1. Introduction

Novel optical properties recently found in two dimensional atomic crystals (2DAC) such as strong photoluminescence [1,2], valley polarization [3], coherence [4], and giant non-linear response [5], enable these two dimensional materials to be used as building blocks for optical components small enough to be integrated with modern semiconductor technologies. Many 2DACs lack inversion symmetry granting them strong non-linear optical response because higher order terms in the power series expansion of the polarization field within individual monolayers fail to get cancelled out by adjacent crystal layers found in their bulk counterpart [6]. This allows odd number of layers of 2DACs to act as photon exchange hubs wherein photons can be exchanged between the various frequency components of the electromagnetic field in a single quantum-mechanical process [7]. Second harmonic generation (SHG) is a fundamental non-linear quantum mechanical photon exchange process converting two photons of frequency ω into a single photon of frequency 2ω. This frequency doubling process is technologically important for the generation of lasers in the blue or UV regions [8] and general tunable laser sources which at present are only made with bulky non-linear crystals.

2DAC made of monolayer transition metal dichalcogenides (TMD) materials (i.e. MoS₂, MoSe₂, WS₂) have shown comparable and even stronger SHG efficiencies than common non-linear crystals due to their lack of inversion symmetry [5,9,10]. Recent advances in SHG of monolayer materials have shown giant second harmonic (SH) response [10], crystal orientation dependence [11,12], and electrical control of SHG [13]. The non-linear optical response of nanomaterials can be increased further by placing them inside an optical cavity which provides enhanced electromagnetic field density and an increased number of cavity photon passes, increasing the likelihood of SH conversion [14–17]. In this Letter, we use a monolithic microcavity with a single resonance at the pump wavelength to enhance SH emission at room temperature from a monolayer of MoS₂ embedded within the resonator structure and characterize the SH enhancement in the presence of the microcavity.

2. Fabrication

We used a planar Fabry-Perót microcavity such as the one shown schematically in Fig. 1(a) consisting of distributed Bragg reflectors (DBR) for both the bottom and top mirrors to enhance interaction between the two-dimensional material and the fundamental beam inside the cavity. Each Bragg reflector consist of alternating layers of SiN_x and SiO₂ and on a glass substrate. The pairs of SiN_x and SiO₂ layers had thickness of ${\lambda }_0/4n$, with ${\lambda }_0$ being the cavity wavelength near 800 nm and n is the refractive index of the material. The top and bottom DBRs were designed to be highly reflective between 700 – 1000 nm where the pump wavelength lies while being transparent at the second harmonic wavelength,$\raisebox{1ex}{${\lambda }_0$}\!\left/ \!\raisebox{-1ex}{$2$}\right.\approx 400nm$. Plasma enhanced chemical vapor deposition (PECVD) was used to grow the DBR layers with the measured refractive index for SiN_x and SiO₂, being n = 2.23 and n = 1.5, respectively. The bottom DBR mirror was grown on a glass substrate with 5 SiN_x and SiO₂ pairs, and with the top DBR mirror grown with 4 pairs so the overall structure is more transparent from above. The MoS₂ layer was sandwiched between two quarter wavelength thick SiO₂ layers forming the cavity and was grown between the DBR mirrors. The first SiO₂ cavity layer was grown on the bottom DBR mirror with PECVD. The MoS₂ was then transferred to this bottom cavity layer while the top SiO₂ cavity layer and the top DBR were sequentially grown above the transferred MoS₂ layer to complete the microcavity structure.

 

Fig. 1 Linear Characterization of DBR-based microcavity with embedded MoS₂. (A) Schematic of microcavity with DBR mirrors consisting of interleaving layers of SiNx and SiO₂ and SiO₂ cavity layers. (B) Brightfield image of MoS₂ flakes. (C) Reflectivity of the microcavity showing the cavity resonance at the fundamental wavelength (~800nm).

Download Full Size | PDF

Monolayer MoS₂ was grown on a 300 nm SiO₂ on Si substrate at 650°C by ambient pressure chemical vapor deposition (APCVD). High purity solid precursors of MoO₃ and S powders were adopted for the synthesis. Prior to the growth, Perylene-3,4,9,10-tetracarboxylic acid tetrapotassium salt (PTAS) solution were uniformly coated on the substrate surface as seeding promoters to enhance the nucleation and growth of the monolayers. The CVD growth of monolayer MoS₂ is carried out in 1 inch furnace with a growth time of 40 minutes and a heating rate of 15°C/min. After the growth, the residual reactants were removed with an Ar gas flow of 100 sccm and the quartz tube was took out of the furnace to rapidly cool down the entire system. More details on the synthesis were presented in previous papers [18,19]. The CVD-grown monolayers were removed from SiO₂/Si substrate by directly immersing the samples into the solution of DI water at room temperature. Immersed monolayers detached from the substrate and float on the DI water surface which were then fished out with the bottom DBR sample. PECVD growth of the top SiO₂ cavity and top DBR layers were carried out on the transferred sample to complete the microcavity. Figure 1(b) shows the optical microscope image of the MoS₂ flakes after traferring to the microcavity structure.

3. SHG calculations and measurement

A completed microcavity with 5 and 4, DBR periods (5,4) for the bottom and top mirrors respectively, shows a cavity mode at ${\lambda }_0=818$nm represented as a dip in the reflection spectrum as seen in Fig. 1(c) The microcavity shows high reflectivity, (> 90%) between 700 – 1000 nm with the cavity resonance at ${\lambda }_0$. Although low reflectivity is desired at the second harmonic frequency, we observed slightly higher than predicted by simple transfer matrix calculations due to thickness and refractive index variations in the DBR.

The (5,4) microcavity has a measured unloaded q-factor $Q_0$ = 74, where $Q_0=\frac{{\omega }_0}{\Delta \omega }$ as obtained from reflectivity shown in Fig. 1(c). Using the relationship between q-factor and cavity finesse, $ℱ$ (Eq. (1), the resonant field enhancement of the pump laser inside a cavity that is perfectly impedance matched (all incident radiation is coupled into cavity) is given by [20],

Second harmonic generation emitted from monolayer MoS₂ was measured at room temperature during different stages of the microcavity fabrication process: bare as grown MoS₂, MoS₂ transferred to bottom cavity and DBR layers, and MoS₂ embedded within a completed microcavity. The observed optical microscope SH images are shown in Fig. 2(a), 2(b) and 2(c). Optical measurements were obtained with a femtosecond Ti:sapphire laser (Spectra Physics Mai Tai) tuned to the cavity mode wavelength, 818 nm, and focused onto the sample in a reflection setup with a 100 × objective (0.85 NA). Average power used 1 mW, (peak power, 39 W/cm²), at 80 Mhz repetition rate. Emission was collected after passing through a short pass filter and detected with an avalanche photodiode in a scanning confocal setup. The stage was scanned while exposed to the laser creating scanned confocal images. Power measurements were obtained by taking the average photon counts of the area of a single CVD triangular flake (~40x40 μm²) from each SHG confocal image.

 

Fig. 2 Second Harmonic Generation (SHG) emission from MoS2 at different microcavity fabrication stages. Scanning SH confocal images of (A) as grown, (B) water transferred, and (C) fully embedded MoS2. (D) Power dependence of SH signal from MoS2 sample at the three fabrication stages. Quadratic linear fit with a slope of 2.17 is displayed for the “Full Cavity” power dependence plot.

Download Full Size | PDF

Second harmonic power emission from MoS₂ samples observed during each microcavity fabrication stage show the expected quadratic dependence on input pump power as shown in Fig. 2(d). Although the majority of the monolayer MoS₂ was successfully transferred to the bottom DBR and cavity layers, this method did not preserve the crystallinity of the MoS₂ over large areas. The MoS₂ layers can be seen to be merged having with different crystal boundaries in direct contact as can be seen in Fig. 2(b) and 2(c).

The highest second harmonic emission for transferred samples are observed from the crystal boundaries as opposed to the seed of the MoS₂ monolayer located at the center of the MoS₂ flake for as grown CVD samples as has been previously reported [5,10]. Although the transferred MoS₂ samples lose their triangular crystalline form as seen in Fig. 2(b) and 2(c) during the water transfer, the overall SH emission is comparable and slightly greater than the emission from as grown samples. This could be due to the increase of higher emitting edge or crystalline boundaries for transferred samples or the incrased reflectivity of the bottom DBR as compared to the as grown samples. Monolayer MoS₂ samples fully embedded within a (5,4) cavity shows an enhancement of ~10 over as grown MoS₂ samples as shown in Fig. 2(d), which roughly matches SHG enhancement predicted by Eq. (1).

There was no background SHG from the cavity as sites without MoS₂ flakes showed no SHG near powers used for MoS₂ measurements (~1 mW avg. power, ≤ 50 W/cm²). However at higher input powers, (8.5 mW avg. power, ≥ 100 W/cm²), the bare cavity showed measurable but considerably lower SHG (≤ 1%) as compare to active MoS₂ sites. These isolated SHG emission sites could be associated with PECVD-grown Si-nanocrystal impurities in the dielectric layers or could originate from tiny MoS₂ flakes that were scattered during the wet transfer process. Interestingly, SHG can also be observed away from the central cavity mode due to the losses in the DBR mirrors. Exciting the MoS₂ regions when pumping off the cavity resonance, (876 nm), the MoS₂ regions shows ~25% of the SHG compared with on resonant excitation, but the emission is only visible from the crystalline borders unlike the emission seen at resonance as can be seen from Fig. 3(a). Shown in Fig. 3(b) is the reflectivity of the microcavity and the off resonabnt pump (red arrow).

 

Fig. 3 Fully Embedded MoS2 site excited off resonance. (A) SH Confocal image of embedded MoS₂ excited off resonance at 876 nm. (B) Reflection plot of cavity resonant at 818 nm with the red arrow indicating the pump wavelength.

Download Full Size | PDF

In summary, using a relatively low-Q cavity we showed enhanced SH emission of a 2D material semiconductor inside a DBR-based optical microcavity. Enhancing only the fundamental mode we produced an SH emission of over one order of magnitude. Further improvements to this design could be made using a polymer-based transfer method to keep the crystallinity of the MoS₂ intact, reducing the mode volume of the cavity, exciting SHG near the resonance of an excitonic state of the 2D material, and investigating doubly resonant cavity designs [21,22].