Boosting Wavelength‐Selective Absorption and Photocarrier Collection in NiO/ZnO Transparent Photovoltaic Heterojunctions by Plasmonic Ag Nanowire Top Electrodes

Global climate change has compelled many to rely on photovoltaic (PV) technology to meet the highly demanding energy needs of urban areas. In particular, transparent photovoltaic (TPV) devices can be utilized for building windows, not only supplying electric energy but also improving the overall thermal efficiency of a building. In this work, NiO/ZnO wide‐bandgap oxide TPV heterojunctions are fabricated with Ag nanowire (NW) top electrodes and their PV characteristics are investigated. Special attention is paid to the contributions of surface plasmon (SP) excitation in AgNWs to the PV performance of the TPV device. Light polarized perpendicular to the AgNW axis induces a localized SP resonance in AgNWs at a wavelength of 400 nm, as shown by optical measurements and calculations. The investigation on how the plasmonic AgNWs affect real‐space electric potential distributions and local current‐voltage characteristics of the TPV devices uses Kelvin probe force microscopy and current‐sensing atomic force microscopy, respectively. The spatial redistribution and transport of photogenerated charge carriers strongly depend on the polarization as well as the wavelength of incident light. The results demonstrate that the AgNW‐based top electrodes boost the wavelength‐selective absorption and the effective collection of photocarriers in TPV devices.


Introduction
The challenge of global climate change has driven us to attempt intensive research efforts to develop eco-friendly, sustainable energy sources in highly populated urban areas.Photovoltaic (PV) technologies are the most widely used renewable energy sources, and the integration of PV modules in buildings seems to be inevitable to meet the rapidly increasing energy needs in cities.According to an early study, windows contribute to 32% of the total energy load for heating, ventilation, and air conditioning in residential and commercial buildings in the US, which causes significant energy supply issues in cities. [1] Typically, conventional PV modules are opaque because PV devices are designed and fabricated to absorb sunlight as much as possible.Therefore, the installation of these conventional PV modules will be possible only on a small portion of buildings (like rooftops).4][5][6][7] Windows in buildings serve several purposes: improving thermal insulation, controlling indoor solar heat gain, utilizing daylight, and addressing aesthetic issues. [2,3][4][5][6][7] In addition, TPVs can be used specifically in self-powered low-power mobile electronics and smart windows, which require only 2-5% power conversion efficiency (PCE). [4]he electric energy in a TPV cell, which selectively transmits visible (VIS) light, is acquired through the absorption of near-infrared (NIR) and/or ultraviolet (UV) light.][11][12][13] The large bandgap energy of WOS allows for the selective absorption of UV light from the sun.The materials used in WOS-based TPV devices are affordable, abundant on Earth, and extremely stable.[11][12][13] The PCE will undoubtedly decrease if some sunlight is permitted to pass through the TPVs.The maximum achievable efficiency of a UV-selective TPV cell was restricted below 7%. [5][11][12][13] Clearly, it is imperative to investigate the critical factors to limit carrier generation and/or collection in TPVs.
[6][7] TCEs must be effective charge collectors to reach maximum power output, in addition to being a limiting factor for device transparency.In order to maintain visual clarity, TPVs require both the front and back electrodes to be transparent.26] Top electrodes composed of AgNWs are frequently employed in thin film solar cells due to their notable mechanical flexibility as well as their excellent optical and electrical properties. [14,15][21][22] Through time-resolved photoluminescence spectroscopy, Zamkoye et al. observed a shorter exciton lifetime in the presence of AgNWs compared to the reference ITO/ZnO substrate. [23]This suggests that a strong electric field near Ag-NWs can lead to rapid charge separation.However, just a few nanoscopic electrical investigations have shown how AgNWs contribute to carrier generation and carrier collection in adjacent semiconductor layers.Because AgNWs have a diameter of several tens of nm and SP-mediated light confinement occurs in their vicinity, in-depth research requires nanometrology with high spatial resolution.][29][30] In our earlier work, we investigated lightinduced surface potential changes in TPV devices, which could be directly related to the output voltage at the local region, using Kelvin probe force microscopy (KPFM). [11]Singh et al. reported the tunneling-mediated charge transfer process from Ag-NWs to ZnO thin films in their ZnO/Al 2 O 3 /AgNW heterostructures, using KPFM and current-sensing atomic force microscopy (C-AFM). [27]Therefore, systematic and complementary studies employing KPFM and C-AFM can help us clarify how the plasmonic AgNWs can influence carrier generation and collection in the TPV devices under light illumination.
In this work, we fabricated NiO/ZnO-heterojunction UVselective TPV devices with AgNW-based top electrodes and investigated the plasmonic benefits of the top electrodes.The most popular WOS materials, NiO and ZnO, were chosen for their guaranteed long-term stability and well-established scalable fabrication processes.To evaluate the PV performance of our TPV device, we carried out macroscopic characterizations, such as optical transmittance and electrical transport measurements.Light-induced surface potential modifications and local photocurrent characteristics, especially their wavelength-and polarizationdependence, were studied using KPFM and C-AFM, respectively.These nanoscopic investigations allow us to reveal the contributions of the plasmonic AgNWs in our TPV device.

Results and Discussion
Figure 1a shows the cross-sectional schematic diagram of our complete TPV device: a NiO/ZnO heterojunction with the ZnOcoated AgNW (Z-AgNW) top electrode.In order to set up the TPV capabilities, the cell is established by forming n-type ZnO (100 nm) and p-type NiO (30 nm) thin films.11][12] High optical transparency and electrical conductivity are required for all electrodes utilized in the fabrication of the TPV device.To collect hole carriers, the bottom electrode is made of fluorinedoped tin oxide (FTO) on a glass substrate.A TiO 2 thin film works as the back surface layer to support the performance of the carrier collection.AgNW networks with a diameter of 20-40 nm are utilized as the top electrode for carrier collection (Figure S1, Supporting Information).Our earlier work shows that coating of a thin ZnO layer (≈12 nm) on AgNW can protect AgNWs from thermal and electrical stress. [13]The fabrication of the TPV device, particularly the conformal coating of the AgNWs with ZnO thin films, was confirmed by a high-resolution transmission electron microscopy image in Figure 1b and the elemental mapping (Figure S2, Supporting Information).The detailed fabrication processes can be found in the Experimental Section and our earlier publication. [11,12]For comparison, two kinds of top electrodes, bare AgNW networks and ZnO-coated AgNW networks (referred to as Z-AgNW, hereafter), were prepared on the NiO/ZnO heterojunctions.Figure 1c shows the optical transmittance spectra of our complete TPV device with the Z-AgNW top electrode and a NiO/ZnO heterojunction with a bare AgNW top electrode.Our devices allow the transmission of visible light, as expected from the wide bandgap nature of the constituent materials, including NiO, ZnO, TiO 2 , and glass substrates.The devices with bare AgNW and Z-AgNW electrodes have average visible transmittances of 62% and 59%, respectively (Figure S3, Supporting Information).6][7][8] The NiO/ZnO heterojunctions exhibit rectifying diode behaviors in the measured current density versus voltage (J-V) characteristics, as shown in Figure 1d.This indicates the formation of pn-junctions, as desired.The light illumination can cause notable down-shifts of the J-V plots in the fourth quadrant, confirming the PV power generation of our devices.Figure 1e shows the output electric power density versus voltage characteristic of the devices.The maximum output power of the complete TPV device with the Z-AgNW electrode was as large as 115 μW cm −2 .In comparison, the NiO/ZnO heterojunction with the bare AgNW top electrode had an output power of 105 μW cm −2 .The light source consisted of three light-emitting diodes (LEDs) with wavelengths of 365, 400, and 520 nm, where significant PV effects are expected from our earlier report. [11,12]The heterojunction with the bare AgNW top electrode exhibited the open-circuit voltage (V OC ) of 528.1 mV, short-circuit current density (J SC ) of 0.466 mA cm −2 , fill factor (FF) of 42.53%, and PCE of 1.04%.The measured PCE of our complete TPV device with the Z-AgNW electrode was as large as 1.15%.The complete TPV device exhibits improved V OC (547.9 mV) and J SC (0.485 mA cm −2 ) compared to the device with AgNW.This enhancement could be attributed to several factors, including the lower electrical resistance of the top electrode, reduced dark current (Figure 1d), and less recombination of holes at the top surface. [13]Table 1 summarizes the PV performance parameters of the two kinds of devices.
As a comparison, the PV performance of our complete device was studied under AM 1.5G illumination (Figure S4a, Supporting Information).The V OC , J SC , FF, and PCE were estimated to be 0.497 V, 0.501 mA cm −2 , 42.5%, and 0.106%, respectively.The external quantum efficiency and integrated current data confirmed the UV-selective PV characteristics of our TPV device (Figure S4b, Supporting Information).The photon balance check was performed and the results could confirm the validity of independent spectral measurements of our TPV device (Figure S4c, Supporting Information). [31]he relative transmittance spectrum of Z-AgNW exhibits two distinct dips at 350 and 400 nm.The finite-difference timedomain (FDTD) calculations can reproduce the measured transmittance spectra, as shown in Figure S5a,b (Supporting Information).According to the calculation, two dips were observed, similar to the measured spectrum in Figure 2a, when the linear Table 1.PV performance of the NiO/ZnO heterojunctions with AgNW and Z-AgNW electrodes under the illumination of light.The light source consists of a set of three light-emitting diodes (LEDs) with a total optical power density of 10 mW cm −2 : 365-nm LED with a power density of 5 mW cm −2 , a 400-nm LED with a power density of 3 mW cm −2 , and a 520-nm LED with a power density of 2 mW cm −2 .V OC and J SC represent the open-circuit voltage and the short-circuit current density, respectively.polarization of light was perpendicular to the AgNW axis (Figure S5a, Supporting Information).Whereas one dip near 350 nm only appeared in the spectrum under parallel light polarization (Figure S5b, Supporting Information).The perpendicular light polarization can cause localized surface plasmon (LSP) excitation in AgNWs. [22,23]Therefore, the 350-nm dip, independent of the light polarization, can be attributed to near-bandgap absorption in the ZnO layer.The 400-nm dip, which appears only under perpendicular light polarization, can be originated from LSP resonance in the AgNWs.The relative transmittance spectrum of our complete TPV device (Figure 2b) shows that the Z-AgNW top electrode can significantly reduce the light intensity transmitted through the TPV device at wavelengths shorter than 550 nm.This suggests that Z-AgNW can enhance the light absorption in the TPV device in a broad wavelength range.The transmittance reduction can also point to increased short-wavelength absorption in Z-coated AgNW on top of the NiO/ZnO heterojunction.
Figure 3 shows the calculated cross-sectional electric field intensity of our TPV device (NiO/ZnO w/Z-AgNW) under the illumination of linearly polarized light with wavelengths of 350, 405, 450, 520, and 550 nm.For the sake of simplicity, a periodic array of AgNWs with a diameter of 20 nm and a period of 1 μm was considered for the FDTD calculations (Figure S5c, Supporting Information).When the device is illuminated by light polarized perpendicular to the AgNW axis, a strongly concentrated electric field appears around the AgNW.In particular, the polarization-dependent field confinement is prominent under light with a wavelength of 405 nm.This suggests that the LSP resonance occurs at a wavelength of ≈400 nm.The LSP resonance at ≈400 nm is further supported by the local minima of the measured (Figure 2a) and calculated (Figure S5a, Supporting Information) transmittance spectra.The calculated distributions show that the polarization-dependent field enhancement near the AgNWs also can appear at 450, 520, and 550 nm, which are somewhat far from the LSP resonance wavelength.When the wavelength is >405 nm, the electric field intensity around the AgNW is less intense and less confined.The electric field intensity for parallel polarization is noticeably lower than that for perpendicular polarization.The AgNW-induced strong electric field intensity under perpendicular polarization extends into the surrounding ZnO layer and the NiO/ZnO heterojunction.This suggests that the polarization as well as the wavelength of incident light can affect the photo-generation of charge carriers near AgNWs.]/e (e: electron charge). [29,30]In order to investigate the roles of the AgNWs, two regions (1 and 2) are chosen in the CPD D map of Figure 4b: 1 and 2 correspond to the top surface regions without and with underlying AgNWs, respectively.The measured CPD D at Region 2 is larger than that at Region 1.The function (WF) of Ag is smaller than the WF of ZnO, and hence AgNW lowers WF measured at the ZnO layer on AgNW, as illustrated in Figure 5a,b. [32]As a result, the underlying AgNWs increase the measured CPD D , as shown in Figure 4b.
The illumination of 405-nm light increases the CPD measured at the whole surface (Figure 4c).Using a linear polarizer, the polarization of incident light was adjusted to be aligned along the vertical or horizontal direction of the CPD map.The light-induced CPD change is called surface photovoltage (SPV): SPV≡CPD L − CPD D .Positive (negative) SPV signals indicate positive (negative) surface charging. [11,29,30]The SPV signals measured at Region 1 and Region 2 clearly show polarization dependence: the SPV under vertical polarization is larger than that under horizontal polarization (Figure 4d).As discussed above, LSP can be excited when the light polarization is perpendicular to the AgNW axis.Thus, light with vertical polarization can cause the LSP excitation at the AgNWs oriented in a nearly vertical direction (Figure 4a-c).The generation of electron-hole pairs can be boosted by the LSP-induced enhanced electric field intensity (i.e., increased number of photons), raising the measured SPV.The SPV maps obtained using 450 and 520 nm light sources also showed polarization dependence, even though the SPV magnitude becomes smaller at longer wavelengths (Figure S6a,b, Supporting Information).In Figure 3, it can be noted that the light vertically polarized to the AgNW axis with the wavelengths of 405, 450, and 520 nm can increase the electric field intensity in the NiO/ZnO heterojunction (Figure 3).The photovoltaic effects in the NiO/ZnO heterojunction separate the photo-generated electron-hole pairs and result in light-induced surface charging.The SPV results are consistent with this expectation: Plasmonic AgNWs can help carrier generation over a wide range of wavelengths (not just close to the LSP resonance wavelength).
Figure 5a-d shows the suggested band diagrams at the regions without and with the underlying AgNW in our TPV device.We can explain the AgNW-induced CPD contrast (Figure 4b) using the dark state band diagrams in Figure 5a,b.The band diagrams for the heterojunctions without and with AgNW when illuminated by light are shown in Figure 5c,d, respectively.For the sake of clarity, the ZnO layer atop AgNWs will be referred to as ZnO(t) and the ZnO layer in the NiO/ZnO heterojunction as ZnO(n).In the NiO/ZnO heterojunction without AgNW, most of the incident photons can be absorbed in the NiO/ZnO rather than the thin ZnO(t) layer in the Z-AgNW top electrode.In the NiO/ZnO heterojunction, the photons produce electronhole pairs, and the built-in potential at the heterointerface separates these electron-hole pairs.As illustrated in Figure 5c, the photo-generated holes and electrons in the pn-junction will move toward NiO and ZnO(n), respectively, resulting in positive charging at the surface.This can explain the positive SPV data in Figure 4b-d.Figure 3 suggests that in the surface region on underlying AgNWs, ZnO(t) and AgNWs in Z-AgNW (rather than the underlying pn-junction) can absorb a significant amount of the incident short-wavelength light.The photo-generated electrons can be transferred from ZnO(t) to AgNW, according to the expected interfacial band alignment (Figure 5b).As a result, the remaining holes at ZnO(t) lead to the positive SPV signal (Figure 5d).The LSP excited by a vertically polarized light can increase the amount of photo-generated charges near AgNWs, leading to large SPV signals, as shown in Figure 4b-d The polarization-dependent SPV signal at Region 2, which is directly on AgNW, provides convincing evidence of the LSP contribution to the carrier generation in our device (Figure 4a-d).
The LSP excitation considerably increases the amount of photogenerated electron-hole pairs and the resulting SPV magnitude by drastically concentrating the electric field near AgNW.It should be noted that the SPV signals at Region 1 also depend on the light polarization (Figure 4a-d).While not directly on AgNWs, Region 1 is near vertically aligned AgNWs.According to the FDTD calculations (Figure 3), the polarizationdependent enhanced electric field appears within a few tens of nm range near AgNWs.Thus, the AgNW-enhanced electric field cannot directly increase the number of electron-hole pairs in Region 1.Under vertically polarized light, the surface region on the underlying AgNWs can be positively charged.Such a region can attract electrons from the neighboring regions and lead to positive charging.Therefore, the polarization-dependent SPV signals at both Region 1 and Region 2 can indicate the carrier collection capability of the AgNWs in the Z-AgNW top electrode.
Figure 6 shows the dark and light current-voltage (I-V) characteristics of our complete TPV device (NiO/ZnO with Z-AgNW), measured using C-AFM.All the C-AFM measurements were carried out at an identical selected region on one specific AgNW.The polarization of incident light was adjusted to be parallel and per-pendicular to the AgNW axis.The local I-V curves were obtained using light sources with wavelengths of 405, 450, and 520 nm, like the CPD maps (Figure 4c; Figure S6a,b, Supporting Information).The photocurrent measured under vertically polarized light is much larger than that under horizontally polarized light.In contrast, the photocurrent measured at the region without underlying AgNWs did not exhibit a distinct polarization dependence (Figure S7, Supporting Information).It should be noted that the photocurrent measured at the region without AgNW was several times smaller than that at the region on AgNW.The transmittance spectra (Figure 2a,b) and the electric field distributions (Figure 3) of our TPV device show that the LSP excitation in Ag-NWs improves the optical absorption.The light-induced CPD maps (Figure 4a-d) provide convincing evidence of the enhanced photo-carrier generation.The polarization-dependence and the increase of the measured photocurrent near AgNWs (Figure 6;  In the SPV measurements, the SPV signal measured under the illumination of 405-nm light (Figure 4d), whose wavelength was close to the LSP resonance wavelength of the AgNWs, was larger than those of 450-and 520-nm light (Figure S6a,b, Supporting Information).In contrast, the photocurrent measured at 520-nm light was larger than those at 405-and 450-nm light (Figure 6).When the wavelength of incident light is close to the LSP resonance wavelength of AgNW, the strongly concentrated electric field appears in the Z-AgNW top electrode (Figure 3).Consequently, the most of incoming photons are absorbed in AgNW rather than the underlying NiO/ZnO heterojunction.The lightinduced charge transfer from ZnO(t) to AgNW results in surface charging, as illustrated in Figure 5d.It should be noted that such interfacial charge transfer does not cause current flow from the top electrode to the bottom electrode of our device.Figure 3 suggests that incident light with wavelengths much longer than 400 nm can be mainly absorbed in the NiO/ZnO pn-junction and the photocurrent is generated by the photovoltaic effects in the heterojunction, as illustrated in Figure 5c.The I-V curves in Figure 6 show the photocurrent in a closed circuit of our device, consisting of the Z-AgNW top electrode, the NiO/ZnO pnjunction, and the FTO bottom electrode (Figure 1a).
According to Figure 3, the electric field intensity distribution exhibits clear distinction depending on the polarization direction of incident light at wavelengths not only close to the LSP resonance of 405 nm but also 520 nm.As discussed above, the photovoltaic effects in the NiO/ZnO heterojunction can be mainly responsible for the photocurrent measured by C-AFM.Near the LSP resonance conditions, AgNWs, rather than the underlying NiO/ZnO heterojunction, can absorb the majority of incoming photons.The resulting photon-to-carrier conversion leads to light-induced surface charging (Figure 4c,d) instead of photocurrent generation.Even though 520-nm light is far from the LSP resonance, polarization-dependent absorption is expected from the electric field intensity distributions in Figure 3.This readily explains why the photocurrent measured at 520-nm light exhibited a significant polarization dependence (Figure 6).These results show that the AgNWs can contribute to enhanced absorption and photocarrier generation in the short-wavelength visible range as well as the UV near LSP resonance region.

Conclusion
This work successfully demonstrates the beneficial role of plasmonic AgNW top electrodes in the enhanced generation and collection of photocarriers in wide-bandgap oxide-based TPV devices.NiO/ZnO heterojunctions with Z-AgNW top electrodes enable UV-selective absorption and a high average visible transmittance of ≈60%.The PCE value of our TPV device was estimated to be 1.15% under the illumination of UV and green light sources.We could investigate the SPV and photocurrent characteristics using KPFM and C-AFM, respectively.In order to clarify the plasmonic contributions of the AgNWs, we paid careful attention to the wavelength-and polarization-dependence of the measured SPV and photocurrent.Both the SPV and photocurrent increased, when the light polarization was perpendicular to the AgNW axis.Optical measurements and calculations showed that the LSP resonance in the AgNWs occurred at a wavelength of ≈400 nm.The largest SPV magnitude (≈100 mV) was observed under 405-nm light illumination: the light-induced charge transfer at the ZnO/AgNW interface led to the surface charging.All of the photocurrent data, obtained at 405-, 450-, and 520-nm light illumination, exhibited notable polarization dependence.These results strongly suggested that the AgNWs enhanced the wavelength-selective absorption and collection of the photocarriers.AgNW-based transparent top electrodes can be used not only for PV devices but also for many kinds of devices.Therefore, our work can provide a novel approach to exploring high-performance optoelectronic and energy devices with AgNW electrodes.

Experimental Section
Device Fabrications: Magnetron sputtering deposition system (4-inch) was used to fabricate the NiO/ZnO-heterojunction TPV devices on fluorine-doped tin oxide (FTO)/glass substrates.Prior to the deposition, substrates were cleaned using ultrasonication in the sequential bath of acetone, methanol, and distilled water.After the cleaning, the substrates were dried by blowing N 2 gas.ZnO and NiO thin films were deposited using 4-inch ZnO and Ni targets (purity: 99.99%, iTasco), respectively.For the growth of ZnO, an RF power of 300 W was applied to the target at a working pressure of 5 mTorr and 50 sccm of flowing Ar.A DC power of 50 W was applied to the target for the deposition of NiO while flowing 20/5 sccm of Ar/O 2 gas at a working pressure of 3 mTorr.The growth rates for the ZnO and NiO thin films were 6.7 and 2 nm min −1 , respectively.After the sputtering processes, 90 μl of AgNW ink in isopropanol solvent (NanoInk) was spin-coated at 2500 rotations per minute for 60 s. on the NiO/ZnO heterojunction.Following the spin coating, rapid thermal annealing treatment of the AgNWs was carried out at 110 °C for 2 min at a pressure of 5 × 10 −3 Torr.Finally, the top ZnO layer was deposited on the device using the sputtering system at an RF power of 50 W.All the thin films were deposited at room temperature.
Optical Measurements: The transmittance spectra were obtained on a Hitachi spectrophotometer U-3900 equipped with a 60 mm BaSO 4 -coated integrating sphere due to the thinness of the samples.The transmission was measured by scanning from 300 to 800 nm, under illumination from a tungsten lamp from 340 to 800 nm and a D2 lamp for wavelengths below 340 nm.The transmittance spectra of the bare glass substrate, the ZnOcoated AgNWs on the glass, the NiO/ZnO-heterojunction without the top electrode, and the NiO/ZnO-heterojunction with the Z−AgNW top electrode were measured.
Finite-Difference Time-Domain (FDTD) Calculations: The optical transmittance spectra of the samples were simulated using a commercial FDTD program (Lumerical FDTD Solutions, Ansys).The optical constants of glass, [33] Ag, [34] ZnO, [34] and NiO [35] were obtained from the literature.A cross-section of the calculation structure is shown in Figure S5c,f (Supporting Information).At the FDTD simulation box boundaries, perfectly matched layer boundary conditions were employed in the out-of-plane direction, while Bloch (periodic) boundary conditions were used in the inplane directions.A plane wave source was placed 0.9 μm from the NiO surface and the simulation was performed over the wavelengths from 300 to 800 nm.
Current-Sensing Atomic Force Microscope (C-AFM) and Kelvin Probe Force Microscopy (KPFM) Measurements: Atomic force microscopy (AFM) systems (XE100 and NX10, Park Systems) in glove boxes were employed for the nanoscopic electrical characterizations.Current-sensing AFM (C-AFM) measurements were carried out with wear-resistant conductive diamond probes (AD-2.8-AS,Adama Innovations) with a radius of curvature of 10 nm, a spring constant of 2.8 N m −1 , and a resonant frequency of 75 kHz.Kelvin probe force microscopy (KPFM) technique using the AFM systems enabled to measurement of contact potential difference (CPD) at the sample surface.In the amplitude-modulation KPFM mode, an AC voltage with the amplitude and the frequency of 2 V and 17 kHz, respectively, was applied to the Pt/Ir-coated Si probes (NSG01Pt, NT-MDT).The work function of the probe was calibrated against a highly ordered pyrolytic graphite (HOPG, SPI Suppliers) reference sample in order to determine the work function of a sample.CPD data were acquired after 10 min of UV light (wavelength: 375 nm) exposure with a power density of 8.1 mW cm −2 on the sample surface to remove surface gas adsorbates.Before starting the KPFM and C-AFM measurements, the samples were kept in the dark for at least an hour after the UV-exposure.All the measurements were carried out in glove boxes purged with dry N 2 gas to avoid artifacts caused by ambient gas adsorption.

Figure 1 .
Figure 1.a) Cross-sectional schematic diagram and b) transmission electron microscopy images of our complete TPV device: the NiO/ZnO heterojunction with the ZnO-coated AgNW (Z-AgNW) top electrode.c) Optical transmittance spectra (light blue and violet curves) and the photopic response of the human eye (green curve), d) current density versus voltage, and e) output power density versus voltage plots of the complete TPV device (violet curve) and NiO/ZnO heterojunctions with bare AgNW (light blue curve) top electrodes.

Figure
Figure 2a,b shows the relative transmittance spectra of the Z-AgNW top electrode on a glass substrate and the complete TPV device, respectively.The former is the transmitted light intensity (I T ) of a glass with the Z-AgNW electrode (I T [Z-AgNW/Glass]) divided by I T of a bare glass substrate (I T [Glass]).The latter is the ratio of I T of the complete TPV device (I T [NiO/ZnO w/Z-AgNW]) and I T of a NiO/ZnO-heterojunction device without any top electrode (NiO/ZnO): IT[NiO/ZnO w/ Z-AgNW]/IT[NiO/ZnO].The relative transmittance spectrum of Z-AgNW exhibits two distinct dips at 350 and 400 nm.The finite-difference timedomain (FDTD) calculations can reproduce the measured transmittance spectra, as shown in FigureS5a,b (Supporting Information).According to the calculation, two dips were observed, similar to the measured spectrum in Figure2a, when the linear

Figure 2 .
Figure 2. Relative transmittance spectra of a) the ZnO-coated AgNW electrode on a glass substrate and b) the complete TPV device (NiO/ZnO w/Z-AgNW).The former and the latter were calculated against a baseline of the bare glass substrate and a NiO/ZnO-TPV device without the top electrode (NiO/ZnO), respectively.

Figure 3 .
Figure 3. FDTD-calculated electric field (E) intensity distributions of the complete TPV device (NiO/ZnO w/Z-AgNW) under the illumination of light polarized perpendicular (top row) and parallel (bottom row) to the AgNW axis.Wavelengths of 350, 405, 450, 520, and 550 nm were used at a normal incidence angle.E 0 indicates the electric field magnitude of the incident light.
. All of the discussions above indicate the interfacial band alignment at the ZnO/Ag interface of the Z-AgNW electrode as well as the photovoltaic effects in the NiO/ZnO heterojunction can contribute to the measured SPV data of our device.

Figure 4 .
Figure 4. a) An AFM image and b,c) CPD maps of our complete device in the dark and under linearly polarized light illumination with a wavelength of 405 nm.The directions of the electric field of incident light are denoted by the black arrows in the CPD maps under light illumination.The two orthogonal directions for the left and right maps in (c) will be referred to as vertical and horizontal, respectively.d) SPV per power density of incident light for Region 1 and Region 2: Each region is indicated in the dark-state CPD map in (b).Region 1 represents the ZnO layers without an underlying AgNW.In contrast, Region 2 indicates the ZnO layer surface directly on AgNW.

Figure 5 .
Figure 5. Schematic band diagrams at the regions a,c) without and b,d) with AgNW in our TPV device.The ZnO(n) layer in the pn-junction (100 nm) is much thicker than the NiO(p) layer in the pn-junction (30 nm) and the ZnO(t) layer in the top electrode (12 nm).Thus, the potential gradient in the NiO(p) and ZnO(t) layers are ignored.Compared with the dark states in (a) and (b), light illumination can cause the redistribution of photo-generated charges (open circle: electron, filled circle: hole) depending on the built-in potential in the pn-junction and the band offsets at the interfaces.The SPV signals are indicated as red arrows at the top surfaces in (c) and (d).

Figure 6 .
Figure 6.Current-voltage (I--V) characteristics of our complete TPV device (NiO/ZnO w/Z-AgNW) under 405-, 450-, and 520-nm light illumination.The dark-state I-V curves are also measured for comparison.The measurements were done on a region directly on one specific AgNW, as shown in the AFM image in the first I-V plot.The linear polarization of the incident light was adjusted to be perpendicular and parallel to the AgNW axis.

Figure S7 ,
Figure S7, Supporting Information) demonstrate the efficient carrier collection capability of our TPV device.In the SPV measurements, the SPV signal measured under the illumination of 405-nm light (Figure4d), whose wavelength was close to the LSP resonance wavelength of the AgNWs, was larger than those of 450-and 520-nm light (FigureS6a,b, Supporting Information).In contrast, the photocurrent measured at 520-nm light was larger than those at 405-and 450-nm light (Figure6).When the wavelength of incident light is close to the LSP resonance wavelength of AgNW, the strongly concentrated electric field appears in the Z-AgNW top electrode (Figure3).Consequently, the most of incoming photons are absorbed in AgNW rather than the underlying NiO/ZnO heterojunction.The lightinduced charge transfer from ZnO(t) to AgNW results in surface charging, as illustrated in Figure5d.It should be noted that such interfacial charge transfer does not cause current flow from the top electrode to the bottom electrode of our device.Figure3suggests that incident light with wavelengths much longer than 400 nm can be mainly absorbed in the NiO/ZnO pn-junction and the photocurrent is generated by the photovoltaic effects in the heterojunction, as illustrated in Figure5c.The I-V curves in Figure6show the photocurrent in a closed circuit of our device, consisting of the Z-AgNW top electrode, the NiO/ZnO pnjunction, and the FTO bottom electrode (Figure1a).According to Figure3, the electric field intensity distribution exhibits clear distinction depending on the polarization direction of incident light at wavelengths not only close to the LSP resonance of 405 nm but also 520 nm.As discussed above, the photovoltaic effects in the NiO/ZnO heterojunction can be mainly responsible for the photocurrent measured by C-AFM.Near the LSP resonance conditions, AgNWs, rather than the underlying NiO/ZnO heterojunction, can absorb the majority of incoming photons.The resulting photon-to-carrier conversion leads to light-induced surface charging (Figure4c,d) instead of photocurrent generation.Even though 520-nm light is far from the LSP resonance, polarization-dependent absorption is expected from the electric field intensity distributions in Figure3.This readily explains why the photocurrent measured at 520-nm light exhibited a significant polarization dependence (Figure6).These results show that the AgNWs can contribute to enhanced absorption and photocarrier generation in the short-wavelength visible range as well as the UV near LSP resonance region.