1. Introduction

In a conventional laser, scattering of the gain medium in a cavity is deemed highly undesirable, since it usually results in huge loss even if not strong enough to prevent lasing action from occurring. However, if the scattering of a gain medium is strong enough that it traps the light traversing within the medium with a longer path, lasing action can occur without a cavity. This so-called random laser was smartly conceived by Letokhov just a few years after its conventional counterpart was invented [1]. Although random lasing action ubiquitous in the universe we are in for variety of wavelengths, its first demonstration in the laboratory setting on the earth by Lawandy et al came almost three decades behind [2]. In recent years, many materials were found feasible in designing and implementing random lasers, such as laser crystal powders [3], liquid crystal [4], translucent ceramics [5], and polymer thin films [6,7], among many others [8]. The appearance of various random lasers raised the hope for implementing nanolasers [8,9], especially when surface plasmon (SP) based lasers (spasers) were conceived and successfully demonstrated [10,11]. Very interestingly, visible SP supported 2D electron gases are formed at the interfaces of two oxides, one of which is of ferroelectrics and the other transparent conductive conductors [12,13]. Without any metal constituents, the SP supported platform is highly promising in plasmonics [14–16], specifically related to this work for designing spasers based random lasers. In our previous work, it was found the huge charge accumulation on the grain boundaries contributes to the random lasing and weak localization of light observed in powdered Nd³⁺ doped (Pb, La) (Zr, Ti) O₃ ceramics [17]. Actually, with different cutting surface of ferroelectric oxides, the strong spontaneous polarization can lead to a nanoscale positive or negative charge dual layers with high enough areal charge density on the surface which can support SP from visible to UV [17,18]. Refer to [12–16], the electron density accumulated near the surface of ferroelectrics to screen the polar field is really high enough to support visible SP, providing that enough free electrons exist in the highly doped samples. Very lately, it was found that random lasing can be enhanced greatly with SPs excited around the scatters [19,20], offering good opportunities in designing more compact nanolasers. In general, a plasmonic enhanced random laser system consists of metallic nanostructures in which SP was generated and laser dye as gain medium. In that work, random lasing threshold was controlled by altering the metallic-dielectric core-shell nanoparticles [21]. Resonant wavelength was tuned by adjusting the type of laser dye [22]. Moreover, directional random laser was realized by a cylindrical to turn the pumping light into a pump stripe [23].

As reported in [16], excitation of SPs in Fe doped lithium niobate (Fe:LN) can strengthen scattering dramatically. Therefore, once rare-earth doped LN is ground into powder, the SP strengthened scattering might lead to random lasing with lower threshold. In this work, Nd: LN is of ferroelectrics with large spontaneous polarization [24]. To screen the spontaneous polarization field, charges would accumulate on the grain boundaries of Nd: LN powder. Under illumination of pumping light at 808 nm, the SPs at frequency corresponding to 808 nm can be excited firstly owing to the roughness of the particles’ surface, and this leads to the initial strong down conversion emission and relatively weak upconversion emission, which can also excite SPs at longer and shorter wavelengths, respectively. Consequently, the SP strengthened scattering will ease random lasing action. The weak localization of light within the powdered system will result in spectral redistribution of light emission over time. To figure out the influence of SPs on random lasing, in this work, spontaneous emission spectra, dynamic curves, and slope plots were measured. To delve into the mechanism, a white light source was chosen as signal light and the variation of reflective light intensity was detected under different pumping powers. The obtained spectra and scatter plots confirmed that SP based random lasing action indeed occurred on the grain boundaries of powdered Nd: LN. This work opens up the feasibility of realizing plasmonic enhanced spasers based random lasers in ferroelectric oxide materials.

2. Experimental results

To study plasmonically enhanced random lasing in Nd: LN crystal, some powdered Nd: LN samples were prepared. The particle diameters of the powdered Nd: LN varied from about 20 µm to 70 µm, to which 1.5 wt% Nd³⁺ ions were introduced in the forms of Nd₂O₃. For comparison, the samples were coated in half with ZnSe thin film with electron beam evaporation (Model LJ-550E, from LJ-UHV Technology, Inc.) under a condition of accelerating voltage at 8.0 kV and substrate holder rotary speed at 20 rpm, and deposition rate at 3.0 Å/s. Coated ZnSe film was characterized as 134 nm with a stylus profilometer (Model XP-100, Ambios Technology, Inc.)

Our experimental setup is exhibited in Fig. 1(a), the Nd: LN powder was fixed on a transparent glass slide with epoxy glue which can withstand high temperature for ease in pumping from broad incident angle. A continuous pumping light which was centered at 808 nm, with 6.0 nm bandwidth from a diode laser (Changchun New Industries, MDL-N-808) was focused by a convex lens of 63 mm in focal length on the specimen to excite SP and hence enhance random lasing action. The beam diameter on the specimen surface was 0.8 mm. The emission was detected by a UV-visible-NIR spectrometer (Ocean Optics, HR4000CG-UV-NIR) in the direction perpendicular to the specimen. The angle between the pumping light and normal of the specimen (toward to spectrometer) was set at 35°. An 808 nm notch filter (with 20 nm bandwidth) was placed in front of the detector to prevent the pumping light from entering the spectrometer. Under pumping power at 0.12 W, only two down emission bands centered at 887 and 900 nm were observed, as shown in Fig. 1(b). When pumping power was above 0.12 W, as the pumping laser was turned on, down conversion emissions centered at 876, 887, 900, 913 and 924 nm emerged simultaneously and their intensities raised with increasing pumping power, corresponding to the energy level transition from ⁴F_3/2 to ⁴I_9/2 of Nd³⁺ ions. Five peaks were observed owing to the Stark level splitting of ⁴I_9/2. For comparison, the spectrum of the other half of the specimen without ZnSe film was measured, as shown along with that from the ZnSe coated half in Fig. 1(c). The emission intensity of the half of the specimen without ZnSe film was stronger than that coated with ZnSe film, partly because the down conversion emission was generated directly by Nd³⁺ ions. For the part of the specimen coated with ZnSe film, a portion of the pumping light was blocked by ZnSe film, resulting in reduction of pumping light reaching to Nd³⁺ ions. In addition, the light emission from particles can be cut lower in passing through the ZnSe/LN interfaces, not excluding the additional loss caused by the strengthened SP induced scattering due to ZnSe coating on the LN particles. Refer to [14] and [16], the reflective light was dramatically reduced owing to SPs’ excitation. The forward scattering near the normal could be strengthened greatly [16]. This is desirable in realizing random lasing. It is also expected that weak localization [17,25–28] comes into play when scattering is strong enough. In fact, the strong SPs induced scattering is based on photorefractive (PR) phase gratings [29]. The shorter is the wavelength, the stronger is the PR effect. The inset of Fig. 1(c) shows the spectral comparison of powdered Nd: LN and Nd: LN coated with ZnSe film from 808 nm to 858 nm. The spectra of powdered Nd: LN were normalized to make their spectral difference clearly. It is obvious that Nd: LN coated with ZnSe film had a blue shift of 3 nm, which remains elusive so far. To get a whole picture regarding the photoluminescence spectrum redistribution, we have to turn our attention toward upconversion emission, which falls in the visible band.

 

Fig. 1. (a) Schematic of experimental setup in studying plasmonic enhanced random lasing action in powdered Nd: LN coated with ZnSe film; (b) Down conversion emission obtained from the specimen at θ=35° under different pumping powers; (c) Comparison of down conversion emission from powdered Nd: LN with/without ZnSe coated. The inset is a magnified image of (c) from 808 nm to 858 nm.

Download Full Size | PDF

Considering the upconversion emission intensity would be orders of magnitude weaker than the down conversion emission, it was necessary to replace the 808 nm notch filter with a 700 nm short pass filter to insure both the pumping light and down conversion emission were blocked. With increasing pumping power, upconversion emission emerged gradually and grew stronger afterwards, as seen in Fig. 2(a). Four emission peaks centered at 526, 536, 551, and 559 nm were simultaneously observed, assigned to the transitions from ⁴G_9/2 to ⁴I_11/2 (526 and 536 nm) and ⁴G_7/2 to ⁴I_9/2 (551 and 559 nm). Besides, emission bands centered at 659, 666, 674, and 698 nm were also observed with increasing pumping power. For clarity, one spectrum under pumping power at 4.69 W from 645 nm to 710 nm was shown in the right inset of Fig. 2(a). It can be seen that the emission bands were nearly featureless, with small peaks centered at 659, 666, 674, and 698 nm, whose intensities were much weaker.

 

Fig. 2. (a) Multiwavelength laser emission spectra under different pumping powers obtained from powdered Nd: LN. The inset left is the photograph taken from the surface of the specimen. The inset right is a magnified image of (a) from 645 nm to 710 nm under pumping power at 4.69 W; (b) Multiwavelength laser emission spectra obtained from the melted point. The inset is the photograph taken from the melted point on the specimen; (c) Energy level diagram of Nd: LN crystal.

Download Full Size | PDF

One intriguing phenomenon was observed when we continued to raise the pumping power. Under pumping power at 7.10 W, the specimen was melted locally and the emission color was transformed apparently from originally greenish (from 521 to 562 nm) to yellowish (from 573 to 615 nm). Different emission bands centered at 599, 632, and 686 nm emerged, corresponding to the transitions from ⁴G_7/2 to ⁴I_11/2, ²H2_11/2 to ⁴I_9/2, and ⁴G_7/2 to ⁴I_13/2, respectively. Compared to the peaks centered at 660, 667, 675, and 686 nm after the sample was melted (refer to Fig. 2(b)), it was found that the peak at 686 nm did not exist before the sample was melted, while the peaks at 660, 667, and 675 nm existed before the sample was melted (considering the measurement error was ± 1 nm). The morphology of the pumping spot before and after melting was checked with a microscope and is exhibited in the inset of Figs. 2(a) and 2(b). Before local melting, the size of Nd: LN particles ranged from 20 µm to 70 µm. Once it was melted, the powder reunited into polycrystal form with no individual particles seen by naked eyes. The energy level diagram corresponding to the upconversion emission of Nd³⁺ ions is exhibited in Fig. 2(c). Major upconversion emission peaks transformed from 526, 536, 551, and 559 nm into 599 and 686 nm after melting. This transformation is related to change of crystal structure in powdered Nd: LN, as is confirmed by the following experiments.

To further investigate the mechanism, typical dynamic curves corresponding to the emissions centered at 526, 551, 1083, and 887 nm under pumping power at 4.75 W were measured, as shown in Fig. 3(a). The emission peak at 1083 nm was assigned to the energy level transition from ⁴F_3/2 to ⁴I_9/2. As the pumping light was turned on, the emission peak at 887 nm emerged immediately, but followed by a rapid drop and meanwhile the emissions centered at 526, 551, and 1083 nm were being built up, as shown in the set of Fig. 3(a). Soon afterwards (about 0.2 s), the emission peaks at 551 and 1083 nm began to decline while the 526 nm emission peak was on the rise until it reached a plateau. During the process, the energy of emission peak at 887 nm transferred to emissions centered at 526, 551, and 1083 nm, then the energy of emissions centered at 551 and 1083 nm transferred to the emission peak at 526 nm, revealing a typical upconversion luminescence. In fact, the rising time for the upconversion emissions and the decaying time for the down conversion emissions were both in milliseconds. However, in our experiment, because the upconversion emission was relatively weak and the spectrometer system used in this work was relatively low in sensitivity, we had to set longer integrating time (0.2 s) to get sizable signal with a better signal to noise ratio. As a result, the dynamic curves shown in Fig. 3(a) were presented in time scale of seconds. Dynamic curves of laser emission intensities under the pumping power from 3.20 W to 7.40 W, with an increment 0.30 W, are shown in Fig. 3(b). Under pumping power at 6.50 W, the emissions centered at 551 and 1083 nm began to decline. The emission peak at 526 nm followed this trend under higher pumping power at 7.10 W. A logarithmic plot of emission powers at 526, 551, and 1083 nm versus pumping power is shown in Fig. 3(c). Intensity of 526 nm was the weakest in the three emission bands at first, while it exceeded the intensity of 1083 nm at 4.10 W and exceeded 551 nm at 6.80 W. The slopes of three fitting lines are 3.90, 2.90, and 2.77, respectively. Compared to their initial slopes 2.50, 1.34, 1.27 exhibiting normal upconversion emission, the enhanced slopes revealed more complex multiphoton process [30]. When the slopes increased, the efficiency in generating upconversion photons was raised. In this case, the photons of different frequencies were trapped longer time in the illuminated area, consequently, the effective path length of the upconversion photons was increased, resulting in random lasing action. The increased path length raises the interaction length of the photons of different frequencies, therefore, the multiphoton processes become more complex. Two thresholds were indicated in Fig. 3(c). The first threshold at 4.10 W apparently corresponds to commencement of random lasing action [8,9] and the second threshold at 6.80 W corresponds to dominance of weak localization of light [25–28], as indicated in Fig. 3(c). To verify the occurrence of weak localization after the second threshold, angular distribution of upconversion emission at 526 nm under pumping power at 7.00 W was measured, as shown in Fig. 3(d). The obtained figure with half sharp peak centered at the zero backscattered angle was regarded as a typical feature of weak localization of light [25–28]. When weak localization of light dominates, the photons can be trapped within the gain medium for a longer time, which increases probability of being absorbed. This explains why the light emission was reduced. When more and more photons were absorbed, more heat was generated, leading to melting of the sample ultimately. Therefore, the angular distribution could serve as an indication of the occurrence of weak localization of light after the second threshold.

 

Fig. 3. (a) Dynamics of upconversion emission intensities upon turning-on the pumping light at 4.75 W. The inset is a magnified image of (a) from 1 s to 2 s; (b) Dynamics of upconversion emission intensities with increasing pumping power from 3.20 W to 7.10 W; (c) Slopes of upconversion emissions at 526, 551, and 1083 nm; (d) Angular distribution of upconversion emission at 526 nm.

Download Full Size | PDF

To figure out the changes in luminous properties of melted Nd: LN, dynamic curves of melted Nd: LN corresponding to the emissions centered at 522, 591, 660, and 686 nm under pumping power from 2.25 W to 6.15 W, with an increment 0.30 W, were measured and are shown in Fig. 4(a). The inset of Fig. 4(a) shows the dynamic curves under pumping power at 5.25 W. The emission peak at 887 nm emerged as soon as the pumping light was turned on, but decreased rapidly with increasing emission peaks at 591, 686, 660, and 1060 nm. After 6.8 s, the 1060 nm emission peak began to decline, while the other emission peaks kept going up until their intensities reached a plateau. The emission peak at 591 nm emerged quite late and much weaker under lower pumping power, yet it rose pretty quickly and finally exceeded the intensity of 686 nm. All the intensities of emission peaks increased with the pumping power except the emission peak at 1060 nm. A logarithmic plot of emission intensities versus pumping power related to emission peaks at 591, 660, and 686 nm is shown in Fig. 4(b). The slopes of the emissions centered at 591, 660, and 686 nm are 4.35, 4.88, and 2.75, respectively. Compared to the slopes before powdered Nd: LN was melted, their slopes indicated that another type of random lasing action occurred. One hypothesis is the lattice structure of Nd: LN transformed from single crystal to polycrystal. To verify this hypothesis, the down conversion emission peak at 887 nm before and after powdered Nd: LN was melted was measured, as shown in Fig. 4(c). The specimen was pumped at 5.39 W under the same condition. After its melting, the Stark levels cannot be resolved because the crystal field differs from grain to grain. One can see that among the original five sharp peaks, only two sharp peaks on the shorter wave side remained and the rest peaks were fused into a featureless one with much lower intensity, even if they both ranged from 872 nm to 947 nm.

 

Fig. 4. (a) Dynamics of upconversion emission upon turning-on the pumping light with increasing pumping power from 2.25 W to 6.15 W. The inset is the dynamics of upconversion emission under pumping power at 5.25 W; (b) Slopes of emissions at 591, 660, and 686 nm; (c) Down conversion emission before and after the specimen was melted at 5.39 W.

Download Full Size | PDF

The spectral redistribution can originate from shift of the gain spectrum in the melted sample due to changes in the local environment of the rare-earth ion. Consequently, some of the decay rates were reduced and others were increased. Because the weak localization of light played an important role in the Nd: LN system, its contribution to the decaying rate changes cannot be excluded.

3. Analysis and discussion

To figure out the transformation of random lasing action in powdered Nd: LN, lattice structure of Nd: LN was illustrated, as shown in Fig. 5(a). The lattice structure of Nd: LN is described by space group $C_{3v}^6$, where the lithium atom lies above an oxygen layer and the niobate atom lies 3.008 Å away from lithium atom [31]. It is widely accepted that in LN single crystal Nd³⁺ ions are incorporated mostly by occupying the sites where the lithium atoms reside because the Li-O bonds are weaker than the Nb-O bonds. Totally, there are three types of non-equivalent Nd³⁺ ions that have established in the LiNbO₃ crystal distinguished by the distance between Nd³⁺ ion and original lithium atom ranging from 0.35 Å to 0.45 Å (0.4 Å shown in Fig. 5(a)), resulting in different types of charge-compensating defects. To estimate whether it is possible for random lasing and weak localization of light to occur in our system, it is necessary to calculate the scattering mean free path ${l_s}$ and the scattering cross section ${C_{sca}}$ . Generally, the scattering cross section is presented as

 

Fig. 5. (a) Lattice structure of Nd: LN in the ferroelectric phase; (b) Absorption spectrum of a bulk Nd: LN slab between 500 and 900 nm.

Download Full Size | PDF

In LN crystal, 1.5 wt% Nd³⁺ ions were introduced in the forms of Nd₂O₃, resulting in three types of non-equivalent Nd³⁺ ions in the LiNbO₃ crystal distinguished by the distance between Nd³⁺ ion and original lithium atom, corresponding to three types of charge-compensating defects. Electrons or holes could be trapped in these defects. Once pumped, the electrons or holes trapped in the defects could serve as scattering centers, which is vital for random lasing and weak localization to occur. The defect concentration in this work is 2.95×10²⁰ cm^-3. Then the shortest scattering mean free path ${l_s}$, only decided by the average defect site distance, is about 1.50 nm, provided that all defects can trap charge carriers to serve as scattering centers. In Fig. 3(d), the half angular width is 32.4°, then the transport mean free path ${l_t} = \frac{{{l_s}}}{{1 - \cos \varPhi }}$ is estimated to be 2.61 nm, $\varPhi $ is the angular width. The scattering cross section under pumping power at 7.00 W is 1.72×10⁻¹⁸ cm². However, in an actual case, only a small portion of defects can be occupied by charge carriers, then the average scattering mean free path should be much larger than this number. Theoretically, the angular width of the backscattering cone is predicted to be in order of ${\raise0.7ex\hbox{$\lambda $} \!\mathord{\left/ {\vphantom {\lambda {{l_t}}}} \right.}\!\lower0.7ex\hbox{${{l_t}}$}}$, where $\lambda $ is the wavelength. As shown in Fig. 3(d), the wavelength is 526 nm and the half angular width is 32.4°, corresponding to a transport mean free path ∼ 4.65×10⁻⁷ m, which is about 1.7 times smaller than the pumping light centered at 808 nm. This estimation suggests that it is possible for random lasing and weak localization to occur in our system. Furthermore, comparing with ${l_s}$ estimated by the defect concentration, it is easy to conclude only about 0.3% defects resulting from Nd³⁺ ions can trap electrons or holes and serve as scattering centers. In this case, the scattering cross section is 7.80×10⁻¹⁵ cm².

Figure 5(b) shows the absorption spectrum of bulk Nd: LN slab from 500 to 900 nm in wavelength. Major absorption bands were centered at 599, 752, and 813 nm. In our work, 808 nm was chosen as the pumping wavelength, close enough to the absorption band at 813 nm, considering the laser bandwidth of 6 nm.

In order to compare the variation of lattice structure before and after locally melting in powdered Nd: LN specimen, photoluminescence spectra of bulk Nd: LN slab and Nd³⁺ doped (Pb, La) (Zr, Ti) O₃ (Nd: PLZT) ceramics powder were studied. The Nd: PLZT powder from which the specimens were prepared consisted of 65 mol. % lead zirconate plus 35 mol. % lead titanate and 10 mol. % lanthanum, to which 1.5 mol. % Nd³⁺ ions were introduced in the forms of Nd₂O₃, respectively. In Fig. 6(a), it is obvious that the spectrum of bulk Nd: LN overlapped that of powdered Nd: LN well with regard to the emission peaks at 526, 536, 551, and 559 nm but dozens of times weaker in terms of intensity, indicating that surface-volume ratio had great influence on photoluminescence intensity. Noticeably, the emission spectrum of melted Nd: LN was very similar to that of powdered Nd: PLZT, as seen in Fig. 6(b). The Nd: LN and Nd: PLZT were pumped at 4.20 W and 2.05 W, respectively. The emission peak of melted Nd: LN at 590 nm overlapped with that of Nd: PLZT while the emission peak at 686 nm exhibited a blueshift of 4 nm in Nd: PLZT. This similarity indicates, once again, after powdered Nd: LN was melted, it became a polycrystal which contains abundant charge-compensating defects. In a polycrystal, such as Nd: PLZT, random lasing is caused by photoinduced multiply scattering and optoenergy storage process resulting from vacancies and defects [30]. In short, the lattice structure of powdered Nd: LN transforms from single crystal to polycrystal after melting, leading to the variation in upconversion, down conversion spectra and their dynamic curves. These transformation characteristics could serve as indications to evaluate the lattice stability of Nd: LN at different temperatures.

 

Fig. 6. (a) Upconversion emission spectra obtained from powdered Nd: LN and bulk Nd: LN slab at 5.01 W and 3.59 W; (b) Upconversion emission spectra obtained from melted Nd: LN and powdered Nd: PLZT ceramics at 4.20 W and 2.05 W.

Download Full Size | PDF

It is well-known that random lasing action is caused by strong scattering in the gain medium. To understand the underlying mechanism of random lasing and weak localization of light in the powdered Nd: LN specimen before and after melting, firstly we have to take a close look at particle surfaces or grain boundaries [24]. According to ab initio theory, for a Z-cut LN slab, its positive Z-face is with as high as ${\sigma _z} = 0.7C/{m^2}$ spontaneous polarization charge density [24]. And for a Y-cut sample, its Y-cut plane, though nominally nonpolar, has areal charge density ${\sigma _y} = {\sigma _z}/3 = 0.23 \,C/{m^2}$ [24]. Here ${\sigma _y}$ and ${\sigma _z}$ are areal charge densities for Y- and Z-planes, respectively. These areal charge densities were consistent with earlier theoretical results [32] and with that obtained experimentally [33,34]. When Nd: LN single crystal was ground into particles, some portion of the particles’ surface was equivalent to Z-cut, some portion was equivalent to Y-cut, some portion was equivalent to X-cut, yet most portion was in between of these typical surfaces. Because highly doped with Nd element, the rich free electrons and holes can drift near the particle’s surface to screen the polar field, since the entire system tended to be electronically neutral macroscopically. Therefore, in the portion of surfaces close to positive Z-cut or Y-cut face, 2D electron gas was formed which is of visible SPs supportive [12,13]. When the down conversion and upconversion light impinged on the related area, SPs can be excited, and associated light scattering was definitely one of the major contributors to the random lasing and weak localization of light observed in our material system. In addition, the strong scattering resulted from vacancies and defects in its polycrystalline structure also made important contribution to those described above [30]. We analyze the influence of surface electron accumulation briefly as following. In general, LN is one of well-known ferroelectric oxides [24]. Below its Curie temperature, the cation in the LN unit cell does not locate exactly in the oxygen plane or the octahedron center. As a result, charge separation occurs and hence LN exhibits non-zero polar charges due to strong spontaneous polarization (refer to Fig. 7).

 

Fig. 7. Schematic diagram of spontaneous polarization in powdered LN single crystal.

Download Full Size | PDF

As stated above, different portions of the particles’ surface are with different densities of positive or negative polar charges. To screen the fields of these polar charges, electrons or holes are accumulated near the polar charge in a nanoscale range, forming 2D electron gases or 2D hole gases (nanoscale positive or negative charge dual layer in nature). Concerning SPs’ excitation, only 2D electron gases need to considered. According to Drude Model [18]:

In fact, charge accumulation on grain boundaries and its great impact on the ferroelectric ceramics (polycrystals in nature) were discussed two decades ago [35]. Lately, 2D electron gases formed at the interface of total insulating oxides due to polar catastrophe [36] are investigated intensely to circumvent hinder of huge loss associated with metal uses in conventional plasmonics. The nanometric 2D electron gases can be modulated by applying an external electric field on MOS structure [37]. Over index refraction change measured experimentally can serve as support to our physical picture used in explaining the SP based random lasing and weak localization of light in our work.

To confirm that SPs were really excited in the accumulative 2D electron gases on the grain boundaries and strong photoinduced scattering indeed played an important role, a straightforward experiment was performed as follows. The reflection spectrum versus pumping power with a white light source and spectrometer was measured. As shown in Fig. 8(a), a white light source was focused by a converging lens (20 mm in focal length) on the powdered Nd: LN sample, and the reflective light was collected by a spectrometer. The incident spot (beam size 0.3 mm in diameter) was totally covered by the pumping light (beam size 0.6 mm in diameter) to make sure the variation of the reflection spectrum was induced by the pumping light. The angle between the white light source and reflective light was set at 65°, and the angle between pumping laser and white light source was set at 30°. The reflection spectra of the seeding white light source versus pumping power are exhibited in Fig. 8(b). Notice that the reflection spectrum in the entire visible band was reduced remarkably with increasing pumping power. This reflection reduction was obviously due to the enhancement of scattering. The accumulated charge on the grain boundaries can serve as scattering centers, accompanied with change in refractive index. Once pumping laser was incident to this area, scattering would be enhanced remarkably and reflective intensity would be reduced simultaneously. Figure 8(c) shows the change in intensities at 580, 630, and 660 nm with increasing pumping power. Intriguing three step-like reduction as well as two thresholds was observed as the pumping power increased. It is believed that after the first threshold at 2.32 W, random lasing action occurred [8,9] and after the second threshold at 4.72 W, weak localization [25–28] began to play a dominant role.

 

Fig. 8. (a) Experimental setup in studying reflection spectrum versus pumping power; (b) Reflection spectra of the seeding white light source under different pumping powers; (c) Reflective intensities at 580, 630, and 660 nm with increasing pumping power.

Download Full Size | PDF

The above experimental results hinted that somehow the reflective intensity of powdered Nd: LN specimen was reduced in a broadband with increasing pumping power. In other words, the effective refractive index difference at air/LN interface was lessened with increasing pumping power. This confirms that SPs were really excited in the accumulative 2D electron gases on the particle’s boundaries and they played key roles in the random lasing action and weak localization of light observed in powdered specimen. In this solid system, the SPs and gain functions were integrated together without any extra foreign metal constituents introduced. This is highly desirable in designing spaser [10] based random lasing system, which possesses great potential in making compact light source, since SP can reduce cavity length by slowing down the phase velocity, i.e., shortening the wavelength. Moreover, the LN crystal is well known as “silicon in photonics” [38] and one of the best nonlinear optical materials. The SPs’ involvement offers a very good opportunity in linking conventional photonics to nanophotonics for natural compatibility.

4. Conclusions

In conclusion, random lasing action and weak localization of light were evidenced in powdered Nd: LN both before and after its photoinduced melting locally. Based on strong spontaneous polarization field in powdered Nd: LN and electron accumulation owing to the screening effect, 2D electron gases can be formed at air/LN interfaces on the particle surfaces. According to Drude model, SPs of visible and near-IR were indeed excited on the grain boundaries. Once SPs were excited, the refractive index around SPP in the specimen would be raised in orders of magnitude. The SPs based plenty of scattering centers thus formed and according to the concentration of scattering centers, the scattering mean free path and the scattering cross section were estimated, revealing it is possible for random lasing action and weak localization of light to occur. observed, which was consistent with the three step-like reflection intensity change obtained experimentally. Moreover, the transformation of the photoluminescence witnessed before and after melting was originated from changes in the local environment of the rare-earth ion and lattice structure variation. This work presents an alternative material system in designing spaser based random lasers and an approach to estimating the stability of crystal structure by monitoring the random lasing features.