The State‐of‐the‐Art Solution‐Processed Single Component Organic Photodetectors Achieved by Strong Quenching of Intermolecular Emissive State and High Quadrupole Moment in Non‐Fullerene Acceptors

A bulk‐heterojunction (BHJ) blend is commonly used as the photoactive layer in organic photodetectors (OPDs) to utilize the donor (D)/acceptor (A) interfacial energetic offset for exciton dissociation. However, this strategy often complicates optimization procedures, raising serious concerns over device processability, reproducibility, and stability. Herein, highly efficient OPDs fabricated with single‐component organic semiconductors are demonstrated via solution‐processing. The non‐fullerene acceptors (NFAs) with strong intrinsic D/A character are used as the photoactive layer, where the emissive intermolecular charge transfer excitonic (CTE) states are formed within <1 ps, and efficient photocurrent generation is achieved via strong quenching of these CTE states by reverse bias. Y6 and IT‐4F‐based OPDs show excellent OPD performances, low dark current density (≈10−9 A cm−2), high responsivity (≥0.15 A W−1), high specific detectivity (>1012 Jones), and fast photo‐response time (<10 µs), comparable to the state‐of‐the‐art BHJ OPDs. Together with strong CTE state quenching by electric field, these excellent OPD performances are also attributed to the high quadrupole moments of NFA molecules, which can lead to large interfacial energetic offset for efficient CTE dissociation. This work opens a new way to realize efficient OPDs using single‐component systems via solution‐processing and provides important molecular design rules.


Introduction
[6][7][8] Although BHJ OPDs produce excellent device performance with high responsivity and detectivity due to the large donor/acceptor (D/A) interfacial areas, in terms of commercialization of OPDs, the use of two or more different materials as the photoactive layer has many limitations, such as complicated optimization procedures, [9,10] low reproducibility of the film quality, [11] and unavoidable broadband absorption due to multi-component blend systems. [2][14] The PM effect is typically realized by charge tunneling via an extremely precise amount of trap states [13,15] by challenging synthesis of double-cable molecules [16] or resonant cavities architectures, [17] resulting in relatively high fabrication complexity and operability compared to normal OPDs.At the same time, high reverse voltage (>15 V) is required, with adverse consequence on dark current, and trap-induced tunneling slows down photodetection speed.In this context, a singlecomponent photoactive layer would be ideal to overcome these limitations.A few single-component OPD studies have been reported in recent years for solution-processed or evaporated organic small molecules such as pyrrolopyrrole derivatives and phthalocyanines, but their device performance is inferior to BHJs or planar-heterojunctions due to the lack of interfacial energetic driving force for exciton dissociation. [18,19]he development of non-fullerene acceptors (NFAs) enables not only efficient organic BHJ photovoltaics (OPVs) [20,21] but also high-performance broadband [22,23] or near infrared BHJ OPDs. [24]Particularly in OPVs, Y6, a widely studied highperforming NFA with a distinctive molecular structure of A-DA'D-A, paves the way for the rapid advancement of OPV performance; [20] and recently, a power conversion efficiency (PCE) > 19% has been reported using its derivative. [25]Such high performance in Y6 and Y6 derivative-based OPVs has been attributed to many structural and photophysical properties, which include distinctive Y6 molecular packing structures which influence the degree of intermolecular interactions between neighboring NFA molecules, with the formation of emissive and delocalized excitons. [26]A barrierless charge generation is also proposed in BHJ OPVs due to a gradual change of electrostatic potential induced by the high quadrupole moment of Y6.It is also reported that Y6 exhibits free charge photo-generation in its neat film, which is attributed to strong electronic coupling between exciton and charge transfer (CT) states and energetic offsets between CT states formed by differently - stacked Y6 pairs. [27]evertheless, OPV devices fabricated with Y6 used as a singlecomponent photoactive layer show poor performance. [27,28]n this study, we utilize these distinctive properties of Y6 in single-component OPDs and elucidate the molecular origins of electronic states and interfacial energetics that are key to the photophysical processes such as charge generation, recombination, and extraction.In particular, the impact of the molecular structures (e.g., planar vs twisted) and molecular properties (e.g., quadrupole moment, intermolecular interaction, and packing) of NFAs on these processes is identified using planar Y6 versus twisted O-IDFBR [29] and other representative NFAs with planar molecular structure but different quadrupole moment along - stacking direction (Q  ) values and D/A character such as IDIC [30] and IT-4F. [31]y doing so, herein, we successfully demonstrate highperformance single-component OPDs using Y6 and IT-4F and identify the origin of superior device performance by comparison with the other NFAs O-IDFBR and IDIC.Y6 and IT-4F devices show great enhancement of external quantum efficiency (EQE) by ≈40 times (for Y6) under reverse bias, reaching ≈50% at −5 V.In addition, these devices show excellent photocurrent linearity, low dark current density (≈10 −9 A cm −2 ), high responsivity (≥0.15A W −1 ), high specific detectivity (>10 12 Jones), and fast photo-response time (<10 μs), comparable to state-of-the-art BHJ OPDs.By measuring bias-dependent photoluminescence (PL) spectra, we show that most of the photocurrent generation in Y6 and IT-4F devices originates from dissociation of the emissive intermolecular CT excitonic (CTE) states in reverse bias, with a clear correlation between EQE increase and PL quenching efficiency.We also confirm much stronger intermolecular interactions, and in particular, the ultrafast formation of CTE states within Y6 films by comparing solution versus film spectroscopic data of Y6 and O-IDFBR.Further, we find the strong correlation between the excellent device performance of Y6 and IT-4F and their high Q  values (quadrupole moment along the molecular - stacking direction).This is attributed to high Q  induced large electrostatic potential differences leading to efficient free charge generation even with small electric field.This work demonstrates the great potential of NFAs for single-component OPD applications and suggests molecular design guidelines for efficient OPD performance using solution-processed NFAs.

Device Structure and Material Properties
We selected various calamitic NFAs, namely O-IDFBR, [29] IDIC, [30] IT-4F, [32] and Y6. [20]These are representative materials with distinctive molecular structures, twisted (O-IDFBR), planar (IDIC and IT-4F), or banana-shaped (Y6).Molecular structures of each NFA and the energy band diagram of the OPDs are shown in Figure 1a,b, respectively.The highest occupied molecular orbital (HOMO) levels of the NFAs are determined by finding the onset of photoemission upon photoexcitation measured by ambient photoemission spectroscopy [33,34] (Figure S1, Supporting Information), while the lowest unoccupied molecular orbital (LUMO) levels are estimated from the HOMO and the optical bandgap.ZnO and MoO x are used as the electron and hole transport layers, respectively, because they are energetically favorable in terms of charge carrier extraction.It is worth noting that energetics extracted from photoemission techniques is considered to be more accurate to describe photogeneration in organic semiconductor devices compared to cyclic voltammetry (CV) results. [35]In particular, the energy levels of strong D/A character based NFA molecules with high quadrupole moments can be strongly affected depending on their molecular orientations in thin films. [36][39][40] Absorption spectra are characterized to identify the light absorption ability of each NFA in neat films (Figure 1c).The absorption range of O-IDFBR extends from 400 to 600 nm, indicating applicability to blue and green light detection.IDIC has a slightly larger optical bandgap compared to that of IT-4F, but their absorption range is quite similar from 500 to 800 nm.Y6 has a relatively broad absorption from 500 to 950 nm.In terms of absorption coefficient, IDIC shows the strongest absorption among the NFAs tested, whereas O-IDFBR shows the weakest.Nevertheless, all NFAs have sufficiently high absorption coefficients of >10 −5 cm −1 , which could indicate suitability for photodetector applications as single component systems.

Effects of Reverse Bias on Photocurrent Generation
To investigate the applicability of those NFAs to singlecomponent OPDs, we firstly investigated the photocurrent generation ability by measuring EQE under various reverse bias conditions from 0 to −5 V. Figure 2a,b represents bias-EQE spectra of O-IDFBR and Y6 OPDs, respectively.At each applied voltage, the electric field strength is comparable in all devices because the active layers have similar thicknesses (i.e., 53 nm for O-IDFBR and Y6 layers, respectively).Despite a slight increase of EQE with reverse bias for O-IDFBR devices, a strong reverse bias (> |5| V) is needed to obtain reasonable OPD performance.On the other hand, Y6 devices show a steeper EQE increase starting at a lower voltage around −3 V. Maximum EQE values near 845 nm reach ≈50% at −5 V for Y6, whereas those of O-IDFBR device only reach ≈36% at the same bias condition.EQE enhancement upon reverse bias for all OPDs is summarized in Figure 2c, by taking their maximum values at specific wavelengths (EQE spectra of IDIC and IT-4F devices are shown in Figure S2, Supporting Information).Interestingly, the trend in EQE increase of Y6 and IT-4F is clearly distinguished from O-IDFBR and IDIC.Note that EQE enhancement of all devices mainly occurs over their main absorption wavelength range as seen in normalized spectra (Figure S3, Supporting Information).As forementioned, we kept the thick-ness of NFA photoactive layers in a similar range of 50-60 nm (Table S1, Supporting Information) to avoid any impact from different external electric field strengths.Therefore, we attribute the greater increase of EQE in Y6 and IT-4F devices to the molecular structure induced effect of these NFAs, as we discuss further below.
To understand the origin of device performances just illustrated and how these are related to the specific NFA choice, we studied the bias dependent optoelectronic properties of the materials in detail.Figure 2d,e shows photoluminescence (PL) spectra of O-IDFBR and Y6 OPDs measured under reverse bias conditions from 0 to −8 V. (PL spectra of IDIC and IT-4F OPDs are shown in Figure S4, Supporting Information.)The PL quenching is not efficient even at high voltage (−8 V) in O-IDFBR devices, whereas strong PL quenching is observed in Y6 devices, activated at around −2 V which is similar to the onset of the EQE enhancement (≈ −3 V).Thus, the EQE increase in all NFA devices can be attributed to dissociation of these emissive species, exhibiting a progressive decrease in PL intensity without affecting the spectral shape (see normalized PL data in Figure S5, Supporting Information).In addition, PL quenching efficiencies (PLQE) are characterized to compare their trends with those of the EQE increase, by using the relation PLQE = (PL V0 -PL V )/PL V0 , where PL V0 is the PL intensity measured at zero bias and PL V the value measured at voltage V. Overall, as for the EQE increase, the trend of PLQE with reverse bias is distinctively different for Y6 and IT-4F on one side and IDIC and O-IDFBR on the other (Figure 2f).Namely, the first pair of molecules shows a continuous increase in PLQE upon increasing reverse bias, with much sharper growth beyond −2 V and reaching up to ≈90% at −8 V. Conversely, IDIC and O-IDFBR show a much weaker bias-induced effect for a mild applied field up to −2 V, reaching only <30% and <50% of PLQE even at high voltage of −8 V, respectively.

Organic Photodetector Performance
Spectral responsivity (R) characteristics of NFA OPDs measured at −3 V are shown in Figure 3a, calculated from EQE spectra using the relation R = EQE•q/(hc), where q is the elementary charge,  is the wavelength of light, h is Planck's constant, and c is the speed of light 1 .Thanks to the higher EQE, Y6 and IT-4F devices exhibit much higher maximum R of 0.151 and 0.169 A W −1 , respectively, compared to those of O-IDFBR and IDIC devices (0.040 and 0.057 A W −1 ), at −3 V applied bias.Figure 3b represents dark current density-voltage (J d -V) characteristics of NFA OPDs.Y6 and IT-4F devices exhibit similarly low J d maintained below 10 −7 A cm −2 even at high reverse bias (−5 V).IDIC and O-IDFBR devices show slightly higher J d (10 −6 -10 −5 A cm −2 ), but still sufficiently low not to significantly affect the photocurrent.In this regard, it seems that the NFA energetics is not the dominant factor to determine J d .In fact, high dark current is expected from Y6 OPDs due to its shallow HOMO and small bandgap, favoring dark hole injection and thermal dark generation.For opposite considerations, a low J d can be predicted from O-IDFBR devices.However, the experimental results suggest that different factors dominate the levels of J d in different materials, probably ascribed to the different morphologies created by intermolecular interactions in different NFAs (e.g., better packing and higher shunt resistance in Y6 and IT-4F), as will be further illustrated in the following sections.The specific detectivity (D*) can be obtained from the relation D* = R/(2qJ d ) 1/2 , assuming the noise of OPDs is dominated by the shot noise in J d . [24]Figure 3c shows spectral D* curves of NFA OPDs characterized at −3 V. Y6 and IT-4F devices exhibit a maximum D* of 2.74 × 10 12 and 3.11 × 10 12 Jones, respectively, an order of magnitude higher than those of O-IDFBR and IDIC devices (4.32 × 10 11 and 3.17 × 10 11 Jones, respectively).In addition, Y6 and IT-4F devices maintain D* > 10 12 Jones even at higher voltages from −2 to −5 V, and these high D* values are attributed to both lower J d and higher R (Figure 3d).Overall OPD parameters measured under −3 V bias condition are summarized in Table S1, Supporting Information.We note that the D* values calculated by using the equation above could be overestimated as only shot noise is considered.Corrected D* (D* Corr. ) by including flicker and thermal noises is shown in Figure S6b, Supporting Information.The Y6 device exhibits an order of magnitude higher D* Corr.(1.82 × 10 9 Jones) than that of the O-IDFBR device (1.22 × 10 8 Jones), showing consistent trends with the original calculation.As most of the reports of solution-processed single-component OPDs use the equation of D* = R/(2qJ d ) 1/2 , it seems not fair to compare device performance using D* Corr. .Based on the D* calculation, our devices (especially Y6 OPD) show the state-of-the-art device performance among solution-processed single-component OPD reports. [41,42]Further, to assess the device performance of particularly Y6 and IT-4F devices, we compared our results with other reports based on state-of-the-art BHJ and PHJ OPDs (Table S2, Supporting Information).Even without incorporating donor materials, singlecomponent Y6 and IT-4F devices showed comparably excellent OPD performance-sufficiently low J d (≈10 −9 A cm −2 ) and high R (>0.15A W −1 ) and D* (>10 12 Jones), indicating great potential of NFAs as single-component systems for OPD applications.

Photophysical Comparison of NFA Films and Solutions
In order to address the origin of the field dependent EQE and PL quenching presented above, we now turn to a detailed comparison of Y6 and O-IDFBR photo-physics; and particularly, how their photo-physics is impacted by film formation.PL spectra of O-IDFBR and Y6 in solution with various concentrations (0.001 to 1 g L −1 ) and as neat films are plotted in Figure S7, Supporting Information.As the solution concentration increases, the PL peak of Y6 gradually red-shifts, whereas the PL peak of O-IDFBR is not changed, indicative of Y6 electronic intermolecular interactions even in concentrated solution.Even more strikingly, Y6 film formation results in a 145 nm red shift compared to dilute solution, coupled with a broadening of the emission peak and an ≈0.27 fold reduction in emission quantum yield (PL quantum yield of 12.6% for 0.005 g L −1 diluted solution to 3.4% for film).An analogous red shift is observed in Y6 absorption (Figure S8, Supporting Information).In contrast, the absorption and PL spectra of O-IDFBR solutions and films are relatively invariant, with the film PL spectrum exhibiting only a 32 nm (blue) shift relative to the solution spectrum.These spectral results are clearly indicative of Y6 film formation inducing large changes in its photophysical properties compared to isolated molecules in solution, with these changes being much less pronounced for O-IDFBR.
The red-shifted, broadened PL of Y6 films compared to solution is suggestive of increased charge transfer character of excited states in Y6, induced by strong intermolecular interactions between Y6 molecules. [31]This conclusion is in good agreement with the recent report of photo-induced charge generation in Y6 films. [27]Further support for this conclusion comes from transient optical analyses of Y6 and O-IDFBR solutions and films.Turning to Y6, in solution, the Y6 TA data are dominated by a single, exponential decay phase with a decay constant indistinguishable from that observed by TCSPC ( ≈ 1 ns), indicative as expected of exciton dominated kinetics, as for O-IDFBR (Figure 4b).In contrast, the TA data for Y6 films are more complex.In particular, an initial (<1 ps), fluence independent decay phase is observed when probing the exciton PIA peak, indicative of an ultrafast, monomolecular exciton relaxation process.This is followed by a slower, fluence dependent decay phase.Unlike Y6 solution data, global analysis of Y6 film data requires a minimum of two components to reasonably fit the data (Figure S11, Supporting Information), including a residual, long lived decay phase.Assignment of these different decay phases is complex, with recent reports considering singlet and triplet states of exciton and CT character. [27,43]Nevertheless, these data, and in particular, our observation of a sub-picosecond, monomolecular decay of Y6 exciton PIA in films, but not solution, is indicative of ultrafast CTE state formation induced by strong intermolecular interactions within Y6 films, consistent with previous literature and our steady-state absorption and PL data discussed above.Intermolecular CTE state formation in Y6 molecules is also confirmed by DFT calculation, which will be discussed in a later section.We note the emission of these Y6 CTE states is strongly quenched in BHJs, indicating they are highly mobile.TCSPC data for Y6 films indicate that the ≈1 ns decay time for these emissive Y6 CTE states exhibits a negligible fluence dependence (Figure S12b, Supporting Information), consistent with our assignment of this emission to the monomolecular decay of CTE states.We note these data do not rule out a model where initially generated excitons undergo ultrafast charge transfer resulting in a bias dependent equilibration with charge transfer states (i.e., a thermal equilibrium rather than quantum mechanical mixing of exciton and charge pair states); although, for simplicity we will re-fer to both cases as CTE states hereafter.In any case, these charge transfer characteristics of Y6 photoexcitations (as well as IT-4F) are correlated with field-dependent charge separation and photocurrent generation, as discussed above, indicating the separation of these Y6 photoexcitations can be substantially enhanced by an electric field.This CTE state formation in Y6 (as well as IT-4F) is correlated with field-dependent charge separation and photocurrent generation, as discussed above, indicating the separation of these CTE states can be substantially enhanced by an electric field.

Charge Carrier Recombination, Transport, and Extraction
We now turn to the consideration of the charge extraction efficiency in our NFA OPDs and whether this may also impact on the field dependent EQE data reported in Figure 2. All devices show excellent photocurrent linearity as a function of light intensity for both reverse bias and short-circuit conditions (Figure 5a; Figure S13, Supporting Information, respectively).O-IDFBR and IDIC devices exhibit a lower dynamic range for photocurrent detection, consistent with their poorer OPD performance: at −3 V; Y6 device shows 87 dB of linear dynamic range (LDR), whereas O-IDFBR device shows 50 dB (Figures S13d and S13e, Supporting Information, respectively).The linear photocurrent behavior under given light intensities is strongly indicative of the absence of non-linear processes such as bimolecular recombination competing significantly with charge transport and extraction. [44]This is particularly striking given the absence of any heterojunction in single-component systems to spatially separate electrons and holes.These data indicate that the field dependence of the EQE does not result from field dependent extraction, but rather from field dependent charge generation, consistent with the bias dependent PL data discussed above.Most strikingly, these data indicate efficient charge extraction, which is field independent for all the OPDs studied, even at short circuit.
To investigate the charge carrier extraction properties further, the photodetection time of NFA OPDs was evaluated under reverse bias using transient photocurrent (TPC) measurements.Normalized TPC data of O-IDFBR and Y6 devices measured at  S3, Supporting Information.)c) Extracted charge carrier density versus light intensity (up to 390% sun equivalent LED illumination) for different bias conditions.The charge carrier density is measured by integrating the current transients recorded upon switching off the LED light.The dashed lines signify a power law fit (≈I  ), revealing a sublinear expected for disordered semiconductors exhibiting a tail state density.d) Laser induced charge carrier density versus LED induced background charge carrier density (up to 3 sun equivalent LED illumination).
−5 V are shown in Figure 5b (see Figures S14-S16, Supporting Information, for non-normalized data and normalized data of IDIC and IT-4F devices, respectively).Y6 devices exhibit almost twice faster rising ( rise ) and falling ( fall ) time, compared to those of O-IDFBR devices.IT-4F devices also show similarly fast  rise and  fall (Figure S16, Supporting Information).In addition, see Figure S17, Supporting Information, for the analysis of the RC time constants demonstrating that the photo-response is not limited by the capacitive discharge.This faster photo-response time for Y6 devices can be attributed to possibly more balanced and sufficiently high mobilities of both holes and electrons as previously established in the literature (see Figure S18, Supporting Information for an estimation of the lower bound of the effective drift mobility). [28,45,46]Overall results indicate that Y6 and IT-4F have a great potential as single-component systems for OPD applications with sufficiently good J ph linearity as well as fast response time under OPD operational conditions.
Charge collection efficiency has previously been discussed as the major limitation to Y6 single component device operation (especially in solar cell operation [47] ) rather than the poor charge generation at forward or reverse bias.Improvements of photocurrent with reverse bias could in that picture be explained by reducing non-geminate recombination during extraction.How-ever, the excellent linearity of photocurrent at all tested biases indicates the absence of non-linear processes (bimolecular recombination), limiting the charge collection efficiency in the devices presented herein (Figure 5a; Figure S13, Supporting Information).It has also been proposed that improving charge extraction aids the generation of free charges in Y6 by potentially shifting a thermodynamic equilibrium between photoexcited states and free charges to favor more charge generation. [28]To investigate the influence of the free carrier density in the device on generation efficiency, small perturbation transient photocurrent measurements were carried out at different background LED light intensity and bias levels.The steady state charge carrier densities in the device were determined by charge extraction measurements.Together, these measurements allowed us to determine the generation efficiency from pulsed laser excitation as a function of the background charge carrier density and bias (see Supporting Information for details of the method).First, we considered the charge carrier density in the device as a function of continuous LED excitation density for different bias voltages, Figure 5c.The increase in the extracted charge carrier density with bias together with the identical slope supported an increase in extracted charge carrier density due to the increased generation efficiency.Figure 5d shows the laser induced charge carrier density versus the background charge carrier density induced by the LED light for a range of background light intensities and a reverse bias ranging from −1 to −5 V.It is evident that the charge carrier density induced by the laser showed a strong bias dependence (see Figure S19, Supporting Information for comparison of the generation profile with the J-V curves), indicative of bias driven charge separation, as discussed above.In contrast, the dependence on the background charge carrier density was negligible, ruling out any significant effects of an equilibrium between photoexcited species and the steady state charge carrier density impacting the efficiency of free charge generation.Only at the highest LED light intensities (>100 mW cm −2 ), does a reduction in the extracted laser induced charge carrier density become evident, attributed to the onset of bimolecular recombination losses started to limit extraction efficiency, as expected.A simple kinetic model had been applied to extract a lower bound of ≈10 −11 cm 3 s −1 for the bimolecular recombination coefficient from these data (see Figure S20, Supporting Information for additional discussion).The low bimolecular recombination coefficient as well as the high charge carrier density are both surprisingly similar to the values reported for BHJs containing Y6 despite the lack of a second component (donor) to spatially separate holes and electrons.These observations further support our conclusion of bias assisted charge generation to be the key determinant of performance for single component Y6 devices studied herein.

The CTE State and NFA Quadrupole Moment
To understand the molecular origin of CTE state formation and the efficient charge generation in single-component NFA systems, we performed molecular simulations by density functional theory (DFT).First, the electrostatic potential (ESP) and its distribution within the molecule of the four NFAs studied were calculated (Figure 6a,b; Figure S22, Supporting Information).A strong A-D-A character with high ESP values is found in IDIC, Y6, and IT-4F NFAs, promoting strong intermolecular interactions between the molecules, consistent with smaller - stacking distance (d-spacing ≈3.5 Å). [27,30,48] In contrast, O-IDFBR shows a weak A-D-A character with low ESP difference leading to poor intermolecular packing, as expected from its twisted structure. [29,49]his is consistent with several studies in the field by X-ray spectroscopy and atomic force microscopy, elucidating the different packing structures of the NFAs compared here.][52] Molecules of the ITIC family (i.e., IT-4F and IDIC) are also reported to be prone to intermolecular order, [47,49] while the amorphousness and poor crystallinity of O-IDFBR are well-established. [53,54]his result confirms that the strong D and A units and strong quadrupole moments within the molecule can foster better intermolecular interactions with neighboring molecules, leading to close packing in thin films, as we have previously shown for ITIC derivatives with different acceptor group substitutions (ITIC, IT-4F, and ITIC-DM). [31]The improved packing quality in turn is expected to assist the formation of intermolecular CTE states, as discussed below.Note that although IDIC shows a relatively high ESP between D and A units (Figure 6a,b), its linear backbone and small molecular dimensions (only ≈26.5 Å compared to >30 Å for IT-4F); and hence, its small separation between D and A units limit its overlap with neighboring molecules.Next, to identify the origin of intermolecular CTE states, the electronic excited states (singlets) are calculated by time-dependent DFT (TD-DFT) for dimers of two closely packed molecules (either face-on or orthogonally) using Y6 as a representative molecule (see Supporting Information for details of the simulation method; Figure S23, Supporting Information for comparing their oscillator strengths).It is evident that the lowest-energy excited state mainly originates intermolecularly from - stacked Y6 molecules and this excited state has a strong CT character (Figure 6c), confirming the intermolecular interaction induced CTE state formation in A-D-A type NFAs.Interestingly, no such intermolecular CTE state is formed between the two molecules packed orthogonally to each other due to the lack of p-orbital overlap (Figure S25, Supporting Information).These intermolecular CTE states are coulombically bound, resulting in radiative recombination at shortcircuit condition.The external electric field can separate these bound CTE states into free charge carriers as proved by the reverse bias dependent PL quenching and concomitant increase of EQE.
In addition to enabling bias assisted charge generation by improved - stacking as described above, electrostatic effects also influence energetic offsets at interfaces which can drive charge generation.While the dipole moments of A-D-A or A-D-A'-D-A type NFAs are negligible due to their symmetric molecular structures, [31,55,56] the quadrupole moment in the - stacking direction Q  (perpendicular to the molecular plane) has previously been shown to affect the energetics within closely packed, ordered films. [57,58]Solution-processed small molecules typically form a mixture of differently oriented crystallites, often with very limited coherence lengths and less order than evaporated films.In our previous study; however, it was demonstrated that differently oriented Y6 molecules (more face-on or more edge-on) can lead to a significant energetic offset (210 meV) between HOMO (or LUMO) levels in thin films.This energetic offset originates from electrostatic effects due to the high Q  of Y6 (Figure 6d) and its tight packing in differently oriented domains. [36]In addition, we observed a significant increase in the surface photovoltage upon illumination for the film exhibiting a mixed morphology.This is indicative of an increase of free charge generation facilitated by the larger density of grain boundaries and the potential steps occurring there due to quadrupole mediated electrostatic effects.See also Figure S28, Supporting Information, for the corresponding EQE spectra.Figure 6e illustrates this important role of Q  values on the energetics of single-component NFA films with mixed domains.
As shown in Figure 6d, IT-4F and Y6 possess a much larger Q  (both > 190 ea 0 2 ) than O-IDFBR and IDIC (<150 ea 0 2 ).The overall trend is consistent with that of the EQE enhancement and the PL quenching upon reverse bias.Surface photovoltage measurements [6,59] also support small but clear photocurrent generation without an external bias in neat NFA films with high Q  values (Figure S29, Supporting Information comparing Y6 and IDIC).The even higher Q  value of IT-4F molecules does not lead to a higher photocurrent at low bias conditions compared to Y6, illustrating the need for the synergistic effects of packing structure and boundaries between differently oriented domains in facilitating photocurrent generation without external bias.In contrast, in O-IDFBR, the moderate Q  is counteracted by its twisted molecular structure and poor aggregation, preventing long range quadrupole-charge interactions and charge separation.IDIC is highly crystalline and possesses a large ESP yet a small Q  .
In summary, we propose the two key photocurrent generation pathways in the NFA materials studied herein: First, in the bulk of crystalline domains, electric field induced quenching of CTE states leads to photocurrent generation in all NFAs studied but the extent of which scales with the degree of D/A character of the NFA and its degree of intermolecular coupling-both are intrinsically linked to the Q  of the NFAs.Second, excitons generated close to a boundary between differently oriented crystalline domains split into free charges in the presence of quadrupole induced interfacial energetic offsets.This is largely dependent on the magnitude of the Q  as well as the orientation, crystallinity, and size of the domains.While it is difficult to decouple the effects of CTE formation and interfacial energetic offsets, we propose that Y6 benefits from both mechanisms for high photocurrent generation, operating as efficient single-component OPD with a superior detectivity at low reverse bias (see Figure 3).

Conclusion
In conclusion, we demonstrate high-performance singlecomponent NFA OPDs by using Y6 and IT-4F and identify the origins of their efficient photocurrent generation by comparison with O-IDFBR and IDIC.The strong electric field induced quenching of CTE states and the large quadrupole (Q  ) induced interfacial energetic offsets are proposed to be the key origins.High specific detectivity of >10 12 Jones is achieved in Y6 and IT-4F devices, enabled by low dark currents and high EQE under reverse bias condition.In addition, these devices exhibit excellent photocurrent linearity and fast photo-response times under electric field.There is a clear correlation between EQE enhancement and PL quenching upon reverse bias, indicating the origin of photocurrent generation is mostly due to dissociation of emissive CTE states.By comparing photophysical properties of O-IDFBR and Y6 in solution and film states, we demonstrate ultrafast intermolecular CTE state formation induced by strong intermolecular interactions in Y6 films.Using a series of molecular simulations, excellent device performance of Y6 and IT-4F OPDs is attributed to their strong calamitic character and high quadrupole moment, providing an additional energetic offset at differently oriented NFA crystallites, which correlates excellently with efficient intermolecular CTE state separation.
From these results, we can draw some general guidelines for the choice of the best NFA candidates for single-component OPDs.As for O-IDFBR, the twisted molecular structure (≈30°dihedral angles between indenofluorene core and benzothiadiazole end groups) limits the ESP localization and Q  value (for example, the planar counterpart O-IDTBR has a larger Q  of 173 ea 0 2 ), while determining a poor crystallinity in solid state. [29,60]The short conjugated backbone and non-halogenated end groups in IDIC also appear to be detrimental for Q  values (namely, Q  increases from 68 ea 0 2 to 154 ea 0 2 for fluorinated IDIC).On the contrary, the 2D-like geometry of Y6 molecule is crucial to establish interactions with more numerous surrounding molecules, resulting in 3D-like molecular arrangement in solid state, [26] while halogenation of Y6 and IT-4F induces charge localization in subunits and larger Q  for a better crystalline packing. [31,61]Overall, this work suggests important molecular design guidelines to achieve high-performance single-component NFA OPDs, with device demonstration as well as systematic identification of the origin of excellent performance.

Figure 1 .
Figure 1.a) Molecular structure of NFAs; O-IDFBR, IDIC, Y6, and IT-4F.b) Energy band diagram of organic photodetector devices.Note that HOMO and LUMO levels of each acceptor are determined by ambient photoemission spectroscopy and optical bandgap.c) Absorption coefficient spectra of neat NFA films.

Figure 2 .
Figure 2. EQE spectra of a) O-IDFBR and b) Y6 OPDs upon applying reverse bias from 0 to −5 V. c) Summary of maximum EQE values as a function of reverse bias for all NFA OPDs.PL spectra of d) O-IDFBR and e) Y6 OPDs upon applying reverse bias from 0 to −8 V. f) Summary of PL quenching efficiency as a function of reverse bias for all NFA OPDs.

Figure 3 .
Figure 3. Various figures of merit of single-component NFA OPDs.a) Spectral responsivity measured at −3 V bias, b) dark J-V characteristics, c) specific detectivity spectra characterized at −3 V, and d) summary of maximum specific detectivity as a function of reverse bias for all NFA OPDs.

Figure 4
represents fluence dependent transient absorption (TA) data of O-IDFBR and Y6 solution and films, and low fluence time correlated single photon counting (TCSPC) kinetics of films (TA transient spectra and global kinetic analyses) are shown in Figures S9-S11, Supporting Information.For both systems, the TA signal is probed at the singlet exciton PIA maximum (1060 nm for O-IDFBR and 930 nm for Y6).O-IDFBR films exhibit progressively slower decay kinetics as the TA fluence intensity is lowered.The TA film data are in good agreement with the ≈1 ns TCSPC decay observed for O-IDFBR in both solution and films (Figure 4a; Figure S12a, Supporting Information, respectively), and with TA solution data, typical of exciton decay dynamics in the presence of exciton-exciton annihilation at the higher

Figure 4 .
Figure 4. Transient absorption (TA) kinetics of film and solution (solution concentration: 1 μg mL −1 ) and TCSPC kinetics of film for a) O-IDFBR and b) Y6.Dashed arrows indicate the decrease of pump fluences from 60 to 2.5 μJ cm −2 .For TA measurements, O-IDFBR and Y6 were excited at 380 and 750 nm, respectively; probed at PIA peak (see TA spectra in Figure S9, Supporting Information).

Figure 5 .
Figure 5. a) Light intensity dependence on photocurrent density (J ph ) of Y6 OPDs.b) Normalized photocurrent transients of O-IDFBR and Y6 OPDs measured at −5 V bias condition (f = 1 kHz).Green (525 nm) and red (633 nm) LEDs are used as excitation light sources, respectively, suitable for large bandgap material (O-IDFBR) and the small bandgap materials (Y6).Here, we define  rise and  fall by the time required for photocurrent to change from 5% to 95% with respect to maximum saturated photocurrents, which is a relatively more restricted condition compared to normal definition (change from 10% to 90%) (See details in TableS3, Supporting Information.)c) Extracted charge carrier density versus light intensity (up to 390% sun equivalent LED illumination) for different bias conditions.The charge carrier density is measured by integrating the current transients recorded upon switching off the LED light.The dashed lines signify a power law fit (≈I  ), revealing a sublinear expected for disordered semiconductors exhibiting a tail state density.d) Laser induced charge carrier density versus LED induced background charge carrier density (up to 3 sun equivalent LED illumination).

Figure 6 .
Figure 6.a) Electrostatic maps of a monomer of IDIC and IT-4F molecules.b) Calculated electrostatic potential (ESP) along the backbone of NFAs.c) Electron (red colored) and hole (blue colored) density plots of the lowest-energy singlet excited states (S 1 ) of two Y6 molecules with face-on packing.d) Calculated quadrupole moment of all NFAs.e) Schematic images showing quadrupole moment effect on energetic offset induced by differently oriented molecular crystallites with high Q  molecules.