1. Introduction

Random laser is a phenomenon discovered in various disordered media, which has attracted significantly theoretical and experimental attentions for its potential applications in sensing, imaging, lighting and medical diagnostics [1–3]. Unlike a conventional laser with a fixed resonator cavity, a random laser relies on multiple scattering of light in a disordered media to sustain an optical gain. Light is scattered in disordered media and simultaneously amplified by the gain medium, and random laser occurs when the total amplification becomes larger than the losses. Typically, amplification in a random laser is achieved by introducing a laser dye into a scattering structure, which codetermines the characteristic of the random lasers [4]. Based on different scattering regimes in disordered media, random laser has two distinct spectral classes: intensity feedback random laser and resonant feedback random laser. Intensity feedback random laser is characterized by a smooth, single-peaked emission spectrum and is explained by the diffusion framework neglecting the interference of light in random systems. However, resonant feedback random laser is characterized by an emission spectrum with sharp peaks, which is explained by framework involved with interference [5,6].

In the past two decades, random lasers have been observed in a variety of materials, including: semiconductors powder [7,8], polymer film [9,10], graphite [11] and metallic nanostructure [12–14]. Liquid crystals (LCs) have dual characters of extraordinarily large birefringence and extremely strong light scattering, and their molecular arrangement is very sensitive to electric field, magnetic field and temperature [15–18]. Therefore, LCs are of particular interest in random lasers. For example, a temperature-tunable random laser was first fabricated by Wiersma et al by using sintered glass and liquid crystal sample [2]. Strangi et al reported the random lasing and weak localization of light in dye-doped nematic liquid crystals for the first time [19]. An all-optically controllable nanoparticle random laser in a well-aligned laser-dye-doped liquid crystal has been realized recently [20]. Fiber, as a kind of high-efficiency and flexible light transmission material, was also introduced in random laser experiments. Recently, random fiber lasers based on Rayleigh scattering and stimulated Raman scattering were studied due to their tunable, narrow bandwidth and high power [21]. Random lasers on the end facet of an optical fiber based on a plasmonic gain channel were achieved [22]. Hu et al. realized coherent random fiber lasers based on the localized surface plasmon resonance of gold nanoparticles in liquid core optical fiber [23]. Random lasing in random network of nanofibers was also reported, which provides an alternative approach for random lasers fabrication by utilizing bio-degradable materials [24].

However, finding a facile and cost-effective strategy to achieve the modes and threshold control of the random lasers is still a challenge. We recently demonstrated the effect of Manganese (II) chloride (MnCl₂) on random lasing in dye-doped polymer-dispersed liquid crystal system [17]. In the previous report, we only compared the emission spectra change at different doping concentration of MnCl₂. Here, we presented the manipulation of random lasers threshold and mode by employing MnCl₂ with different concentration in dye Pyrromethene-597 (PM597)-doped negative liquid crystal (NLC) system. Moreover, the stability of the random systems provided by MnCl₂ was confirmed once again. A linear optical fiber was embedded in PM597-doped NLC without and with MnCl₂ glass capillary systems for the first time. We found that compared with the emission spectra from samples without fiber (see below), low-threshold and fewer-mode random lasing was obtained when fiber was embedded in the samples (see below). This is closely related to the variation of photon lifetime and the coupling degree of the modes to the environment provided by the fiber. The fiber is very flexible, which can also be doped in other random systems except the glass capillaries we used in the experiment. The simple and cost-effective method of controlling random lasers opens a new path for the manufacture of photonic devices based on random lasers.

2. Experiments

A systematic study was carried out regarding the random lasing performance for different random gain-scattering systems. Dye Pyrromethene-597 (PM597, from Exciton Ltd.) dissolved in methylbenzene is chosen as gain medium, whose concentration is 10⁻³ mol/L. Negative liquid crystals (NLC, $n_e=$1.5778, $n_o=$1.4833, $\Delta n=$0.0945, colorless transparent liquid) of average molecular weight (Mw)154, were purchased from Beijing Bayi Space LCD technology Co. Ltd. and used as received. MnCl₂ solution was prepared by dissolving MnCl₂▪4H₂O ($n=$1.985, $m=$6.150 g) in deionized water (1.5 mL) and kept at 80$°C$ for 15 minutes under magnetic stirring, which was added in the following samples when it cooled to room temperature. Dye-doped NLC (DDN) solution was prepared by mixing PM597 (0.2 mL) and NLC (1.5 mL) followed by stirring for 30 minutes. By adding 0.4 mL, 0.8 mL and 1.2 mL MnCl₂ solution in the as-prepared DDN solution with stirring for 30 minutes, three kinds of dye doped NLC and MnCl₂ (DDNM (0.4), DDNM (0.8) and DDNM (1.2)) solutions were obtained. Then the above DDN, DDNM (0.4), DDNM (0.8) and DDNM (1.2) solutions were shaken for 40 minutes by ultrasonic and poured into glass capillaries with 50 mm in length and 5 mm in diameter. A linear plastic optical fiber (POF) with a length of 50mm, core diameter of 120$\mu m$and cladding thickness of 55$\mu m$was embedded in the above DDN and DDNM (0.8) glass capillaries samples, consequently we fabricated DDN with fiber (DDNF) and DDNM (0.8) with fiber (DDNMF) samples. Figure 1(b) illustrated detailed operation process of embedding the fiber. All samples solutions are depicted in Table 1. Figure 1(c) plots SEM image for the cross section morphology of the POF.

 

Fig. 1 (a) Experiment setup for the random lasing measurement. (b) Schematic of random system. (c) The SEM image of the side view of the optical fiber. (d) Fluorescence spectrum of PM597; absorption spectra of MnCl₂ and PM597.

Download Full Size | PDF

Table 1. The ingredient of all the samples.

View Table

Beam from Nd: YAG laser ($\lambda$ = 532 nm, 10 HZ repetition frequency, 8 ns pulse duration) is divided into two parts through a polarizer, a half-wave plate and a beam splitter (see Fig. 1(a)). One is collected by an energy meter while the other is focused onto samples by the half-wave plate and a cylindrical lens. Emission from the sample’s surface is collected with a probe bundled into a spectrometer (with resolution 0.13 nm) to resolve the spectra. Pump position and detection position of the sample are shown in Fig. 1(b). Single-shot spectrum recorded by the spectrometer is obtained from 10 pump pulses. Spectra in Figs. 2(a), 2(b), 3(a), 4(a), 4(b), 5(a) and 5(b) are the ensemble-averaged emission spectra for a sum of single-shot spectrum (sum = 20).

 

Fig. 2 Emission spectra of (a) DDN and (b) DDNM (0.8) versus pump energies. Peak intensity and FWHM versus pump energies for (c) DDN and (d) DDNM (0.8).

Download Full Size | PDF

 

Fig. 3 (a) Emission spectra from DDNM (0.4), DDNM (0.8) and DDNM (1.2) recorded at the pump energy of 12.05$\mu J/c{m}^2$. (b) Peak intensity as functions of pump energies for DDNM (0.4), DDNM (0.8) and DDNM (1.2). Single-shot emission spectra for (c) DDN and (d) DDNM (0.8) while maintaining the pump conditions, where the pump energies for the DDN and DDNM (0.8) system are 26.05 and 11.96$\mu J/c{m}^2$, respectively.

Download Full Size | PDF

 

Fig. 4 The dependence of emission spectra for (a) DDNF and (b) DDNMF on pump energies. Peak intensity and FWHM as functions of pump energies for (c) DDNF and (d) DDNMF.

Download Full Size | PDF

 

Fig. 5 Emission spectra from (a) DDN, DDNF and (b) DDNM (0.8), DDNMF versus pump energies, where the pump energy for (a) and (b) is 20.79 and 11.96$\mu J/c{m}^2$, respectively. (c) The enlarged view of (a). (d) The ensemble-averaged Power Fourier Transform (PFT) curves for a sum of (sum = 50) single-shot emission spectra of DDNM (0.8) and DDNMF recorded at the pump energy of 11.96$\mu J/c{m}^2$.

Download Full Size | PDF

3. Results and discussion

Figures 2(a) and 2(b) depict the evolution of emission spectra versus pump energies for DDN and DDNM (0.8), respectively. For the DDN, a broad spontaneous emission spectrum centered at 591.50 nm with a FWHM (full width at half maximum) of 15 nm emerges at a pump energy of 14.94$\mu J/c{m}^2$. At the pump energy of 17.62$\mu J/c{m}^2$, a single narrow peak located in 591.3 nm with the FWHM of 2.6 nm suddenly appears. More sharp peaks occur as the pump energy increases, which is a signature of the resonant feedback random laser. This originates from the multiple scattering of NLC, which extends the travel-time of photons in the gain medium and ensures the gain greater than the loss to amplify light [14–17]. In Fig. 2(b), similar dependence of emission spectrum on pump energy as observed in Fig. 2(a) is found. Below the pump energy of 4.38$\mu J/c{m}^2$, only a broad spontaneous emission spectrum centered at 597.51 nm is observed. When the pump energy continues to increase (>4.38$\mu J/c{m}^2$), several sharp peaks (with the FWHM less than 1 nm) start to emerge. In fact, besides the multiple scattering of NLC, the resonant feedback random laser in the DDNM (0.8) system also attributes to the multiple scattering provided by the MnCl₂ molecules that not dissolved in the deionized water. Moreover, the MnCl₂ molecules vary the arrangement of the NLC molecules, which changes the photon path in the random system. Additionally, we find in Fig. 2(a) the position of the main peaks changes as the increase of the pump energy when the pump energy is lower than 23.44 $\mu J/c{m}^2$. However, the position of the main peaks in Fig. 2(b) is almost fixed as the increase of the pump energy. We think that the presence of the quasi-periodic spectrum in Fig. 2(b) may be related with the microcavity formed in the glass capillary [25,26].

The dependence of peak intensity and FWHM on pump energies for DDN and DDNM (0.8) is shown in Figs. 2(c) and 2(d), where an abrupt increase of emission intensity appears, indicating the threshold of the random systems. It is found that the threshold of the DDN decreases from 17.62 to 4.38$\mu J/c{m}^2$as doping MnCl₂ in it. We note that the absorption spectrum of MnCl₂ overlaps the fluorescence spectrum and absorption spectrum of PM597 (see Fig. 1(d)), which indicates the energy transfer between MnCl₂ and PM597. Energy from the PM597 molecules is transferred to the MnCl₂ molecules by radiative or radiationless energy transfer process [6,27,28]. Furthermore, the MnCl₂ molecules can directly absorb energy from the pump light. Thus, the localized electric field around MnCl₂ molecules is enhanced, which facilitates the reabsorption of the PM597 molecules and boosts the random lasing. So the energy transfer, together with the enhanced multiple scattering and arrangement change of NLC molecules induced by MnCl₂ molecules, significantly reduce the threshold of the DDNM (0.8) [15, 17].

We now focus on the lasing properties of dye doped NLC and MnCl₂ systems with MnCl₂ 0.4 mL, 0.8 mL and 1.2 mL (DDNM (0.4), DDNM (0.8) and DDNM (1.2)), respectively. Emission spectra from DDNM (0.4), DDNM (0.8) and DDNM (1.2) systems recorded at the pump energy of 12.05 $\mu J/c{m}^2$are depicted in Fig. 3(a), in which the emission intensity and the random lasing modes are different from each other. The DDNM (0.8) system shows more sharp peaks and higher emission intensity. Peak intensities for DDNM (0.4), DDNM (0.8) and DDNM (1.2) systems as functions of pump energies are described in Fig. 3(b). Similar to the DDNM (0.8) system (see Fig. 2(d)), the threshold behavior is again observed in DDNM (0.4) and DDNM (1.2) systems, with the thresholds of 15.04 and 7.39$\mu J/c{m}^2$, which are somewhat higher than that of the DDNM (0.8) system (4.38$\mu J/c{m}^2$). As changing the concentration of MnCl₂ in the dye-doped NLC system, the intermolecular distance between PM597 and MnCl₂ is varied, which will affect the energy transfer efficiency between MnCl₂ and PM597. Additionally, scattering and absorption in the random systems are varied when the concentration of MnCl₂ is increased. Thus, different scattering, absorption and energy transfer efficiency between MnCl₂ and PM597 in the random system determine the superior lasing efficiency and spectrum of DDNM (0.8) system in comparison with other two systems (DDNM (0.4) and DDNM (1.2)).

As we reported before, the doping of MnCl₂ in the random system reduces the number of coupling of lasing modes and stabilizes the system [17]. In a similar way, Figs. 3(c) and 3(d) plot the single-shot emission spectra from the DDN and DDNM (0.8) systems while maintaining the pumping conditions, where the pump energies for the former and latter system are 26.05 and 11.96$\mu J/c{m}^2$, respectively. As expected, obvious shot-to-shot fluctuation owing to a chaotic behavior of the emission spectra is presented in the DDN system (see Fig. 3(c)), in which a large number of modes are coupled. However, emission spectra from the DDNM (0.8) system do not show intense shot-to-shot fluctuation (see Fig. 3(d)), implying the decrease of the number of the coupling modes due to the doping of MnCl₂ [17].

By embedding a fiber in the DDN and DDNM (0.8) systems, two systems (DDNF and DDNMF) with fiber, scattering particles and gain materials were built (see Fig. 1(b)). Figures 4(a) and 4(b) show the effect of pump energies on emission spectra for DDNF and DDNMF systems, respectively. As observed in Figs. 2(a) and 2(b), both systems only manifest a broadband spontaneous emission spectrum at low pump energy. For the DDNF system, a single sharp peak located at 591.8 nm with the FWHM of about 0.3 nm appears above the broad spontaneous emission curve at the pump energy of 8.84$\mu J/c{m}^2$. Furthermore, the number of the sharp peaks increases with the increase of pump energy. Similar resonant feedback random laser actions at high pump energy larger than the pump energy of 3.78$\mu J/c{m}^2$is also shown in the DDNMF system, where the FWHM of some spikes are less than 0.3 nm. The corresponding dependence of peak intensity and FWHM on pump energies for DDNF and DDNMF systems is displayed in Figs. 4(c) and 4(d), where the both systems show clear threshold behavior. The thresholds of the DDNF and DDNMF systems are 8.73 and 3.78$\mu J/c{m}^2$, respectively. It is noteworthy that the thresholds of the DDNF and DDNMF systems are dramatically reduced in comparison with that of the DDN and DDNM (0.8) systems (17.62 and 4.38$\mu J/c{m}^2$, see Figs. 2(c) and 2(d)), which confirms the role of the fiber for the decrease of the threshold in the random gain system.

To interpret the experimental observation better, we take into account the threshold population inversion, which is determined by the balance of gain and loss in the random gain system [29]. The lower threshold of the systems with fiber (DDNF and DDNMF) in comparison with that of the systems without fiber (DDN and DDNM (0.8)) means the decrease of radiative loss and the increase of the dimension of active laser medium with the embedding of the fiber. For a random laser, the radiative loss is inversely proportional to the photon residence time in a random cavity. So we conclude that the DDNF and DDNMF systems have longer photon residence time (see Fig. 5(d)) compared with the DDN and DDNM (0.8) systems. In other words, the embedding of the fiber extends the motion path of the photons in the random gain systems.

To demonstrate the influence of fiber on random lasing modes, we record the emission spectra at the same pump energy for systems without and with fiber (see Figs. 5(a)-(c)). Figures 5(a) and 5(c) plots the emission spectra of DDN and DDNF at the pump energy of 20.79$\mu J/c{m}^2$, it is seen that the number of separated sharp peaks in the DDN is significantly reduced when fiber is embedded in it. Similar decrease of the random lasing modes is again observed in the DDNMF system (in comparison with DDNM (0.8)) at the pump energy of 11.96$\mu J/c{m}^2$ (see Fig. 5(b)). It suggests that emission profile can be easily adjusted by embedding fiber in a random gain-scattering sample. As it is reported by Fallert et al. that confined modes and extended modes are able to coexist in a same sample [4]. Balance of the gain, gain saturation and mode lifetime codetermine which modes are dominant. The decrease of the number of separated sharp peaks in the systems with fiber in comparison with that of the systems without fiber indicates the confined modes are dominant when fiber is embedded in the samples [5]. Since the confined modes have longer lifetime, this further confirms variation of photon residence time provided by the fiber. The increased mode lifetime implies variation of the degree of a mode coupling to its environment and results in different mode distributions.

To extract the variation of the photon residence time triggered by the fiber, the ensemble-averaged Power Fourier Transform (PFT) analysis of a sum (sum = 50) of single-shot emission spectra recorded at the pump energy of 11.96$\mu J/c{m}^2$for the DDNM (0.8) and DDNMF systems is shown in the Fig. 5(d). Fewer well-separated peaks are observed in the DDNMF system, which further verifies that the fiber reduces the number of lasing modes in the system. The length of the equivalent resonant cavity of a random laser is given by [17,30]

Finally, contour maps of shot-to-shot lasing spectra from systems without and with fiber upon different pump pulses are plotted in Figs. 6(a)-(d). Figures 6(a) and 6(c) corresponds to the DDN and DDNF systems when they are pumped by the energy of 20.79$\mu J/c{m}^2$, while Figs. 6(b) and 6(d) corresponds to the DDNM (0.8) and DDNMF systems with the pump energy of 11.96$\mu J/c{m}^2$. Clearly, the lasing modes of the random systems are dependent on the embedding of the fiber. More important, all the systems exhibit photobleaching effect upon shot-to-shot excitation due to the tendency of dyes to photodegrade under intense illumination [31–33]. However, it is found that the rate of photobleaching of the system with fiber is lower in contrast with that of the systems without fiber. The embedding of the fiber results in the formation of larger domains, thus resulting in smaller degradation rates [31–33]. On the other hand, strong localized effects related to the dominant confined modes (see Figs. 5(a) and 5(b)) freeze modes interactions in multimode laser [34]. The above effects lead to decrease of the rate of photobleaching in the system with fiber. Our results suggest a novel approach to designing random lasers properties by embedding a fiber in a random gain-scattering system.

 

Fig. 6 Contour map of shot-to-shot lasing spectra of (a) DDN, (c) DDNF, (b) DDNM (0.8), and (d) DDNMF obtained upon different pump pulses recorded at the pump energy of 20.79 (a and c) and 11.96$\mu J/c{m}^2$ (b and d), respectively.

Download Full Size | PDF

4. Conclusion

To conclude, we demonstrate a simple approach to adjust random laser performance by doping MnCl₂ or embedding a fiber in random glass capillaries systems with gain and scattering particles. First, the doping of MnCl₂ in the dye PM597-doped NLC system dramatically lowers the threshold of the system and enhances the stability of the systems. This is closely related to the energy transfer between the MnCl₂ molecules and the dye PM597 molecules. Moreover, the decrease of the threshold for the systems with MnCl₂ depends highly on the doping concentration of the MnCl₂. Second, emission spectra with lower threshold and fewer modes are observed in the system with fiber in contrast with that in the system without fiber. This attributes to longer photon residence time provided by the fiber. The longer photon residence time is verified by PFT analysis of emission spectra in the systems with and without fiber. Third, slower rate of photobleaching effects of lasing in the contour map of shot-to-shot lasing spectra for the system with fiber in comparison with that for the system without fiber further affirms the effect of the fiber on random lasing. This facile and cost-effective strategy to manipulate random lasers properties opens a new path for the manufacture of photonic devices based on random lasers.