Design of hybrid MoS2/photonic devices compatible with technological constraints

The integration of transition metal dichalcogenide layers into photonic devices is a current challenge in the field of 2D materials. Based on numerical simulations, this work explores the design of devices combining an MoS2 monolayer with planar photonic gratings sustaining localized optical resonances. A special attention is paid to the technological constraints. The optical response of six devices is compared taking into account the limitations imposed by the growth conditions of the MoS2 layer and the processing of the resonant optical gratings. The reported photonic devices composed of grating filters and a backside reflector on silicon and silica substrates exhibit a theoretical absorption by the MoS2 layer between 85 and 99% at 532 nm. The numerical simulations further show that the addition of an Al2O3 encapsulation layer, to protect the MoS2 monolayer, results in an increase of the performance of the devices. These hybrid MoS2 based photonic devices are promising technological platforms for the study of the optical properties of integrated MoS2 monolayers.


Introduction
Transition Metal Dichalcogenides (TMD) are two-dimensional (2D) semiconductor materials with astonishing physical properties [1] thanks to their stunning excitonic binding energy and valley spin properties [2]. Numerous works reported numerical simulations of devices achieving near perfect absorption by stacking these ultrathin dichalcogenide layers on or under metallic layers and antenna [3][4][5][6][7][8][9][10][11][12]. The devices with plasmonic structures on top of a TMD monolayer present the advantage of concentrating the electromagnetic field in the 2D material [13] paving the way for systems exploiting enhanced photoluminescence (PL) emission and/or photocurrent (PC) conversion. Thus, some works report measurements of significant PL enhancement of TMD monolayers in hybrid photonic devices with various metallic 2D structures near the TMD material [14][15][16][17][18][19][20][21][22]. However, the contact between the metal and the TMD material could be responsible for a quenching of TMD emission due to charge transfer [23][24][25]. To avoid this detrimental effect while keeping the ability to demonstrate exceptional absorption [26], the use of less common metals as Gallium [27], or stacked spacing layers [28] or only dielectric materials [29] have been reported.
A general review of the reported experimental demonstrations involving TMDs shows that the transition from isolated lab-scale objects to large-scale integrated devices comes with a significant degradation of the TMD material performances [30]. Given the recent demonstration of the synthesis of high-quality TMD monolayers on a wafer scale [31], one of the key challenges remaining to make TMD-based devices a reality is to develop integration methods compatible with TMD materials [1,32]. Integration of TMD materials can be significantly improved by taking into account the technological and experimental constraints into the design of devices. As TMD materials show limited stability when exposed to air [33], an encapsulation layer is mandatory to protect the TMD material from contact with the atmosphere [34][35][36][37][38]  steps. Similarly, should the TMD material be deposited by CVD methods, the underlying multi-layer coated substrate has to withstand the high deposition temperature (above 400°C).
In this work, we investigate hybrid devices with a MoS 2 monolayer incorporated into grating based photonic cavities. The devices are designed to achieve a maximum optical absorption by the MoS 2 layer at a target excitation wavelength (532 nm in our case). The MoS 2 monolayer is integrated between a reflector and a resonant grating. The several stacks are conceived to protect the MoS 2 materials from degradation induced by the most reactive integration steps, in particular etching processes. Two types of resonant gratings are compared: an infinitely long grating-mode resonant filter (GMRF) and a more realistic cavity resonator integrated grating filter (CRIGF) with finite lateral dimensions. A CRIGF is composed of a grating coupler (GC) between two distributed Bragg reflectors (DBRs) on top of a waveguide layer [39]. Under optical illumination of the GC, a fraction of the incoming light is injected into a singlemode confined in the lateral cavity formed by the two DBRs. As demonstrated in [40], a CRIGF can be viewed as a small foot-print, folded GMRF device. Its small footprint is particularly interesting in the case of ultrathin 2D materials of which lateral size may be too small to allow an efficient coupling with a GMRF device. In this work, the MoS 2 monolayer is incorporated just below this waveguide resonator (figure 1) leading to the absorption of the resonating optical mode by the MoS 2 .

Numerical tools
Hybrid devices made of photonic structures and 2D materials combine micro and nanoscale engineering which, for modelling, implies microscopic meshing with locally sub-nanoscale precision. To maintain a reasonable calculation time, numerical simulations were performed using 2D-FMM methods [41] with an S-matrix algorithm [42]. The calculations are carried out using up to 1200 Fourier components for stacks with an entire CRIGF structure. For the CRIGF-based devices, the incoming light is under normal incidence with transverse electric (TE) polarisation; its intensity is Gaussian-shaped and overlaps only with the central part of the grating coupler. For GMRF-based devices, the calculations were performed for a plane wave at normal incidence, in TE polarization. The complex refractive index of the materials used in the simulations are provided in the Supplementary Information section. Absorption of the here considered dielectric materials (SiO 2 , SiN x and Al 2 O 3 ) is negligible in the wavelength range of interest (500 to 800 nm). Thus, absorption occurs only within the MoS 2 layer and the Si substrate or back side reflector.

Optimal numerical solutions
For each investigated device, the geometrical parameters of the stack were optimized to maximize the absorbed fraction of the incident light. Table 1 summarizes these parameters and the calculated optical absorption by the MoS 2 layer integrated in the GMRF and CRIGF structures compared to a reference of a MoS 2 layer on 90 nm SiO 2 /Si substrate (hereafter named MoS 2 /SiO 2 /Si).   structure that allows the absorption of 99.1% of the incident light at 532 nm in the MoS 2 monolayer as shown in figure 2(c). The near-perfect absorption at 532 nm is obtained owing to the coupling of the resonance of the GMRF structure with the B and C-exciton resonances of the MoS 2 layer. Indeed, the A and B exciton absorption bands are clearly visible around 660 and 610 nm, respectively; whereas the one of the C-exciton arises below 500 nm [14,43]. The absorption of the same device but without the MoS 2 layer is significantly lower. This residual absorption arises from the enhancement of the electromagnetic field near the GMRF structure and the absorption into the SiN x layer. Comparison of the GMRF/Ag reflector device with and without the MoS 2 layer shows that the MoS 2 layer absorbs the main part of the incident light (see the Supplementary Information). Since the experimental properties of MoS 2 are generally studied for flakes reported or grown on a 90 nm SiO 2 /Si substrate (see figure 2(b)), we calculate that 20.9% of the incident light is absorbed in this reference stack at 532 nm, a value consistent with the results reported in the literature [44]. The GMRF with Ag back reflector enhances the optical absorption by the MoS 2 layer by a factor 5 compared to the MoS 2 /SiO 2 /Si reference.
The GMRF-based structure on Ag back reflector shows impressive performances, however its fabrication faces several technological challenges. Firstly, SiN x film is commonly grown using plasma assisted processes under reactive conditions (nitride atmosphere and/or nitride precursors) which may create defects in the atomic structure of the TMD layer during the early stages of the deposition. Secondly, silver (back reflector) is rarely used in microelectronic devices, because of its low thermal stability incompatible with the MoS 2 deposition temperature. The melting temperature of the material used as a back reflector is given in the table 1 as an indication of its thermal stability. One must keep in mind that degradation of the reflector (delamination, phase segregation) occurs at temperatures way below its melting temperature. The incompatibility between the low thermal stability of silver and the MoS 2 deposition temperature can only be solved using a transfer of the MoS 2 layer or by deposition of the Ag layer, after the growth of the MoS 2 layer, combined with a challenging back side thinning down to the spacer layer. Vertical DBRs based on dielectric materials are a potential substitute to metallic reflector. However, their fabrication requires an excellent repeatability (in composition and thickness) of the dielectric deposition steps and the multiplicity of layers which can be a source of stack delamination.
To bypass these technological issues, alternative structures need to be explored and compared to the GMRF based near-perfect absorber structure. Furthermore, the sharp resonance generated in the hybrid photonic device is well adapted to the illumination by spectrally coherent light sources, which can be focused to the active region of the device. To this aim, CRIGF structures, consisting of a resonant grating of finite length and an additional lateral optical confinement have been shown to facilitate high-Q cavity operation under focused laser spot [45]. As such, CRIGFs are also more amenable to lead to high performance devices with a spatiallyrestricted uniformity.

Alternative devices based on practical integration considerations
In this work, the GC of the CRIGF structures is composed of 41 periods matching a beam width of about 12 μm. As the period of the DBRs is half the GC period, considering DBRs composed of 100 periods, the total width of the device is 45 μm at maximum. Figure 3(a) shows the theoretical optical response of three types of hybrid CRIGF/MoS 2 devices.
The first CRIGF/MoS 2 -based design is similar to the previously studied GMRF/MoS 2 design (see figure 2(a)) with the GMRF structure and Ag mirror replaced respectively by a CRIGF structure and a Si substrate (see figure 3(b)). As reported in table 1, the Ag mirror improves significantly the absorption by the MoS 2 layer compared to the same device on a standard silicon substrate (99.1 compared to 85.8%). However, the stack on Si substrate is compatible with the direct deposition of TMD materials at high temperature (around 650°C ). With the CRIGF structure, the absorption by the TMD layer is weaker than with a GMRF structure (82.0 instead of 85.8%). This is due to the finite lateral size of the CRIGF. Indeed, for a GMRF limited to 41 periods (as the GC part of the CRIGF) and using an illumination source with the same gaussian beam waist, the theoretical absorption by the MoS 2 layer in the GMRF decreases to 80.8% (as compared to the initial 85.8%). Thus, when considering realistic lateral dimensions of the GMRF structure, better performances are predicted for the CRIGF structure compared to the GMRF.
The second CRIGF design (see figure 3(c)) is composed of an additional alumina Al 2 O 3 layer intercalated between the MoS 2 monolayer layer and the SiN x guide. The Al 2 O 3 layer is a protective capping layer which could be deposited to prevent contamination and degradation of the MoS 2 layer during the reactive processes. Indeed, Al 2 O 3 is airtight and has a good chemical stability. In addition, Al 2 O 3 is commonly obtained by atomic layer deposition (ALD), a softer process compared to chemical or physical vapour deposition (CVD or PVD). With a 10 nm thick Al 2 O 3 layer and optimized parameters of the CRIGF (see values in table 1), the maximum of absorption by the MoS 2 layer reaches 84.9% due to a better confinement of the optical mode at 532 nm. Hence, the protective Al 2 O 3 integrated in the CRIGF/MoS 2 device is beneficial as it increases the achievable maximum of absorption by the MoS 2 layer.
Finally, a third CRIGF-based design is considered ( figure 3(d)). It involves an inverted fabrication process flow allowing the integration of a metal mirror after the growth of the MoS 2 layer. In this way, the metal is kept below its melting temperature which can be close to the MoS 2 growth temperature (650°C) in the case of Aluminium for example. This reversed design is based on a transparent fused silica substrate which permits a back side optical excitation. After partial etching of the substrate to fabricate the CRIGF structure, the corrugated substrate is covered by SiN x . The MoS 2 layer would then be grown onto the SiN x and almost no change in the nucleation and growth of the 2D material are expected compared to SiO 2 substrate as both dielectrics are amorphous materials. Because of the inversion in the integration steps, the protective Al 2 O 3 layer is incorporated between the MoS 2 layer and the SiO 2 spacer. In this reversed CRIGF-based design, the encapsulation of the MoS 2 layer is essential as the oxidizing conditions under which the SiO 2 is deposited may damage the MoS 2 and introduce defects that strongly alter its optical absorption and emission properties. Finally, an aluminium reflector is added on the top of the device. Despite its slightly lower reflectance at 532 nm, Aluminium is preferred to silver due to its better thermal stability (see table 1). According to the calculations with the reversed CRIGF-based design, the optical absorption by the MoS 2 layer reaches 89.1% at 532 nm. The integration of emerging 2D compounds in 3D devices is challenging considering the potential detrimental impact of the integration processes on the performances of the TMD materials [30]. The latter device anticipates several potential technological fabrication limitations while exhibiting impressive absorption as compared to the other CRIGF devices investigated in this work.

Conclusion
In summary, using numerical simulations, we found that the integration of a 2D MoS 2 layer between a CRIGF or GMRF structure and a reflector enhances the optical absorption by the MoS 2 layer to 85 to 99% as compared to the ≈20% achieved with the standard MoS 2 /90 nm SiO 2 /Si stack. The increase is mainly due to the coupling of the resonance sustained by the CRIGF or the GMRF with the MoS 2 excitonic absorption. The calculated absorption by the MoS 2 layer implemented in the CRIGF-based device is higher than the one reached with the GMRF-based device. Furthermore, the substitution of the Si substrate by an aluminium back reflector and the addition of a capping layer increase the theoretical absorption by the MoS 2 . The designed devices are compatible with the material thermal and chemical constraints and technological processes and their optical response is suitable for studying the optical properties of integrated MoS 2 layers. This investigation of hybrid TMD/ photonic devices is a contribution to the understanding of the impact of integration processes on TMD materials, identified as the main current challenge preventing an ubiquitous use of 2D TMD materials [1,30].