Proceedings SPIE 3796A 02, invited talk, Denver 1999

Laser action in periodically structured polymer films J.-M. Nunzia,b,*, F. Sobelb, D. Gindreb, B. Sahraouib, C. Denisa, V. Dumarchera, C. Fiorinia, B. Pacia, L. Rochaa a

LETI (CEA - Technologies Avancées), DEIN/SPE, Groupe Composants Organiques, CEA Saclay, F-91191 Gif-sur-Yvette Cedex, France b Laboratoire des Propriétés Optiques des Matériaux et Applications, Université d’Angers, 2 Boulevard Lavoisier, 49045 Angers Cedex, France * corresponding author ; fax : 33 1 6908 7679, e-mail : [email protected] ABSTRACT

We present a study of Distributed Feedback (DFB) Laser emission in various polymer materials. This permits efficient control of the stimulated emission in dye doped polymer materials. Confinement and waveguiding effects are evidenced. We also propose a dynamical study of the effect using 4 wave mixing in the saturable absorption regime with stimulated emission. Keywords: organic device, polymer laser, distributed feedback, stimulated emission, degenerate four wave mixing.

1.

INTRODUCTION

Organic devices1 are built using technological tools which permit the patterning and micro-structuring of functional molecules and polymers. For instance, a challenging issue in the field of designing molecular devices for photonic applications was to achieve a complete manipulation of the molecular order. In this respect, azo-dye aromatic polymers have been shown to offer interesting properties for materials structuring using light matter interactions. Dual-frequency irradiation using appropriate combinations of circular beam polarization have indeed been demonstrated to enable a full control of the molecular polar order2,3. More recent results on photoinduced surface-relief gratings have also opened the way to molecular translation control using optical fields4,5. A good understanding of the relevant parameters inherent to the molecular translation processes is still a key issue for the optimization of such opto-mechanical effect, but a simple microscopic model accounting for the essential features of photoinduced surface-relief grating formation has been developed6 and partly verified experimentally7. Control of the molecular order is mandatory for the realization of efficient organic devices used for optical applications, but it can reveal important as well as concerns devices such as light emitting diodes8 or photovoltaic cells9. In this respect, polar orientation of molecules has permitted a control of the potential drop at the polymer - metal interfaces8 and a control of the permanent field inside a polymer structure outpacing the limited extension of the depletion layer9. Another challenging issue today is to pattern and micro-structure organic devices in order to control the emission properties of polymer thin films such as the one used for electroluminescent diodes. A control of the radiation modes of electroluminescent diodes has already been demonstrated using planar microcavities10,11,12. We propose here the use of a periodical excitation in a distributed feedback (DFB) scheme to investigate the luminescence and laser properties of dyedoped polymer thin films13. The so-called organic lasers14-18 have recently been revisited with an aim at building an organic semiconductor laser diode19.

2.

PRINCIPLE OF OPERATION

The name of distributed feedback laser (DFBL) is due to Kogelnik and Shank13 that in 1971 have succeeded to release laser emission in a bulk dye laser using a spatially periodical optical-pumping scheme. The feedback produced by this periodic pumping scheme is distributed in all the laser medium instead of being localized on two mirrors. The present work summarizes the principle but here, the laser dye is deposited within a thin polymer film. The dye is a high gain medium because it has a large absorption spectrum and it is easily saturable. Optical pumping easily permits to reach the stimulated emission regime. Using an interferometer, the pump beam becomes periodically structured in space, we thus get feedback through an array of bright and dark fringes which excite the polymer film. So we have an amplification of light by the gain

1

Proceedings SPIE 3796A 02, Denver 1999

grating, a sharp selection of the spectral lines in the film by the feedback, and an optical confinement if the film can act as a waveguide. Last point was previously evidenced in the case of stimulated emission in a thin organic slab-crystal20. Several poly (methyl methacrylate) (PMMA) and poly (N-vinyl carbazole) (PVK) polymer films were chosen as samples for the experiments. Polymers are doped with usual laser dyes (Rhodamine 6G – Rh6G, or DCM ) and deposited by spincoating onto glass or quartz substrates. The choice is determined by the index of refraction of the film, in order to realize a film that plays the role of a waveguide. Films thickness is determined by solvent concentration and rotation speed of the spin-coater. The materials used have interesting properties for our purpose : they are highly non linear in order to respond to the interference pattern and the stokes shift between absorption and fluorescence bands is large. The absorption spectra show that a wavelength of 532 nm falls inside the absorption band, and that linear absorption above 600 nm is very weak. The experimental setup for DFBL characterization is shown in figure 1. A pulsed mode-locked YAG:Nd laser was used to produce frequency-doubled 35 ps pulses (FWHM) at 532 nm. The energy of the transmitted beam after a Glan-Taylor polarizer is varied by means of a half-wave plate (λ/2). An afocal system coupled to a diaphragm permits shaping of the beam profile. A cylindrical lens is used before the interferometer in order to focalize the pump, thus increasing its energy density, and to improve guidance by a better gain confinement. The feedback which is realized by distribution of the gain in the film permits a guided laser emission into the film. The signal is directed onto a spectrometer by means of a plastic optical fiber. The spectral detection is achieved by a grating spectrometer. Spectra are recorded using a Hamamatsu charge coupled device (CCD) camera cooled by a three stage Peltier. The heart of the setting is the interferometer. We have used two different configurations. The first one (Setup 1) uses a interferometer with two separated arms21 (Fig.1-a) that has the advantage to render the sample immobile. The angle α is varied by changing the mirrors position. A photography of this device is shown in figure 2.

Fig. 1 : Schematic diagram of the experimental setup for DFBL characterization. A cylindrical lens is used before the beam splitter in order to enable efficient pumping over a long narrow stripe. The emissive polymeric structure is spin-coated on glass or silica substrates. Emission is collected by a plastic optical fiber and analyzed by a grating spectrometer. On the right side are schematized the two interferometers used for building the interference pattern in the thin film. The angle α is varied by changing the mirrors positions in setup 1 (a) or by rotating the device in setup 2 (b).

The second one (setup 2) is a Loyd-mirror interferometer (Fig.1-b), where a mirror is placed perpendicularly to the film. The pump beam is splitted into two parts. The first one goes directly onto the film, and the other one is reflected by the mirror. Both interfere in the polymer film. This permits easy adjustment over a broad interval of the incidence angle α between the pump beam and the thin film. The interferometer can rotate around its axis, keeping by symmetry the same angle for the two interfering beams. The period Λ of the interference pattern, is given by :

λp

Λ = 2n sin α

(1)

where λp is the wavelength of the pump. For Rh6G doped films, the optimal incidence angle permitting excitation at the fluorescence maximum is close to 45°. A rotation of two degrees allows to sweep all the fluorescence band.

Proceedings SPIE 3796A 02, Denver 1999,

2

Proceedings SPIE 3796A 02, Denver 1999

Fig. 2 : Photography of the interferometer setup. We can see both green beams arriving onto the film. Incidence angle is varied by changing the mirrors position.

It is easy to see how this setup can produce the feedback of a laser. The periodic structure can be seen as a superposition of microcavities distributed along the illuminated zone. In this pumping scheme, both gain and cavity are produced by the same absorption mechanism, that makes the system very simple. All these microcavities have a length equal to that of the interfringe. Longitudinal modes being able to propagate have to verify the relationship : nλ p

λ = m sin α

(2)

where n is the refractive index of the medium and m is the diffraction order of the grating. We see that the emitted wavelength λ depends on the incidence angle of the pump beam onto the film. By modifying this angle α, one is able to tune the emission wavelength of the laser and thus to make a tunable laser, in the spectral limits of the fluorescence band of the dye used. With this type of pumping, DFB may have two origins : i) ii)

a gain grating, an index of refraction grating.

The first one is due to the alternance of large and low gain domains in the interference region. The second one results from the optical Kerr effect, inducing a nonlinear index which can be written in the linear intensity regime (far from saturation) :

n = n0 + 3.

3 χ ( 3) 4ε 0 n 02 c

I

(3)

LASER THRESHOLD AND TUNING

To check for laser emission, we use a film of PMMA doped with Rh6G. Film thickness in 1.9 µm. Practically, stimulated emission linewidth gets smaller as the pump energy increases, down to 0.5 nm under appropriate DFB conditions. Laser action threshold is 500 nJ.mm-2 with a pump-pulse duration of 35 picoseconds. Rotating the interferometer around its axis permits to change the incidence angle of the pump, and so the grating period. This permits to find the so-called Bragg condition : that is to select the emission wavelength by grating diffraction. We can thus tune the laser wavelengths all over the stimulated emission band. Figure 3 shows the evolution of the laser emission line as a function of the angle imposed on the interferometer. On the left part of figure 3 (PVK/DCM, thickness 1.9 µm on a quartz substrate, using setup 1), we get a tunability of the laser line on a spectral band broader than 40 nm. On the right part of figure 3 (PMMA/Rh6G, thickness 400 nm on a glass substrate, using setup 2) we get a tunability of the laser line on a spectral band of 30 nm.

Proceedings SPIE 3796A 02, Denver 1999,

3

Proceedings SPIE 3796A 02, Denver 1999

Fig. 3 : Laser wavelength as a function of the tuning angle α in two spin-coated thin films : PVK-DCM (thickness : 1.9 µm) on the left (setup 1) and PMMA-Rhodamine 6G (thickness : 0.4 µm ) on the right (setup 2). The pump source is a frequency doubled (λp = 532 nm) picosecond Nd-YAG laser (800 µJ/cm2). Tuning range is 40 nm in the configuration on left and 30 nm in the configuration on right.

We were able to adjust the refractive indexes of the polymer film and substrate. Several trials have been undertaken in this respect. Signal with a quartz substrate can be 100 times larger as with a glass substrate when polymer host is PMMA. A strong reduction of the laser threshold is observed when waveguiding conditions are met. Such effect reveals the importance of wave guiding.

4.

WAVEGUIDE APPROACH AND SPECTRAL CONTENT

The influence of the film thickness was investigated using a PVK/DCM film on a glass substrate. Figure 4 shows the spectrum of the DFB laser at a fixed angle, that is with a fixed grating period. For film thicknesses from 215 nm to 1600 nm, we see the number of laser modes increasing : from one to three laser modes appear within the stimulated emission band.

Fig. 4 : Evolution of the laser emission spectrum with the film thickness. In the case of a monomode film (a) (e = 215 nm), only one laser line appears. Increasing the thickness, the number of laser lines increases accordingly : two in (b) (e = 740 nm) and 3 in (c) (e = 1600 nm). The background spectrum is due to stimulated emission.

There appears that all laser lines in figure 4 peak at different wavelengths. Additionally, the non periodic spectral gap between laser lines cannot be attributed to the free spectral range (FSR) of a cavity. Therefore, these laser modes are not longitudinal modes. We varied the grating period linearly and made a cross section of the emission spectrum for each angle. We get the plot in Figure 5. For a thickness of 2.9 µm, seven modes appear. Each laser mode can be tuned over a 40 nm range. The stimulated emission bandwidth limits the number of laser modes. We also performed a theoretical analysis of the modal content in the waveguiding structure.

Proceedings SPIE 3796A 02, Denver 1999,

4

Proceedings SPIE 3796A 02, Denver 1999

Bragg conditions strictly select the propagation constant β of the waveguide modes undergoing DFB laser emission. β is related to the grating period Λ, which is related to the incidence angle α and to the pump wavelength λp as (in the order 2 of the grating) :

β = 4πλsin α p

(4)

Fig. 5 : Plot of laser mode emission wavelengths as a function of the incidence angle a. Film thickness is 2.9 µm. Seven modes appear, with a 40 nm tuning range. The stimulated emission bandwidth limits the modal content.

Waveguided modes can be found within the usual theory of planar asymmetric waveguides. The transcendental equation which gives the transverse wave vector is : tan( k x e ) =

where

p0 + p2

(5)

 p p  k x 1− 0 2 2  kx  

pi2 = β 2 − (k 0 ni )

2

with i = 0,2 ,

k x2 = (k 0 n1 ) − β 2 , 2

(6) (7)

and k0 =

2π . λ0

(8)

We get the dispersion curves by a simulation using appropriate parameters : index of refraction of the layers and thickness of the film. With equation (4), we can plot the dispersion curves as a function of the incidence angle. Figure 6-a shows the dispersion curves of our waveguide in the 550 to 650 nm range. Such plot is usually used to get the guided modes. In usual studies of planar waveguides, we launch a definite wavelength, and we predict the group velocities or propagation constants of the various propagating modes. Here, the problem is analyzed in a different way : Bragg condition on the DFB imposes a definite propagation constant β and we get all the laser wavelengths corresponding to the modes which are amplified inside the stimulated emission band. Figure 6-b shows the superposition of our numerical simulation of waveguided modes with the

Proceedings SPIE 3796A 02, Denver 1999,

5

Proceedings SPIE 3796A 02, Denver 1999

experimental data for laser emission. The spectral range is restricted to the experimentally investigated domain. The agreement between theoretical curves and experimental results is excellent. Only one waveguide mode is not laser active, because it falls outside of the stimulated emission band. So, we got evidence for two important characteristics of polymer DFB lasers : i) grating period determines the longitudinal propagation constant, ii) propagation constant is restricted by waveguiding conditions inside the stimulated emission band.

a

b

Fig. 6 : Dispersion curves of our 2.9 µm-PVK/DCM waveguide between 550 and 650 nm on the left part. Superposition of the simulated dispersion curves with the experimental data from figure 5 on the right part.

Other spectral effects appear in these investigations. We studied the laser threshold of the various modes for different grating lengths. As a matter of fact, as a cylindrical lens was used in order to reduce the width of the pump impact down to a narrow confining stripe of 20 µm-width, using a diaphragm, we could also select the grating length on the sample. We observe a splitting of some laser modes by increasing the grating length or the pump intensity. Figure 7 shows cross sections of the laser emission spectrum at a fixed angle, the intensity increasing downwards, for two grating lengths of 1.3 (a) and 3.5 mm (b). Sample is the same as in figure 5. Up to six lines are observed at such intensities. There is no spectral shift on increasing the grating length, nor the pump intensity. This was expected because the grating period of 380 nm does not change at all. The broad continuum is due to stimulated emission. More laser modes than in figure 5 are lying over the large stimulated emission band and the first saturated mode is not in the center the band.

Fig. 7 : Intensity and grating length dependence of the laser spectrum in the 2.9 µm-PVK/DCM sample of figure 5. Grating period in the polymer is 380 nm. On the left side, grating length is 0.62 mm (a) and 0.74 mm (b). Pump intensity increases downwards. The right side shows the evolution (splitting) of a single laser line when grating length increases, for a pump intensity of 3.5 GW/cm².

We also observe a spectral splitting of the laser line above 2 GW/cm²-pump intensity. The laser spectrum divides in two adjacent lines whose splitting increases with grating lenth. The mode splitting increases also with pump intensity. The spectral gap between both splitted peaks reach a maximum at 3.5 GW/cm². The splitting effect does not affect equally all the laser modes. As discussed previously by Kogelnik and Shank,22 splitting is due to a competition between gain and nonlinear index gratings. It is theoretically explained within the framework of the coupled mode theory.

Proceedings SPIE 3796A 02, Denver 1999,

6

Proceedings SPIE 3796A 02, Denver 1999

For larger pump intensities (above 3.5 GW.cm-2), we see some « ghost modes » (figure 7-b) whose emission wavelength is no more tunable. We attribute these modes to a permanent grating which is photo-induced into the polymer film. A sample observation (500 nm-PVK/coumarin film pumped at 355 nm) was made using an atomic force microscope (AFM). The relief image is shown in figure 8. We see a permanent surface relief grating with a depth of about 100 nm in the photo-damaged region.

Fig. 8 : AFM image of a photo-damaged 500 nm-PVK/coumarin film pumped at 355 nm. A permanent relief grating with ≈ 100 nm-depth is observed in the laser pumped region.

5.

TEMPORAL STUDY

Fig. 9 : Experimental setup for temporal characterization of DFB laser emission using degenerated four wave mixing at 532 nm. We have set a degenerate four wave mixing (DFWM) experiment in order to study the temporal behavior of our DFB lasers. The original idea is to use angles between the so-called probe beam and one of the pump beams (pump #1) which correspond to Bragg grating conditions for DFB laser action. The experimental setup is shown in figure 9, the laser is the same picosecond Nd :YAG at 532 nm. The 3 incoming beams have independent optical delay lines. Detection is achieved with a photomultiplier tube coupled to a spatial filter used in phase-conjugation conditions.23

Proceedings SPIE 3796A 02, Denver 1999,

7

Proceedings SPIE 3796A 02, Denver 1999

We first studied the intensity dependence of the DFWM-signal using pump-probe angles falling outside of the Bragg grating conditions for DFB laser action. Sample was a 1.9 µm-thick PMMA/Rh6G-film. Time delay between incoming beams was set to coincidence. We get a regular cubic χ(3) response for pump intensities as low as 0.3 MW/cm². The onset of saturation appears for 30 MW/cm²-pump intensity. We thus see that the DFB-laser experiments reported above were essentially performed in the strong saturation regime. Figure 10 shows the temporal response of the DFWM-signal on delaying pump-beam #2 with respect to the two other ones. Angles did not match the Bragg grating conditions for DFB laser action. Two intensity regimes were investigated. Dashed curve corresponds to 60 MW/cm²-pump intensity and solid curve to 0.5 MW/cm². Several oscillations with a period T = 40 ps appear at low pump intensity. They may be attributed to the acoustic oscillations which can be induced either by impulsive thermal effect or by electrostriction.24 In PMMA, speed of sound is reported to be vs = 2700m/s.25 The T = 40 ps period observed in figure 10 is too short to correspond to a thermally-generated acoustic wave. It can thus be attributed to an acoustic wave generated by electrostriction and whose dispersion relation verifies : Λ =2vsT. We thus get a period of 216 nm for the grating, which is consistent with our experiments. At large pump-intensity in figure 10, the time dependence at delay zero is broader, owing to saturation of the nonlinearity. Also, the decay is faster owing to stimulated emission effects. Larger pump intensities are difficult to investigate because of sample photo-damage

Fig. 10 : Temporal dependence of the DFWM signal on delaying pump #2 with respect to the other ones. Dashed curve is for a 60 MW/cm² pump-intensity and solid curve for 0.5 MW/cm². Period of the oscillations is about 40 ps.

6.

CONCLUSION

We have successfully implemented a tunable distributed feedback (DFB) laser configuration which is now available for further evaluation of various laser active polymer materials. Waveguiding of the DFB laser emission has been studied in detail. We have introduced the DFWM technique as a tool to characterize the temporal properties of DFB laser emission. More experiments are in progress : spectral narrowing under DFB-laser conditions should be studied shortly using the socalled Kerr ellipsometry technique.25

7.

ACKNOWLEDGMENT

The authors are grateful to R. Chevalier and J.P. Lecoq for their technical assistance in Angers. We thank Doctors Geoffrey Gale for the 2 mirror DFB setup and Sylvain Magne for the spectrometer, both used in Saclay. Work performed at LETI was partly supported by ESPRIT - LTR project 28580 (LUPO).

Proceedings SPIE 3796A 02, Denver 1999,

8

Proceedings SPIE 3796A 02, Denver 1999

8. 1. 2. 3. 4. 5. 6. 7.

8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26.

REFERENCES

J.-M. Nunzi, F. Charra, C. Ngo, “ Les composants organiques ”, in Les Nouvelles du LETI, 59, 1, (1995) C. Fiorini, F. Charra, J.-M. Nunzi, P. Raimond, J. Opt. Soc. Am. B, 14, 1984 (1997) A.-C Etilé, C. Fiorini, F. Charra, J.-M. Nunzi, Phys. Rev. A, 56, 3888 (1997). P. Rochon, E. Batalla, A. Natansohn, Appl. Phys. Lett., 66 ,136 (1995). D.Y. Kim, L. Li, J. Kumar, S.K. Tripathy, Appl. Phys. Lett., 66, 1166 (1995). P. Lefin, C. Fiorini, J.-M. Nunzi, Pure Appl. Opt., 7, 71 (1997) ; P. Lefin, C. Fiorini, J.-M. Nunzi, Opt. Mater., 9, 323 (1998). C. Fiorini, N. Prudhomme, A.-C. Etilé, P. Lefin, P. Raimond, J.-M. Nunzi, Macromol. Symposia, 137, 105 (1999) ; C. Fiorini, N. Prudhomme, G. de Veyrac, I. Maurin, P. Raimond, J.-M. Nunzi, “ Molecular migration mechanism for laser induced surface relief grating formation ”, submitted to Synthetic Metals (1999). F. Nüesch, L. Si-Ahmed, B. François, L. Zuppiroli, Adv. Mater., 9, 222 (1997). C. Sentein, C. Fiorini, A. Lorin, J.M. Nunzi, Adv. Mater., 9, 809 (1997) ; C. Sentein, C. Fiorini, A. Lorin, J.M. Nunzi, P. Raimond, L. Sicot, Synthetic Metals, 102, 989 (1999). N. Takada, T. Tsutsui, S. Saito, Appl. Phys. Lett., 63, 2032 (1993). T.A. Fisher, D.G. Lidzey, M.A. Pate, M.S. Weaver, D.M. Whittaker, M.S. Skolnick, D.C.C Bradley, Appl. Phys. Lett., 67, 1355 (1995). H.F. Wittmann, J. Gruner, R.H. Friend, G.W.C. Spencer, S.C. Moratti, A.B. Holmes, Adv.Mater., 7, 541 (1995). H. Kogelnik, C.V. Shank, Appl. Phys. Lett., 18, 152 (1971). M. Kuwata-Gonokami, R.H. Jordan, A. Dodabalapur, H.E. Katz, M.L. Schilling, R.E. Slusher, S. Ozawa, Opt. Lett., 20, 2093 (1995). N. Tessler, G.J. Denton, R.H. Friend, Nature, 382, 695 (1996). F. Hide, M.A. Diaz-Garcia, B. J. Schwartz, M. R. Anderson, Q. Pei, A. J. Heeger, Science, 273, 1833 (1996). A. Schülzgen, Ch. Spiegelberg, M.M. Morell, S.B. Mendes, B. Kippelen, N. Peyghambarian, M.F. Nabor, E.A. Mash, P.M. Allemand, Appl. Phys. Lett., 72, 269 (1998). V.G. Kozlov, G. Parthasarathy, P.E. Burrows, S.R. Forrest, Y. You, M.E. Thompson, Appl. Phys. Lett., 72, 144 (1998). D.G. Lidzey, D.D.C. Bradley, S. F. Alvarado, P. F. Seidler, Nature, 386, 135 (1997). D. Fichou, S. Delysse, J.M. Nunzi, Advanced Materials, 9, 1178 (1997). G.M. Gale, P. Ranson, M. Denariez-Roberge, Appl. Phys B, 44, 221 (1987). H. Kogelnik, C. V. Shank, J. Appl. Phys.,43, 2327 (1976). J.M. Nunzi, F. Charra, Pure and Appl. Opt., 7, 501 (1998). J.M.Nunzi, D.Grec, J. Appl. Phys., 62, 2198 (1987). Polymer handbook, J. Bandrup and E.H. Immergut, eds., (Wiley-Interscience, New York, 1967). S. Delysse, J.M. Nunzi, C. Scala-Valero, Appl. Phys B, 66, 439 (1998).

Proceedings SPIE 3796A 02, Denver 1999,

9