Flexible Photodiodes Based on Nitride Core/Shell p-n Junction Nanowires.

A flexible nitride p-n photodiode is demonstrated. The device consists of a composite nanowire/polymer membrane transferred onto a flexible substrate. The active element for light sensing is a vertical array of core/shell p-n junction nanowires containing InGaN/GaN quantum wells grown by MOVPE. Electron/hole generation and transport in core/shell nanowires are modeled within nonequilibrium Green function formalism showing a good agreement with experimental results. Fully flexible transparent contacts based on a silver nanowire network are used for device fabrication, which allows bending the detector to a few millimeter curvature radius without damage. The detector shows a photoresponse at wavelengths shorter than 430 nm with a peak responsivity of 0.096 A/W at 370 nm under zero bias. The operation speed for a 0.3 × 0.3 cm2 detector patch was tested between 4 Hz and 2 kHz. The -3 dB cutoff was found to be ∼35 Hz, which is faster than the operation speed for typical photoconductive detectors and which is compatible with UV monitoring applications.

Introduction

The ultraviolet A (UVA) spectral region (320–400 nm) is a major component of the UV solar radiation, which corresponds to more than 99% of UV light that reaches the earth surface. At high dose, UVA sunlight generates a severe oxidative stress in cells and is harmful for human health; however, moderate exposure to UVA can be favorable, in particular stimulating the vitamin D generation.[1] In this context, the development of wearable UV sensors helping people to balance their sun exposure is today an important societal challenge.[2] In addition to high sensitivity and spectral selectivity, wearable UV sensors need to be lightweight and flexible in order to be easily incorporated in skin patches or integrated on clothes. Autonomous operation without external bias is also an important requirement for this application. Today, there exist a large number of commercially available UV sensors;[3] however, they remain rather bulky and cannot be integrated directly on human clothes or skin.[4] Today, nanomaterials are extensively investigated for UV sensing as a way to enable high sensitivity combined with a possible integration on fabric and self-powering.[5,6]

Wide bandgap nanowire (NW) photodetectors can successfully respond to the above-mentioned specifications for a wearable UV sensor. NW photoconductors based on nitride or oxide materials were reported to exhibit a very high sensitivity thanks to the separation of photogenerated carriers in radial direction.[7−11] NW arrays are also reported to enhance the light absorption by light trapping[12] and antireflection[13] effects, which can lead to a high sensitivity even for a small amount of active material. The use of wide bandgap semiconductors naturally provides spectral selectivity by the bandgap absorption cutoff without the use of additional optical filters. In particular, ternary InGaN alloys can be employed to precisely tune the absorption cutoff to the desired wavelength (for example, in order to adjust the detection to wavelengths of UVA shorter than 400 nm).[14,15] Finally, thanks to their small diameter, NWs can stand large deformations without damage.[16,17] This last property enables their use as an active material for flexible optoelectronic devices.[18,19]

Many realizations of flexible UV detectors have been reported, mainly based on ZnO photoconductors.[20,21] However, photoconductors, despite many efforts,[13] still suffer from long response time constants.[22] The operation speed can be significantly enhanced by changing the operation principle from a photoconductor to a photodiode.[15] In addition, p-n photodiodes can operate under zero bias without any need for external polarization, which is a major advantage for their application as portable UV sensors.

The flexible detector fabrication typically follows one of the two approaches: direct NW growth on plastic substrates[13,20,23] or NW transfer to plastic by different methods such as electro-spinning method,[24] printing,[21,25] dielectrophoretic positioning,[26] etc. The direct growth on plastic severely restricts the choice of the growth techniques (by limiting the growth temperature) and as a consequence the NW materials that can be synthesized. Transfer methods are potentially applicable to any NW material. However, for the majority of transfer methods, the NWs are positioned horizontally on the substrate, which eliminates the benefit of absorption enhancement in a NW ensemble due to light trapping. The initial NW orientation cannot be maintained. Recently, an alternative transfer method yielding vertical NWs has gained a broad interest.[18,19,27,28] It is based on NW embedding in a polymer layer followed by either mechanical peeling of the membrane,[18,27] or under-etching of a sacrificial layer.[19,28] Following this method, a flexible Schottky photodiode based on CVD-grown GaN NWs has been demonstrated.[29] Unfortunately, deep defects in CVD-grown GaN wires yield a strong photoresponse in the visible spectral range, which does not allow for selective detection of UVA. It is desirable to replace the defect-related absorption by the above-bandgap absorption, which can be better controlled in particular by using ternary alloys for tuning the absorption edge. This is the objective of the present study.

In this work we demonstrate for the first time a flexible nitride p-n photodiode. The device employs core/shell p–n junction NWs grown by metalorganic vapor phase epitaxy (MOVPE) on sapphire substrates. InGaN/GaN quantum wells (QWs) are inserted in the active region to extend the detection range to wavelengths longer than the GaN near band edge cutoff. We employ a mechanical peeling transfer method,[18] which allows to maintain the NW orientation. This is of particular importance for the p-n NWs, for which the polarity of the junction should be preserved for all wires contacted in parallel. The NWs were embedded into a polymer layer and the polymer/NW membrane was mechanically peeled-off, contacted and mounted on a piece of a copper tape. A fully flexible transparent contact based on a silver nanowire network was used, which allows bending the detector to a few millimeters curvature radius without damage. Without any external bias, the detector shows a photoresponse at wavelengths shorter than 430 nm with a peak responsivity of 0.096 A/W at 370 nm. The operation speed of a 0.3 × 0.3 cm2 detector patch was tested between 4 Hz and 2 kHz. The −3 dB cutoff was found to be ∼35 Hz, which is faster than the operation speed for typical photoconductive detectors and which is compatible with UV monitoring applications. Angular dependence of the photoresponse was analyzed.

Experimental Section

Nanowire Growth

The core–shell NWs containing GaN/InGaN multiple quantum wells (MQWs) were grown on c-sapphire substrates. An MOVPE growth method has been employed using a closed coupled showerhead reactor. The approach to grow the self-assembled c-GaN wires on sapphire is based on an in situ thin SiN layer deposition acting as a partial mask. The vertical growth was favored by using a high flux of silane (200 nmol/min), a low V/III ratio (50), a high temperature (1040 °C), and a high pressure (800 mbar) as described in detail in ref (30). A first section of about 5 μm is grown with silane injection under trimethylgallium (TMGa) precursor and ammonia (NH3) flux leading to a spontaneous SiN passivation of the wire sidewalls that maintains the wire geometry[31,32] and to a high n+-type doping of the wire core (donor concentration around ∼1020 cm–3 [33]). By stopping the silane flux, a second nonintentionally doped section (1018 cm–3 [34]) is grown to reach a length of about 25 μm. Thirty radial InGaN/GaN QWs are subsequently grown around the top unpassivated wire section corresponding to the nonintentionally doped part at 400 mbar under the injection of triethylgallium (TEGa), trimethylindium (TMIn), and ammonia at 720 °C for the InGaN wells (In-content target about 18%) and 900 °C for the GaN barriers. The nominal thickness is 3 and 10 nm for the wells and barriers, respectively. A final p-GaN thick shell of about 150 nm is grown at 920 °C with a biscyclopentadienyl-magnesium precursor (Cp2Mg) followed by a dopant activation annealing performed at 750 °C for 20 min in N2 atmosphere (acceptor concentration around ∼1016–1017cm–3 [34]).

The schematic of Figure a illustrates the core/shell NW internal structure, while Figure b presents an SEM image of an as-grown NW array illustrating the wire morphology. The average wire height is 25 ± 5 μm, and the diameter is around 1–2 μm. The typical wire density is about 106 cm–2. We note that this growth procedure yields MQWs not only on the m-plane lateral sidewalls, but also on the top c plane, with a different QW thickness and In content.[35]

Figure 1

(a) Schematic of the core–shell MQW NW structure. (b) SEM image of the NW array. (c) Transversal cross-sectional STEM-HAADF images taken along the c-zone axis evidencing the core–shell structure on m-plane hexagonal facets (wire slice prepared by FIB technique). The increasing magnification of TEM images reveals that the shell is composed of 30 × InGaN (7 nm)/GaN (22 nm) MQWs coated with a thick p-GaN layer. (d) μPL of a single NW measured at 4 K.

The internal wire structure was probed by scanning transmission electron microscopy (STEM) in high-angle annular dark-field (HAADF) imaging mode. A thin slice perpendicular to the wire axis was prepared by focused ion beam (FIB) in order to observe the wire cross-section in transversal direction. Figure c shows STEM-HAADF images at three increasing magnifications taken along the c-zone axis on the same wire transversal slice. A clear core–shell structure is observed around wire sidewalls corresponding to m-planes facets evidencing the presence of 30 InGaN/GaN MQWs followed by a p-type GaN shell. The thickness of InGaN wells in dark contract is estimated to be 7 ± 1 nm while the GaN barriers are 22 ± 2 nm. The thickness of the p-GaN surrounding shell is measured to be ∼175 nm. We observe that the external interface of the wells (corresponding to the InGaN/GaN interface) is significantly rougher compared to the internal well interface (corresponding to the GaN/InGaN interface), as generally observed for InGaN/GaN heterostructures, especially with thick wells. The In-content in the well has been estimated close to 15%[36] instead of the targeted 18% due to lower incorporation of In on the m-plane surfaces.[37] In Figure c, we observe that the MQW structure can be disturbed by dislocations and stacking faults originating from the first QW interface that propagate across the whole shell heterostructure. These morphological properties of the core–shell structure are consistent with the previously reported core–shell structures grown under similar growth conditions for photovoltaic applications.[38]

The optical properties of NWs were investigated by low temperature microphotoluminescence (μPL) spectroscopy. In order to avoid excitation of multiple NWs, the NWs were cut from their substrate by an ultrasonic bath and dispersed with a low density on a Si substrate. The μPL characterization was carried out at 4 K by exciting single NWs at 375 nm wavelength by a continuous wave laser diode. The luminescence signal was collected by a HR460 spectrometer and a charge coupled device (CCD) camera. Figure b) shows a typical μPL spectrum of a single NW. The spectrum presents two contributions. The main peak at 3.07 eV (403 nm) is attributed to the radial QWs on the m-plane. A weak shoulder peak observed at a lower energy around 2.8 eV (442 nm) originates from the axial QWs at the top part of the NW as previously reported in ref (35).

Flexible Photodetector Fabrication

The fabrication of flexible photodetectors started by forming an ohmic contact to the p-GaN shell. A photoresist layer was spin-coated on the as-grown NWs to protect the highly n-doped GaN base of the NWs. Then a semitransparent 3 nm/3 nm Ni/Au metallic layer was deposited by e-beam evaporation on the p-GaN shells protruding from the resist. The metal was lifted-off and the sample was annealed at 400 °C in air for 10 min. The presence of this thin Ni/Au layer allows an ohmic contact to form to the p-GaN shells. The NWs were then encapsulated into polydimethylsiloxane polymer (PDMS) with an average thickness of 25 μm. The composite nanowire/PDMS membrane was peeled-off from the sapphire substrate and flipped upside down to metalize the bottom part of the NWs. An ohmic Ti/Al/Ti/Au (10 nm/30 nm/10 nm/200 nm) contact was deposited on the NW highly n-doped bases. Then the device was again flipped upside-down and transferred to a flexible substrate (copper tape). The transparent top contact was deposited by spin-coating on the surface using a suspension of silver nanowires to achieve a flexible network connecting the p-GaN shells. The contact area is approximately 0.3 × 0.3 cm2. The fabrication process is schematically illustrated in Figure a. The SEM image (Figure b) shows a bird’s eye view of the device. The distribution of silver nanowires is uniform. The GaN NWs are well connected by silver nanowires as illustrated in a close-up SEM image of Figure c. The density of protruding GaN NWs connected by the silver nanowires is ∼1.2 × 106 cm–2. Figure d presents a photo of the device and illustrates its flexibility.

Figure 2

(a) Schematic representation of the fabrication steps: encapsulation in PDMS and peel-off of the membrane; deposition of the back metal contact; deposition of the top transparent contact composed of a silver nanowire mesh. (b) Bird’s eye view SEM image of the top surface of the detector. (c) Top view SEM image of an individual nitride NW contacted with silver nanowires. (d) Device photo illustrating its flexibility.

We note that the main challenge in the fabrication process is the realization of a flexible transparent ohmic contact. For the bottom opaque contact, a standard metallization is used, which provides an ohmic contact to the n-GaN NW bases. Metal contacts were shown to well sustain bending deformations for curvature radiuses of several millimeters.[39−43] Therefore, they are commonly used for flexible LEDs and photodetectors.[44,45] However, standard metal contacts are opaque and do not allow for light coupling into the NW photodiode. Following our previous developments, we have chosen a silver nanowire mesh as a top transparent contact.[18] This type of contacts is today intensively studied in the literature as an alternative to ITO, which combines mechanical flexibility with optical transparency and low resistance.[46,47]

Flexible Photodetector Characterization Techniques

The electrical characteristics of the flexible photodetector were investigated using a Janis probe station and a Keithley 2636 source-meter at room temperature. The spectral dependence of the photocurrent was analyzed by illuminating the device with a wavelength tunable xenon lamp light coupled with a Jobin Yvon Triax 180 spectrometer. A calibrated photodiode sensor was used to measure the light source output intensity, which was then used to normalize the photocurrent spectra. The photo responsivity of the detector was calibrated using a continuous wave laser diode emitting at 375 nm wavelength. In addition, electron beam induced current (EBIC) measurements were performed to probe the top contact properties. EBIC maps were collected at room temperature using an acceleration voltage of 20 kV in a Hitachi SU8000 SEM controlled by Gatan Digiscan system as described in ref (48).

Results and Discussion

Electrical Characterizations

A current–voltage (I–V) curve in the dark of the fabricated flexible photodetector is displayed in Figure a. The curve exhibits a typical rectifying behavior with a diode knee voltage around 2.3 V, after which the forward current increases steeply. Under reversed bias, the leakage current is negligible up to −3 V. The diode-like I–V characteristic validates the fabrication procedure, showing that the PDMS membrane provides a good electrical insulation and that there is no pronounced Schottky contact between the p-GaN shell and the silver nanowire contact, which would otherwise block the forward current. The I–V measurements were repeated after 50 cycles of device bending down to a radius of 5 mm. No modification of the I–V curve was observed.

Figure 3

(a) I–V curve of the flexible photodetector. (b) Top-view SEM image of the detector contacted with silver nanowires and (c) the corresponding EBIC map (bright contrast corresponds to the induced current in nitride NWs).

To probe the electrical connection between the silver nanowires and the nitride NWs and to evaluate the long-range conductivity of the silver nanowire top contact, EBIC mapping was performed on the flexible photodetector. Figure b,c presents the SEM image and the EBIC map (under zero bias) of the top surface of the detector. The electron beam arriving at the sample surface creates electron–hole pairs in the core/shell nitride NWs, which are separated by the internal electrical field of the p–n junction and then collected by the electrodes.[48] The induced current appears as a bright contrast in Figure c. The region probed by EBIC in Figure is located 1 mm away from the metal bonding pad, and there is no current spreading grid on top of the silver nanowire network. No significant attenuation of the EBIC signal with distance from the bonding pad is observed, which proves that the carrier transport by the silver nanowires network is efficient. By comparing the SEM and the EBIC maps, we evaluated the number of nitride NWs protruding from PDMS (and thus contacted by silver nanowires) and the number of NWs generating EBIC signal. As an example, the positions of the NWs are highlighted with green hexagons in the upper right part of the maps. The number of NWs generating EBIC signal corresponds to 85% of all protruding NWs. This value gives an estimation of the yield offered by the present silver nanowire contacting technique.

Electro-optical Characterizations

The inset of Figure a shows the I–V curves of the detector in the dark and under illumination with UV light (λ = 370 nm, 1.26 × 10–8 W power) in logarithmic scale. Under zero bias, the photocurrent is negative, the corresponding responsivity is ∼0.1 A/W. Under forward bias, the photocurrent changes sign at ∼0.25 V. This relatively low value may be related to some defects in the active region.

Figure 4

(a) I–V curve in the dark and under illumination with λ = 370 nm. (b) Power dependence of the responsivity. (c) Frequency response of the detector. Inset shows the current trace under zero bias in response to a square light pulse. (d) Angular dependence of the responsivity at different angles from θ = 0° to 45°. Insets show schematics explaining the illumination configuration under normal and oblique incidence.

The power dependence of the responsivity has been probed by illuminating the detector with an Ar2+ ion laser (λ = 244 nm) and varying the incident power density by almost 4 orders of magnitude (from 5 × 10–6 to 7 × 10–3 W/cm2). As shown in Figure b, at moderate excitation power densities (up to ∼3 × 10–3 W/cm2), the photocurrent presents almost a linear dependence with power. The responsivity slightly decreases from 0.084 to 0.057 A/W in this region. At high power densities the photocurrent dependence becomes sublinear and the photocurrent saturates. Correspondingly, the responsivity decreases down to 0.03 A/W at 7 × 10–3 W/cm2. Similar power dependence has been reported in axial junction NW photodiodes.[49] In thin film p–n photodiodes the response is linear up to higher power densities (0.2 W/cm2 [50]). The saturation at high excitation is related to the screening of the built-in field by the photogenerated carriers.

The detector operation speed was analyzed. First, the detector response to a square light pulse (λ = 370 nm) was measured under zero bias. The inset of Figure c shows the temporal current trace. The flexible photodetector presents a fast response with a rise and decay switching time below 0.1 s (which is the time resolution of our measurement system). To further investigate the device operation speed, the frequency dependence of the photocurrent was measured at zero bias using a mechanically chopped illumination and a lock-in detection. Figure c shows the detector frequency response from 4 to 2000 Hz. The −3 dB cutoff frequency of the large area (0.3 × 0.3 cm2) device is ∼35 Hz. This operation speed is higher than the typical values for NW photoconductors[8,11] and is comparable to the one of the axial p-i-n NW photodiodes.[49] We note that it is compatible with the potential application of UV monitoring.

The angular dependence of the photodetector responsivity was investigated. The photocurrent was collected under zero bias for incident angles θ between 0 and 45 deg for a power density of 0.001 W/cm2 (i.e., in the range where the photocurrent shows an almost linear dependence on the incident power as shown in Figure b). As reported in Figure d, the responsivity increases with the incident angle. The responsivity is ∼0.066 A/W for the laser light perpendicular to the sample surface, whereas it increases to 0.18 A/W for a 45 deg tilt. This phenomenon can be explained by the enhanced light harvesting when the sample is rotated. Indeed, the density of nitride NWs obtained by self-catalyzed MOVPE growth is low, they cover only ∼3.8% of the total detector surface. At normal incidence, only a small part of the NW surface (mainly their top surface) absorbs light, as illustrated in the inset of Figure d. The effects of light concentration reported for nanowire solar cells[51] do not play important role for the wire diameters in the 1–2 μm range that are used in this study. When the photodetector is tilted, the light is absorbed not only by the top surface but also by the NW side walls, which increases the portion of the absorbed light and induces a stronger photocurrent for the same power density.

The spectral response of the detector was analyzed. Figure displays the room temperature photocurrent spectra in logarithmic scale under zero bias and under reverse bias of −0.5 V, respectively. The low-energy onset of the photocurrent (i.e., the energy value, for which the photocurrent clearly dominates over the noise) is around 2.88 eV (430 nm). At higher energies, the photocurrent increases by more than 1 order of magnitude, reaching its maximum at 3.35 eV (370 nm) and then slightly decreasing. The bandgap of the radial QWs is estimated to be close to 3.03 eV (410 nm) at room temperature (according to μPL measurements discussed above and accounting for the Varshni shift with temperature). Therefore, the low-energy onset value of the photocurrent is below the bandgap of the radial QWs. This is most likely due to the contribution to the photocurrent of the axial QWs, which have a smaller bandgap value due to both their higher In concentration and to the quantum confined Stark effect induced by the internal electric field along the c-axis. Nevertheless, the overall spectral response matches quite well the UVA spectral domain, which validates the possible application of the developed photodiode for UVA sensing.

Figure 5

Room temperature photocurrent spectrum in logarithmic scale under zero bias and under reverse −0.5 V, respectively.

The peak responsivity of the detector reaches its maximum value of 0.096 A/W at 3.35 eV. We note that the density of the active NWs is quite low in the present device; a higher value can be reached with a higher density NW array. A decrease of the photocurrent for energies above 3.4 eV is observed (the responsivity drops to 0.063 A/W at 4.1 eV). This decrease is typical also for thin film nitride photodiodes and can be attributed to the high absorption coefficient of GaN above the bandgap, which makes the absorption depth small. The carriers are generated close to the surface of the p-doped GaN layer and have a large probability to recombine on surface states before being collected. The same explanation can be applied in the present case for the enhanced light absorption close to the top NW surface. Under a reverse bias of −0.5 V, the spectral shape remains similar whereas the photocurrent signal increases. The peak responsivity becomes 0.157 A/W. In agreement with theoretical modeling presented in the next section, this signal enhancement is attributed to the increase of the electric field in the active region facilitating the carrier extraction from the QWs. For energies above the GaN bandgap the signal increase under reverse bias can also be due to a slight broadening of the space-charge region in the p–n junction.

Modeling of Current Generation in the Active Region

The photocurrent generation in the InGaN/GaN p–n junction QWs has been modeled using the nonequilibrium Green function formalism. This model is described in details in ref.[52] This approach considers the quantum character of electrons, which cannot be neglected in the present device with an active region containing QWs, since in QWs the confinement and the tunneling both control the electronic transport. Moreover, as we show, the electron–phonon scattering has also a key role in the photodetector operation.

The first step of the calculations is the determination of the potential profile in the entire active region in the dark. Figure a displays the conduction band and valence band profiles for 30 InGaN/GaN MQWs with a nominal composition of 18%. The calculations are performed for zero external bias and for −0.5 V reverse bias. Zero of energy corresponds to the Fermi level in the p-GaN layer. The built-in field of the p–n junction is distributed homogeneously over the 29 left QWs which are empty in the dark. Only the first QW next to the n-GaN layer is filled with electrons. The electron density in this well is 1018 and 1.3 × 1018 cm–3 at −0.5 and 0 V, respectively.

Figure 6

(a) Conduction band and valence band potential profiles in the entire active region for external biases of 0 V and −0.5 V. Photocurrent spectra in conduction band versus position in a representative QW under illumination for (b) V = 0 V, illumination at 3 eV; (c) V = 0 V, illumination at 3.3 eV, and (d) V = −0.5 V, illumination at 3.3 eV. For the current the color legend is in Arb. Unit. In panels (b–d) arrows schematically represente physical phenomena involved in the extraction of carriers. The vertical arrows represent the interband interactions (photon absorption/emission). The curved arrows represent the intraband interactions (phonon absorption/emission). The horizontal arrows represent the diffusion process.

Due to the computational burden of the method, the photocurrent is only calculated for one representative QW. The procedure is described in ref (53). To illustrate the evolution of the photocurrent with the photon energy, Figure b,c presents the current maps as a function of position and energy in the conduction band of a QW in an unbiased p–n junction under illumination with 3 eV (i.e., deep in the InGaN QW) and with 3.3 eV (i.e., slightly below the barrier bandgap), respectively. The calculation predicts that the responsivity increases by a factor of 21 when the photon energy increases. This prediction is in qualitative agreement with the experimental results shown above, for which the photocurrent signal increases ∼9 times between 3 and 3.3 eV (cf. Figure ). As shown by the quantification of the spectrum presented in Figure a, when electrons are generated deep in the well, the scattering with phonons controls the extraction (the optical phonon energy is 92 meV). The interaction with phonons is schematically illustrated with curved arrows in Figure b–d. In this case, a large number of phonon-absorption is necessary to extract the photogenerated carriers from the QW. This current is then expected to increase with temperature. In contrast, as shown in Figure c for excitation with a photon energy of 3.3 eV, the electrons can be withdrawn from the well without a large number of phonon scatterings. The electronic extraction is then easier and the current is larger.

To simulate the photodetector response under bias, the photocurrent for a reverse bias of −0.5 V has been calculated. Figure a shows the increase of the electric field in the whole active region with the reverse bias. The corresponding current spectrum for the incident photon energy of 3.3 eV is shown in Figure d. The model predicts the current increase by a factor of 1.34 under bias, which is in qualitative agreement with the experimental value of 1.61 (cf. Figure )). By comparing the spectra presented in Figure c,d, it is clear that the reverse bias facilitates the direct extraction of photogenerated carrier without any scattering with phonons. However, the tunneling is not observed since the barriers are too thick.

Conclusions

In conclusion, a flexible nitride p–n photodiode has been demonstrated using core/shell p–n junction NWs containing InGaN/GaN QWs. The polymer membranes embedding active nitride NWs were contacted using a transparent silver nanowire mesh to ensure high mechanical flexibility. Without any external bias, the detector shows a photoresponse at wavelengths shorter than 430 nm with a peak responsivity of 0.096 A/W at 370 nm. The −3 dB cutoff frequency for a 0.3 × 0.3 cm2 detector patch was found to be ∼35 Hz, which is compatible with UV monitoring applications. The detector photoresponse increases for an oblique incidence angle due to an increased light absorption.