Polarized stimulated emission of 2 D ensembles of plasmonic nanolasers

been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1002/adpr.202000083. This article is protected by copyright. All rights reserved Polarized stimulated emission of 2D ensembles of plasmonic nanolasers Nikita Toropov 1 *, Aisylu Kamalieva 2 , Anton Starovoytov 1 , Sajid Zaki 3 , Tigran Vartanyan 1


Introduction
The extensive development of nanoscale coherent optical sources is the current trend of research in the field of nanophotonics, directed towards the development of photonic integrated circuits capable of augmenting today's integrated electronic circuits. [1] The creation of nanometersize lasers for the visible and infrared spectral ranges is impossible with traditional designs since the size of the resonator in a conventional laser cannot be smaller than its emission half-wavelength.
This problem has been addressed by Bergman and Stockman 2 who suggested the new feedback mechanism in which the usual mirrors are replaced with metal nanostructures supporting localized plasmon oscillations. Similar to lasers, these devices were initially called 'spasers' and later termed 'plasmonic nanolasers', [2][3][4][5][6][7][8][9] which have been in focus for the last ten years. [10] Plasmonic nanolasers and spasers are commonly confused, but these two terms are not synonymousspasers generate plasmons [2,11] while plasmonic nanolasers emit light [12] with wavelengths that are larger than their own sizes.
The plasmon-based sources overcome the diffraction limit, have faster (up to THz) switching capabilities in comparison to semiconductor counterparts, exploiting stimulated recombination, [13] limited to switching rates of the order of GHz. The spaser-based nanolaser is the smallest, biocompatible, and non-toxic tool, and can be incorporated in living tissues for a large number of biomedical in vivo and in vitro applications. [14][15][16][17][18] To date, several implementations of spasers and plasmonic nanolasers have been investigated: nanospheres, [12] nanodisks, [19] nanoflakes, [20] nanowires, [21] nanorods, [22] nanotori; [23] a topological spaser [11] and free electron excited spaser [24] were proposed. Organic dyes, [12] carbon nanostructures, [25] transition metal dichalcogenides, [26] as well as their hybrids [27,28] can act as a gain medium. The properties of subwavelength lasers are constantly improving; several groups have dealt with low-threshold for lasing, [29][30][31][32] multiwavelength lasing, [33] and high external quantum efficiency. [34] Despite the cogent results that spaser-based nanolaser have demonstrated, there are several negative responses to Accepted Article plasmonic nanolasers, particularly because the lasing can be interpreted as a stimulated emission in a random laser [35,36] or as an amplified spontaneous emission. [37][38][39][40][41][42][43] The first part of the experiment described in this article details plasmonic nanolasers and the reliability of different interpretations of their emission, as in conventional lasers. First, a lasing input-output characteristic has a threshold-like behavior. Second, it is characterized by a narrow emission band. Coherent optical sources can also have a distinguished polarization and directivity.
For plasmon nanostructures, possibly enhanced-spontaneous radiation in an amplifying medium can cause spectral narrowing and, as a result, can be misinterpreted as stimulated emission. The presence of the spectral narrowing and nonlinear input-output dependence is not enough to interpret the results as the manifestation of spaser-based lasing. Therefore, it is necessary to study at least one of the characteristics of radiation, such as polarization, and we are aiming to take this into account in the second part of the work, as well as an analysis of radiation patterns of the emission.
So, we propose plasmonic nanolasers based on a polymer-dye composite of 2D ensembles of silver nanoparticles where random lasing cannot be obtained. Silver nanoparticles in ensembles act as nanoresonators; a PMMA layer serves as a buffer layer preventing the uncontrolled fluorescence quenching of dye molecules by metals; dye molecules represent gain media. In contrast to previous studies, this work demonstrates a combination of the aforementioned lasing features. Thus, for the first time, the proposed 'plaser' implementation provides comprehensive evidence of the plasmonic nature of the stimulated emission.

Preparation of ensembles of silver nanoparticles with polymer-dye thin film
The laser dye coumarin 481 (C 14 H 14 F 3 NO 2 ) was chosen as the gain medium because it has a high quantum yield of fluorescence, which is important to compensate for dissipation losses in the metallic resonators. The absorption and fluorescence spectra of this dye are well-overlapped with the Ag NPs plasmon resonance spectrum, a key factor to the implementation of the plasmonic-Accepted Article based nanolaser. A coumarin dye layer was deposited on the surface of the samples with 2D ensembles of silver nanoparticles via the evaporation technique. The same layers were prepared on a clean quartz substrate to distinguish the NPs effect.
For the creation of 2D ensembles of plasmonic nanostructures, silver was chosen because among plasmonic materials in the spectral range investigated, it demonstrated minimal damping of plasmonic oscillations caused by losses. The physical vapor deposition technique in a vacuum chamber PVD-75 by 'Kurt J. Lesker' with a residual pressure level of ~10 -7 Torr was used to obtain samples as granular films consisting of silver nanoparticles, with the equivalent thickness of 10 nm on quartz substrates. After, the samples were additionally thermally annealed at 220 °C. More details on the experimental setup, growing of nanoparticles and micrographs were given previously in refs. [44,45] and in Supporting Information. Using these micrographs and measuring the concentration of dye molecules dissolved from the surface, we estimated the surface density of plasmonic nanoparticles which has an order of 10 10 cm -2 , their lateral sizes in average were of 60 nm in diameters with a big distribution, the thickness of dye layers, about 700 nm and the ratio of 2000 molecules per a nanoparticle. We considered that fluorophores located close to the metallic surface of NPs experience reducing or even are entirely quenched of their fluorescence. [46,47] Usually to minimize this effect, a dielectric layer can be used 47 . In this work we use a polymethyl methacrylate (PMMA) dielectric layer deposited on the surface of Ag NPs through the spin-coating technique. Following the work, [48] the thickness of PMMA layers was set to 10 nm.

Optical characterization of the prepared nanostructure
At first, all samples were characterized with absorption spectroscopy in the range of 300-700 nm.

Accepted Article
This article is protected by copyright. All rights reserved To register the input-output characteristics, the experimental scheme was assembled as schematically shown in Figure

Accepted Article
This article is protected by copyright. All rights reserved namely because at pump pulse energies larger than 2.4 mJ, the input-output characteristic goes into saturation. Contrary to this, for the thin coumarin film with Ag NPs the emission continues to grow after the threshold. Its growth rapidly becomes much faster than at lower pump. The same behavior of the input-output characteristics was observed at 436 nm excitation. However, the threshold in this case was almost 2 times lower (0.6 mJ) due to the effect of spectral coincidence of coumarin and silver plasmonic bands. It should be noted that in contrast to random lasers, the maximum intensity of the stimulated emission is independent of excitation wavelengths and remains constant, 491 nm.

Radiation pattern of the 2D ensembles of plasmonic nanoparticles with coumarin
The input-output characteristics plotted in Figure 2 were measured when the detector was positioned at 45 degrees from the substrate in the far field. For this, the detecting aperture was tilted at different angles from -60 to +60 degrees where the initial position of the detector was perpendicular to the exciting beam. The whole polar distribution of the angular variation of the intensity of the coumarin thin film with and without silver nanoparticles was not able to be measured because of weak signals, with the exception in this direction. According to well-known results, [49,50] the radiation pattern of a perpendicularly oriented dipole placed on a plane dielectric interface, is similar to that observed in our experiments. In particular, the absence of radiation in the direction parallel to the interface is in accord with the picture presented, by which the observed radiation is a sum of contributions of individual plasmonic nanolasers that act independently of each other. On the other hand, this observation contradicts all mechanisms that involve scattering of amplified radiation among many nanoparticles on a surface. Polarization measurements described in the next section further support the conclusion that plasmonic dipole sources are normally oriented to the surface.

Accepted Article
This article is protected by copyright. All rights reserved

Polarization of the stimulated emission
Another characteristic of the stimulated emission is polarization because spontaneous fluorescence is not polarized. We demonstrated above that the nonlinear dependence of the input-output characteristic and the spectral narrowing of the emission can be observed for plasmonic-based nanolasers. Similar results were analyzed in ref.
, [51] about realizing directional random lasing, but we do not observe the same phenomenon when the PMMA-coumarin composite is excited.
Moreover, it cannot demonstrate lasing since we observe the saturation of the emission.

Accepted Article
This article is protected by copyright. All rights reserved a) b) Figure 3. (a) Schematic representation of the polarization measurements (Ppolarizer) and (b) pairwise comparison of the emission of coumarin layers with Ag NPs at position 1 (blue spectra) and at position 2 (orange spectra) In accord with the previously described results, the emission intensity of the structure with Ag NPs increases linearly under threshold pump energies, and continues to grow at larger pump energies with even larger rates than at lower pump energies if the polarizer is in Position 2.
Contrary to that, if the polarizer is in Position 2, the fluorescence intensity at larger pump pulse energies goes into saturation. Analysis of these results shows that the stimulated emission radiation of the samples with Ag NPs are linearly polarized. That is to say, we show one more proof of the statement that the emission observed from the samples with silver nanoparticles is due to localized plasmon oscillations oriented perpendicular to the substrate surface.

Conclusion
We reliably demonstrated the stimulated emission of the 2D ensembles of plasmonic nanolasers, where the nanoscale layer of organic dye molecules acts as a gain medium and silver nanoparticles with sizes of ~10 nm arranged in the monolayer act as nanoresonators providing feedback. A set of stimulated emission features for these composites was ultimately measured and explained. The first signature of the stimulated origin of the emission is the nonlinear input-output characteristics obtained: at pump energies larger than 2.4 mJ a separate emission band over the wide fluorescence background appears in the spectra of the coumarin samples with silver nanoparticles; and grows faster than at lower pump energies. At the same time, the bare thin coumarin layer emission goes into saturation in the same range of pump energies. The second signature of the stimulated emission is the narrowness of its spectrum. The stimulated emission is linearly polarized in the scattering plane, while the fluorescence is unpolarizedthe third signature of stimulated emission. The threshold of lasing is lower at pumping, with wavelengths corresponding to a better overlap of

Accepted Article
This article is protected by copyright. All rights reserved molecular absorption bands and plasmonic oscillation frequencies. Finally, the radiation angular dependence shows the prominent direction of the stimulated radiation. Ultimately these features, combined with 2D geometry of the samples, negate interpretations of the emission based on random lasers or amplified spontaneous emission concepts. Thus, the results obtained put an end to many years of discussion about the nature of stimulated emission of the lasers with plasmonic nanostructures.

Supporting Information
Supporting Information is available from the Wiley Online Library.

Conflict of interest
The authors declare no conflict of interest.