Tunable Encapsulation and Doping of Monolayer MoS2 by In Situ Probing of Excitonic Properties During Atomic Layer Deposition

Here, it is shown that in situ spectroscopic ellipsometry (SE) is a powerful method for probing the effects of reactant adsorption and film formation on the excitonic properties of 2D materials during atomic layer deposition (ALD), thus allowing optimization of both film growth and opto(electronic) characteristics in real time. Facilitated by in situ SE during ALD on monolayer MoS2, a low temperature (40 °C) process for encapsulation of the 2D material with a nanometer‐thin alumina (AlOx) layer is investigated, which results in a 2D/3D interface governed by van der Waals interactions rather than chemical bonding. Charge transfer doping of MoS2 by AlOx is found to be an interfacial phenomenon that initiates from the earliest stages of film formation, but saturates upon deposition of a closed layer. However, the lack of chemical binding interactions at the 2D/3D interface enables physical removal of the AlOx that results in a reversal of the charge transfer doping effect. Overall, it is demonstrated that in situ SE of 2D materials during ALD can precisely probe the impact of film formation on sensitive optoelectronic characteristics of 2D materials, which is of key importance in the development of integrated 2D/3D systems.


Introduction
2D layered materials are attracting intensive interest in fundamental and applied science, with groundbreaking performance characteristics achieved in applications across different fields, including catalysis, [1] photonics, [2] optical sensing, [3] and morethan-Moore (MtM) electronics. [4] Because 2D van der Waals (vdW) semiconductors retain a high carrier mobility even at the monolayer limit, where short channel effects are largely ALD on 2D materials are largely missing. With few exceptions, [12] film nucleation on 2D materials has been studied by transferring the sample to an analysis chamber and characterizing the surface after ALD (half)cycles. [7b,13] While this approach has provided important mechanistic insights into growth, it is less suitable for understanding interactions of the 2D material within reactive environments or for rapid development of complex growth processes. Here, we address this gap via the implementation of in situ spectroscopic ellipsometry (SE) during ALD to gain real-time insights into film nucleation on the 2D surface, while simultaneously probing the impact of the adlayer on the dielectric function and excitonic properties of the 2D material. We use this approach to enable encapsulation of CVD-grown monolayer MoS 2 with a continuous and nanometer-thin aluminum oxide (AlO x ) layer using trimethylaluminum (TMA) in combination with ozone at 40 °C. The low temperature of the process facilitates nucleation [13a,14] due to physical adsorption of TMA on MoS 2 [15] and minimizes crystal defect creation in the 2D layer.
We find that the optoelectronic characteristics of 2D monolayers are highly sensitive to film formation, with adlayer charge transfer (or defect modulation doping) [8d,16] starting from the very first ALD cycle and increasing during several successive cycles. However, film-induced changes to excitonic absorption saturate following the formation of a nanometer-thin alumina coating, confirming that charge transfer doping is defined by the abrupt interface. Complementary ex situ elemental analysis by X-ray photoelectron spectroscopy (XPS), as well as combined atomic force microscopy (AFM) and optical spectroscopy measurements, indicate the formation of a weakly interacting alumina layer that is physisorbed on MoS 2 . In addition to introducing n-type doping in monolayer MoS 2 via adlayer charge transfer, [8d,16a] we find that a continuous ≈18 Å-thin AlO x film protects the 2D material and enables metallization while suppressing defect formation, which is of relevance for fabrication of tunneling contacts. Thus, this study provides a new route for tailoring the electronic properties at 2D/3D interfaces via low-temperature ALD, along with a generalizable approach to understanding film formation and accelerating process development by means of in situ SE at early stages of growth.

Results and Discussion
To investigate film deposition in real time, we utilized the experimental configuration illustrated in Figure 1a, which comprises an in situ spectroscopic ellipsometer coupled to an ALD chamber equipped with an ozone generator. Using SE, we measured the changes in phase, Δ, and amplitude, Ψ, from the incident polarized light reflected off a Si substrate Figure 1. In situ spectroscopic ellipsometry of monolayer MoS 2 during ALD. a) Schematic illustration of the ALD setup, including in situ spectroscopic ellipsometry light source and detector, and the sample structure. b) Representative Ψ and Δ spectra of a bare Si/SiO 2 substrate (black circles and squares) and a monolayer MoS 2 covered substrate (red circles and squares). The inset shows the two structures used to model the measured data. The positions of the MoS 2 A, B, C, and D exciton features are indicated. c) Real-time monitoring of the adlayer thickness on MoS 2 during the TMA/O 3 process at 40 °C (black line and circles, left axis) and simultaneous changes of the MoS 2 B-and C-exciton absorption energies (right axis). d) Plots of the extinction coefficient as a function of wavelength for monolayer MoS 2 , extracted from the analysis of SE data, following different in situ treatments. The inset shows a zoom-in of the spectral region in the range of the A-and B-exciton optical transitions. www.advmatinterfaces.de covered with monolayer MoS 2 (Figure 1b) in the spectral range from 340 to 1700 nm. Continuous monolayers of MoS 2 ( Figure S1, Supporting Information) with lateral dimensions larger than the spot size of the focused light from the ellipsometer (≥5 × 8 mm 2 ) were used to ensure that film nucleation was exclusively probed on the vdW surface. To omit optical interference and facilitate spectral modeling, we employed Si substrates with an ≈4 nm-thick silicon dioxide (SiO 2 ) layer, which also served to block charge transfer between Si and MoS 2 . The SE data were modeled with a three-layer structure consisting of Si, SiO 2 , and monolayer MoS 2 (Figure 1b, inset), which enabled changes in the dielectric function of MoS 2 to be determined as a function of growth parameters and time. Using the well-established dielectric functions of Si and SiO 2 , along with the known parameters for the monolayer MoS 2 thickness (t MoS2 = 7 Å, measured by atomic force microscopy) and highfrequency permittivity (ε ∞ = 6.1), [17] the optical response of MoS 2 in the relevant spectral range is well described with four uncorrelated Tauc-Lorentz oscillators (Supporting Information S2), [18] characteristic of the A, B, C, and D excitons of monolayer MoS 2 ( Figure 1b). [19] While the dominant optical transitions of MoS 2 occur in the range of 340-800 nm, the near-infrared spectral range (1100-1700 nm) exhibits a strong sensitivity to the thickness of growing AlO x without a significant contribution from the 2D material ( Figure S2, Supporting Information), thereby allowing simultaneous quantification of the excitonic properties of MoS 2 and the AlO x thickness, as detailed in the Supporting Information S2.
We then applied the ellipsometric model to probe changes in the adsorbate thickness and the MoS 2 dielectric function in situ, during each stage of the ALD cycle, as well as across multiple cycles (Figure 1c,d). This study focuses on low-temperature (40 °C) ALD of alumina, which provides a benchmark material and is of relevance to 2D materials electronics. Such a low-temperature process is desired to enable the physisorption of TMA on the otherwise unreactive basal plane, while also avoiding the possibility of sulfur vacancy generation and suppressing deleterious side reactions of monolayer MoS 2 with ALD reactants. Ozone (O 3 ) was used as the oxidant instead of water because the latter easily forms a physisorbed multilayer at such low deposition temperatures. We note that at elevated temperatures (>100 °C), O 3 spontaneously decomposes into O 2 and oxygen radicals (O*), whereas it is stable at lower temperatures. [20] Importantly, there was no measurable change in the SE spectra after consecutive exposure of the MoS 2 to O 3 ( Figure S3, Supporting Information), confirming that at the given reaction conditions (40 °C, p O3 ≈ 20 Torr), there is little chemical interaction between O 3 and MoS 2 . Similar to alumina growth on an oxidized substrate (e.g., Si/SiO 2 ), the measured thickness increased after the introduction of TMA and decreased after exposure to O 3 (Figure 1c), indicating the decomposition of methyl-ligands and the generation of hydroxylated surface groups suitable for reaction in the subsequent ALD cycle. [20][21] However, metal-organic precursor adhesion on the vdW surface is still weaker than on an initially hydroxylated surface and results in delayed nucleation on MoS 2 for the TMA/O 3 process ( Figure 1c and Figure S4, Supporting Information).
We next examined the influence of the ALD film formation on the MoS 2 optical properties. By tracking ALD in real-time with in situ SE, we refined the growth parameters, including the precursor dose, background pressures, and purge times required to reach saturation after the introduction of gasphase reactants. After dosing TMA, we observed a constant shift of the Δ spectrum without any measurable changes in Ψ ( Figure S3, Supporting Information) or in the peak energies of the excitonic transitions associated with MoS 2 (Figure 1c). The observation that the exciton properties are not altered suggests the formation of a weakly interacting (physisorbed) molecular adsorbate layer that has a negligible impact on the dielectric environment and carrier concentration in MoS 2 . However, after sequential exposure to both TMA and ozone, the characteristic exciton absorption peaks redshifted and the Δ signal decreased over the entire spectral range (Figure 1c and Figure S2, Supporting Information). The measured redshift and reduced strength of the extinction coefficient of MoS 2 ( Figure 1d) after exposure to TMA and ozone suggest an increased carrier concentration associated with the formation of AlO x . [22] While similar changes have been previously reported via ex situ measurements of MoS 2 monolayers capped with thicker AlO x layers, [8d,23] our results shed light into the underlying mechanisms of these changes and their evolution during the early stages of film formation. In particular, the reduced intensities and red-shifts in the excitonic absorption features are resolved even after the very first TMA/O 3 cycle (Figure 1c). With subsequent cycles, the magnitudes of these changes increase stepwise, and eventually saturate at an AlO x film thickness of ≈15 Å (Figure 1c and Figure S5, Supporting Information), which approaches the minimum thickness typically required to achieve a continuous oxide overlayer, [24] and accomplished here without the use of a detrimental plasma treatment. We note that the decreased strength of the B exciton feature after nine ALD cycles leads to significant fitting uncertainty, which manifests as apparent noise and drift of the plotted data. However, the saturation characteristics are well resolved via analysis of the C optical transition (Figure 1c).
The strong responsivity of the MoS 2 dielectric function to the initial TMA/O 3 cycles, as well as its saturation after repeated cycling ( Figure S5, Supporting Information), indicates that changes to the excitonic absorption are dominated by interfacial processes, such as charge transfer doping ( Figure S10, Supporting Information). Indeed, electrical transport measurements confirm the increase in electron concentration in MoS 2 monolayers following ALD deposition ( Figure S11, Supporting Information). Importantly, these in situ SE measurements provide strong evidence that AlO x adlayer-induced doping stems largely from the charge distribution directly at the MoS 2 /AlO x interface, as previously suggested by Pop and co-workers. [8d] As such, these results provide the intriguing opportunity to controllably adjust the carrier concentration in the 2D semiconductor with the ALD cycle number while maintaining an ultrathin coating within the length scale of charge carrier tunneling.
X-ray photoelectron spectroscopy (XPS) analysis confirmed the formation of aluminum oxide on continuous monolayer MoS 2 following ALD ( Figure S6, Supporting Information). To assess how the TMA/O 3 process affects the chemistry of monolayer MoS 2 , we compared X-ray photoelectron spectra of a bare monolayer of MoS 2 and a monolayer of MoS 2 capped with www.advmatinterfaces.de an 18 Å-thin AlO x coating (Figure 2). To within the measurement error, these results reveal equal molybdenum to sulfur ratios (Mo:S) of 1:1.98 and 1:1.96 for the uncoated and the AlO xcoated MoS 2 film, respectively (Table S2, Supporting Information). Importantly, the absence of nonstoichiometric Mo x S y peaks [25] in the photoelectron spectra suggests a physisorbed alumina adlayer on MoS 2 devoid of interface chemical bonds and without the introduction of lattice defects (e.g., sulfur vacancies), to within the measurement sensitivity. Noticeably, molybdenum trioxide (MoO 3 ) is present in both films at similar concentrations of 7.23% and 6.51% for the bare and AlO x -coated MoS 2 , respectively. The detected MoO 3 is characteristic of the commercially-obtained continuous monolayer MoS 2 films and likely stems from parasitic MoO 3 that crystallized on the growth substrate during CVD (Supporting Information S1, Supporting Information), as well as partial oxidation during prolonged air exposure. Nevertheless, we find that the ALD process reported here does not result in additional oxidation of monolayer MoS 2 .
To elucidate the effect of the ultrathin AlO x adlayer on the surface morphology and optoelectronic properties of MoS 2 at the micrometer scale, while also avoiding the typical inhomogeneities such as grain boundaries that characterize the continuous 2D film discussed above, we then utilized CVD-grown isolated monolayer MoS 2 flakes ( Figure S1b, Supporting Information). As previously reported, such flakes are also characterized by an absence of MoO 3 . [25] Atomic force microscopy images before and after encapsulation of the same monolayer MoS 2 flake (Figure 3) show a continuous and smooth ALD coating with a root mean square (rms) roughness, R rms , of 2.2 Å, which agrees well with reported roughness values of AlO x -coated MoS 2 on Si substrates. [24,26] Both monolayer MoS 2 and the AlO x adlayer on the single layer sheet conform to the topography of the underlying Si/SiO 2 substrate (R rms = 2.3 Å). The fact that the roughness did not increase following alumina deposition provides further support for the formation of a continuous AlO x coating.
To determine the AlO x thickness and examine the bonding strength of the ALD film to the MoS 2 monolayer, we performed contact mode AFM. Within a defined range of forces exerted by the AFM tip (F ts ≥ 250 nN), the ALD AlO x film can be selectively removed from the MoS 2 , as indicated by the restored phase contrast (Figure 3c,d) and optical properties (Figure 4). In comparison, substantially larger forces (F ts > 800 nN) are required to abrade the AlO x coating from the Si/SiO 2 substrate ( Figure S8, Supporting Information). We note that much larger forces (>2 µN) were reported to scratch the MoS 2 itself. [27] The obtained thickness of the AlO x layer on monolayer MoS 2 of  www.advmatinterfaces.de ∼18 Å following 12 cycles of TMA and O 3 at 40 °C (Figure 3e) corresponds to an average growth-per-cycle (GPC) of 1.5 Å on the vdW surface and ≈2.2 Å on the SiO 2 , which is similar to reported GPC of ≈2 Å on oxide surfaces for an ozone-based ALD AlO x process at 30 °C. [28] The finding that AlO x growth proceeds similarly on both continuous MoS 2 films, which possess MoO 3 impurities, and on MoS 2 flakes, which do not possess MoO 3 impurities, suggests that the presence of Mo oxides does not play a crucial role in the ALD process.
The weak interaction and damage-free removal of AlO x off the 2D surface by contact mode AFM is further confirmed by Raman and photoluminescence (PL) spectroscopy measurements ( Figure 4). Raman spectroscopy indicates an increase of the carrier concentration for the AlO x -coated MoS 2 . [29] In particular, we observed a significant redshift of the A´1 mode, Δω(A´1) = -0.99 ± 0.19 cm −1 , which translates to an increase in the electron concentration of 4.4 ± 0.8×10 12 cm −2 , [29] for the AlO x -coated monolayer MoS 2 (Figure 4a, green line) compared to the uncoated MoS 2 (Figure 4a, black line). Despite the small AlO x thickness, the observed Raman shift is comparable to reported doping-related AlO x -induced shifts in monolayer MoS 2 encapsulated with thicker AlO x . [8d] In addition to the changes of the A´1 mode, the E´ mode is blueshifted after MoS 2 monolayer capping with AlO x . Because the E´ mode of monolayer MoS 2 is nearly insensitive to changes in the carrier concentration, [29] its peak position and width allow for quantification of changes of strain within the monolayer, largely decoupled from changes in the carrier concentration. The observed shift of the E´ mode is relatively minor (Δω(E´) = 0.19 ± 0.08 cm −1 ) indicating negligible strain (<0.1%) [30] following encapsulation with AlO x .
Comparing the PL spectra obtained before (Figure 4b, black line) and after AlO x deposition (Figure 4b, green line) reveals an intensity decrease and a redshift of the MoS 2 A and B exciton emission lines (Table S3, Supporting Information). The observed changes in exciton luminescence are consistent with the spectral changes obtained by in situ SE ( Figure 1) and suggest n-type doping of MoS 2 [31] via charge transfer from the amorphous alumina adlayer, which has been reported to introduce electrons from occupied defect states within AlO x (e.g., oxygen vacancies and hydrogen interstitials) [32] into the MoS 2 conduction band. [8d] Together with the immediate impact of AlO x on the MoS 2 dielectric function measured by in situ SE, this finding highlights the important role of the interface on charge transfer and adlayer modulation doping.
Remarkably, physical removal of the AlOx adlayer using an AFM tip results in near-complete recovery of the MoS 2 Raman and PL spectra to their initial peak positions, linewidths, and intensities ( Figure 4). These results provide further confirmation that the ALD AlO x /MoS 2 interface reported here is characterized by weak electrostatic interactions at the 2D/3D interface. Indeed, the finding that AlO x -induced changes to the optoelectronic properties of MoS 2 are reversible upon removal of the coating indicates that the ALD process does not introduce a significant concentration of defects into the MoS 2 and that few chemical bonds between MoS 2 and AlO x are formed. Here, we note that the initial PL intensity was not fully restored after AlO x removal. However, this can be explained with the finite lateral dimension of the scratched window (≈4 µm 2 ). While this size is larger than the size of the laser spot (≈0.6 µm 2 ), it is smaller than the area of exciton diffusion (≈7 µm 2 ), which we estimated with the reported neutral exciton diffusion length of 1.5 µm for monolayer MoS 2 . [33] Thus some suppression of PL intensity is induced by the adlayer-doped regions outside the scratched window.
PL and Raman spectroscopy of individual monolayer MoS 2 flakes on Si/SiO 2 substrates further demonstrate the protective function of the nanometer-thin AlO x layer during metallization ( Figure 5). After evaporation of a 3 nm-thick gold (Au) layer directly onto monolayer MoS 2 , the initially strong photoluminescence (Figure 5a, black line) is nearly completely quenched (Figure 5a, red line), which is consistent with prior findings. [34] In contrast, the PL signal for the AlO x -protected MoS 2 with the gold layer atop is enhanced (Figure 5a, blue line) relative to the bare MoS 2 luminescence (Figure 5a, black line). The observed increase in the PL intensity can be explained with the electrostatic screening of charged defects within MoS 2 (e.g., vacancies) and the dielectric surrounding by the gold coating. [35] Additionally, the PL intensity of the gold-coated 2D material can increase through plasmonic enhancement, [36] though given the small thickness (3 nm) of the gold layer, this effect likely plays a minor role. For the case of the gold-capped alumina-coated MoS 2 , the stronger dielectric screening reduces the band gap as a consequence of band gap renormalization, resulting in an additional red shift of the exciton peak energies (Table S3, Supporting Information). [37] For the case of a 3 nm gold layer applied directly onto the surface of monolayer MoS 2 without AlO x present (Figure 5b, red line), we observed a softening of both Raman modes as a consequence of strain (4.1%) and charge transfer effects, [38] as well as a splitting of the A´1 mode, which has previously been attributed to the formation of gold islands on MoS 2 . [39] Indeed, www.advmatinterfaces.de the 3 nm thin gold coating forms a dense film of gold nanoparticles ( Figure S9, Supporting Information) rather than a smooth and continuous layer. While it has been reported that reactions between gold and MoS 2 are energetically unfavorable, [40] strong interactions between Au und MoS 2 are commonly observed and can lead to interfacial Mo-S bond formation and introduce lattice defects [41] and introduce lattice defects. [6] Here, such strong interactions between Au and MoS 2 are observed, as indicated by the measured XPS linewidth broadening of the Mo 3d core levels ( Figure S7, Supporting Information). In contrast, for the AlO x -protected MoS 2 monolayer, the energy positions and linewidths of the Raman modes (Table S4, Supporting Information) are hardly affected by Au deposition, demonstrating protection of the 2D material from metal adlayerinduced strain in the metal-insulator-semiconductor (MIS) structure ( Figure 5, blue line). Thus, the AlO x adlayer allows the desirable optoelectronic properties of the MoS 2 to be maintained, even at the ultrathin limit, making it a highly effective encapsulant.

Conclusion
In summary, we demonstrated in situ SE on 2D semiconductors as a powerful method for simultaneously understanding film nucleation and its influence on the optoelectronic characteristics of the 2D material during ALD. Excitonic absorption in 2D materials is sensitive to lattice defects and changes in the carrier concentration, which could thus be probed in real time during alumina deposition on a continuous film of CVD-grown monolayer MoS 2 that was used as the model 2D semiconductor. Here, we find that charge transfer doping of the MoS 2 occurs from the very first cycle of AlO x formation in the sub-monolayer regime and progresses with increasing cycle number. However, its saturation within the ultrathin film regime (∼15 Å) points to a modulation doping mechanism defined directly at the MoS 2 /AlO x interface. From the insights gained by in situ SE, we applied an ALD process for the encapsulation of 2D materials with a weakly bonded (physisorbed), continuous, and ultrathin AlO x layer at 40 °C. By implementing this AlO x coating as a spacer layer, we prevented defect formation in the 2D crystal during metallization and enabled photoluminescence enhancement for the Au/AlO x /MoS 2 system instead of quenching. Thus, the presented approach is promising for producing tunneling contacts or for seeding thicker dielectrics of relevance to the integration of 2D materials in electronic devices. Because the ALD process is not limited to the surface chemistry of MoS 2 , the ultrathin alumina encapsulant can be applied as a general interlayer that facilitates integration of 2D materials with bulk materials. Beyond the demonstrated ALD process of this study, we highlight that in situ SE during deposition enables critical insights into the impact of film growth on the optoelectronic properties of 2D materials, starting from the very earliest stages of nucleation. Such an approach can thus greatly accelerate the development of new coatings and processes of relevance for controlling the properties of 2D materials and their integration into functional systems.

Experimental Section
Atomic Layer Deposition: MoS 2 was coated with AlO x in a hot-wall ALD reactor (Fiji G2, Veeco CNT) in continuous flow mode. AlO x films were grown using TMA (electronic grade, 99.999%, STREM Chemicals) as the precursor and Ar (99.9999%, Linde) as the carrier gas during the first half-cycle at ≈0.09 Torr. The reactor wall and chuck temperatures were controlled to 40 °C. Importantly, the thermal energy at 40 °C is below the O 3 decomposition energy and the heat of desorption for sulfur vacancies in MoS 2 , thus substantially lowering the likelihood of undesired chemical reactions of the 2D surface in the presence of O 3 . [42] During the second half-cycle, O 3 was remotely generated using oxygen (99.9999%, Linde) and supplied with 150 ms pulses and peak pressures of ≈20 Torr. Each cycle of the TMA/O 3 process followed the sequence:  (same as shown in a) before (black squares) and after gold evaporation (red circles) on a bare MoS 2 flake, as well as before (green triangles) and after gold deposition (blue diamonds) on an AlO x -protected MoS 2 monolayer. A 532 nm laser at 295.2 µW power, focused through a 50× objective, was used for the measurements. The spectra are plotted with a vertical offset for clarity.

www.advmatinterfaces.de
bandgap (<800 nm), AlO x adlayer formation affects both the surface optical properties and the MoS 2 excitonic absorption. By acquiring SE data in the visible and the near-infrared range (1000-1700 nm), in which there is no absorption by MoS 2 , we were able to measure the optical properties and thickness of the adlayer independent from its effect on the MoS 2 optical properties ( Figure S2, Supporting Information).
X-Ray Photoelectron Spectroscopy: XPS data were acquired in the hybrid lens mode at pass energy of 10 eV and a take-off angle of 0° with a Kratos Axis Supra setup equipped with a monochromatic Al Kα X-ray source (photon energy = 1486.7 eV) operated with an emission current of 15 mA. The beam area was set to ≈2 × 1 mm 2 using the slot collimation mode. The binding energy of the hemispherical analyzer was calibrated with in situ sputter-cleaned silver, gold, and copper standard samples. Thereby, the kinetic energies of the Ag 3d (1118.51 eV), Au 4f (1402.73 eV), and Cu 2p (554.07 eV) core levels were referenced with an accuracy of 25 meV to the known peak values. The instrumental broadening (0.30 eV) was determined by fitting of the measured Ag 3d core level spectrum of the silver calibration sample with a Voigt function.
Atomic Force Microscopy: AFM measurements were carried out with a Bruker Multimode V microscope (Billerica, MA, USA) in ambient conditions using NSG30 AFM probes (TipsNano) with a nominal tip radius of 8 nm, typical resonance frequency of 320 kHz, and force constant of 40 N m −1 . Height images over 10 × 10 µm 2 and 1 × 1 µm 2 scan windows were acquired at a scan rate of 0.5 Hz with 512-point sampling. The rms roughness was determined from 1 µm 2 images. The ultrathin ALD coating was removed from the surfaces of monolayer and multilayer MoS 2 using a PtSi-coated AFM tip (Nanosensors) in contact mode at forces < 100 nN. The tip-sample force, F ts , in hard contact with the surface was calculated using Hooke's law, F ts = ks, where the measured deflection, s, was determined from a force-distance curve and the spring constant, k, was obtained from a thermal tune.
Raman and Photoluminescence Spectroscopy: Raman and PL spectroscopy measurements were carried out using a 532 nm green laser with the excitation power was fixed at 295.2 µW. Spectra were acquired with a 50× objective and spectrally resolved with an 1800 grooves/mm grating using an inVia Reflex Raman spectroscopy setup (Renishaw, United Kingdom). Measured Raman spectra were referenced to the 520 cm −1 mode of the Si substrate. The PL spectra were collected with a 10 s integration time using Renishaw's SynchroScan method of acquiring wide-range spectra. PL and Raman peaks were approximated with Voigt functions.

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