Nanoparticle-doped electrospun fiber random lasers with spatially extended light modes

Complex assemblies of light-emitting polymer nanofibers with molecular materials exhibiting optical gain can lead to important advance to amorphous photonics and to random laser science and devices. In disordered mats of nanofibers, multiple scattering and waveguiding might interplay to determine localization or spreading of optical modes as well as correlation effects. Here we study electrospun fibers embedding a lasing fluorene-carbazole-fluorene molecule and doped with titania nanoparticles, which exhibit random lasing with sub-nm spectral width and threshold of about 9 mJ cm^-2 for the absorbed excitation fluence. We focus on the spatial and spectral behavior of optical modes in the disordered and non-woven networks, finding evidence for the presence of modes with very large spatial extent, up to the 100 micrometer-scale. These findings suggest emission coupling into integrated nanofiber transmission channels as effective mechanism for enhancing spectral selectivity in random lasers and correlations of light modes in the complex and disordered material.


Introduction
Random lasers are complex photonic devices which rely on multiple scattering of light in microscopically non-homogeneous media with optical gain [1]. When relevant interference effects are present for the scattered light, some degree of spatial localization can be predicted for optical modes, which reduces inter-mode coupling [2]. These phenomena have also been associated to the presence of very narrow (full width at half maximum, FWHM, down to sub-nm) lasing peaks [3]. Such sharp resonances are largely independent oscillations, typically observed in various systems which spontaneously embed random cavities, such as semiconductor powders [3,4], clusters of colloidal nanoparticles [2], and conjugated polymer films [5]. Methods to achieve so-called resonant feedback random lasing from independent modes include using excitation beams with high directionality, and designing weakly diffusive media as widely studied with solutions of lasing dyes with titania scatterers [2,6]. These aspects have motivated a debate about how much the modes of random lasers spread in the disordered medium, with various possible configurations proposed, ranging from highly localized states with exponentially decaying amplitude to extended modes related to scattering resonances [3,4,[7][8][9].
Correlated emission and non-locality may also arise in random lasers, associated to mode competition and interplay through open cavities leading to extended enough and eventually overlapping light states [2,6,8,10]. In this case, the propagation of light is more likely to be describable as a diffusive process [6,11], with a characteristic photon transport mean free path, in analogy with random walking particles. The 3 interaction, until only cooperative modes survive and condensate into a single wavefunction. Such mode-locking regime is promising for applying these devices to several types of photonic chips, including platforms with all-optical control [2,10].
Random lasing has been achieved in a wide range of three-dimensional (3D) systems, largely using inorganic nanostructures and powders (e.g., ZnO [12,13], GaN [12], BaSO 4 [14] etc.), photonic glasses [15], or dispersions of elastic scatterers in solutions with lasing dyes [2,6,8,10]. Organic crystals [16,17] and epitaxial nanowires [18], biopolymers [19,20], as well as conjugated polymers [5,21], have also been found to efficiently work as random lasing materials. Affordable, large scale and easy processing procedures make organic building blocks, and their nanocomposites at the solid-state [13,22], especially sound for building low-cost and versatile devices, whose possible applications might span from speckle-free imaging schemes [23] to thermal [1], bio-chemical [24], and biological [25] sensing and diagnostics. In addition, some organic nanostructured systems, such as light-emitting polymer nanofibers [26], would allow unexplored aspects of mode condensation to be probed, since in quasi-one dimensional organic filaments a guided way of propagation for the lasing modes is added to diffusion and scattering.
In this work we analyze the spatial mode behavior in random lasing from disordered, non-woven networks of nanoparticle-doped polymer fibers based on a molecular active material with optical gain. This system undergoes lasing with very narrow emission peaks as typical of resonant feedback, and simultaneous presence of spatially extended modes, whose spreading might be supported through emitting regions connected by polymer nanofibers. Detailed space/spectrum cross-correlation  Electrospinning is performed in air at ambient conditions, and fibers are collected on quartz substrates placed on top of the metal collector surface. The setup used for SSCC is schematized in Fig. 1. A 250 µm spot from the Nd:YAG laser is obtained by an iris pin-hole, a subsequent 3.3× expansion with a telescope system and a focalization onto fibers (L1 lens in Fig. 1). Part of the excitation is redirected (BS1) and monitored in real-time by a pyroelectric detector (J3-09, Coherent). The radiation emitted by fibers is imaged onto the entrance slit of the monochromator with a 5× magnification through an objective lens (L2) and a focusing lens (L3). The collected radiation is imaged by the Peltier-cooled CCD with a 1024×256 array of 26×26 µm 2 , vertically-binned pixels allowing emission spectra to be collected for each coordinate along the direction defined by slit long axis. This leads to a resolution on the sample surface, parallel to the Y-direction, of about 10 µm (Y S coordinate, calculated by taking into account the magnification applied). Moving the L3 lens along the X direction perpendicular to the direction of the laser beam propagation and to the slit long axis (Y), with steps of 100 µm through a translation stage, leads to a resolution of 20 µm for the corresponding coordinate at the sample surface (X S ). The whole excitation field is reconstructed by the composition of 15 maps which are then shown by using contour lines as color-fill boundary. Image processing involves calculating a set of values by interpolating the slope (steepness) intensity between two adjacent pixels, which is in turn defined as the ratio of rise (difference in the intensity values) and run (pixel width). Spatial information is simultaneously acquired with far-field fluorescence imaging. To this aim a beam splitter (BS2 in Fig. 1) sends a part of the emitted radiation to a 405 nm longpass filter

Results and Discussion
Our devices are built by disordered samples of electrospun PS fibers doped with a Fl-Cz-Fl. The molecular structure of Fl-Cz-Fl is shown in Fig. 2(a). Fl-Cz-Fl dispersed in a PS matrix features a broad absorption band at 3.54 eV, a fluorescence spectrum with clearly resolved vibronic bands from 3.12 eV to 2.77 eV, and a corresponding high quantum yield of 0.86 [28]. In the nanocomposite mat, Fl-Cz-Fl serves as gain material, whereas additional scattering can be provided in the system by doping with TiO 2 nanoparticles. Electrified Fl-Cz-Fl/PS solution jets biased at 13 kV are used to deposit fibers on quartz substrates, which are in turn placed onto the metal collector.
The resulting morphology of a network of randomly oriented Fl-Cz-Fl/TiO 2 -doped PS fibers is shown in Fig. 2  The fibers have typical diameter ranging from 2 to 5 µm, and form a dense nonwoven mat. In the dark-field micrograph in Fig. 2(b), with illumination from the top fibrous surface and signal being collected in back-scattered configuration, the different focusing conditions highlight the 3D character of the network, and light-scattering can be appreciated from the bright spots decorating fibers along their length, which is given by incorporated particles. These are better inspected by STEM, unveiling clusters distributed in the organic filaments, with mutual distances from a few to many tens of µm, as displayed in Fig. 2(c). 9 cylindrical or ring-shaped cavities formed during fiber deposition, is characterized by well-defined and stable spectral features determined by the cavity geometry (i.e., fiber segment length or ring perimeter), here the overall number and intensity of the lasing modes is found to change from shot-to-shot, as displayed in Fig. 3 and in Fig. 4(a). In particular, in Fig. 3 we show two couples of single-shot emission spectra obtained with excitation fluence 17 mJ cm -2 and 42 mJ cm -2 , respectively. The inter-mode shot-toshot variability in the emission is not related to instabilities in the excitation fluence.
Indeed, fluctuations of the pumping laser are below 2% (evaluated as the ratio between standard deviation and average value) in the time interval typical of spectroscopic measurements (a few minutes), whereas the measured shot-to-shot intensity variations of the fiber emission is in the range 5-20%, similarly to other random laser systems [31]. These fluctuations have been observed in various systems at the solid state, both in resonant-feedback [4,16,32], and in intensity-feedback random lasers [13], and  A broader and less spiky spectrum is found by increasing the excitation area, attributable to a larger number of activated lasing modes. In fact, controlling the spatial shape of the excitation beam constitutes an effective method for tailoring the spectral features of the random lasing emission, and for eventually selecting specific modes by active control [33].
The spatial distribution of modes, and the SSCC [5] is studied in our work by recording the dispersed spectrum by a CCD array, with resolution on the sample emission plane of ~10 µm/pixel, in binned configuration along the detector vertical direction (Y, parallel to the slit long axis, as detailed in the Experimental Section). A typical map of the spectrally-and Y S -resolved emission is given in Fig. 5(a), showing that most of the emission comes from the inner part of the imaged region (Y S  310 µm  50 µm in Fig. 5(a)), and that brighter lasing spots are appreciable in the intensity map (e.g., at Y S  290 µm and 340 µm). The overall spectrum from the excited  nanofibers is obtained by the line-to-line sum of the intensity data from Fig. 5(a), performed for each single wavelength. This is shown in Fig. 5(b), where we highlight three sample wavelengths (  ,   ,   ) whose optical mode spatial profiles will be considered in the following. We then collect the emission from many vertical sections of the entire excited region, according to a detailed SSCC method [5]. This is carried out by translating the collecting lens (L3 in the scheme of Fig. 1) along the direction, composed along the X direction as a function of wavelength (Fig. 5(c)). Here X S is the Y S -perpendicular coordinate on the sample. The map evidences that most of the random lasing peaks are distributed along a few tens of µm along X S , which is of the order of the delocalization measured for conjugated polymer films [5]. Far-field fluorescence micrographs highlight unique features of random lasing nanofibers ( Fig. 5(d)). Following patterns much more complex than in clusters of nanoparticles [2,34], in random media with bubble structure [35] and in conjugated polymer films [5], emission is not found to occur from all the illuminated field of the 13 cavities in competition in the electrospun mat, and that emission from the lightemitting fibers is also detected well far away from the directly excited area, due to waveguiding of light along the organic filaments. In Fig. 5(d), the bright filaments are clearly visible due to outcoupling of waveguided photons from the lateral surface of fiber bodies, which is supported by Rayleigh scattering from surface roughness or embedded nanoparticles. As opposed to breaks at the tips of nanofiber waveguides [29,36], here the scatterers might redistribute a fraction of radiation also forwardly, namely along the longitudinal axis of the organic filaments. The self-absorption of the gain molecule, potentially hindering waveguiding, is below 5×10 2 cm -1 in our fibers due to the significant Stokes shift of Fl-Cz-Fl (about 0.55 eV). In principle, scattering of the incident UV excitation light by the electrospun fibers could also contribute to the broadening of the excitation laser spot when it impinges onto the fibrous sample [35]. However, given the micrometer size of the fibers (2-5 µm), which is an order of magnitude larger than the excitation wavelength (0.355 µm), such effect is not expected to be dominant in our material. For instance, calculating the angular dependence of the light scattering form factor, f(θ), for cylindrical polymer fibers [37], evidences that for ka>>1 (a is the fiber radius and k=2π/λ is the wavevector of the incident light) most of the incident light is scattered at forward angles < 60°. Instead, light scattering at angles around 90° (i.e. within the plane of the fiber network), which can contribute to spatial broadening of the excitation spot on the fibrous samples, becomes relevant for ka≈1, namely for fiber with size below 100 nm.
Integrating each (, Y S ) image (as that in Fig. 5(a)) over the whole width of the CCD (i.e., over wavelengths), and composing the resulting profiles obtained for each where the entirety of modes (i.e., at all wavelengths and across the whole emitting region) is compelled. Isolated cavities cannot be distinguished in Fig. 6(a) Here, different profiles are found, corresponding to differently shaped cavities, different degrees of delocalization within the excited region, and possibly different coupling extent for resonances across the complex material. The localization length of the mode at  i , ℓ loc-λi , defined as the length of the portion of the system in which the amplitude of the optical state differs appreciably from zero, can be estimated from each mode map, by the inverse participation ratio, IPR λi (how many sites the optical state is distributed over), as [10,38]: where ( , ) is the λ i -mode intensity. IPR λi are found to be below 10 -4 μm -2 , indicating spread states. The resulting localization lengths (155 µm, 123 µm, and 148 µm for   ,   , and   , respectively), much larger than in cluster of nanoparticles [4,10], lead to the conclusion that modes broadly extend over the network of fibers. The overall spatial broadening of these modes, partially overlapping in space, is of the order account the extra contribution of fibers to spectral selection, with some wavelengths possibly being better supported by waveguiding and formation of loops with relatively lower optical losses [34]. Hence, the corresponding modes might benefit from a higher lifetime, thus lasing from spectral regions where a relatively lower gain is measured for Fl-Cz-Fl/PS nanofibers. Overall, electrospun non-wovens are likely to sustain mode coupling and non-locality in the fiber random laser, selecting specific wavelengths from a background of uncorrelated and spatially-decoupled modes, which could be attributed to internal waveguiding along fibers [36].

Conclusions
In conclusion, we show electrospun fibers with optical gain and light-scattering properties, which features random lasing with sub-nm spectral width as well as modes with very high spatial extent, up to the 100 µm-scale. The emission patterns allow in principle the guiding behavior of the fibers to be linked with their spectral features, active cavities, and morphology. This is an interesting outlook for future experiments.
Many applications might be opened by coupling modes from random lasers, possibly with different degrees of localization, with other sites of the same complex material or with external receivers through efficient transmission channels as provided by integrated nanofibers. Platforms especially benefiting from such architectures would include chemical and biochemical sensors, which would exploit electrospun non-wovens with ultra-high surface-to-volume ratio to detect agents affecting random lasing emission, and promptly transmit information from extended optical modes to coupled detectors. Also, fibers could be used as input components triggering emission, by altering the oscillation frequency of spatially-defined regions in the disordered material [8] in a controlled way.