Tailoring Photoluminescence from MoS2 Monolayers by Mie

Mar 5, 2019 - We experimentally investigate coupling of the photoluminescence (PL) from monolayers of MoS2 to Mie-resonant metasurfaces consisting of ...
0 downloads 0 Views 6MB Size
Subscriber access provided by Washington University | Libraries

Article 2

Tailoring photoluminescence from MoS monolayers by Mie-resonant metasurfaces Tobias Bucher, Aleksandr Vaskin, Rajeshkumar Mupparapu, Franz J. F. Löchner, Antony George, Katie E. Chong, Stefan Fasold, Christof Neumann, Duk-Yong Choi, Falk Eilenberger, Frank Setzpfandt, Yuri S. Kivshar, Thomas Pertsch, Andrey Turchanin, and Isabelle Staude ACS Photonics, Just Accepted Manuscript • DOI: 10.1021/acsphotonics.8b01771 • Publication Date (Web): 05 Mar 2019 Downloaded from http://pubs.acs.org on March 6, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

Tailoring photoluminescence from MoS2 monolayers by Mie-resonant metasurfaces Tobias Bucher,∗,†,‡ Aleksandr Vaskin,†,‡ Rajeshkumar Mupparapu,†,‡ Franz J. F. Löchner,†,‡ Antony George,¶,‡ Katie E. Chong,§ Stefan Fasold,†,‡ Christof Neumann,¶ Duk-Yong Choi,k Falk Eilenberger,†,‡,⊥,# Frank Setzpfandt,†,‡ Yuri S. Kivshar,§ Thomas Pertsch,†,‡,⊥,# Andrey Turchanin,¶,‡,@ and Isabelle Staude†,‡ †Institute of Applied Physics, Friedrich Schiller University Jena, 07745 Jena, Germany ‡Abbe Center of Photonics, Friedrich Schiller University Jena, 07745 Jena, Germany ¶Institute of Physical Chemistry, Friedrich Schiller University Jena, 07743 Jena, Germany §Nonlinear Physics Centre, Research School of Physics and Engineering, The Australian National University, Canberra, ACT 2601, Australia kLaser Physics Centre, Research School of Physics and Engineering, The Australian National University, Canberra, ACT 2601, Australia ⊥Fraunhofer Institute for Applied Optics and Precision Engineering IOF, 07745 Jena, Germany #Max Planck School of Photonics, Germany @Jena Center for Soft Matter (JCSM), 07743 Jena, Germany E-mail: [email protected]

Abstract We experimentally investigate coupling of the photoluminescence (PL) from monolayers of MoS2 to Mie-resonant metasurfaces consisting of silicon nanocylinders. By

1

ACS Paragon Plus Environment

ACS Photonics 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 23

a systematic variation of the nanocylinder diameter, we sweep the metasurface resonances over the excitonic emission band of monolayer MoS2 . We observe strong enhancement, as well as spectral and directional reshaping of the emission. By a comprehensive optical characterization we unveil the different physical factors including electronic, photonic, and mechanical influences responsible for the observed PL changes. Importantly, we show that by geometrical tuning of the nanocylinder resonances the emission can be tailored from occuring under very large angles to being directed out of the substrate plane. Our results highlight the need and potential of controlling not only the photonic but also electronic and mechanical environmental factors for tailoring PL from TMD monolayers by integrating them in nanophotonic architectures.

Keywords Light-emitting metasurfaces, Mie-resonances, dielectric nanoantennas, transition metal dichalcogenides, 2D materials, excitonic emission Van-der-Waals stacked transition metal dichalcogenide (TMD) crystals attracted massive research attention over the past years due to their unique physical properties when thinned down to a few- or monolayer sheet. 1–4 In the few-layer regime, their mechanical, electronic, and optical properties not only remarkably deviate from the properties of respective bulk materials but also strongly depend on the number of layers forming the crystal stack. Importantly, it was shown theoretically and experimentally that some TMDs (MX2 , where M = Mo, W and X = S, Se) transition from indirect to direct band gap semiconductors when approaching the monolayer phase 5,6 , show strong excitonic effects 7–9 as well as large spin-orbit splitting 10 , and exhibit circular dichroism at the K/K’ valleys 11–14 . These unique properties make TMDs versatile emitters, which can be harnessed for light emitting photonic applications by integrating them with resonant photonic nanostructures. The interaction of light with localized emitters such as fluorescent molecules, nitrogen va2

ACS Paragon Plus Environment

Page 3 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

cancy centers in diamond, and semiconductor quantum dots placed in proximity to photonic and plasmonic nanostructures has been widely investigated 15–19 . Similar studies have been extended to TMD monolayers with plasmonic nanostructures for enhancing their interaction with light 20–25 . Recently, plasmonic nanostructures attracted further interest owing to their capabilities of routing bi-chiral emission of TMD monolayers, targeting valleytronics applications 26–28 . However, plasmonic nanostructures may also induce strong local strain 29,30 and can mediate significant injection of hot carriers into TMD monolayers, thus quenching the emission 31,32 and altering the crystalline phase 33 . Therefore, alternative material platforms for nanophotonic applications based on TMD monolayers are highly desirable. Resonant dielectric metasurfaces 34,35 are promising candidates in this respect. Metasurfaces consist of two-dimensional arrays of dielectric nanoresonators with a size, geometry, and spacing such that multipolar Mie-type resonances are combined with little absorption in the visible regime. Single dielectric nanoresonators provide only moderate emission enhancement of emitters placed in their proximity as compared to their plasmonic counterparts. However, their dense arrangement in a resonant dielectric metasurface can result in a strong collective response which can be tailored to simultaneously enhance the light-matter interaction and efficiently manipulate emission directivity 36,37 . Previous work has been done on integrating TMD monolayers with resonant dielectric nanostructures. In particular, Cihan et al. 38 highlight the role of multipolar electric and magnetic resonances of single silicon nanowires to control the MoS2 emission directionality and spectrum, whereas Chen et al. 39 show enhanced directional emission from WSe2 monolayers integrated with a multi-resonant silicon waveguide-grating structure. Furthermore, Zhang et al. 40 have considered Fano resonances in dielectric nano-hole arrays to simultaneously enhance the absorption and emission of MoS2 monolayers while achieving unidirectional emission. In this work, we explore the potential of resonant all-dielectric metasurfaces to manipulate the light interaction with molybdenum disulphide monolayers (1L-MoS2 ) by alteration of 3

ACS Paragon Plus Environment

ACS Photonics 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

their photonic and electric environment as well as their surrounding topography. We realize a hybrid system constituted of 1L-MoS2 placed on top of silicon nanocylinder metasurfaces by means of a wet transfer process. Figure 1 displays a conceptual sketch of the resulting structures. Photoluminescence (PL) spectroscopy, microscopy, and back-focal-plane imag-

Figure 1: Sketch of a 1L-MoS2 flake (the crystal structure is shown in the inset) placed on top of a silicon nanocylinder metasurface on a glass substrate. ing were performed to investigate the interaction of the resonant dielectric nanostructures with the 1L-MoS2 . Verification of the monolayer phase as well as further characterization was done by means of Raman spectroscopy and second-harmonic measurements. The multidimensional analysis allows us to isolate different environmental factors that may influence the emission characteristics of the 1L-MoS2 .

Sample fabrication We fabricated hybrid structures of 1L-MoS2 supported by silicon nanocylinder metasurfaces using a polymethylmethacrylate (PMMA) assisted wet-transfer process 41,42 as depicted in Figure 2a. First, 1L-MoS2 flakes were synthesized by chemical vapour deposition (CVD) from solid precursors using silicon wafers capped with a 300 nm thermal-oxide layer as substrates 43 (see Methods for details on the growth process). Our growth process results in densely spaced 1L-MoS2 flakes with edge lengths typically reaching up to 60 µm and is widely scalable for large area coverage. Figure 2b shows an optical microscope image of as-grown 1L-MoS2 4

ACS Paragon Plus Environment

Page 4 of 23

Page 5 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

Figure 2: (a) Scheme of the PMMA-assisted wet-transfer process. (b) True-color optical microscope image of 1L-MoS2 flakes as CVD-grown on the sacrificial thermal-oxide capped silicon wafer. (c) Top-view scanning electron micrograph of a silicon metasurface on glass used as target substrate in the fishing step. (d) True-color optical microscope image of a silicon metasurface after fishing and removal of the PMMA transfer-layer. (e) Cross-sectional scanning electron micrograph of the sample shown in (d) and for a region covered by the monolayer molybdenum disulphide flake. flakes which are clearly visible as triangles (single crystal) or clusters. The formation of clustered flakes can occur due to merging of adjacent 1L-MoS2 flakes by either forming grain boundaries (polycrystalline 1L-MoS2 flake) or an overlapping region (bilayer MoS2 ). The latter case was excluded in our experiments by observing the visual change in contrast in optical microscope images for bilayer MoS2 as compared to 1L-MoS2 . Note that the color impression of the 1L-MoS2 flakes in Figure 2b originates from thin film interferences in the thermal-oxide capping of the growth substrate, where the blue color is determined by the thickness of the layer. Next, silicon nanocylinder metasurfaces were fabricated on standard 5

ACS Paragon Plus Environment

ACS Photonics 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

microscope coverslips (borosilicate glass) using an electron-beam lithography based process (see Methods for details on the fabrication process). Our samples consist of square arrays of silicon nanocylinders with constant height H = 188 nm, lattice constant Λ = 560 nm, and varying diameter D = 200 nm to 350 nm. Note that in contrast to previous works employing nanocylinder arrays 44 , the nanocylinders are designed such that they support optical resonances throughout the visible and near-infrared range including the emission band of 1L-MoS2 with a centre wavelength of 660 nm. In order to assess the multipolar composition of the considered metasurface resonances, we performed numerical simulations of the metasurface transmittance, showing very good agreement with experimentally measured spectra (see Supporting Information). Then, a multipole decomposition of the local currents inside the nanoresonators was performed, 45 revealing that the resonances overlapping with the emission band of 1L-MoS2 are characterized by electric dipolar as well as significant electric and magnetic quadrupolar contributions (see Figure S4b). Therefore, in the following we refer to them as resonances with significant quadrupole contribution. A scanning electron micrograph (SEM) of a typical fabricated sample is shown in Figure 2c. Finally, for transferring the 1L-MoS2 from the growth substrate to the target metasurfaces, a silicon wafer with as-grown 1L-MoS2 flakes was spin-coated with PMMA and immersed into a weak potassium hydroxide (KOH) solution to separate the silicon oxide layer and the PMMA. Eventually, the wafer sinks down in the immersion leaving the PMMA layer (plus superficially embedded 1L-MoS2 flakes) floating on the surface. Subsequently, the transfer layer can be fished from the surface by catching it with the target substrate, i.e. the fabricated metasurface sample. In the final step, the PMMA layer was removed in a critical point dryer using acetone as solvent and liquid carbon dioxide as supercritical fluid 42,46 . In Figure 2d a top-view optical microscope image of a fabricated silicon metasurface (dark square) after the successful transfer of 1L-MoS2 flakes (cyan triangles) is shown. Note that the structural integrity of the 1L-MoS2 flakes is largely preserved during the transfer. Further, we analyzed the placement of the 1L-MoS2 flakes on the silicon metasurface by taking an SEM of a 6

ACS Paragon Plus Environment

Page 6 of 23

Page 7 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

cross-section through the hybrid structure, milled using a focused ion beam (see Figure 2e). To improve the quality of the cross-sectional cut, we deposited platinum on top and below the transferred 1L-MoS2 flake by means of electron-beam induced deposition (visible as grainy regions on top and in between the silicon nanocylinders). The 1L-MoS2 flake is lying flat on top of the metasurface providing physical contact of the 1L-MoS2 flakes with the nanocylinders, while it is freely suspended in the region between adjacent nanocylinders. The suspended layer appears thicker in the cross section image than expected from a 1LMoS2 flake which is likely due to organic residue and Si/SiO2 debris accumulated during the focused ion beam milling process. The monolayer phase of the transferred MoS2 flakes was verified by Raman spectroscopy and second-harmonic generation microscopy (see Supporting Information for details). All measurements in this study were performed under ambient conditions.

Photoluminescence enhancement To investigate the effect of the metasurfaces on the emission properties of 1L-MoS2 first, we compare the A-excitonic emission of single flakes of 1L-MoS2 situated on top of a metasurface with that of flakes placed on the bare substrate. To exclude possible variations of the emission properties between different flakes, we consider only flakes which are partly situated on top of the metasurface and partly on the substrate. PL microscopy and spectroscopy measurements were performed using a commercially available confocal laser-scanning microscope setup (PicoQuant, MicroTime200). For excitation, a 532 nm pulsed laser with a repetition rate of 80 MHz and pulse lengths of 100 ps was focused on the sample using a 40x/0.65 NA objective resulting in an estimated spot diameter of 2r = 2λ/(NA · π) ≈ 0.52 µm. The average laser power in front of the objective was ≈ 5 µW. The same objective was used to collect the PL in reflection configuration. In the detection path the reflected laser light was blocked using a 550 nm longpass filter. An additional (685±35) nm bandpass filter was used

7

ACS Paragon Plus Environment

ACS Photonics

for acquiring spatial PL mappings. Measured PL mappings of 1L-MoS2 flakes transferred onto four different silicon nanocylinder metasurfaces with varying diameter (240 nm, 260 nm, 275 nm, and 300 nm, as measured from SEMs) are shown in Figure 3a. Each mapping shows Max

D = 240nm

D = 260nm

MS

30 µm

Substrate

D = 275nm

MS

30 µm

Substrate

D = 300nm

MS

30 µm

(b)

Substrate

MS

30 µm

Substrate

PL substrate (×3) PL metasurface

1 4.5 0.5 4 0 1 3.25 0.5 2.75 0 1 2 0.5 1.5 0 1 0.75 0.5 0.25 0

(Bandpass 650 nm < λ < 720 nm)

PLmetasurf. /PLsubstr. Transmittance (a.u.)

D = 240nm 50

225 25 ×3 200 0 50 D = 260nm 162.5 25 137.5 ×3 0 50 D = 275nm 100 25 ×3 75 0 50 D = 300nm 37.5 25 ×3 12.5 0 630 670 710 Wavelength (nm)

PLmetasurf./PLsubstr.

Counts (a.u.)

(a) Min

PL intensity (a.u.)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 23

Figure 3: (a) Spatial PL mappings of 1L-MoS2 on silicon nanocylinder metasurfaces shown for different samples of varying diameter D. The metasurface areas are indicated by white dashed lines. (b) PL spectra of the same 1L-MoS2 flakes as shown in (a) measured on top of the silicon nanocylinder metasurface (red curves) and on bare substrate (blue curves) with each measurement position being indicated by a blue circle in (a), respectively. The values for the bare substrate are multiplied by 3 for better visibility. For each metasurface the relative emission enhancement PLmetasurf. /PLsubstr. (green curves) as well as the transmittance spectrum (dotted black curves) are also plotted. The black dashed line denotes the spectral position of the resonance with significant quadrupole contribution of the nanocylinders. Note that the enhancement curves were smoothed by a three-pixel binning of the wavelength resolution of the spectrometer. a fairly uniformly enhanced PL signal in the area of the silicon nanocylinder metasurfaces (inside white dashed lines). Typical enhancement factors range from 5 to 8. Further, we measured PL spectra of the same four 1L-MoS2 flakes. Each measurement position is indicated by a blue circle in the respective PL mapping in Figure 3a and was chosen to exclude grain boundaries or damaged regions of the 1L-MoS2 flakes. The resulting spectra 8

ACS Paragon Plus Environment

Page 9 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

are presented in Figure 3b. When situated on a bare substrate (blue curves), we observe emission centered at around (668±5)nm. In contrast to 1L-MoS2 on bare substrate, 1L-MoS2 placed on top of the silicon nanocylinder metasurfaces (red curves) shows not only enhanced PL but also a spectral reshaping of the emission band. We observe a spectral broadening and a shift of the emission band maxima to shorter wavelengths around (655 ± 1)nm. The shift becomes eminent as a pronounced peak in the relative PL enhancement spectra (green curve), defined as the ratio PLmetasurface /PLsubstrate . Additionally, the flat but non-vanishing tails of the relative enhancement spectra are related to a uniformly enhanced PL in the whole range of the emission band. Note that the enhancement curves were smoothed using a three-pixel binning of the wavelength resolution of the spectrometer for better visibility in the plot. Next, we aim to identify the physical origin of the enhancement and spectral reshaping by considering photonic influences from the resonant metasurfaces, strain modulations from the nanostructured topography, and electronic influences by unintentional doping from the substrate materials. We performed additional measurements and data analysis in order to check for their individual contributions to the observed changes in PL. First, we quantified the reshaping by employing a multi-Lorentzian fitting scheme to the measured PL spectra from 1L-MoS2 supported by bare substrate, as well as by silicon nanocylinder metasurfaces of varying diameters. For all samples, two dominant contributions can be identified and resulting fit parameters of a double-Lorentzian model show peak positions at (655 ± 2)nm and (670 ± 3)nm and full-widths at half maximum (FWHM) of (17 ± 2)nm and (38 ± 7)nm, respectively. The well reproduced peak positions and widths throughout all samples suggest a global reason common to all samples leading to the observed reshaping. In agreement with previously reported values 5 , we attribute the peak at the shorter wavelength to neutral A-excitonic (X) emission of 1L-MoS2 . Due to the red-shift of the centre wavelength and the broadening of the peak width, we attribute the second peak to negatively charged A-trionic (T) emission induced by unintentional n-doping mediated by the substrate material 47,48 . 9

ACS Paragon Plus Environment

ACS Photonics

Even though strain modulations can lead to similar changes (red-shift and broadening) in the emission spectrum of 1L-TMDs 49 , we exclude macroscopic strain as dominant cause for the occurrence of the second peak by Raman spectroscopy (see Supporting Information). We quantified the intensity contributions of the exciton species IX and of the trion species IT by integrating the respective Lorentzian contributions within their FWHMs. Their relative spectral weight in the overall observed emission is defined as wX/T = IX/T /(IX + IT ). It was measured for 1L-MoS2 on bare substrates as well as on top of different silicon nanocylinder metasurfaces and is shown in Figure 4a. For 1L-MoS2 on bare substrates, we observe

(b)

1

Enhancement η

(a) Spectral weight w

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 23

T substrate X metasurf.

0.5 T metasurf. X substrate

10

X T

2

101 100

0 210

250 290 330 Diameter (nm)

210

250 290 330 Diameter (nm)

Figure 4: (a) Relative spectral weights wX/T of exciton (red curves) and trion (blue curves) species in 1L-MoS2 situated on silicon nanocylinder metasurfaces (solid lines) and glass substrate (dotted lines). (b) Enhancement factors of excitonic (red curve) and trionic (blue curve) contributions. The lines were added as guide to the eye. a dominant trion species whereas for 1L-MoS2 supported by the metasurfaces both exciton and trion species are of the same order of brightness. Across all samples, the spectral weight of the exciton species gets increased up to wX (metasurface) = 0.65 and a relative enhancement of the excitonic species up to wX (metasurface)/wX (substrate) = 14 is observed which can be explained by the fact that the metasurfaces provide a regular pattern of pillars supporting the 1L-MoS2 wherever they are positioned directly above individual nanocylinders but suspending the 1L-MoS2 in between nanocylinders of adjacent unit cells. Hence, the freestanding 1L-MoS2 is not efficiently n-doped and exhibits dominant neutral excitonic emission. From purely geometric considerations, an even higher relative enhancement of the 10

ACS Paragon Plus Environment

Page 11 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

excitonic species would be expected, pointing towards residual doping of the freestanding 1L-MoS2 due to sample impurities. We further calculated the absolute enhancement factors substrate metasurface of the exciton (red curve) and trion (blue curve) populations /IX/T ηX/T = IX/T

mediated by the silicon nanocylinder metasurfaces as shown in Figure 4b. A drastic enhancement of the absolute exciton emission up to a factor of 170 is observed, strongly exceeding the aforementioned relative enhancement up to a factor of 14 due to the repopulation of exciton and trion species. However, as shown in 50 , the surface roughness of amorphous substrates can strongly influence the radiation efficiency of 1L-MoS2 . By random microscopic strain modulation on the amorphous surface, the band structure of the supported 1L-MoS2 can locally be modified resulting in a transition from a direct to an indirect band-gap material. Accordingly, freestanding 1L-MoS2 shows an up to two orders of magnitude higher radiation efficiency, leading to the strong enhancement observed in our experiments. For the trionic emission we observe an absolute enhancement of up to 11 despite the relative depopulation of the trion contribution in 1L-MoS2 situated on top of the metasurfaces. Following the explanation above, this might partly suggest a better surface quality of the nanostructured silicon compared to the amorphous glass substrate which, as a result of the etching process during the fabrication of the silicon nanostructures, is expected to have a high surface roughness. Further, we also need to consider the photonic influence of the substrate itself which, due to its higher refractive index, provides a higher density of photonic states compared to air. Consequently, the emission efficiency can be increased but, more importantly, the emitted radiation is mainly directed into the substrate halfspace 51 . This effect is reduced for 1L-MoS2 situated some distance away when being supported by the metasurfaces and hence leading to an increased emission into the air halfspace, i.e. the collection side in our experiments. We also checked for a possible photonic influence on the observed change of the PL spectra of 1L-MoS2 mediated by the resonant silicon metasurfaces. However, from Figure 4b no pronounced systematic dependency of the observed enhancement factors on the nanocylinder diameter and supported resonances is apparent. We attribute this to a combi11

ACS Paragon Plus Environment

ACS Photonics 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

nation of two factors. Firstly, the radiative rate enhancement is expected to be rather low for the considered sample geometry 52 . Secondly, slight differences in the 1L-MoS2 sample quality might be introduced during the wet transfer process making a comparison of the absolute enhancement values across many flakes and samples difficult. Note that slight changes in the weak photoluminescence signal emitted from 1L-MoS2 on bare substrate can lead to stronger variations in the calculated enhancement factors as the emission from 1L-MoS2 on bare substrate is used as reference.

Directional shaping of emission While the resonant-photonic influence of the metasurface on the emission spectra is weak, its influence on the 1L-MoS2 emission pattern remains to be investigated. We analyzed the directionality of emission by back-focal plane (BFP) imaging of the PL from the hybrid structures consisting of 1L-MoS2 on silicon nanocylinder metasurfaces. BFP images were recorded using a home-built setup with a 532 nm continuous wave excitation laser and a 100x/0.73NA focusing objective. The average laser power at the sample position was 0.8 mW. The same objective was used for collecting the signal in reflection. The signal was then coupled into a dedicated lens system for imaging the back-focal plane of the objective onto a charge-coupled device (CCD) camera. The collected light was filtered using a (660 ± 5) nm bandpass filter. The experimental results for the same four metasurfaces as presented above are shown in Figure 5. For all measurements we observe a 4-fold symmetric pattern in the measured BFP images of 1L-MoS2 on top of silicon nanocylinder metasurfaces reflecting the symmetry of the underlying nanostructures. However, strong changes occur for the different nanocylinder diameters with respect to the actual angular intensity distributions. To further understand the measurement results, we analyzed the BFP pattern regarding the influence of the resonators (nanocylinder diameter) and the lattice (square array arrangement). Starting with the influence of the lattice, we analyzed the presence of propagating lattice modes for dif-

12

ACS Paragon Plus Environment

Page 12 of 23

3800

Counts (a.u.)

D=260nm

-0.73

D=300nm

Counts (a.u.)

0.73 D=275nm

2620 2200

0 1520 1950

0

900

1150

ky /k0

0.73 D=240nm

ky /k0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Photonics

4250

Page 13 of 23

-0.73 -0.73

0 kx/k0

0.73

-0.73

0 kx/k0

0.73

Figure 5: Measured back-focal plane images of the photoluminescence emitted by 1L-MoS2 flakes on top of silicon nanocylinder metasurfaces with diameters D = 240nm, 260 nm, 275 nm, and 300 nm collected with a (660 ± 5)nm bandpass filter. ferent emission directions. For a fixed wavelength (λ = 660 nm) as defined by the bandpass filter used for the BFP measurements, we calculated the number of propagating lattice modes for different emission directions (θ, ϕ) and respective points in the back-focal plane (kx , ky ) = k0 (sin(θ) cos(ϕ), sin(θ) sin(ϕ)). Regions of the back-focal plane with a constant number of propagating lattice modes accurately describe the underlying pattern in the measured BFP images (see Supporting Information for details). However, this simple analytical consideration cannot make any predictions for the amount of light which is emitted into the directions (θ, ϕ) due to the presence of the resonant scatterers. To this end, the actual farfield pattern from each unit cell needs to be considered which will be dominated by the scattering properties of the nanocylinders. As seen in the measurements, for increasing nanocylinder diameter D the PL gets redirected more strongly towards the collection objective until the main lobe of the emission is directed normal to the metasurface by sweeping the resonance

13

ACS Paragon Plus Environment

ACS Photonics 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

with significant quadrupole contribution of the nanocylinders into spectral overlap with the 1L-MoS2 emission band maximum. Hence, the radiation pattern directed into the air half space can be tailored from emission under very large angles (off-resonant case) to emission directed preferentially out of the substrate plane (resonant case).

Conclusion We have demonstrated experimentally the integration of CVD-grown 1L-MoS2 with silicon nanocylinder metasurfaces and have studied the influences of the electronic and photonic environment on the spectral and directional properties of the emitted photoluminescence. The observed spectral reshaping and strong enhancement of the measured photoluminescence spectra have been attributed to the charge transfer mediated repopulation of excitons and trions as well as the freestanding nature of 1L-MoS2 on silicon nanocylinder metasurfaces. Most importantly, by tuning a metasurface resonance with significant quadrupole contribution to spectrally overlap with the emission band maximum of 1L-MoS2 , a preferential emission out of the metasurface plane could be achieved. By this, we have demonstrated an efficient platform for accessing the strong emission from freestanding TMD monolayers being directed out of the substrate plane allowing for light collection with simple low-NA optics, a crucial step for integrating TMD monolayers in large-scale photonic applications.

Methods CVD growth of monolayer MoS2 1L-MoS2 flakes were grown by chemical vapour deposition on thermally oxidized silicon substrates (Sil’tronix, oxide thickness 300 nm, roughness