1. Introduction

Initial reports of lasers with mirrorless feedback due to scattering [1,2] triggered considerable experimental research interest. This involved incoherent light amplification in diffusive media with gain [3], which basically is amplified spontaneous emission (ASE), as well as coherent random lasers [4,5]. Conventional lasers require accurate fabrication and extremely precise alignment of the resonator. In contrast to that, resonant random lasing is enabled by randomly formed closed-loop cavities [6–9] of high refractive index dielectric or metallic nanostructures inside a gain medium. This permits an easy and cheap fabrication of random lasers and thus makes them extremely attractive for a manifold of potential applications, starting from displays [10] and lighting devices, to smart sensors and medical diagnostics [11–14].

A combination of laser dyes, such as Rhodamine 6G (R6G), as gain medium with noble metal nanoparticles (NPs) as scattering centers is of particular interest. Noble metal NPs exhibit localized plasmon resonances in the visible spectral range, which allow them to scatter incoming light more efficiently than their dielectric counterparts of similar dimensions, made out of high refractive index dielectrics such as TiO₂ or ZnO. The spectral position of the plasmon resonance strongly depends on material, shape and environment of the NPs, so that the variation of one or more of these parameters allows spectral tuning of the plasmon resonance to overlap the emission spectrum of the desired active medium. Plasmonic resonances change the local density of optical states in the close vicinity of the NPs and yield strongly enhanced and highly confined optical fields close to the NPs’ surface. Thus, plasmonic NPs modify the non-radiative and radiative transition rates of nearby dye molecules [15,16]. Most of the research activities are focused on modifying the fluorescence yield or enhancing Raman signals with NPs, whereas the modification of stimulated emission as well as the role of the field enhancement to overcome laser thresholds is less well studied. ASE [17,18] and coherent random lasing [19–22] with metallic NPs were chiefly demonstrated with silver NPs, apart from a few studies on ASE [23] and coherent random lasing [24,25] with spherical gold NPs, because silver nanostructures have less losses than golden ones. However, gold nanostructures feature much lower chemical degradation and thus are more stable in ambient conditions.

In this letter, we report that star-shaped gold NPs (nanostars) outperform gold NPs of conventional shapes, such as spherical gold NPs (nanospheres) and prolate gold NPs (nanorods) as scattering centers in random lasers composed of R6G-doped polymer thin films. Complex-shaped NPs, such as nanostars, provide multiple plasmon resonances overlapping with the emission spectrum of R6G and, in addition, extremely high field enhancements, strongly localized at their spiky tips [26]. Our experimental results deliver major hints that these peculiar qualities can improve the feedback mechanism for random lasing and increase the efficiency in terms of emission intensity. Additionally, we reveal that the nanostar-based random laser is operating at a threshold fluence in the range of 0.9 mJ/cm², showing multiple lasing modes with linewidths of 0.2 nm or even below, since the spectral resolution is limited by the experimental setup.

2. Device fabrication and experiment

For fabrication of the random lasers, R6G (Sigma Aldrich) and polyvinylpyrrolidone (PVP, MW ~29000, Sigma Aldrich) were used without further treatment. The particle concentration of gold nanospheres and nanorods (both Nanopartz Inc.) was controlled by gentle centrifugation and redispersion in deionized water (DI). The nanostars were wet-chemically synthesized with a seed-mediated growth method [27]. Figure 1(a) displays scanning electron microscope (SEM) images of the differently shaped NPs. The concentrations of the different solutions of NPs were adjusted so that the optical densities were the same at 590 nm, see Fig. 1(b). This assured comparable scattering properties in the interesting spectral region despite inherent size inequalities of the differently shaped NPs. We chose the gold NPs to serve as multiple scattering centers for random lasing, with three disparate shapes so that their plasmon resonances overlapped well with the emission of R6G. Each NP solution was mixed 1:1 with a stock solution consisting of PVP (150 g/L) and R6G (6 mM) in ethanol, under continuous stirring to avoid NP aggregation. A reference solution containing no NPs but the same respective concentrations of R6G and PVP was prepared as well (called reference). Glass slides (Menzel, 1 mm thick) were cleaved to a size of approximately 2 cm x 2 cm with a diamond cutter, in order to obtain a neat edge of the sample. The thin films were spin coated from solution at 3000 rpm onto the pre-cleaned glass slides, and cured at 60 °C overnight. The PVP constituted a matrix for the R6G gain medium and immobilized the embedded NPs in the dried thin film. The resulting film thickness amounted to 320 ± 10 nm for every sample and, for the sake of comparability, all samples were prepared on the same day. Figure 1(c) shows the extinction and the normalized steady-state fluorescence spectra of the reference film. The various thin films containing NPs of the particular shape as well as the reference film were then examined and compared in random lasing measurements.

 

Fig. 1 (a) Electron micrographs present gold NPs of different shapes. (b) Extinction spectra of the differently shaped NPs in aqueous solution. The NP concentrations were adjusted so that the optical densities were the same at 590 nm. (c) Extinction and emission spectra of a thin film containing R6G as gain medium. The vertical green line indicates the wavelength of excitation. (d) Illustration of the excitation stripe and the detection of emission from the edge of the thin film.

Download Full Size | PDF

The experimental setup resembled the one known from the variable stripe length method [28]. A diode-pumped, passively Q-switched solid state laser (CryLaS GmbH, pulse length shorter than 1.3 ns) was used to excite the thin films at a wavelength of 532 nm, i.e. near the absorption maximum of R6G, with single pulses to prevent photobleaching of the gain medium. The excitation beam was shaped into a thin stripe by two crossed cylindrical lenses and focused in the plane of two razor blades. In order to establish a 3 mm long, 55 µm wide excitation stripe on the sample, the razor blades selected the homogeneous central part of the beam. This was projected onto the sample by an imaging lens to reduce Fresnel diffraction and to ensure a constant pump fluence along the stripe. The cleaved edge of each sample was positioned close to the rim of the excitation stripe, as illustrated by the illumination and detection geometry in Fig. 1(d). The emission, accumulated alongside the stripe due to lateral waveguiding and coupled out of the edge of the thin film, was focused onto the entrance slit of a Czerny-Turner spectrometer (Newport MS260i, 25 cm focal length) equipped with a Peltier-cooled charge-coupled device camera (Andor iVac). A 532 nm notch filter was used to suppress the detection of scattered excitation light.

3. Results

Figure 2 summarizes the measurements for the various samples excited by single pulses at a fixed pump fluence of 1.2 mJ/cm². The different curves in each subfigure represent emission spectra recorded at several excitation regions shifted along the edge of each film. The emission of the reference film, see Fig. 2(a), is primarily characterized by steady-state fluorescence from a R6G film and ASE [29]. It does not show any coherent lasing behavior and exhibits increased emission (red and pink curves) with reduced linewidth only at a few regions of excitation. The emission spectra of the thin film with embedded nanospheres are presented in Fig. 2(b). Surprisingly, we do not observe an increased emission or reduced linewidth of the fluorescence emission for any excitation region. Gold nanospheres rather seem to be detrimental for ASE or random lasing. Figure 2(c) displays the emission of a thin film containing nanorods. We observe augmented ASE for some excitation areas along the sample, i.e. enhanced emission at around 570 nm with reduced linewidth (red, blue and navy blue curves). Occasionally, sharp features (green curve) emerging out of the ASE can be observed for some specific regions on the thin film. The overall emission intensity is higher, too, compared to the previous cases of the reference and the film containing nanospheres. In the case of a thin film containing nanostars as scattering centers, ultrasharp spectral lines characteristic for coherent random lasing are frequently observed, see Fig. 2(d). The overall emission intensities and the peak heights relative to the ASE are much stronger for a film with nanostars than for films with NPs of any other studied shape.

 

Fig. 2 Emission characteristics of thin films containing R6G and gold NPs of different shapes. For each film, spectra were recorded at the same excitation fluence of 1.2 mJ/cm² from various regions shifted along the edge of the film. (a) The reference film without NPs shows ASE in some cases. (b) Films with nanospheres do not even show ASE at this fluence. (c) Films containing nanorods show ASE and occasionally spectrally narrow peaks. (d) Films comprising nanostars expose lasing with high emission intensities and spectrally narrow peaks in several cases.

Download Full Size | PDF

The observed ASE for the reference film can be enabled by a waveguide mechanism in the thin PVP-film, which channels the photons towards the edge of the sample. This ASE might be augmented at a few excitation regions by incidental scattering events owing to defects in the film such as PVP aggregates or air inclusions [30]. Apart from that, in the studied films with gold NPs, multiple scattering events [10] may provide sufficient coherent feedback for random lasing. However, no amplification of the emission takes place for the film with nanospheres at a pump fluence of 1.2 mJ/cm², although they exhibit a plasmon resonance spectrally overlapping with the emission of R6G. It is likely that the inherent absorption of the gold nanospheres and their ability to quench fluorescence in their vicinity constitute crucial loss channels. The local field enhancement is rather moderate for gold nanospheres, so the losses dominate over a possible gain due to feedback from multiple scattering events. In contrast to nanospheres, the nanorods bare larger field enhancements which are closely confined nearby their round caps. Nanorods efficiently scatter the photons emitted by the R6G and, moreover, the high local fields provide an additional excitation enhancement of the molecules. This can cause a significant increase in the effective emission rate [16,31] and an overcoming of the laser threshold, in contrast to the case of nanospheres. The nanorods may therefore provide sufficient gain for lasing when forming a resonator due to their random arrangement. The spectrally narrow peaks observed in Fig. 2(c) hint towards resonant lasing modes. Photons are kept via multiple scattering events in a single lasing mode long enough to be coherently amplified in the gain region, before they are coupled out. The complex-shaped nanostars exhibit multiple plasmon resonances [32], spectrally overlapping with the emission of R6G. Strongly enhanced fields are highly localized at their spiky tips [32] and effectively augment interactions with nearby molecules [33]. The accordant random lasing behavior of films comprising nanostars in Fig. 2(d) is the most pronounced of all studied films. Hence, the superiority of the nanostars over NPs of other shapes may be explained, at this stage, by coherent feedback due to multiple scattering events and by superior field enhancements at the nanostars’ tips, which locally induce an additional gain to overcome the laser threshold.

Next, the lasing behavior of the thin film containing nanostars was examined in more detail. The emission spectra for increasing fluences of single excitation pulses are displayed in Fig. 3. All spectra were taken at the same excitation region of the film and sorted into Figs. 3(a) and 3(b) for low and high pump fluences, respectively. Starting from a fluence of 0.61 mJ/cm² (black curve), up to a fluence of 0.85 mJ/cm² (blue curve) in Fig. 3(a), the emission spectrum is rather broad and centered at 575 nm, showing no lasing features. At a fluence of 0.97 mJ/cm² and onwards, a single, narrow peak 1 emerges at 575.0 nm out of the broad fluorescence background along with some weaker peaks at 573.5 nm and 575.8 nm. The peak intensities rapidly increase with the fluence while their linewidths remain narrow. For instance, the spectral width is only 0.2 nm for the major peak 1 at 575.0 nm, which constitutes the resolution of the spectrometer (grating 1200 l/mm, entrance slit 100 µm). The measured values therefore represent an upper limit and the actual linewidths of the lasing modes might be well below. The emission spectra for stronger pump fluences at the same excitation spot are shown in Fig. 3(b). Random lasing intensities and number of peaks grow further, unveiling the existence of several other prominent lasing modes (denoted by arrows 2 and 3 in Fig. 3(b), peak 1 is the same in Figs. 3(a) and 3(b)). The laser emission of the additional modes occurs at several spectral positions and becomes more intense at high fluences. Remarkably, peak 2 even gets more intense than peak 1 at the highest fluence of 1.58 mJ/cm². It is noteworthy that most of the peaks in Fig. 3, albeit their relative intensities alter, remain at the same spectral position for all applied excitation fluences and do not show any degradation even after several excitation pulses. This provides a strong hint that structural deformation of the nanostars as well as photobleaching of the gain medium can be excluded. The inset in Fig. 3(b) depicts the intensity evolution of the three dominant peaks of the random laser as a function of pump fluence. A nonlinear behavior is clearly observed for peak 2, as well as a threshold fluence for lasing at 1.21 mJ/cm² for peak 3. The nonlinearity in the intensity of peak 1 is not that obvious, but the threshold fluence can be estimated from the emission spectrum in Fig. 3(a) to lie between 0.85 mJ/cm² and 0.97 mJ/cm².

 

Fig. 3 Emission of the random laser containing nanostars for increasing excitation fluences. (a) Spectrally narrow peaks emerge above a threshold fluence, indicating the onset of coherent random lasing. (b) For higher fluences, more lasing modes arise. Inset: Nonlinear behavior of the intensity of the indicated peaks (1, 2, 3) with increasing fluence. Note that peak 1 is the same in (a) and (b).

Download Full Size | PDF

The appearance of the major peak 1 in Fig. 3(a) with the extremely narrow linewidth is indicative of a single, coherently lasing mode due to an efficient feedback mechanism. It arises at the lowest threshold fluence and represents the mode for which the compensation of losses is achieved first. The intensity of this peak 1 is, however, surpassed by peak 2 at the highest fluence in Fig. 3(b). The crossover from single-mode to distinctly multimode behavior in random lasers has been attributed to the spatial extent of the modes [34] and their relative strengths to strongly nonlinear interactions of spatially overlapping lasing modes [35]. The characteristic that the spectral position of each peak is independent of excitation fluence becomes apparent in that the excitation stripe for the measurements of Fig. 3 is at a fixed region on the film. The random arrangement of nanostars in the gain region thus does not change for each single-pulsed excitation but is encoded in the random lasing.

4. Conclusion

In summary, we have investigated coherent random lasing from thin films containing R6G and plasmonic gold NPs of different shapes. We found that nanospheres seem to hinder the occurrence of random lasing modes. In contrast, nanostars offer the best performance of random lasing in terms of emission intensity and number of coherent modes, outperforming even nanorods. We infer that a significant increase of the molecular excitation rate together with an emission rate enhancement are the dominant mechanisms for nanostar-assisted random lasers, considering the huge field enhancements at the spiky tips of the nanostars [26] and the associated plasmonic hotspots. Both, the low threshold for lasing of around 0.9 mJ/cm² and the narrow linewidth below 0.2 nm suggest that the nanostars provide an efficient coherent feedback for random lasers.