Active SERS Platform Based on 2D Materials-Flexible Nanotip Array

1. Introduction

Surface enhanced Raman scattering (SERS) is an enhanced version of Raman scattering enabled by metallic nanostructure. It is now well documented that enhanced Raman signal i.e., SERS is attributed to electromagnetic and chemical enhancement. While the chemical enhancement mainly comes from the interaction between the molecule and metallic substrate, the architecture of the substrates plays an important role towards electromagnetic enhancement[1,2]. Metallic nanostructure decorated substrate[3] such as nanospheres or nanoparticles[4,5], nanorods[5,6], nanowires[6], nano-stars[7,8], nanoscale roughened films[9,10] have enabled application in biosensing[11,12], pesticide detection[13], chemical analysis[14]. Apart from metallic nanostructure, plasmonic structures such as nanogap have also shown great SERS enhancement[15,16,17]. However, a dynamic SERS platform pivoted by tunable electromagnetic field localization is still lagging. Recently, enormous research has been focused to develop highly sensitive SERS platform based on two-dimensional (2D) materials[15,18,19]. Among 2D materials, transition metal dichalcogenides (TMDs) are an emerging class of materials with superior applications in opto- electronics. In particular, their dynamic response to an external perturbation such as mechanical strain[20], electrostatic gating[21], pressure[22] makes them a great candidate for a flexible SERS platform[8,9]. Amongst various external perturbation methods, mechanical strain is an easy and exciting way to control the electronics properties, band gap, phonon dispersion of TMDs facilitating applications in tunable photonics devices[23], flexible electronics[21], thermoelectric energy conversion[24], field effect transistor[25] and catalysis[26]. MoS₂-metal hybrid structure has been reported to show improved optical properties[27]. When 1L MoS₂ is decorated on nanostructure by means of mechanical exfoliation or chemical vapor deposition (CVD), intrinsic strain is generated on the monolayer, which dynamically changes the band-gap and its phonon properties.

This kind of system can have a dynamic response by laser pressure[25], which lacks reconfigurable functionality and controllability. Moreover, most of these studies are particularly focused on opto-electronics device application. Hence, an active system that can dynamically control the SERS intensity of 2D materials needs to be designed. Here, we report a system consisting of mechanically exfoliated 1L MoS₂ on flexible nanotip array. We could dynamically control the electromagnetic field that is confined in between the two neighboring nanotips by applying external mechanical strain, which would then tune the SERS intensity of 1L MoS₂, paving the way for a tunable 2D material-based SERS platform. In this paper, we mostly focused on tuning the Raman intensity of 1L MoS₂ using the proposed flexible nanotip array sample. However, the hybrid platform of 1L MoS₂/nanotip sample could also be extended to sense other molecules.

2. Materials and Methods

To realize the dynamic SERS platform, we fabricated metallic nanotip array on polyethylene terephthalate (PET) substrate. The detailed fabrication is schematically described in Figure 1(a). Briefly, at first, a large area (5 mm x 5 mm) anodized aluminum oxide (AAO) template containing hole arrays of ~70 nm diameter with interpole distance about ~100 nm is transferred to PET[28] (see SI1). Using the same AAO template, we fabricated two different types of samples which are (1) low-aspect ratio nanotip (2) high-aspect ratio nanotip. A direct deposition (normal to substrate) of Ag metal (~100 nm) leads to low-aspect ratio nanotip array[29] whereas low angle deposition (2 degree) of metal on AAO template leads to high-aspect ratio nanotip array (Figure 1(a) and SI2). After metal deposition, we stripped AAO template from the substrates using a Kapton tape to get the final sample containing nanotip array. Figure 1(b), (c) shows scanning electron microscopy (SEM) images of direct and angle deposited Ag nanotips, respectively, which confirms a good coverage of nanotip (see SI3 for cross-sectional images). Angle-deposited nanotips have higher coverage with ~70 nm diameter and shorter inter-tip distance of ~20 nm, as shown in Figure 1(d). In case of direct deposition, Ag nanoparticles cannot be deposited at the corner of AAO template due to shadow effect of nanoholes[30] (see SI2). However, in angle deposition process, Ag nanoparticles are well deposited on corner of AAO template, resulting in Ag nanotips with high coverage. Finally, 1L MoS₂ is transferred to Ag nanotip/PET and to bare PET substrate (SI4).

3. Results and Discussion

Figure 2(a) shows the optical microscope (OM) image of monolayer (1L), bilayer and few-layer molybdenum disulfide (MoS₂) flakes transferred on Ag nanotip with some part overlapping Ag metal film region. The FESEM image of the sample (area inside yellow dashed line box) is shown in Figure 2(b). High magnified FESEM image shown in Figure 2(b) confirms that 1L MoS₂ is transferred well on to Ag nanotips. The height profile extracted from atomic force microscope (AFM) image (SI5) further confirms the thickness of MoS₂ to be ~0.6 nm which is in good agreement with the previous reported data[31,32,33]. At first, we did finite element method (FEM) electromagnetic simulation of low and high-aspect ratio nanotip array, shown in Figure 2(c). For the low aspect ratio sample, the field is mostly confined to the tip region which would affect the Raman intensity of 1L MoS₂. For high-aspect ratio, since the tips are closer, the field around these tips interact with each other. This eventually confines more light in between the tips, thereby enhancing the Raman signal even more as compared to low aspect ratio nanotip sample. This enhancement is demonstrated experimentally as well. The Raman spectroscopy of the sample is measured using a commercial confocal Raman setup having 532 nm excitation laser with 0.5 mW power (see SI6). Figure 2(d) presents the Raman spectra of 1L MoS₂ on bare PET substrate, low and high aspect ratio Ag nanotip array sample. All the data are deconvoluted with three Lorentzian peaks to extract peak position, full width at half maxima (FWHM) and intensity. The Raman spectra of 1L MoS₂ on bare PET has two first-order vibrational peaks i.e., E¹_2g peak at 384.27cm^-1 due to in-plane vibration of molybdenum (Mo) and sulfur (S) in opposite direction, and A_1g peak at 405.82 cm^-1 due to out-of-plane vibration of S[34,35]. The difference between these two peaks is ~21.55 cm^-1, which is in accordance with the previous data for 1L MoS₂[36]. A third peak at 451.76 cm^-1 is 2LA (longitudinal acoustic phonon) mode and is observed due to double-resonance Raman (DRR) scattering[37].

For SERS intensity, we checked the intensity of the Raman signal of E¹_2g, A_1g and 2LA peak. The intensity for 1L MoS₂/high-aspect ratio nanotip is enhanced by factor of ~2.7 times compared to bare 1L MoS₂/PET (SI7). To further demonstrate our proposed tunable SERS platform, we used 1L MoS₂ transferred on high aspect ratio flexible nanotip array sample since it offers enhanced intensity as compared to low-aspect ratio nanotip array in its flat condition (Figure 2(d)).

The schematic diagram of the method to tune the SERS intensity of the sample is presented in Figure 3(a). As the sample is on a flexible substrate, we can apply both compressive (inward bent) and tensile (outward bent) strain to change the overall gap width in between nanotips which ultimately increase or decrease the localized field intensity (Figure 3(a)). The strain-controlled field confinement eventually manipulates the SERS intensity of 1L MoS₂. To avoid any artifact due to change in experimental condition, we fabricated all the sample at the same time with similar dimension and measured under similar bending condition. Furthermore, laser was focused on the same spot on 1L MoS₂ for all the measurements at different bending condition (see SI6). The detailed calculation of bending strain is presented in SI9. The sample was bent systematically with a strain magnitude of 0.82 %,1.22 %, 1.57 % in both inward (negative sign) and outward (positive sign) direction. Figure 3(b) and SI10 present the strain dependent Raman spectra of 1L MoS₂/PET and 1L MoS₂/nanotip. The extracted peak positions (mode peak), FWHM and intensities are summarized in Figure 3(c).

At flat condition, for 1L MoS₂/PET, the peak positions of E¹_2g, A_1g and 2LA are 384.27 cm^-1, 405.82 cm^-1, 451.76 cm^-1, respectively. For 1L MoS₂/nanotip, the respective peaks positions are shifted to 384.04 cm^-1, 403.44 cm^-1 and 454.06 cm^-1. While we observed a small shift (-0.06 %) in E¹_2g peak position, A_1g peak is shifted by -0.58 %. Since the E¹_2g peak is insensitive to the substrate features, small peak shift is due to the additional strain induced by nanotip. A huge change in A_1g peak is due to the charge doping from nanotip to 2D material[38,39]. The 2LA peak is red shifted by 0.5%. At flat conditions, both FWHM and SERS intensity of MoS₂/nanotip is increased because of enhanced strain and field localization in between nanotips.

The mode peak E¹_2g of 1L MoS₂/PET is shifted by +0.3 % and -0.4 % when it is under compressive and tensile strain of 1.57 %, respectively as observed by other researchers[40,41]. When 1L MoS₂/nanotip is under compressive strain, E¹_2g is red shifted by 0.4 % which is opposite of what we observed in 1L MoS₂/PET. To explain this anomaly, we consider that at flat condition 1L MoS₂ is fixed on top of nanotip as a tent without slippage. Under compression the contact area between MoS₂ and metal increases. As MoS₂ approaches the metal surface it takes the shape of the nanotip, and tensile strain is only exerted on 1L MoS₂ at the tip of the metal[25] which explains the anomalous redshift of E¹_2g peak under compressive strain. There is a very small change (-0.02%) in E¹_2g peak due to tensile strain on MoS₂/nanotip. The out-of-plane vibration mode (A_1g) is shifted by -0.18 % and -0.42 % for compressive and tensile strain of same magnitude (1.57 %) for 1L MoS₂/PET, respectively. However, it is shifted by -0.31 % (compressive strain) and 0.02 % (tensile strain) for 1L MoS₂/nanotip. When 1L MoS₂/nanotip is bent inwardly, 1L MoS₂ becomes much closer to Ag nanotip, thereby causing n-type doping of 1L MoS₂ as confirmed by the redshift of A_1g peak. For 2LA peak, the shift is about ~-1.0 % in both bending directions in 1L MoS₂/PET sample. The redshift is due to a change in lattice constant which affects the band structure of the material[42,43]. The charge doping and strain also affects the line width of the spectra. FWHM of these modes are inversely proportional to phonon lifetime. We observe that on PET and nanotip, FWHM of 2LA mode increases(decreases) with compressive(tensile) strain which is in good agreement with previous results[44]. Under compressive strain, A_1g and E¹_2g peak broadening is observed due to electron-phonon scattering[45] and tensile strain[46], respectively on nanotip array. On the other hand, there is an increase in phonon lifetime in 1L MoS₂/PET substrate, which leads to decrease in FWHM. However, for outward bending, the effect of uniaxial tensile strain is dominant rather than doping, leading to increase in FWHM of both the optical phonons in 1L MoS₂/nanotip[47,48]_.

In terms of SERS performance, the strain-controlled intensity change of Raman modes is also investigated. For 1L MoS₂/PET, the intensities of all mode peak are decreased significantly by ~75 % for both directions of bending. This decreasing tendency is not desirable for a highly sensitive tunable SERS platform. On the other hand, for MoS₂/nanotip, the intensities of the mode peaks E¹_2g, A_1g and 2LA are tuned from -45.6 %, -40 %, -33.3 % to +186.7 %, +216.6 %, +374.3 % by simply applying strain from +1.57 % (tensile) to -1.57 % (compressive). Please note that it is surprising to observe that the tuning range is in increasing order from E¹_2g to 2LA, which needs further study. We performed finite element method (FEM) simulation of structural and electromagnetic field of the nanotip array, shown in Figure 3(d), (e). From the structural simulation, it is found that there is enormous strain localization in between the tips when bending strain is applied, which might cause the inter-tip gap width to increase or decrease. The electromagnetic field simulation confirms the increase or decrease of field confinement when the nanotips are closer or far-away (Figure 3(e)). In pursuit of enhancing the SERS intensity, we deposited ~10 nm diameter Ag nanoparticle on 1L MoS₂/nanotip array to fabricate nanoparticle/1L MoS₂/nanotip geometry. These nanoparticles decorated on 1L MoS₂/nanotip would not only exert additional strain but also enhance the field confinement in the gap created by it. The schematic diagram of the sample is presented in Figure 4(a). Shown in Figure 4(b) is the SEM image of Ag nanoparticle deposited on bare nanotip array on PET. From the image, it is observed that the nanoparticles are decorated homogeneously on the surface of the tip. The strain dependent Raman spectra for nanoparticle/1L MoS₂/nanotip sample (SI11) are shown in Figure 4(c), SI2. We applied compressive strain to the sample to enhance the Raman signal. We observed E¹_2g mode split into two peaks at 372.85 cm^-1 and 381.66 cm^-1, suggesting additional strain due to nanoparticles at flat condition[20]. The extracted parameter corresponding to E¹_2g(M), E¹_2g, A_1g and 2LA mode is summarized in Figure 4(d). Similar to 1L MoS₂/nanotip, there is minor change in peak position of E¹_2g and A_1g. However, E¹_2g(M) and 2LA peak shifted by -0.73 % and +0.6 %, respectively. The FWHM of E12g and 2LA peak is increased whereas it decreased for E¹_2g(M) and A_1g. It is interesting to point out that the intensity of SERS signal is greatly enhanced as compared to 1L MoS₂/nanotip with inclusion of nanoparticles. The significant enhancement in intensity of Raman peaks is due to stronger field localization between nanoparticle and nanotip separated by 1L MoS₂. Beyond -1.57 % of strain, the intensity, peak position and FWHM of all the modes are reverted to their initial values. This might be due to the damage or sliding of MoS₂ by excessive strain[25]^.

Enhancement factor is an important parameter to validate the SERS performance (SI13). The experimental enhancement factor (EF_Exp) is calculated by:

$${EF}_{Exp}=I_{SERS}/I_{Reference}\left(\text{SI11}\right)$$

(1)

where $I_{SERS}$ is the surface enhanced Raman intensity (in our case nanoparticles/1L MoS₂/nanotip and 1L MoS₂/nanotip), $I_{Reference}$ is the Raman intensity of bare 1L MoS₂ on PET substrate. The highest enhancement factor is calculated to be 12, 20, 34.1 for E¹_2g, A_1g, 2LA mode, respectively, at – 1.57 % (inward) strain for nanoparticles/1L MoS₂/nanotip. To find out the strain dependent inter-tip gap width change, the simulated enhancement factor for different gap width is normalized and compared with normalized EF_exp, shown in Figure 5(b) (SI14 and SI15). From the data, it is concluded that there is an overall gap width increase (decrease) of 10 nm for inward (outward) bending. The modulation of depth of enhancement factor is of the order 10³ as compared to lowest enhancement factor of 1L MoS₂/nanotip at outward bending (Figure 5(c)). We could also achieve initial intensity at flat after bending experiment (SI16), showing reconfigurability of the platform. We checked the intensity at various spots in nanoparticles/1L MoS₂/nanotip sample (SI17). We observed similar SERS signals from five different spots, suggesting a large area SERS platform. A comparative study between our work and reported work is presented in SI18 and the respective graph is shown in Figure 5(d). Although some of the work reported a good enhancement factor, our work shows the ability to control the enhancement factor from low to high for first and second order Raman signal.

4. Conclusions

In summary, we demonstrated a mechanically active SERS platform based on nanoparticles/1L MoS₂/nanotip and 1L MoS₂/nanotip fabricated on PET substrate. We recognized three different mode E¹_2g, A_1g and 2LA in Raman spectra of MoS₂ on bare substrates and nanotip array. The peak position and FWHM are affected by the external strain induced on it. More importantly, the SERS intensity is enhanced (reduced) significantly with inward bending (outward) of 1L MoS₂/nanotip sample. To enhance the signal even more, silver nanoparticles are deposited on 1L MoS₂/nanotip sample, forming nanoparticles/1L MoS₂/nanotip geometry. With the external inward bending, this sample offered enhancement factor as high as 35. The overall enhancement factor is tuned from 1.23 to 35, showing an excellent tunability of the SERS platform. The change in intensity (or enhancement factor) is due to the change in gap width which is confirmed by FEM structural and electromagnetic simulation. Our work may find applications in developing future tunable devices for energy, sensing, and opto-electronics.