Interfacial Electrostatic‐Interaction‐Enhanced Photomultiplication for Ultrahigh External Quantum Efficiency of Organic Photodiodes

A photomultiplication‐type organic photodiode (PM‐OPD), where an electric double layer (EDL) is strategically embedded, is demonstrated, with an exceptionally high external quantum efficiency (EQE) of 2 210 000%, responsivity of 11 200 A W−1, specific detectivity of 2.11 × 1014 Jones, and gain–bandwidth product of 1.92 × 107 Hz, as well as high reproducibility. A polymer electrolyte, poly(9,9‐bis(3′‐(N,N‐dimethyl)‐N‐ethylammoinium‐propyl‐2,7‐fluorene)‐alt‐2,7‐(9,9‐dioctylfluorene))dibromide is employed as a work‐function‐modifying layer of indium tin oxide (ITO) to construct an EDL‐embedded Schottky junction with p‐type polymer semiconductor, poly(3‐hexylthiophene‐diyl), resulting in not only advantageous tuning of the work function of ITO but also an enhancement of the electron‐trapping efficiency due to electrostatic interaction between exposed cations and trapped electrons within isolated acceptor domains. The effects of the EDL on the energetics of the trapped electron states and thus on the gain generation mechanism are confirmed by numerical simulations based on the drift–diffusion approximation of charge carriers. The feasibility of the fabricated high‐EQE PM‐OPD especially for weak light detection is demonstrated via a pixelated prototype image sensor. It is believed that this new OPD platform opens up the possibility for the ultrahigh‐sensitivity organic image sensors, while maintaining the advantageous properties of organics.


DOI: 10.1002/adma.202104689
describing the number of conducting charges that can be induced by one incident photon. In the classical theory, the photoconductive gain can be expressed as where τ r is the minority carrier recombination lifetime, and τ transit is the carrier transit time. [13,14] This theory is related to intrinsic gain generation and explains that a transit time shorter than the recombination time allows the photogenerated carriers to circulate within the circuit multiple times, generating gain. More recently reported alternative theory predicts that the high gain observed in the experiments can originate from other external effects such as defects, surface conditions, and surface depletion regions that may localize excess minority carriers, resulting in accumulated excess majority carriers creating gain, or extrinsic gain generation mechanism. [15] In this case, the gain (for a p-type photoconductor) is determined by where Δp and Δn are the excess hole and electron densities, respectively, and μ p and μ n are the hole and electron mobilities, respectively. Despite the differences in detail, both theories predict that high gain and thus EQE of PM-OPD would be possible by enhancing the trapping efficiency of minority carriers, which can either prolong the carrier recombination lifetime (intrinsic gain), hole-to-electron mobility ratio, or excess carrier density ratios (extrinsic gain). In conventional p-type PM-OPDs, where the photoactive layer primarily comprises an electron donor material, such as poly(3-hexylthiophene-diyl) (P3HT), with a small number of acceptor molecules (typically 1 wt%), such as [6,6]-phenyl-C 71 -butyric acid methyl ester (PC 71 BM), the trapping efficiency of minority carriers or electrons is a function of the energetics of trapped electrons within isolated acceptor molecules used for PM-OPDs because an energetically deep electron trap can suppress the electron detrapping rate. Then one can speculate that employing donor-acceptor combinations with a larger lowest unoccupied molecular orbital (LUMO)

Introduction
Recently, research regarding photomultiplication-type organic photodiodes (PM-OPDs) with external quantum efficiencies (EQEs) exceeding 100% have been actively conducted. [1][2][3][4][5][6][7][8][9][10][11][12] Because the main application fields of OPDs are expanding not only to cameras, but also to finger/vein/iris recognition sensors and bio-signal sensors, which are operated primarily by weak light sources, PM-OPD can be a promising candidate for the commercialization of OPDs based on its self-signal amplification behavior. The high EQE of PM-OPDs is a result of photoconductive gain generation, where the gain is an index www.advmat.de www.advancedsciencenews.com offset would result in increased electron trapping efficiency and thus a higher EQE. There have been several reports that introduce various acceptor molecules instead of conventional PC 71 BM with larger LUMO offset; [7][8][9] however, a clear correlation between LUMO offset and EQE was not found. In fact, the EQE of high-performance PM-OPDs with various combinations of donor-acceptor molecules remained ≈100 000%, with a maximum value of 150 000%. [3,[6][7][8][9][10] One possible explanation for this phenomenon is that trapped electrons in localized PC 71 BM can find various energetically more favorable escape routes other than LUMO offset, such as structural defects or chemical defects, through which trapped electrons can reach the transport level of the donor polymer. This implies that the electron trapping efficiency cannot be significantly improved in a bimolecular donor-acceptor system without external force.
In this study, a new device architecture for extremely highperformance PM-OPDs by introducing an electric double layer (EDL) adjacent to the photoactive layer of a PM-OPD is presented, such that strong electrostatic interactions can occur between intentionally introduced positively charged interfacial molecules and trapped electrons within the acceptor molecules, which can significantly increase the trapping efficiency. A polymer electrolyte, poly(9,9-bis(3'-(N,N-dimethyl)-N-ethylammoinium-propyl-2,7-fluorene)-alt-2,7-(9,9-dioctylfluorene))dibromide (PFN-Br) was introduced as a work-function-modifying layer of indium tin oxide (ITO). It was revealed that PFN-Br not only reduced the work function of ITO to achieve effective Schottky junctions with P3HT, but also efficiently enhance the trapping efficiency, which can be ascribed to electrostatic interactions between the positively charged quaternary ammonium groups and trapped electrons within the isolated PC 71 BM domains. In addition, preferential and sequential alignments of each counterion (bromide anion), cation, and hydrophobic polymer backbone along the vertical direction allowed a highly ordered face-on structure of the upper-deposited P3HT, thereby reducing the defect states within the P3HT crystalline domains and thus enabling additional stabilization of trapped electron states. Drift-diffusion simulation clearly shows the formation of deeper effective trap states owing to the synergetic contributions of the introduced EDL. Several additional experiments with various EDLs were conducted to prove the operating mechanism of the suggested PM-OPD with an unprecedentedly high EQE of 2 210 000%. In the last part, a demonstration of a prototype ultrasensitive organic image sensor is provided, which is especially advantageous against weak illumination. Figure 1a shows the device structure suggested in this study. The photoactive layer consists of a P3HT:PC 71 BM (100:1, w/w) bulk heterojunction, which is typical for PM-OPDs where PC 71 BM is localized within the P3HT matrix without forming electron percolation pathways. PFN-Br was deposited onto precleaned hydrophilic ITO substrates by using methanol/isopropyl alcohol (2:1, v/v) as a processing solvent. Considering the protic solvent nature of methanol and its solvation ability for halide ions, [16] it is assumed that the bromide anions can be preferentially adsorbed onto the high-energy ITO surface. Subsequently, a solid-state EDL layer would be formed by the upper deposition of the PFN backbone with electrostatic interactions between the pre-adsorbed bromide anions and quaternary ammonium cations. [17][18][19] This can be confirmed via ultraviolet photoelectron spectroscopy (UPS) measurements for both bare ITO and ITO/ PFN-Br surfaces, as shown in Figure S1, Supporting Information. An apparent shift in the secondary cut-off region of the UPS spectrum to a higher binding energy was observed in the ITO/PFN-Br surface compared with the bare ITO surface. Such a significant shift in the work function (≈0.6 eV) implies the sequential adsorption of bromide anions followed by quaternary ammonium cations onto the ITO surface; therefore, permanent electric dipoles pointing outward from the ITO surface were formed, as reported previously. [17][18][19] Nevertheless, considering the rotational freedom of the polymer backbone and the operating bias of the PM-OPD, which provide a positive bias to the ITO side, molecular reorientation of charged alkyl chains or migration of dissociated ions enable the partial exposure of quaternary ammonium cations to the uppermost surface ( Figure 1b). [19][20][21][22][23][24][25][26] The operation mechanism of a typical PM-OPD can be summarized as follows.  www.advmat.de www.advancedsciencenews.com to the counter electrode by the applied bias. The trapped electrons near the ITO interface accelerate the additional hole injection from the ITO electrode via band bending such that the hole injection barrier can be overcome and injected holes transit to the electrode until electrons are detrapped and recombine with free holes, thereby generating a gain. [2,5,27] In other words, the PM-OPD operates based on two key mechanisms: 1) a trapped electron-assisted photoconductive gain generation and 2) a trapped electron-assisted band bending of the Schottky junction to enable hole injection from ITO to the photoactive layer. Therefore, the trapped electron states must be stable to achieve an efficient gain generation. In this context, it is speculated that surface-exposed quaternary ammonium cations can stabilize the trapped electron states within PC 71 BM via electrostatic interactions, as schematically shown in Figure 1c. Figure 2a shows the EQE spectra of the optimized PM-OPD under various bias conditions. An unprecedentedly high EQE of 2 210 000% at 630 nm was observed when a reverse bias of 15 V was applied with a maximum ever reported responsivity of 11 200 A W −1 ( Figure S2, Supporting Information). We believe that the exponential increase of EQE as a function of bias is due to the synergetic contributions of various phenomena: as the bias increases, 1) the hole transit time decreases because

The Effects of Electrostatic Interactions on PM-OPDs
, where d is the thickness of the active layer, μ is the hole mobility, and V is the applied voltage, 2) more number of holes can tunnel into the active layer from cathode because of higher electric field, and 3) ion rearrangement within PFN-Br and the amount of exposed quaternary ammonium cations increase, resulting in the increase in the number of trapped electrons. This can be also confirmed by the Richardson-Dushman equation where J i is the injection current density, Δφ is the average energy barrier change of the Schottky junction, k is the Boltzmann constant, T is the temperature. Therefore, considering close relationship between Δφ and number of trapped electrons as well as their lifetime, the exponential increase of EQE as a function of bias can be explained.
The corresponding J-V characteristics are summarized in Figure S3, Supporting Information. To date, the highest EQE of PM-OPD was ≈150 000% and most other works also remained near ≈100 000%. [3,[6][7][8][9][10] The observed EQE in this work which is at least ten times higher than previous results should be attributed to the dramatically different electron trapping mechanism. At this stage, it is argued that electrostatic interactions between the cations of PFN-Br and trapped electrons at the PFN-Br/ P3HT:PC 71 BM interface enabled such a high EQE. Electrostatic interaction effects were also observed in the acceptor-free PM-OPD structured as ITO/PFN-Br/P3HT/MoO 3 /Ag, which exhibited EQE values of 320% (at 375 nm), 2450% (at 345 nm), and 5480% (at 375 nm) under reverse biases of 5, 10, and 15, respectively ( Figure S4, Supporting Information). Far smaller EQE values of the acceptor-free PM-OPD compared to those of the P3HT:PC 71 BM-based PM-OPD can be attributable to the absence of effective electron trap sites and lack of exciton dissociation ability. As explained above, the EQE of PM-OPD is primarily determined by the electron trapping efficiency and the flux of electrons emitted from the traps via thermally activated excitation, or electron detrapping flux, can be described by where n t is the density of the filled electron traps, τ d is the electron detrapping time constant, and ΔE is the effective electron trap energy. [28] Because the Coulombic attractive interaction energy between a single positive charge and an electron with 1 nm displacement can be as high as several hundreds of millielectron volts ( Figure S5, Supporting Information), it is speculated that the detrapping rate of electrons can be significantly suppressed by interfacial electrostatic interactions because of the increased ΔE (based on Equation (4)). In reality, the positive charge encountered by a trapped electron may not be +1, which can result in a lower attractive interaction energy than that suggested in Figure   www.advmat.de www.advancedsciencenews.com Figure S6a, Supporting Information, which shows that PEIE also effectively functioned as a work function modifier of the ITO surface by ≈1 eV. However, the PM-OPD with PEIE yielded a significantly lower EQE with a maximum of only 62 800% at 615 nm, which is more than 30 times lower than that of PFN-Br, implying that the ionic nature of PFN-Br is crucial for enhancing the EQE of PM-OPD ( Figure S6, Supporting Information). Despite such a high EQE, the PM-OPD with PFN-Br maintained low dark currents of 1.30 × 10 −7 and 4.27 × 10 −7 A cm −2 at −0.1 and −1 V, respectively, owing to the well-defined Schottky junction, which efficiently prevents charge injection under dark conditions. Accordingly, the current spectral densities under dark condition and various light intensities indicated a reasonably low noiseequivalent power (NEP) of 1.42 × 10 −15 W Hz −0.5 at −15 V as shown in Figure S7, Supporting Information. Using the NEP values extracted at each bias condition, the specific detectivity (D*) was calculated, as displayed in Figure 2b. An exceptionally high peak D* of 2.11 × 10 14 Jones was measured at 630 nm; it is the highest reported D* value in visible-detecting OPDs hitherto, primarily owing to the high EQE and responsivity. With the NEP as the lowest measurable light power, the linear dynamic range (LDR) of the PM-OPD was measured, as shown in Figure S8, Supporting Information, where ≈200 dB of LDR was recorded under various bias conditions. In addition, to confirm that the unprecedentedly high EQE of the suggested PM-OPD platform is reliable, deviceto-device reproducibility was tested by measuring OPD performance of 25 independently fabricated PM-OPDs with PFN-Br.
The EQE values measured from these PM-OPDs are summarized in Figure 2c. Of the 25 PM-OPDs, 20 devices showed an EQE of over 1 500 000% with an average EQE of 1 979 000%, implying that the suggested PM-OPD platform is sufficiently reliable.

Microstructural Analyses
Another synergetic contribution of using PFN-Br for the formation of EDL can be found from the morphological improvement of the upper-deposited photoactive layer. In previous studies, although several combinations of donor-acceptor materials were tested, only a similar level of EQE was observed despite the different LUMO offsets, which may be ascribed to the imperfect crystalline ordering of donor polymer semiconductors, resulting in trapped electrons within localized PC 71 BM to escape via energetically more favorable interfacial defect states present in the less ordered region of polymer semiconductors. In this regard, the more ordered and less structural defect morphology of polymer semiconductors can also be advantageous for suppressing electron detrapping and thus increasing the gain. Based on the Fowkes method, the surface energies of three interlayers (poly(3,4-ethylenedioxythiophene):poly(styrene sulfonate) (PEDOT:PSS), which is most widely used for PM-OPD, PEIE, which is introduced as another interfacial dipole layer, and PFN-Br) and P3HT were measured using a sessile drop of water and diiodomethane, as shown in Figure S9, Supporting Information. The contact angles and surface energies obtained are summarized in Table S1, Supporting Information; making it clear that PEDOT:PSS and PEIE rendered relatively more hydrophilic surfaces with a surface energy of 71.32 and 70.97 mJ m −2 , respectively, whereas the PFN-Br resulted in a relatively less hydrophilic surface with a surface energy of 45.35 mJ m −2 . Because the P3HT layer surface energy was obtained as low as 23.09 mJ m −2 , the cohesive energy between P3HT domains must be lower than the adhesion energy between P3HT and the substrates. In this case, a highly oriented and continuous P3HT crystalline domains can be more easily formed on PFN-Br layer, because of the weaker interaction between P3HT and the substrate. Figure 3a compares 2D grazing-incidence X-ray diffraction (2D-GIXD) patterns of P3HT:PC 71 BM (100:1, w/w) films deposited onto PEDOT:PSS, PEIE, and PFN-Br. The corresponding line-cut profiles along the q xy and q z axes are displayed in Figure S10, Supporting Information. Based on the peak positions of (100) and (010), the d-spacing and π-π stacking distance values of the films were estimated, as summarized in Table S2, supporting Information. In all cases, features of simultaneously developed edge-on and face-on orientations were observed, whereas P3HT on PFN-Br showed stronger preferential face-on orientations from the pole figure analysis for the (010) peak of the three films (Figure 3a), presumably owing to more efficient π-π interactions between the π-conjugated backbones of P3HT and PFN-Br. Paracrystalline disorder (g) which is related to the structural disorder in an imperfect crystal and hence the electrical trap states of a polymer film can be calculated using 2D-GIXD data. [29,30] The paracrystalline disorder parameter for π-π stacking in the outof-plane direction (g (010),z ), which is directly related to the vertical charge transport in diode geometry, can be calculated by using the single peak-width estimation method as follows.
where Δq is the width of the diffraction peak and q 0 is the center position of the peak. The extracted paracrystalline disorder parameters of P3HT:PC 71 BM (100:1, w/w) films on PEDOT:PSS, PEIE, and PFN-Br were 13.58%, 10.77%, and 8.18%, respectively. For clarity, rest of paracrystalline disorder parameters (for lamellar stacking in the out-of-plane direction, lamellar stacking and π-π stacking in the in-plane direction) were also handled ( Figure S11 and Table S2, Supporting Information) and it is concluded that the P3HT:PC 71 BM film on PFN-Br overall exhibited the highest paracrystallinity. Detailed explanation regarding paracrystalline disorder parameters is described in the Supporting Information. The 2D-GIXD results clearly confirmed the preferential face-on molecular orientation of the P3HT:PC 71 BM (100:1, w/w) film on PFN-Br with apparently low paracrystalline disorder, which can result in relatively less amorphous or more ordered regions. To evaluate the actual hole trap state density distribution in the PM-OPDs with PFN-Br or PEDOT:PSS, thermal admittance spectroscopy analysis was conducted, as shown in Figure 3b. The P3HT:PC 71 BM (100:1, w/w) film on PFN-Br showed a relatively low hole trap density of 10 15 to 10 16 cm −3 eV −1 over the entire range, whereas P3HT:PC 71 BM (100:1, w/w) film on PEDOT:PSS exhibited a higher hole trap density of 10 16 to 10 18 cm −3 eV −1 , which is consistent with the 2D-GIXD studies. Although we believe that the electrostaticdriven interfacial contribution is a major source of stabilized electron states for high-EQE PM-OPDs with PFN-Br, the highly ordered morphological feature of the upper-deposited P3HT can also be a reasonable contribution to the observed high gain.

Theoretical Interpretation Based on Numerical Simulations
In order to elucidate the effects of the strategically employed EDL on the trapping efficiency, a drift-diffusion simulation was conducted to experimentally fit obtained data and find relevant fitting parameters. To model the injection, transport, generation, and trapping of carriers, the drift-diffusion approximation was used similar to various previous works, and Fluxim software was used for the actual numerical simulation. [31][32][33][34][35] For simplicity, a single-level trap state was assumed, as used in various theoretical analyses of charge transport. [36,37] For boundaries, typical conditions introduced by Schottky were used, which assumes constant electron and hole concentrations at the semiconductor/electrode under an external electric field and illumination. [37,38] Because electrons are highly localized owing to small contents of PC 71 BM, hole and electron mobilities were set as 1 × 10 −4 and 5 × 10 −8 cm 2 V −1 s −1 , respectively, which are consistent with hole-only and electron-only space-charge-limited current (SCLC) measurements as summarized in Figure S12, Supporting Information. Another important assumption used for the simulation is that the Schottky contact between ITO/ PFN-Br and P3HT can be tuned to a pseudo-Ohmic junction because of the nearby trapped electrons under illumination, as suggested in previous studies. [6][7][8] Therefore, different work function values were used for dark and illumination condition. The optical parameters of each constituting layer were obtained either by our own measurements or from references. All other fitting parameters were carefully set based on several relevant references and our own measurements. [32,34,39] The details of the simulation parameters are summarized in Table S3, Supporting Information, which are identical to those of the actual measurements. Figure 3c compares the responsivity versus voltage (R-V) relations of PM-OPDs with PFN-Br and PEIE with both experimental and simulated data. In the case of PM-OPD with PFN-Br, trap energy depth of 0.88 eV with respect to the LUMO of P3HT was used, while an apparently smaller trap energy depth of 0.56 eV was enough to reproduce the R-V relation of PM-OPD with PEIE. This indicates that the electrostatic interaction is a direct source of more stabilized trapped electron states, and thus significantly enhances the gain. This simulation also suggests that the design and introduction of a more efficient EDL can yield an even higher EQE.

Other Electrolytes for Electrostatic Interactions
Furthermore, we fabricated a PM-OPD using another ionic polymer electrolyte to prove that the interfacial electrostaticinteraction-based strategy for improving the EQE can be universally applied in other ionic work-function-modifying layers. Hence, we introduced a cationic polymer electrolyte,  in Figures S13b and S13c, Supporting Information, respectively. The PM-OPD with PTB7-NBr exhibited a high EQE of ≈272 600% at 565 nm under a reverse bias of 15 V, which is apparently higher than that of conventional PM-OPDs, demonstrating the validity of the electrostatic interactions between quaternary ammonium cations and trapped electrons. Relatively smaller EQE of the PTB7-NBr-based PM-OPD owes to the mutual positions of ionic alkyl chains; they are on the opposite side in PTB7-NBr while they are on the same side in PFN-Br. Therefore, the dipole moment of PTB7-NBr should be smaller than that of PFN-Br, leading to relatively weaker electrostatic interactions. Another example that is expected to yield the same effect is an ionic surfactant. We used cetyltrimethylammonium bromide (CTAB) as a work-function-modifying layer of ITO, and a PM-OPD with CTAB was also constructed in the same manner ( Figure S14, Supporting Information). The PM-OPD with CTAB also revealed a quite high EQE of ≈356 800% at 580 nm under −15 V, demonstrating the effect of interfacial EDL. Nevertheless, imperfect coverage of CTAB layer just by spin-coating, which leads to a weaker dipole and thus less effective electrostatic interactions, inevitably resulted in a relatively smaller EQE than the PM-OPD based on PFN-Br. The performances of PM-OPDs based on various interlayers are summarized in Table 1.

Response Speed of PM-OPDs
One disadvantage of the PM-OPD is its relatively low operational speed, resulting in a significantly limited −3 dB cut-off frequency. [3,4] According to the traditional intrinsic gain generation mechanism of Equation (1), the gain is directly proportional to the carrier lifetime, and because the carrier lifetime should directly affect the response time of the photodiode, the PM-OPDs often demonstrate slower responses compared with conventional OPDs. The relation between the carrier lifetime and response time of the PM-OPD can be obtained by solving the rate equation for free and trapped holes.
where τ 0 is the response time, and p and p t are the free and trapped hole carrier densities, respectively. Details regarding the mathematical processes used to attain Equation (6) are available in the Supporting Information. According to the multiple trap and release model, [40] most of free holes can be assumed as trapped holes in the localized trap states, and a small number of holes on the transport bands originating from the trap states by thermal energy can contribute to the electric current. Therefore, it can be assumed that the trapped hole density (p t ) is higher than the free hole density (p), and it can be concluded that the response time of the device should be longer than the carrier lifetime. In this work, the optimized PM-OPD with PFN-Br rendered a reasonably fast response time of 402 μs, as confirmed by the −3 dB cut-off frequency of 870 Hz in Figure 3d and the transient photoresponse curve in Figure S15, Supporting Information. A far shorter lifetime is expected from Equation (6), which cannot support the observed large gain in this work according to the intrinsic gain generation mechanism described by Equation (1). It is speculated that the observed rather fast temporal response of the high-EQE PM-OPD with PFN-Br can be explained by the extrinsic gain generation mechanism of Equation (2). According to this mechanism, the gain is not only proportional to the carrier lifetime but also proportional to the relative ratio of carrier mobilities and excess charge carrier densities. In other words, a short carrier lifetime does not necessarily result in low gain; rather, the gain generation mechanism is a complex function of carrier lifetime, transit time, defects, and surface conditions. The observed lower defect states of the suggested PM-OPD with PFN-Br (Figure 3b) presumably enabled optimized combinations of the above-mentioned parameters determining the gain without seriously prolonging the carrier lifetime. This differs significantly from the conventional PM-OPDs structured as ITO/PEDOT:PSS/P3HT:PC 71 BM (100:1, w/w)/Al, which demonstrates a small bandwidth of less than 10 Hz while exhibiting a high EQE of ≈42 000% at 610 nm under a reverse bias of 15 V ( Figure S16, Supporting Information). Owing to the observed rather fast temporal response of the optimized high-EQE PM-OPD, an unprecedentedly high gain-bandwidth (G-B) product of 1.92 × 10 7 Hz was observed, of which both the gain and bandwidth were measured under the same light intensity of 9.96 × 10 −6 W cm −2 . G-B products of previous PM-OPDs where both measurements of the gain and bandwidth were conducted under the same illumination condition are summarized in Table 2 and one can find out that such a high G-B product has not been reported. [3,10,41,42] The G-B product of the proposed PM-OPD is ≈50 times higher than the previous record.

Image Sensor Demonstration
Considering the exceptionally high EQE of >2 000 000% and responsivity of > 10 000 A W −1 for the suggested PM-OPD, we prepared prototype organic image sensors consisting of 10 × 10 pixelated PM-OPD arrays. Each pixel feature size was set to 50 μm × 50 μm, defined by both the top electrode deposited via a shadow mask and patterned bottom ITO electrode. For comparison, arrays of conventional BHJ OPDs structured as ITO/PEDOT:PSS/P3HT:PC 71 BM (50:50, w/w)/Al and conventional PM-OPDs structured as ITO/PEDOT:PSS/P3HT:PC 71 BM (100:1, w/w)/Al were prepared. Subsequently, we illuminated the prepared OPD arrays with a relatively weak intensity light (520 nm, 6.40 × 10 −7 W cm −2 for the BHJ OPD and PFN-Brbased PM-OPD arrays; 600 nm, 5.06 × 10 −7 W cm −2 for the PEDOT:PSS-based PM-OPD array) through a star-shaped shadow mask. Next, the photocurrent measured by each pixel was recorded, translated to responsivity, and visualized, as displayed in Figure 4. Note that relatively lower responsivity values were observed at the vertices of the star shape because the pixels are not right at the vertices as displayed in the blueprints of the pixel arrays and star-shaped shadow mask ( Figure S17, Supporting Information). All the prepared prototype organic image sensors were successful to read a star-shaped optical signal with high precision. Nonetheless, significantly clearer images could be obtained from arrays of high-EQE PM-OPDs developed in this study, owing to high EQE and thus high responsivity.

Conclusion
A new OPD platform for exceptionally high EQE and high responsivity was proposed. Unlike conventional PM-OPDs, where isolated acceptor molecules are solely responsible for the electron trapping mechanism, in this study, we additionally introduced interfacial electrostatic interactions between the surface cations of the PFN-Br interlayer and trapped electrons to realize a more efficient trapping mechanism. In addition, owing to the surface-energetic advantages of PFN-Br, long-range ordered preferential face-on stacking of P3HT was achieved, resulting in a remarkably low paracrystalline disorder, which can also contribute to the stabilized trapped electron states. Such synergetic contributions of PFN-Br EDL enabled an unprecedentedly high EQE of 2 210 000% as well as an unexpectedly high responsivity of 11 200 A W −1 and high specific detectivity exceeding 10 14 Jones. Numerical simulations based on the drift-diffusion approximation of charge carriers were introduced to explain the effect of the charged interfacial layer on the trapping mechanism. Other interfacial ionic layers which can form solid-state EDL were further tested to verify the effects of electrostatic interaction on stabilizing trapped electron states and thus enhancing gain. The feasibility of the proposed high-EQE PM-OPD platform was tested using fabricated prototype image sensors with 10 × 10, 50 × 50 μm pixel arrays.
Device Fabrication: To fabricate PM-OPDs, ITO-patterned glass substrates were cleaned by sequential sonication in a Mucasol solution,  www.advmat.de www.advancedsciencenews.com distilled water, acetone, and isopropyl alcohol, followed by UV-ozone treatment for 30 min. To prepare interlayers of each device, PFN-Br and PTB7-NBr were dissolved in methanol:isopropyl alcohol (2:1, v/v) at a concentration of 0.2 mg mL −1 , and CTAB was dissolved in distilled water at a concentration of 0.4 mg mL −1 . The PEIE solution was diluted in 2-methoxyethanol with a concentration of 0.35-0.4 wt%. All interlayer solutions (PFN-Br, PTB7-NBr, CTAB, PEIE, and PEDOT:PSS) were spincoated on top of the UV-ozone-treated ITO substrates at 5000 rpm, followed by thermal treatment at 140 °C for 20 min to evaporate the residual solvent. The P3HT:PC 71 BM (100:1, w/w) blend was dissolved in DCB at a concentration of 40 mg mL −1 . The solutions were stirred at 80 °C for 24 h, then spin-coated at 1000 rpm on top of each interlayer, and thermally annealed at 150 °C for 20 min in a nitrogen-filled glove box. Molybdenum oxide (MoO 3 )/Ag or Al electrodes were deposited onto the active layers by thermal evaporation under high vacuum. The photoactive area of the fabricated devices was 0.09 cm 2 .
Thin Film Deposition and Characterization: The samples for UPS and 2D-GIXD measurements were fabricated with ITO-deposited glass and Si ++ /SiO 2 substrates, respectively, which were deposited with the interlayer solutions as described in the device fabrication section. The UPS measurements were performed with an ESCALAB 250Xi system under high vacuum and an ultraviolet source of He I (21.2 eV) was used. 2D-GIXD measurements were conducted using PLS-ΙΙ 3C and 9A beamlines at the Pohang Accelerator Laboratory (PAL) in Korea.
Device Characterization: J-V characteristics and EQE spectra were measured using a combination of Keithley 2450 SourceMeter and Oriel Cornerstone 130 1/8 m monochromator, in control of homemade LabView programs. The noise current values that were used for the calculation of the specific detectivity were evaluated at frequency (>10 Hz) just over the cut-off frequency, where the white noise starts to dominate. For NEP measurements, a Stanford Research SR570 current pre-amplifier and Agilent 35670A spectrum analyzer were used with a chopped light source of which the light intensity is controlled with neutral density (ND) filters. [43,44] Specific detectivity was calculated using the definition where q is the elementary charge, λ is the wavelength of incident light, A is the photoactive area, EQE is the EQE, h is the Planck constant, c is the speed of light, and i noise is the noise current. [38,45,46] The LDR was measured using Keithley 2450 SourceMeter and two different light sources with various optical filters: monochromatic light (520 nm) from a 300 W Xe arc lamp of the above-mentioned monochromator for light intensities below 4.03 × 10 −4 W cm −2 , and a laser (520 nm) for light intensities up to 1.32 × 10 −1 W cm −2 , and calculated using the equation = LDR 20log max NEP P P (8) where P max is the maximum value of the detectable power density and P NEP is the power density extracted from NEP. [45,46] Due to the current limit of the measurement (|I| ≤ 1.05 A), deviation could not be observed in the photocurrent measurements at −10 and −15 V. The −3 dB cut-off frequency and transient photocurrent were measured using a TDS5052 digital phosphor oscilloscope (Tektronix) and an optically filtered laser diode in connection with a Tektronix AFG310 arbitrary function generator. The SCLC analyses were performed with the geometry of ITO/PEDOT:PSS/P3HT:PC 71 BM (100:1, w/w)/MoO 3 /Ag and ITO/PEIE/ P3HT:PC 71 BM (100:1, w/w)/LiF/Al for hole-only and electron-only devices, respectively. The theoretical fitting was provided by the SCLC equation.
where ε is the relative permittivity of the thin film, ε 0 is the permittivity of free space, μ is the charge carrier mobility, and d is the thin film thickness. [47]

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