Plasmonic Photoresistor Based on Interconnected Metal‐Semiconductor Grating

Metal‐semiconductor nanostructures in various configurations are extensively used in photodetection, photocatalysis, and photovoltaics. For photodetection purposes, the working principle is straightforward; on illumination, generated charge carriers in excess lead to a decrease in resistance. Notably, using an interconnected metal‐semiconductor grating, it is observed and now reported an opposite response, an increase in the resistance. Such photoresistors are fabricated through wrinkle structuring and oblique angle material deposition methods. It is found that the controlled wrinkling leads to large‐area 1D periodic structures with coexisting cracking perpendicular to the grating direction—such cracks are used as connections between the two‐point contact measurement through the associated gold layer deposition. An enhanced current reduction is further observed on photoexcitation for an additional deposition of an amorphous titania layer. Subsequently, a discussion on the mechanisms and interaction between hot electron injection, charge carrier recombination, and thermalization is presented. Supported by numerical modeling, the angle‐resolved plasmonic modes with the photoresistance can be correlated. The ease of layered deposition of the materials allows one to extend the studies on cavity‐based structures with sandwiched titania layers as hotspots. This simple, scalable, and robust fabrication method thus promises an efficient routeway toward photosensor development in which plasmon‐mediated hot electrons play a crucial role.


Introduction
The range of applications of plasmonic nanostructures has expanded exponentially in the last couple of decades due to their Plasmonic 1D gratings play a crucial role in device formation for many applications in the fields of sensing, [23] polarizing spectral filtering, [24] structural color filtering, [25] bandgap tuning, [26] beam steering, [27] and others. Since the grating can be considered as an ensemble of individual nanobars, the resonant wavelength for the plasmonic grating undergoes redshift compared to that of a single nanobar; thus, a change from evanescent to radiative diffraction modes results in broader plasmonic linewidths. [28] It should be noted that, depending on the width of these metallic nanobars, both propagating surface plasmons (SPP), [29] as well as localized surface plasmon resonance (LSPR) [30] can be excited. Notably, these metallic gratings with narrow slit widths also allow additional slit modes [31] that combine with the plasmon modes on the surface to provide extraordinary transmission (EOT) in the far field. [32] Such plasmon modes favoring EOT in metallic grating have also been investigated to demonstrate hot electron-enabled photodetection below the semiconductor bandgap. [29] In any case, the generation of hot carriers or nonthermal electrons on light illumination is negligible compared to the thermalized carriers. On shining a metallic surface with photon energy, three processes coexist; excitation of an electron-hole pair, electron-phonon collision resulting in lattice heating, and electron-electron interactions leading to thermalization. Among such processes, most of the absorbed power results in a change in the electron distribution near the Fermi energy, i.e., overall heating of the structure, rather than producing the desired high-energy electrons. [33] For metallic nanostructures, the localized surface plasmons produce highly intense electric fields that boost the nonthermal electron-hole pair generation (since the plasmons decay through Landau damping [34] ); the energy can be dissipated again via scattering, in terms of heat or through an electron (hot-carrier) transfer into an adjacent semiconductor matrix. [35] Generally, 1D plasmonic gratings are fabricated on a largescale through preferred techniques like physical vapor deposition and mechanical lift-off, [36] nanoimprint lithography, [25] softlithography, [37] and others. These approaches require 1D mask/ template generation that can be easily formed over a large area through laser interference lithography. [2,38] In contrast, serial writing techniques like direct-laser writing [39] and electronbeam lithography [35] possess the capabilities of producing structures with control over shape and geometry. However, these come at the cost of the limited patterned region, extensive fabrication time, as well as excessive energy consumption. [40] Hence, affordable 1D structures through favorable fabrication schemes are desirable for real-life device manufacturing. Interference lithography-mediated 1D-mask formations, followed by metal deposition and lift-off, result in plasmonic grating over a large area; [36] yet, the number of procedural steps involved can be a constraint, because of the needs to be precisely optimized to ensure successful lift-off and pattern transfer. Particularly in this regard, 1D pattern formation through wrinkling on a soft polymeric substrate (PDMS; polydimethylsiloxane) [41,42] not only evades the multistep fabrication process but also enhances the uniformity in the range of tens of centimeters square area as well as offers scalability to macroscopic areas. [43] Benefitting from the large-scale wrinkled geometry and ease of metal deposition, grating coupled plasmonics has emerged in the last few years to promote a plethora of applications. [44,45] The present work builds on a large-area fabrication scheme to realize plasmonic grating-based photoresistors on a flexible substrate. By combining the wrinkling-assisted grating formation with oblique-angled deposition (OAD) of metal, we utilize the versatility of the well-known route to attain various geometries of different complexity levels. [46] As a core part, we discuss the photoresponses of the proposed metal-semiconductor geometry, which consists of an amorphous TiO 2 semiconductor layer over concave gold (Au) nanobars, supported on a flexible PDMS substrate. [47] We use cyclic photoexcitation to describe the underlying charge transfer mechanism, which is measured with two metal contacts parallel to the 1D grating structure. Note, the lattice structure between the metal contacts is conductive due to cracks perpendicular to the grating and the subsequent metallic film deposition. Furthermore, we correlate the charge transfer mechanism with numerical simulations to conclude the observation of the photoresistance properties. The observed photoresponse, defined as the difference in the current values between the "off" and "on" states, is compared with known plasmonic photodetectors. [13] Our fabrication approach provides additional benefits, which are relevant for forming cavity modes and is suitable for easier accessing of plasmonic hot spots. [48]

Charge Transfer Properties of the Proposed Photoresistor
We start by describing the concave Au/TiO 2 grating in Figure 1, where the structure consists of three components. First, a wrinkled PDMS grating that serves as a substrate and has a periodic (sinusoidal) surface profile. In general, the wrinkling technique provides an accessible periodicity range from submicron to micron; [49,50] however, wavelengths up to the millimeter range are also conceivable. [51] Nevertheless, we aim for a periodicity in the lower range of visible light (≈600 nm) because the damping of gold is lower at this wavelength. Additionally, when coupled with the plasmonic properties, the diffraction modes from the grating can reduce the damping by collective resonances. Second, a thin metal grating (thickness 30 nm) supports the excitation of the plasmonic modes for hot electron generation. The metallic grating is conductive because of the cracks (formed due to mechanical stress during the plasma oxidation process) that get interconnected through the metal deposition. The wrinkled substrate and the deposited metal form the concave Au grating. Third, a 50 nm thin titania layer is further deposited to characterize the recombination of the charge carriers. We measure the conductivity in the proposed photoresistor architecture (i.e., the concave Au/TiO 2 grating) with a bias voltage that is applied using a two-point contact, as shown schematically in Figure 1a. For this purpose, 60 nm thick gold electrode stripes, parallel to the grating lines, are deposited on top of the final layer. The distance between the two counter electrodes for current measurement is 880 µm.
To validate our experimental findings, we compare our results with a metallic grating that shows no connectivity between the electrodes. Such a metallic lattice consisting of nonconnected strips is fabricated using a combination of lithography and lift-off methods that leads to the formation of metallic photonic crystal slabs (mPhCs) as described in our previous works. [21,36] A detailed description of the fabrication and electrical characterization of the mPhCs is provided in Text S1 of the Supporting Information ( Figure S1, Supporting Information) with piecewise photoresponse measurements for its various components discussed in Figure S2 (Supporting Information). For such mPhCs, the semiconductor (crystalline Titania) waveguide is actively connected to the external circuit using similarly deposited electrodes, whereas the nonconnected metallic (Au) bars resting on top of the waveguide respond to the plasmonic excitation ( Figure S1a, Supporting Information). For an applied bias voltage, the dark current values range in the order of 10 −11 Ampere, due to the high resistivity of the crystalline TiO 2 (1.76 TΩ, Figure S2e,iii, Supporting Information). On illumination with resonant frequencies, the plasmonically excited Au nanobars decay by generating kinetic hot electrons that get injected into the semiconductor matrix after crossing the Schottky barrier ( Figure S1a, Supporting Information). This increases the electron density of the Titania waveguide, as a result of which the measured photocurrent values increase ( Figure S2f,v, Supporting Information), as reported earlier. [12,36] For our proposed photoresistor, we have carried out similar piecewise photoresponse measurements ( Figure S3, Supporting Information) on individual structures by considering each of them as an active part of the external circuit; correspondingly, their resistance values are provided in Figure 1b. As expected, the uniformly deposited 30 nm Au layer shows a very low resistance (3.08 Ω, Figure 1b,i) in comparison to the similarly deposited highly resistive 50 nm titania layer (17 GΩ, Figure 1b,ii). However, for a combination of both layers, the presence of an additional TiO 2 layer (50 nm) on top of the Au layer does not affect the overall resistivity (4.99 Ω, Figure 1b,iii) and hence doesn't disconnect the conductive gold layer from Figure 1. a,i) Schematic of a plasmonic photoresistor with micron range interconnecting channels fabricated by PDMS wrinkling. b) Current-voltage (I-V) characteristics of various structures deposited on wrinkled PDMS: i) 30 nm of gold (Au) film, ii) 50 nm of amorphous titania (aTiO 2 ) film, iii) Au /TiO 2 films of total thickness (30 + 50) nm, iv) interconnected concave Au grating of 30 nm thickness, and v) concave Au/TiO 2 grating, with total thickness (30 + 50) nm. c) Charge transport through the interconnected metallic grating for the i) 'off' and ii) 'on' states, depicted schematically along with the band diagram.
the external circuit. It should be noted that such combinations are insignificant to plasmonic excitations, which motivate the inclusion of concave Au/TiO 2 grating for designing the proposed photoresistor. The plasmonically significant concave Au-grating, when included in the external circuit, manages to provide low resistance (0.39 kΩ, Figure 1b,iv) due to the interconnections between the metallic bars. However, for the proposed photoresistor with the concave Au/ TiO2 grating, one order higher resistance (2.64 kΩ, Figure 1b,v), is observed. This is because the connection to the external circuit, being solely dependent on the in-between interconnections, can be somewhat hindered by the TiO 2 deposition on top.
Thus, we suggest the following charge carrier transport, described schematically in Figure 1c, as per our hypothesis. For the "off" state, applying a bias voltage measures a dark current (I off ) of the order 10 −4 Ampere due to the ease of flow of electrons through the interconnected metallic bar and the in-between thin TiO 2 layer (Figure 1c,i). For the "on" state, illumination with frequencies near the plasmonic excitation of the metallic nanobars produces hot electrons through nonradiative decay that possesses sufficient energy to overcome the Schottky barrier and get transferred to the conduction band of the adjacent TiO 2 layer. Note that we used crystalline titania in the mPhCs structure, which indicates that the importance of amorphous titania has been underestimated, as reported by Liang and coworkers. [52] Since the interconnected Au grating itself is connected to the external circuit, the holes generated due to charge separation recombine with the electrons coming from the external voltage source, resulting in reduced current values (I on ). The amorphous TiO 2 layer acts as an electron scavenger, without which the generated electron-hole pairs could have again recombined within the metallic nanobars, thus failing to exhibit a higher reduction in current. This is visible in Figure S3d,v,e,v (Supporting Information) for the cases of concave Au grating and concave Au/ TiO 2 grating. Additionally, radiative decay of the incident energy manifests in electron-electron scattering leading to thermalization, and electron-phonon collision resulting in lattice heating. Such mechanisms increase the overall resistance of the concave Au/TiO 2 grating, thus showing a drop in the measured current under photo excitation. Unlike a wide variety of photodetectors that record an increase in photocurrent, the proposed photoresistor, therefore, exhibits quite the opposite i.e. a decrease in current values on photoexcitation.

Fabrication of the Concave Au and Au/TiO 2 Grating
For the proposed photoresistor, we use a PDMS substrate produced by the wrinkling method. The preparation of such wrinkled substrates involves in situ plasma oxidation, operated at microwave frequency [50] using oxygen as a process gas. The details of the PDMS grating fabrication are provided in the experimental section. Briefly, the oxidation of the prestressed substrate leads to a more rigid SiO 2 layer on top of the PDMS surface. After the mechanical stress is released, the SiO 2 layer forms a wrinkled structure. Further details about the wrinkling process and corresponding atomic force microscope (AFM) images are documented in Figure S4, Supporting Information. Such 1D patterning can possess irregularities like y-defects and cracks, [53] often regulated with the controlled strain release pattern. When reaching the initial length, the lateral expansion of the substrate causes cracks perpendicular to the wrinkle direction. [54] After gold layer vapor deposition, these cracks provide connectivity with electrical characterization comparable to that of a continuous gold layer. The fabricated PDMS grating has an average structural periodicity and grating amplitude of ≈627 nm and ≈137 nm, respectively as confirmed by statistically analyzed AFM micrographs. Figure 2 shows the fabrication of the proposed photoresistor using the well-known OAD method. The deposition angle (defined as the angle between the grating normal and incoming material flux) and deposition rate are the only variables for the OAD technique to control the shape and geometry of these concave Au gratings, whereas a more generalized technique named Glancing Angle Deposition (GLAD) also involves rotation of the substrate plane. For OAD, we fixed the deposition angle at 65° and adjusted the asymmetric deposition from both sides to form concave Au grating on top of the patterned PDMS surface. Previous work with deposition from a single side has resulted in asymmetric structures [44,46] that exhibit anisotropic diffraction properties. In this scenario, the two-step deposition from both directions leads to the symmetrical curvature (Figure 2b,i). We applied an extra carbon layer to increase the contrast in the FIB cross-section. During the deposition process, the thickness of the Au layer was maintained at 30 nm from both directions. As shown in Figure 2b,ii, we deposited an additional 50 nm TiO 2 layer, normal to the surface, which resulted in a continuous high-indexed layer above the concave Au grating (without the carbon layer). The uniform deposition of the TiO 2 layer confirms the continuity between the regions above and in between two consecutive concave gold bars. Figure 2c highlights the presence of cracks on the PDMS surface, before and after the Au deposition. Hence, it is confirmed that the grating structure bars are separated and only connected by metal-filled cracks. The wrinkling technique combined with the OAD technique provides centimeter-scaled grating structures with diffraction modes in the visible range (see Figure 2d) suitable for conductivity measurements; here one can estimate the benefits of these Au-deposited unavoidable cracks in terms of interconnections between two electrodes deposited parallel to one another. Because of these, we can directly compare and analyze the physical mechanisms in contrast to the mPhCs that have parallel but nonconnected plasmonic grating. [36] The choice of parallel electrode direction is thus essential in the current study to differentiate between the positive photoresponse (mPhCs) and negative photoresponse (concave Au/TiO 2 grating). However, previous studies [44] with Au nanowires on PDMS wrinkles show resistive measurements using the perpendicular orientation of metallic grating between electrodes to report longitudinal sheet resistance of values as low as 15 Ω/sq.

Plasmonic Modes of the Concave Au and Au/TiO 2 Grating
The optical characterization of the concave Au/TiO 2 grating is undoubtedly essential for a proper correlation of the plasmonic resonance to the photoresponse measurements. First, we study the concave Au grating (without the titania layer) to identify the fundamental plasmonic resonances. For this, we modeled an Au grating with a periodicity of 630 nm, bar width of 400 nm, and thickness of 30 nm using the finite-difference in time-domain (FDTD) simulation mode. Figure 3a compares the transmittance from the FDTD simulation (Black curve) with the experimentally measured data (red curve) recorded using a UV-vis-NIR spectrometer. In the experimental section, one can find more details about theoretical modeling and spectroscopy. The optical characterization is carried with light, polarized perpendicular to the Au grating structures, i.e., the transverse magnetic (TM) polarization. Since the plasmon resonances at the Au grating occur only at TM polarization, the light polarization parallel to the grating (transverse electric, TE) remains noneffective. [28] The agreement between simulation and experiment allows a closer look at the electric field distribution and the associated surface charge densities necessary for electrical resistance or conductance. While the broader mode (at 1,300 nm) corresponds to the pure plasmonic modes due to nanobars of width ≈400 nm, the sharper mode (at 600 nm) represents the diffraction-mediated plasmonic modes, similar to surface lattice resonance (SLR) [18] from a 1D periodic array. Compared to single nanobar, SLR mode shows much less damping and could be suitable for enhanced hot electron injection. We would like to mention that the diffraction mode occurs at a lower wavelength than the grating period because of the effective refractive index of PDMS, metal, and air. The dependence of the position of the resonance dip on the periodicity, metallic bar width, and metallic bar thickness have also been explored via FDTD simulations and are presented in Figure S5 and Text S3, (Supporting Information). From this modeling, we can conclude that the wavelength of the diffraction mode for the cover and substrate depends on the periodicity but is independent of the bar width and bar thickness while considering a constant periodicity. The bar width, on the other hand, has direct relevance to the position of the broad pure plasmonic mode and its interaction with the diffraction mode to determine the diffraction-mediated plasmonic mode. Moving a step further, by including the high-indexed TiO 2 layer over the Au concave grating, one can expect a change in the supported resonant modes observed from the recorded transmittances in Figure 3b. Both the dominant resonant modes get red-shifted, which can be confirmed directly from the simulation spectra. The diffraction mode is shifted from 596 nm to 626 nm, whereas the LSPR mode gets shifted from 1,319 nm to 1,459 nm. For the experimental counterpart, we fabricated both concave Au gratings and the concave Au/TiO 2 grating, separately. These distinct samples show reproducibility of the grating fabrication; however, it leads to minor deviations in the transmission spectrum. The changes can be attributed to certain factors such as differences in i) structural variables, ii) material properties, and iii) the presence of defects. For the Concave Au grating, the position of the diffraction mode matches quite effectively (tolerance of ± 10 nm) due to the fixed wrinkle periodicity between the model and experiment. The broad plasmonic dip due to Au grating is found to be shifted for the experimental counterpart, which can be caused by differences in the dielectric properties of the modeled Au grating and the experimentally fabricated one. Similar conditions are also implied when materials for the titania layer are considered. The broadening of the experimental LSPR peaks is again caused by a probable difference in the structural variables as well as due to the inclusion of defects in the form of cracks in these fabricated structures. However, the nature of the curves remains the same for the theoretical modeling as well as the experimentally obtained transmittances. In both of these cases, the concave grating structures involving the noble metal Au (with filled d-shells) enable the interband transition, showing a transmission maximum at ≈500nm due to the existence of bulk absorption modes which can be readily spotted in Figure 3a,b as well as in relevant plots in Figures S2 and S3  Next, we study the electric field and surface charge distribution in Figures 3c,d, and Figure S6 (Supporting Information) at selected wavelengths to better understand charge carrier transport. For the Au grating structure at 596 nm, the diffraction mode resulting in diffraction-mediated plasmonic excitation causes radiation leakage through the cover region, along with simultaneous excitation of the edges of the concave metallic bars. On the other hand, the pure LSPR mode at 1,319 nm can be identified from the strongly localized electric field, existing only at the edges. The surface charge distributions in Figure S6 (Supporting Information) also show that the pure plasmonic mode at 1,319 is dipolar, whereas those at 596 nm and the 762 nm mode are quadrupolar. However, these quadrupolar modes couple differently to the far field and are, also observed in the transmittance intensity. For the Au/TiO 2 structure, the diffraction efficiency is increased because of the involved high-indexed grating; the diffraction mode for the metal-cover interface at 626 nm records intensified fields in the cover and along the metal edges. The dipolar mode at 1,459 nm shows characteristic field localization, as observed previously at 1,319 nm for the Au grating case. An additional dip at 753 nm is observed, which is also recognized in the experiment at ≈866 nm, showing mode confinement in the TiO 2 layer without any excitation of the metallic edges. Such field confinement may correspond to a cavity mode because of a sufficiently thick titania layer. However, this mode is leaky through the metalsubstrate interface and is of lesser importance in relevance to the present study. Thus, supported by the FDTD simulations, we can effectively distinguish between the SLR dipolar, quadrupolar, and cavity modes.

Angle-Resolved Photoresponse of Concave Au/TiO 2 Grating
Our experimental photoresponse measurement setup consists of discreet light sources to excite the calculated modes selectively. The LED light source covers a range of wavelengths, including excitation near the titania bandgap (405-505 nm) and excitation of the plasmonic resonances (590-880 nm). Furthermore, a customized setup makes it possible to collect angle-resolved photoresponse data. More details about the setup and the photoresponse measurements can be found in the experimental section and the Supporting Information (see Figure S1 and Text S1, Supporting Information). The corresponding optical modes are discussed in the dispersion diagram provided in Figure 4a. The angle-resolved transmittance study shows the diffraction mode (i.e., RA; for details, see Figure S5, Supporting Information) in the air (cover) and PDMS (substrate) with the starting   [55] ) at 626 nm and 900 nm, respectively. This diffraction mode is directly observed from the FDTD simulated spectra (Figure 4a,i), corresponding to the effective refractive indices toward the cover and substrate side. These modes split into branches with positive and negative slopes on increasing the angle of incidences. [21,36] The intermediate points with the plasmonic modes are noteworthy since these show less attenuation and certain field distributions due to their hybrid character (i.e., plasmonic + photonic diffraction). [46] One can also see these intersections in the experimental dispersion diagram (Figure 4a,ii) in terms of the intensity variations that are more clearly visible from the corresponding magnified plot (Figure 4a,iii), where the wavelength span corresponds to the range of the LED light sources. The diffraction-mediated plasmonic modes are also observed in the dispersion diagram for the concave Au grating structure without the titania layer (see Figure S7, Supporting Information). Next, Figure 4b contains angle-dependent photoresponse of the structure, specific to different excitation wavelengths. As a basic operating principle of a photoresistor, we observe a decrease in photocurrent when our structure is irradiated with light. The theoretical work of Dubi and Sivan indicates that the contribution of hot charge carriers is very small compared to thermalized charge carriers. [33] We, therefore, assume that the absorbed photons mostly flow into heating, with very few percentages going into the generation of hot charge carriers. Due to the additional free charge carriers in an amorphous titania layer, we observe a magnified thermalization effect. In an upcoming section, we investigate how significant the thermalization and plasmonic effects are concerning concave Au grating, concave Au/TiO 2 grating, and a cavity-based grating structure. For the present measurement with concave Au/TiO 2 grating, it is observed that in all of the measurements, the photocurrent saturates within 150 seconds and shows an exponential profile. From previous experience, we know that exposures near the titania bandgap lead to a direct photocurrent. [12,36] We observe this behavior at the LED light sources of 405 and 505 nm wavelengths. We can see the contribution of the metal-semiconductor grating at the angle-resolved current reduction at the wavelengths 590, 652, and 780 nm. However, it should be noted that the drop in current values also includes the thermalization effect in the background; apart from the loss of electrons due to transfer, the increase in resistance is also caused by electron-electron scattering.
The concave Au/TiO 2 grating contribution becomes clear as we want to quantify its efficiency. For normal photodetectors, such quantification is carried out in terms of "incident photon conversion efficiency (IPCE)" which is defined as the rate of conversion of an incident photon into an electron. Such IPCE is always observed through increased photoconductivity, i.e., increased current values due to the inclusion of photoconverted electrons into the active circuit. For our concave Au/TiO 2 grating, quite the opposite occurs where plasmonically excited electrons leave the active circuit into the adjacent amorphous TiO 2 layer. The holes recombine with the externally applied electrons, thus resulting in a decrease in the current  values. As mentioned earlier, due to the thermalization of the electrons, the resistivity of the Au grating also increases, which also affects the dropping of the current values under photoexcitation. Thus, in contrast to the previously reported plasmonic photodetectors where external or internal quantum efficiencies (EQE or IQE) are measured by considering only the plasmonically photoconverted electrons, our concave Au/TiO 2 grating is involved in different physical phenomena. For the sake of simplicity, we define this difference in current values between the "off" and "on" states as the measured photocurrent I ph = |I on −I off |. Further, we calculate photoresponsivity (PR) defined by PR = J ph /P in , (J ph is the modulation current density and P in is the power of the incident light per unit area) to quantify "Current Modulation Efficiency (CME)," using approaches similar to IPCE calculation. Note that this CME should not be confused with EQE or IQE as it involves separate chargetransport mechanisms, unlike the conventional photodetectors. More details about the angle-resolved CME calculation are provided in Text S6 (Supporting Information). Figure 4c,i-iii shows the angle-resolved I ph , PR, and CME as heatmaps calculated using Origin Software by using the values listed in Tables S1,S2, and S4. Between Figure 4a,iii, and 4c,iii, a close analogy can be observed; the regions of the plasmonic absorption (dark region corresponding to the transmittance minima) can be mapped to regions with higher CME. For example, the negative order diffraction mode for the substrate (RA sub ) at 45°@780 nm corresponds to a relatively higher CME value. However, for the case of 405 nm, other mechanisms might take part in the CME calculation. The excitation wavelength being close to the bandgap of the titania layer leads to increased electron-hole pair generation, which can be related to the observed enhancement in the I ph values. Additionally, the negative diffraction mode for the cover region (RA cov ) propagates along the titania surface and hence offers increased light-matter interaction with enhanced CME values for the case 45°@405 nm. One should also note from Figure 4a,iii that the observed transmittance maxima near 500 nm is due to interband transition that also affects the radiative losses [56] and thus plays a significant role in determining the CME at such lower wavelengths. Thus, the CME, calculated by considering the I on and I off values taken at the stationary state of illumination, is mostly an indicator of thermalization along with the plasmonic effect, which can be adjusted by the wavelength and angle of incidence.

Analysis of Thermalization Using Various Metal Semiconductor Gratings
So far, we have shown that the thermalization effect and plasmonic excitation can occur on metallic/semiconductor grating structures, providing a photoresponse in terms of a drop in current values, i.e., it acts like a photoresistor. We would now like to demonstrate that we can enhance such an effect by introducing a cavity structure. For this purpose, we use our OAD method to fabricate a concave metal-semiconductor-metal (MSM) grating. The sandwiched titania layer between two parallelly deposited concave metallic nanobars (in the vertical direction) has a thickness of 10 nm. A 50 nm titania layer is further deposited onto the concave MSM grating as a final layer. Figure 5 shows the time-dependent current signal for the concave Au, Au/TiO 2 , and the MSM/TiO 2 grating carried out using a similar photocurrent measurement setup. Figure 5a,i-iii shows the SEM-focused-ion-beam (FIB) cross-section of concave Au, Au/TiO 2, and MSM/ TiO 2 grating, with their photo-responses, compared in Figure 5b,i-iv,5c,i-iv, and 5d,i-iv, respectively, under normal incidence excitation. For the present setup, since the power of the LED sources varies with the wavelength, it is more sensible to compare the response of the structures separately for each of the wavelengths. Although the dark current values differ for these structures, the current scale in the Y axes is kept to the same span (0.7 mA) to compare the modulation strength directly. On excitation with 405 nm, in the absence of the TiO 2 layer, the concave Au grating exhibits photoresponse with a reduction in current values by 0.4 mA during the switching-on period. In comparison, on adding the TiO 2 layer, a higher modulation strength of 0.56 mA is observed, exhibiting an enhancement of 40%. Since compared against the same excitation LED, the enhancement on the inclusion of TiO 2 can be considered independent of the incident source power.
Instead, such enhancement of the modulation strength is wavelength-dependent; for 625 nm LED excitation, the modulation strengths for the concave Au grating and concave Au/ TiO 2 grating are 0.15 and 0.45 mA, respectively, thus showing an enhancement factor of 200%. This increment can be attributed to the plasmon-induced charge transfer of electrons from the metallic bars into the adjacent amorphous TiO 2 layer, in the backdrop of the thermalization effect. Due to the high resistivity of the TiO 2 layer, the dark current value decreases with its inclusion. As confirmed earlier, the circuit for current measurement is closed through the interconnected metallic bars, regardless of the TiO 2 layer. We want to highlight that the thermalization effect by including the plasmonic grating as an active circuit component has resulted in a modulation current (in mA) that is several orders higher in magnitude than the other plasmonic photodetectors (generally operating in nA). [13,57] At wavelengths near the plasmon resonance of these metallic bars, nonthermal hot electrons can be produced, resulting in charge transfer from the nanobars into the adjacent amorphous TiO 2 . A contrast between Figure 5b,iv,c,iv carries the confirmation of such charge transfer processes that show enhancement in the modulation strength compared to the contrast found between Figure 5b,i,c,i. The current response also depends on the bias voltage; Figure S9 and Text S6 of Supporting Information display a linear relationship between the applied voltage (−3V to 3V) and measured current for the concave Au/TiO 2 grating under illumination with different sources. For the present comparison in Figure 5, the bias voltage is kept constant at 1 V for all of the cases.
Unlike the Au/TiO 2 case, the MSM/TiO 2 configuration may have the discontinuity of the TiO 2 layer on the top surface due to increased grating height (now 70 nm instead of 30 nm).
Depositing electrodes on such surfaces allows for recording dark current in the mA range, proving once again that the connectivity due to the cracks in the PDMS layer lowers the overall resistance of the closed circuit. The lower magnitudes in the dark current are brought in by the extra resistance from the high-indexed amorphous TiO 2 layers. However, the presence of an additional gold layer involves more electrons being agitated through the thermalizations (away from plasmon resonance) or removed to the adjacent amorphous TiO 2 matrix through plasmonic hot-electron injection (near plasmon resonances). These factors can be taken into account in consideration of the observed enhancement in the modulation strength compared to the other two cases. However, a quantitative contributory analysis of these factors is currently beyond the scope of the present study. The responsivity of the concave MSM/TiO 2 grating appears to be cascaded, which can be attributed to the multiple scattering between the layered structures and is confirmed by repeated measurements. The ease of fabrication of such complex geometries in the nanoscale opens up the possibility of realizing the hybridization of plasmonic modes [58] as well as optical modes (IMI+MIM = MIMI), [48] where a very thin insulating (I) layer of thickness t≈10 nm can be sandwiched between metal (M) coatings supporting surface plasmon modes. Such hybrid modes are suitable for accessing the plasmonic hotspots for different applications in enhancing surfaceenhanced Raman scattering (SERS) signals and sensing abilities. Thus, with our up-scalable and cost-efficient fabrication approach, one can easily reproduce these structures following the given variables. The reproducibility of the photoresponses from our proposed structures is also confirmed over a large span of time. For example, Figure S3

Conclusion
In summary, we have fabricated conductive grating structures by selective and layered deposition of materials on large-area periodic nanostructures that retain the plasmonic properties while remaining interconnected to serve as an active part of the circuit. Mechanical instability during wrinkle structuring has resulted in 1D lattice structuring in the visible region with vertical cracks. Verified by a two-point measurement with applied bias voltage, these cracks as defects can lead to conductivity after metal deposition. By illuminating this connected lattice structure at several excitation wavelengths, we observe a reduction in the current values, which we can correlate with plasmonic modes and the thermalization effect. Such effect is enhanced by applying an amorphous titania layer due to the material's additional free charge carriers, allowing scavenging of the photoexcited and transferred electrons from the metallic bars. The separated holes in the interconnected metallic bars (being an active part of the external circuit) recombine with the electrons from the externally applied voltages, increasing the circuit resistance. In addition, our cost-effective large-scale fabrication technique allows fabricating of metal-semiconductormetal nanostructures-based complex geometries that further improve the device performance. Overall, the fabricated structures show modulation of photocurrent ranging in milliamperes that are significantly higher by several orders than other metal/semiconductor-based nanostructures, as compared to photoresponses from lithographically fabricated nonconnected metallic lattices. The proposed application as photoresistors can be utilized for electron annihilation on photoexcitation, to record such a huge drop in current values. Such phenomena can find application in photoswitches and logical operations with the benefits of flexibility from our elastomeric substrates. [59] Our method of combining amorphous titania films alongside the plasmonics of the metallic bars opens up new perspectives on the optoelectronic properties of such conjugate systems. [60] Because of the long-range atomic disorder; such amorphous titania can play an essential role in photocatalytic performances, such as water splitting or photodegradation of organic materials, as reported by Liang's group. [52] By linking theoretical modeling, angle-resolved spectroscopy, and angleresolved photocurrent measurement, we have successfully identified the hybrid plasmonic modes and determined their current modulation efficiency (CME). This link and the simple fabrication technique are also crucial for relevant nanophotonic applications, which can use the free hot charge carriers for corresponding physicochemical processes.

Experimental Section
PDMS Grating Fabrication: A commercially available Sylgard 184 kit (Dow Corning, USA) was used to mold a 2 mm thick sheet of PDMS. A mixing ratio of 5:1 (w/w) was maintained for dimethylsiloxane oligomer and Pt-based crosslinking agent. [61] A two-step curing process -room temperature curing for 24 h, followed by curing at 80 °C for 4 h -was followed to retain a consistent elastic modulus. The 24 h slow curing process was done on a leveled plate to maintain uniformity in thickness. The cured PDMS sheet was cut into 4.5 cm × 1 cm stripes and clamped onto a custom-built stretching device for plasma treatment.
A computer-controlled MicroSys apparatus (Roth&Rau, Wüstenbrand, Germany) was used for low-pressure plasma treatments of PDMS surfaces. The cylindrical vacuum chamber, made of stainless steel, had a diameter of 350 mm and a height of 350 mm. The base pressure obtained with a turbomolecular pump was <10 −7 mbar. On top of the chamber, a 2.46 GHz electron cyclotron resonance plasma source RR160 by Roth&Rau with a diameter of 160 mm and a maximum power of 800 W was mounted. The samples were introduced by a load lock system and placed on a fixed holder near the chamber's center. The distance between the sample and the excitation volume of the plasma source was ≈200 mm. The process gas (oxygen 99.999%, Air Liquide) was introduced continuously into the active volume of the plasma source via a gas flow control system. A process pressure of 2 × 10 −2 mbar was adjusted. The PDMS stripes were stretched to 30% of their initial length before oxygen plasma treatment (treatment time 95 s at a microwave power of 800 W for all samples). Figure S4 (Supporting Information) depicts the basic wrinkling preparation procedure.
Thin-Film Deposition: The concave grating was fabricated with the oblique angle deposition technique. Fabricated PDMS grating was mounted on a custom-built sample stage, designed to achieve such deposition configuration. The chamber was pumped down in the range of 10 −7 mbar before the deposition, which resulted in a chamber pressure of ≈10 −6 mbar. A Telemark e-beam source of 7 kV was used to achieve a deposition rate of 0.1 nm s −1 throughout the process for both materials, Au and TiO 2 . During the deposition step of TiO 2 , a 10-rpm rotational speed was used for better homogeneity.
Photoresponse Measurements: Before the photoresponse measurement Au-electrodes of 60 nm thickness were deposited using an e-beam that was aligned parallel to the wrinkle direction, with a channel length of 220 µm and width of 4500 µm. The current was measured with a Beryllium Copper probe (Radius: 25 µm) between the first and fifth electrodes (thus, a total channel area of 0.0099 × 4 = 0.0396 cm 2 ) and was recorded by Keithley SMU (Keithley, 2612B). Different LED lights (Thor Labs), operated through a controller unit, were used as light sources, and the wavelengths were as follows: 405, 505, 590, 625, 780, and 880 nm. The distance was kept constant (2 cm) between the LEDs and the fabricated device; the same was followed for angle-dependent measurements. 1 V DC bias was applied for the complete measurement. A silicon photodetector (S120VC, Thorlabs) coupled with an optical power meter (PM100D, Thorlabs) was used to measure the power of the LED sources.
UV-vis-NIR Spectroscopy: UV-vis-NIR spectroscopic data was captured with transverse electric (TE) and transverse magnetic (TM) polarized light at transmission geometry, employing a Cary 5000 spectrophotometer (Agilent Technologies, USA). The beam spot was fixed at 5 mm × 5 mm for all of the measurements. A specific external accessory, universal measurement accessory (UMA), was used to produce the dispersion curves with different changing variables (polarization direction, plane of incidence). TM and TE polarization were recognized as the plane of incidence along and perpendicular to wrinkles.
FDTD Simulation: A 3D Electromagnetic Simulator operating in the finite-difference time-domain method was used (Lumerical FDTD) to perform the numerical calculations. [62] The sinusoidal grating representing the PDMS block was equated using the 'Custom' structure modifier, where the dimensions were matched to those determined by SEM imaging. To include concave gold grating and titania layer, 'Surface' structure modifiers with 0.25 and 0.22 radii of curvature were considered in the simulation model with material properties from Johnson and Christy, [63] and Sarkar et al., [64] respectively, both fitted using six coefficients, with a root-mean-square (RMS) error of 0.25. A plane wave source (λ = 400 to 1400 nm) was used to simulate the optical response under normal incidence illumination along the z-axis. Monitors with frequencies matching the wavelength span of the source were used to obtain the transmittances. Periodic boundary conditions were set along x and y directions, with perfectly matching layers along the z-axis. BFAST techniques were implied for oblique incidence to obtain the simulated dispersion diagrams. Charge density calculations were recorded using a 'current-charge density' analysis group. All simulations were set to reach an auto-shut-off of at least 10 −5 before reaching 300 fs of the simulation time.
Atomic Force Microscopy (AFM): AFM height images of the grating surface were taken with Dimension Series Fastscan (Bruker-Nano, Santa Barbara, USA) in ScanAsyst method using Nanoscope 9.10 software. The Fastscan-C cantilever (nominal frequency 300 kHz) used had a tip radius of 5 nm and a spring constant of 0.8 N.m −1 . The captured images had a dimension of 20 µm × 20 µm and a resolution of 512 px × 512 px. Further analysis of the topographical images was executed using a programmed Python 3.10 script ( Figure S4, Supporting Information).
Scanning Electron Microscopy (SEM): SEM images were captured at an SE2 detector in a NEON 40 FIB-SEM workstation (Carl Zeiss Microscopy GmbH, Germany). The micrographs were captured at electron high tension (ETH) of 3.5 kV. The cross-section images of the concave gratings were studied using focused ion beam profiling (FIB). The crosssection profiling was done using a two-step process, the deposition method, followed by the mill for depth. In the deposition method, a constant current of 100 pA was maintained, whereas 500 pA for the mill for the depth step due to their near-perfect gaussian step.

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