the quadratic susceptibility, based on a two-level model, leads to a value for d33 of 34 p m / V at a fundamental wavelength of 1.3 I~m. In conclusion, we have shown that a new organic-inorganic composite film in which the chromophore DR1 is covalently bound into the silica network exhibits a large resonant second-order susceptibility (d33 = 55 p m / V for a fundamental wavelength of 1.06 I~m) and excellent room temperature stability. We have obtained different responses depending on the poling temperature and of the curing of the sample. The molecular orientation gets faster when the film is heated at 160° for a few hours, before applying the poling field. Other experiments are in progress to fully characterize the stability of NLO properties at high temperature, to characterize the Pockels electro-optic effect and to thoroughly explain the orientation mechanisms involved. New sol-gel films with other azo dyes, silicon precursors and spacers are in preparation where the dye concentration can be increased to a DR1 : Si ratio of 1 : 1. The observed behaviour of these sol-gel non-linear materials and the various possibilities above-men-

tioned to improve the NLO response are promising for making integrated optical devices. References [1] P.N. Prasad and D.J. Williams, in: Introduction to nonlinear optical effects in molecules and polymers (Wiley, New York, 1991). [2] D.M. Buriand, R.D. Miller and C.A. Walsh, Chem. Rev. 94 (1994) 31. [3] M. Becker, L. Sapachak, C. Xu, L.R. Dalton, Y. Shi, S. Kalluri and W.H. Steier, Chem. Mater. 6 (1994) 1184. [4] G. Puccetti, E. Toussaere, I. Ledoux, J. Zyss, P. Gdesmar and C. Sanchez, Polymer Pepr. 32 (1991) 61. [5] C. Sanchez, B. Lebean and B. Viana, in: SPIE Proceed. 2288 (1994) 227. [6] E. Toussaere, J. Zyss, P. Griesmar and C. Sanchez, Nonlinear Optics 1 (1991) 348. [7] Y. Zhang, Y.P. Cui, C. Wang, P.N. Prasad and R. Burzynski, in: SPIE Proceed. 1560 (1991) 264. [8] J.l. Chen, S. Marturunkakul, L. Li, R.J. Jeng, J. Kunmr and S.K. Tripathy, Macromolecules 26 (1993) 7379. [9] K. Izawa, N. Okamoto and O. Sugihara, Japan J. Appl. Phys. 32 (1993) 190. [10] Z. Yang, C. Xu, B. Wu, L.R. Dalton, S. Kalluri, W. Steier, Y. Shi and J.H. Bechtel, Chem. Mater. 6 (1994) 1899. [11] S. Kalluri, Y. Shi, W. Steier, Z. Yang, C. Xu, B. Wu and L.R. Dalton, Appl. Phys. Letters 65 (1994) 2651. [12] R. Kasemann, S. BrUck, H. Schmidt and L. Kador, in: SPIE Proceed. 2288 (1994) 321. [13] D. Avnir, D. Levy and R. Reisfeid, J. Phys. Chem. 88 (1984) 5956. [14] M. Canva, P. Georges, G. Le Saulx, A. Brun, F. Chaput and J.P. Boilot, J. Non Crystalline Solids, 147-148 (1992) 627. [15] M. Canva, P. Georges, J.F. Perelgritz, A. Brun, F. Chaput and J.P. Boilot, Appl. Optics 34 3 (1995) 428. [16] F. Chaput, D. Riehl, Y. L~vy and J.P. Boilot, Chem. Mater. 5 (1993) 589. [17] D. Riehl, F. Chaput, A. Roustamian, Y. I.Avy and J.P. Boilot, Nonlinear Optics 8 (1994) 141. [18] M. Dumont and Z. Sekkat, in: SPIE Proceed. 1774 (1992) 188. [19] R. Loucif-S~'bi, K. Nakatani, J. Delaire, M. Dumont and Z. Sekkat, Chem. Mater. 5 (1993) 229. [20] M. Ueda, H. Kim, T. lkeda and K. lchimura, Chem. Mater. 4 (1993) 1229. [21] M. Kasha, in: Spectroscopy of the excited state, ed. B.D. Bartoio (Plenum Press, New York, 1976) p. 337. [22] R. Ulrich and R. Torge, Appl. Opt. 12 (1973) 290. [23] M. Born and E. Wolf, Principles of optics (Pergamon Press, Oxford, 1980). [24] J. Jerphagnon and S. Kurtz, J. Appl. Phys. 41 (1970) 1667. [25] H.W. Guan and C.H. Wang, J. Chem. Phys. 98 (1993) 3463.