1. Introduction

Since the experimental demonstrations [1–4], following earlier theoretical prediction [5], the study of Random Lasers (RLs) became an extremely attractive subject [6]. RLs can be broadly defined as a random assembly of elastic scatterers dispersed into an optical gain medium or a random arrangement of micro or nanoparticles that act simultaneously as gain medium and scatterers. Except for the proposal of RL action in cold atoms [7], whereby the gain media and scatterers will be the atom itself, the exploited scatterers have been dielectric or metallic particles with different geometrical forms and micro or nanometric dimensions.

The possibility of obtaining stimulated emission and light amplification in disordered media have made RLs unparalleled sources of both incoherent and coherent electromagnetic radiation, as well as a benchmark for multidisciplinary studies involving highly scattering and nonlinear media. From the laser development point of view [6,8], recent advances have brought in results on how multimode RL [9], resonance driven RL [10], Random Fiber Laser [11], Random Distributed Feedback Fiber Laser [12], RL in solutions containing Stöber silica nanoparticles [13], upconversion UV RL [14], random semiconductor lasers [15] and mode-locking in RL [16]. Plasmonically controlled RL have also been described [17, 18].

Two types of feedback regimes in RLs are identified [6,13]: (a) incoherent feedback, when the size of the scattering gain medium is greater than the photon mean-free-path, and this path, in turn, is greater than the emission wavelength. In this regime, light propagation is diffusive and the probabilistic nature of diffusion means that interference contributes negligibly to the feedback process; (b) coherent or resonant feedback, that occurs when the photon mean-free-path and the emission wavelength have the same order of magnitude making it possible for localization of the radiation field to occur within the structure, allowing for coherent feedback.

In the present paper, we describe a RL system with coherent feedback mechanism in chips fabricated of Polyvinyl Alcohol (PVA) films doped with Rhodamine 6G (Rh6G) dye, supported on a glass substrate containing titanium dioxide nanomembranes (TiO₂-NM). One specificity of the RL system described here is that the scattering medium (TiO₂-NM arrangement) is separated from the gain medium. This configuration differs from the conventional RL schemes and allows independent control of molecules concentration and scatterers density.

2. Experimental details

The first step in the samples preparation was obtaining a cobalt film of ≈10 nm thickness that was fabricated by the electron beam physical vapor deposition method on a soda lime glass substrate. To obtain cobalt nanoparticles agglomeration the film was annealed at 350°C for 20 minutes on a hotplate; afterwards the film was bombarded using an electron beam source to obtain cobalt particles of ≈12 nm diameter.

The TiO₂-NM were grown on the top of the cobalt particles using the Metal-Organic Chemical Vapor- phase Deposition (MOCVD) technique. The MOCVD equipment used has a horizontal reactor operating under 76 torr with temperature adjusted between 550 and 650 °C. N₂ was used as the bubbling/carrier gas (0.6 slm), Ti(OC₃H₇)₄ was the Ti and O source (held at 40 °C) and ferrocene [Fe(C₅H₅)₂] was the catalyst source being held at 50 °C. All the MOCVD tubing lines were heated up to 80 °C to avoid vapor gas condensation during the TiO₂ growth. Before the injection of the Ti/O source within the MOCVD reactor, the ferrocene flow was turned on into the reactor cell during 1 min at the growth temperature. After this stage the substrate surface was exposed to the ferrocene flow and the Ti/O source during 20 min.

The Rh6G/TiO₂/NM/glass samples were prepared by spin coating of a PVA film (thickness of ≈3 µm) containing Rh6G dye (concentration of 10⁻³ M) on the surface of a glass substrate (15 × 5 mm²) that supports theTiO₂-NM. For comparison, additional samples with the same characteristics and dimensions of the above were prepared depositing the PVA film with Rh6G directly on a glass substrate without TiO₂-NM.

RL emission was achieved exciting the sample with the second harmonic of a Nd: YAG laser (532 nm, 6 ns, 5 Hz). The laser beam was focused using a cylindrical lens and the beam shape on the sample was a ≈15 µm × 12 mm stripe. The direction of the excitation beam was perpendicular to the samples’ surface. To analyze the emitted spectrum we used a monochromator equipped with a CCD camera (resolution ≈0.1 nm); the integrated and single shot detection modes of the CCD were used.

3. Results and discussions

Figures 1(a) and 1(b) shows scanning electron microscope (SEM) images of the butterfly-like TiO₂-NM that have a wing tip of ≈10 nm, thickness of few hundreds of nm and surface area of few µm². The crystalline structure of the TiO₂-NM presents both rutile and anatase phases [19].

 

Fig. 1 (a) and (b) show SEM images at different magnifications of TiO₂-NM grown on the surface of a soda lime glass using the MOCVD technique. The butterfly-like nanomembranes have ≈10 nm (the wing tip), few hundreds of nm thickness and few µm² of surface area.

Download Full Size | PDF

Figure 2(a) is a diagram of the experimental setup showing the pumping laser beam and the RL emission directions for a sample with TiO₂-NM. The laser beam was slightly defocused to reduce photobleaching of the Rh6G molecules.

 

Fig. 2 (a) Image showing the experimental (not in scale) scheme of excitation and emission. (b) RL integrated linewidth narrowing as a function of the pumping energy for samples Rh6G/glass (circles) and Rh6G/TiO₂-NM/glass (stars). (c) and (d) are the integrated emission spectra of samples Rh6G/glass and Rh6G/TiO₂-NM/glass, respectively.

Download Full Size | PDF

Figure 2(b) shows that both samples present linewidth narrowing for increasing values of the excitation pulse energy. At low excitation energy the linewidth is ≈60 nm for both samples. For large pumping energy the minimum linewidth observed for the sample without TiO₂-NM was ≈9 nm while for the sample with TiO₂-NM the linewidth reaches ≈4 nm. In both cases most of the emitted light propagates parallel to the film surface that acts as a planar waveguide.

Figures 2(c) and 2(d) exhibits the time-integrated emission spectra and linewidth for the samples consisting of Rh6G/glass and Rh6G/TiO₂-NM/glass for different laser pulse energy. The spectrum of Fig. 2(c) is red shift with respect to the Rh6G spectrum for low concentration: The shift is due to re-absorption of the emitted light by monomers and by molecular aggregates that are present due to the large dye concentrations used [20,21]. Although the spectra of Fig. 2(d) is time-integrated, they show narrow peaks with ≈0.8 nm linewidth for pumping energy larger than 0.14 mJ that is a clear indication of coherent feedback due to light scattering. The blue shift of ≈30 nm observed in the spectrum of the sample with TiO₂–NM [Fig. 2(d)] with respect of sample without TiO₂-NM [Fig. 2(c)] may be associated to the resonant Mie scattering by TiO₂-NM that influence the lasing wavelength [10]. This behavior was shown in monodispersed RL medium that sustain electromagnetic resonances. In [10] the RL wavelength was controlled by choosing the diameters of the spherical scatterers. In the present experiments, since the TiO₂-NM are irregularly shaped, it is difficult to provide a precise quantitative analysis for the results. However, considering the actual TiO₂-NM distribution on the substrate, we made a rough estimate of the scatterers density. For instance Fig. 1(a) shows that 50 TiO₂-NM with ≈2 µm size are spread on an area of 20 × 28 µm² of the glass substrate. Then, applying Mie’s theory, we obtain the kl_s ≈34 where l_s is the mean free path, k = 2π/λ, and λ is the wavelength of emitted light. The value of kl_s indicates that the system is diffusive, but the size of the TiO₂-NM are enough large to support Mie resonances that contribute for coherent feedback.

Figure 3 depicts the behavior of the emitted peak intensity as a function of the pumping pulse energy. Both samples show a noticeable change of emission performance for pumping energy above certain value. For the sample without TiO₂-NM, shown in Fig. 3(a), we attribute the intensity behavior as well as the linewidth narrowing to the amplified spontaneous emission (ASE) [22]. The dashed line shown in Fig. 3(a) corresponds to the function $IASE\propto [e^{\left(\gamma -\alpha \right)}{}^{\iota }-1]$ that fits well the experimental data. γ and α are the gain and loss coefficients, respectively, and l is the pumping stripe length. The emitted light due to ASE is mostly guided within the excitation volume and parallel to pumping stripe. The behavior of the peak intensity emitted from the TiO₂-NM sample cannot be described by the same exponential function as in Fig. 3(a). In this case, we observe a RL threshold at pumping energy of ≈0.2 mJ with a clear change in the slope of the curve represented by the solid line in Fig. 3(b).

 

Fig. 3 Emission peak intensity as a function of the pumping pulse energy: (a) Sample without TiO₂-NM and (b) sample with TiO₂-NM.

Download Full Size | PDF

Figures 4(a) and 4(b) shows the single shot spectra recorded with the spectrometer synchronized to detect only one laser pulse each time. This condition allowed us to take advantage of the maximum resolution of our spectrometer. RL spikes with ≈0.3 nm linewidth were observed with pumping energy larger than 0.14 mJ for the sample with TiO₂-NM as shown in Fig. 4(a). On the other hand, the single shot spectra of Fig. 4(b), obtained for a sample without TiO₂-NM, are smooth being solely due to ASE, without RL action.

 

Fig. 4 (a) Emission spectra close and above to threshold from samples with TiO₂-NM obtained by single shot detection; (b) Emission spectra below and above to threshold from a sample without TiO₂-NM obtained by single shot excitation; (c) and (d) are two spectra of sample Rh6G/TiO₂-NM/glass corresponding to pumping pulse energy of 0.41mJ/pulse and 0.59 mJ/pulse, respectively.

Download Full Size | PDF

To understand the RL emissions from the samples TiO₂-NM we consider that the RL process is due to random cavities formed as consequence of the heterogeneous distance distribution among the TiO₂-NM. It is also related to the diverse TiO₂-NM shapes, sizes and orientation. The irregular configuration of nanomembranes contributes for light scattering in random directions for the pumping light as well as for the fluorescence. Under these conditions, optical interference in various light paths is highly frequent and some light modes can be amplified. This gives rise to spikes in the emission spectra for pumping energy above the threshold. Below threshold, the emission spectrum is smooth and originates mostly in the top surface of the film.

Figures 4(c) and 4(d) show the emission spectra of Rh6G/TiO₂-NM/glass obtained with single shot detection for two excitation pulse energies above the RL threshold.

The presence of spikes in Figs. 4(a), 4(c) and 4(d) is a signature of RL action with coherent feedback and according to the literature [23] this is a consequence of the simultaneous existence of extended and localized light modes in the sample.

4. Conclusion

We demonstrated a new RL scheme with coherent feedback in a sample consisting of a gain medium (PVA film doped with Rhodamine 6G) adjacent to the scattering medium comprised of TiO₂ nanomembranes. The herein reported results open up new avenues to exploit nanomembranes arrangements as active integrated photonic devices. As nanomembranes themselves find a myriad of applications that range from energy production and conversion, to biomedicine [24]. The incorporation of active random media, as demonstrated here, will certainly enhance these applications.