Sensitive near-infrared circularly polarized light detection via non-fullerene acceptor blends

Circularly polarized light (CPL) is widely used for various applications in sensing and imaging1–3. An ongoing challenge is to realize high-quality CPL detection using chiral organic semiconductors, especially in the near-infrared (NIR) region4. Chiral molecules tend to rely on twisted stereogenic moieties; however, conventional approaches to reduce the bandgap of organic semiconductors are based on the use of co-planar backbones that commonly lead to molecular symmetries preventing chirality. Here we report a widely applicable strategy to directly induce chiroptical activity in planar non-fullerene acceptors5–7, which are widely used for high-performance organic photovoltaics and provide a wealth of opportunities to fill the spectral gap of CPL detection in the NIR regime. We demonstrate proof-of-concept circularly polarized organic photodiodes using chiroptically active non-fullerene acceptor blends, which exhibit strong circular dichroism and hence great sensitivity to CPL in the NIR region. Importantly, this strategy is found to be effective in a wide series of state-of-the-art non-fullerene acceptor families including ITIC5, o-IDTBR6 and Y6 analogues7, which substantially broadens the range of materials applicable to NIR CPL detection. Non-fullerene acceptors open the way to sensitive organic photodetectors for detecting circularly polarized near-infrared light.

Circularly polarized light (CPL) is widely used for various applications in sensing and imaging [1][2][3] . An ongoing challenge is to realize high-quality CPL detection using chiral organic semiconductors, especially in the near-infrared (NIR) region 4 . Chiral molecules tend to rely on twisted stereogenic moieties; however, conventional approaches to reduce the bandgap of organic semiconductors are based on the use of co-planar backbones that commonly lead to molecular symmetries preventing chirality. Here we report a widely applicable strategy to directly induce chiroptical activity in planar non-fullerene acceptors [5][6][7] , which are widely used for high-performance organic photovoltaics and provide a wealth of opportunities to fill the spectral gap of CPL detection in the NIR regime. We demonstrate proof-of-concept circularly polarized organic photodiodes using chiroptically active non-fullerene acceptor blends, which exhibit strong circular dichroism and hence great sensitivity to CPL in the NIR region. Importantly, this strategy is found to be effective in a wide series of state-of-the-art non-fullerene acceptor families including ITIC 5 , o-IDTBR 6 and Y6 analogues 7 , which substantially broadens the range of materials applicable to NIR CPL detection. Circular polarization of electromagnetic waves has been used in a wide range of applications including satellite communication 8 , bioimaging 9 and polarization-enhanced sensing technologies 1 . The basic working principle in circularly polarized light (CPL) photodetectors is the ability to discriminate between left-handed (LH) and right-handed (RH) CPL, which can be quantified by the dissymmetry factor (g-factor) 10,11 : where I represents a physical parameter such as absorbance or photocurrent, whereas L/R refers to LH/RH CPL illumination 12 .
To translate CPL signals into electrical signals, the sensing wavelength of the photodetector has to match that of the deployed CPL sources. The successful use of a CPL photodetector, thus, requires a rigorous selection of detection wavelengths (λ det ) with the desired high absorption and photocurrent dissymmetry factors (g abs and g ph , respectively). Bragg reflective metamaterials have been extensively developed for CPL detection beyond >1,000 nm (ref. 13) and towards the mid-infrared regime, mainly used for satellite communications 14 . Similar working principles apply to liquid-crystal-layer-functionalized Letter https://doi.org/10.1038/s41566-023-01230-z band; the CD spectrum of the blend film only shows a weak signal at λ = 240 nm (~20 mdeg; Supplementary Fig. 2) corresponding to R5011 absorption ( Supplementary Fig. 3), which indicates the absence of any induced CD in the ITIC structures. When a mixed spin-coating solvent is used, for instance, in the presence of 10 vol% chlorobenzene (CB) ( Fig. 2a; Supplementary Fig. 4 shows the other solvents), the main absorption band of ITIC becomes broader and slightly redshifts, indicating a larger distribution of electronic states. This is accompanied by the emergence of a strong negative CD signal of approximately -2,000 mdeg (g abs ≈ -0.06 ± 0.01) at λ = 706 nm. The CD spectra show no change subjected to rotation or flip of the thin-film sample (Supplementary Fig. 5), suggesting negligible linear dichroism-linear birefringence effects in the measured CD response 32 . Further optimization of the mixing ratio of the co-solvent ( Supplementary Fig. 6), selection of the other co-solvent ( Supplementary Fig. 5) and blending ratio of the chiral additives (Supplementary Figs. 4 and 7) result in an optimal g abs value of −0.15 (±0.01) ( Supplementary Fig. 7), which is 75-fold larger than that of the very first example of intrinsically chiral organics used for NIR CPL detection 4 . It is worth noting that recent studies have shown that the morphology of many neat or blend NFA films has a strong dependence on the nature of the solvents and additives used for their preparation 33,34 .
The g value for an electronic transition in a chiral compound is defined as 35 where m and μ denote the magnetic and electric transition dipoles, respectively, whereas θ is the angle between them. Since ITIC has a planar backbone conformation and a symmetric structure, the intrinsic molecular rotatory strength R (|m||μ|cosθ) is vanishingly small due to a negligible |m| value and the orthogonality of m and μ (ref. 35). Thus, it can be ruled out that individual ITIC molecules are responsible for the strong chiroptical activity and large rotatory strength observed in the ITIC blends. Instead, to rationalize the strong CD signals appearing in our samples 36 , we need to consider the interchromophoric exciton coupling between adjacent ITIC molecules, as described by equation (3): Here R 1,2 describes the rotatory strength in the case of a dimer consisting of two interacting chromophores 1 and 2 with electric transition dipole μ 1 and μ 2 , and r 12 represents the intermolecular distance vector. As a result of exciton coupling, the magnitude of the dissymmetry factor is no longer restricted by the relative magnitudes of the μ and m dipoles of the individual chromophores as in non-excitonically coupled systems 36,37 . The excited states in such coupled chromophores split into non-degenerate levels (that is, Davydov splitting 37 ), giving rise to LH and RH CPL absorption 37,38 . We note that the observation of split excited states is consistent with other experimental results reported here, including the broadening of the low-energy absorption band and the appearance of bisignate couplets in the CD spectra (Fig. 1a).
To investigate which moieties drive the chirality induction in ITIC, we separately blend R5011 with the core-unit molecule (that is, IDTT) and the end-group molecule (that is, IC). As shown in Supplementary Figs. 8 and 9, IDTT blend films exhibit no CD signal, whereas IC:R5011 blend films show a new CD peak at ~340 nm corresponding to IC π-π* transitions ( Supplementary Fig. 10). This clear difference indicates that it is the IC units that play a dominant role in the formation of chiral configurations in the solid-state ensemble of ITIC molecules. Due to the large steric hindrance of the bulky side chains on the IDTT cores, the π-π stacks of ITIC molecules are predominantly driven by IC-IC interactions owing to the closer spatial and orientational contacts devices, where CPL detection relies on selective reflection instead of intrinsic chiroptical response (that is, circular dichroism (CD)). For cost-effective and miniaturized devices for CPL detection, an increasing number of intrinsically chiral organic and hybrid materials have been investigated. For studies reported in this field over the past decade, CPL detection is mainly accessible in the range of 300-650 nm ( Fig. 1) 12,[15][16][17][18][19][20] .
The key challenge in CPL detection lies in the near-infrared (NIR) region ( Fig. 1 4,21 , which has potential applications in the field of night vision 22 and autonomous vehicles 23 . Although attempts have been made to approach NIR absorption via direct synthesis strategies 24,25 , it remains challenging to achieve strong and intrinsic chiroptical activity at the same time (for example, g abs > 0.1). To reduce the bandgap of organic compounds and achieve the favoured π-π stacking for efficient charge transport, the material design usually relies on achieving co-planar backbones with large fused aromatic moieties and efficient long-axis charge-transfer-type electronic transitions. Unfortunately, these design strategies are usually not compatible with the emergence of strong chiral activity. From the pioneering work on CPL detectors using chiral side-chain polymers 26 to the most recent chiral perovskites [18][19][20] , materials designs for circularly polarized organic photodiodes (CP-OPDs) were based on developing more twisted molecular structures, whose lowest-energy absorption band peaks can hardly reach 650 nm and beyond 12,[15][16][17][18][19][20]26,27 .
An attractive approach to address this dilemma is to make use of the so-called non-fullerene acceptors (NFAs), which are being extensively developed for high-efficiency organic solar cells 5 . Indeed, these NFAs not only show strong absorption peaks up to 1,000 nm (ref. 28) but also display strong intermolecular interactions, which makes it possible to develop chiroptical activities by exploiting exciton coupling among adjacent chromophores, as reported for polyfluorenes 29 .
An archetypal NFA, ITIC ( Fig. 2a) 5 , was selected to demonstrate the emergence of chiroptical activities, on addition of a chiral compound (for example, R5011 (ref. 30); Fig. 2b) that induces a strong excitonic CD with the assistance of the mixed spin-coating solvent. ITIC features a symmetric indacenodithieno[3,2-b]-thiophene (IDTT) core and two 1,1-dicyanomethylene-3-indanone (IC) end groups. Its main absorption band at 700 nm consists of a combination of a π-π* IDTT transition and a charge transfer transition from the IDTT core to the IC ends ( Supplementary Fig. 1) 31 . When deposited from chloroform (CF), both neat (ITIC) and blend (ITIC:R5011) films exhibit silent CD signals ( Fig. 2a and Supplementary Fig. 1 Fig. 8) 39 . Thus, when the π-π-stacking motifs across IC moieties are perturbed by the chiral additive, the entire NFA ensemble is then driven into a chiral configuration, similar to the 'sergeants and soldiers' effect observed in chiral polymer assemblies 40 .
Since exciton coupling in ITIC (occurring through IC intermolecular interactions) is evident from the spectroscopic analysis, it is informative to visualize such chromophoric interactions at the level of polycrystalline thin films. Using two-dimensional (2D) grazing-incidence wide-angle X-ray scattering (GIWAXS), we probed the morphological information associated with the induced CD, to capture the three-dimensional packing structures in the chiral ITIC blends. As shown in Fig. 2c and Supplementary Fig. 11 (2D patterns), all the neat and blend films exhibit almost identical ITIC π-π-stacked (010) diffraction signals in the out-of-plane direction. Compared with the films deposited from CF, when spin coated from the mixed solvent, the neat ITIC films exhibit a slightly more ordered in-plane structure with a diffraction peak at q = 4.83 nm −1 (Fig. 2d) 41 . This is consistent with the reported GIWAXS patterns, where a mesh-like structure is formed 33 , featuring continuous aromatic structures separated by aliphatic domains. On the incorporation of chiral additives, a more ordered in-plane structure is formed with increased molecular separation, indicating a reorientation of the ITIC assemblies.
The absence of disrupted π-π stacks confirms our previous spectroscopic observation, that is, the coupled CD originates in the exciton coupling of rotated ITIC chromophores instead of twisted backbones. Next, we discuss-with the help of time-dependent density functional theory (DFT) calculations-how the three-dimensional assembly of ITIC chromophores amplifies the chiroptical response and results in increased separations within the aromatic-aliphatic-aromatic (lamellar) packing structure.
We start with symmetric ITIC monomers, taken from an experimental crystal structure 42 or optimized at the DFT level. As expected, these monomers exhibit zero rotatory strength (R) and g-factor (Supplementary Fig. 12 and Supplementary Table 2). Twisting the molecules to generate a helical structure has only a minor effect on the calculated g-factors (Supplementary Fig. 13 and Supplementary Table 3); we recall that distorting the planarity of the molecules would impact the structure of the ITIC π stacks driven by the packing of the IC end groups 39 , which would contradict the GIWAXS results. Thus, intramolecular twisting can be reasonably ruled out as the dominant contribution to the induced chirality.
When considering ITIC dimers, although optimized collinear dimers with no in-plane rotations between individual monomers exhibit zero R as expected, the appearance of in-plane rotation angles    (Fig. 3b, Supplementary Fig. 16 and Supplementary Table 6), we hereafter focus on [P]-enantiomers in larger ITIC assemblies. We note that the molecules successively stacked with the same direction of rotation form enantiomerically pure trimers (Fig. 3c), whereas zigzag-shaped shape [P,M]-trimers exhibit negligible g-factors ( Fig. 3d and Table 1). Increasing the number of ITIC molecules in enantiomerically pure stacks results in a g-factor amplification from 8 × 10 −4 (dimers; Fig. 3b) to 2.6 × 10 −3 (tetramer; Fig. 3e and Table 1). Both tetramers assembled through continuous in-plane rotation and out-of-plane stack of two [P]-dimers exhibit improved positive g-factors (Fig. 3e,f), indicating that the promotion of one enantiomeric π stack perturbed by chiral additives is the prerequisite for the enhanced g-factors 43 (Supplementary Figs. 17 and 18 and Supplementary Tables 7 and 8).
In chiral ITIC blends, ITIC molecules are assembled either in the chiral configuration (for example, [M,M]-trimers) or achiral configuration (for example, [P,M]-trimers; Supplementary Fig. 19). To explain the increased in-plane displacement (Fig. 2d) Fig. 21) and associated additional hindrance, the chiral-fragment-embedded ITIC chains are expected to exhibit a larger in-plain displacement. We remark that in a spin-coated film, the packing structures of crystalline and non-crystalline sites of ITICs are more complicated and harder to predict, thus exhausting all the possible chiral assemblies, and this is beyond the scope of this study. However, we believe our proposed structure based on planar ITIC chains would inspire further theoretical studies on the solid-state packing of a strongly chiroptically active π-conjugated system.
The strong CD of the ITIC:R/S5011 blends motivated us to develop an active layer for CP-OPDs, where a donor polymer PBDBT 44 is added to form a bulk heterojunction. The optimized ternary blends based on R5011 and S5011 exhibit mirror CD images and a slightly reduced signal of ~3,000 mdeg (|g abs | ≈ 0.09; Supplementary Figs. 22 and 23 show the detailed optimization), compared with their binary counterparts (CD ≈ 4,000 mdeg and |g abs | ≈ 0.15 at λ = 704 nm; Fig. 4a), which is possibly due to the tail absorption of PBDBT at ~700 nm and perturbation of the interactions between the ITIC molecules and the chiral additives caused by PBDBT.
Proof-of-concept CP-OPDs were fabricated using a device structure of indium tin oxide/PEDOT:PSS/R5011:ITIC:PBDBT/PDINO/Al (see the 'OPD fabrication and characterization' section for more details), where PDINO was selected as the electron transport layer (Supplementary Fig. 24) 45 . The dark current of our CP-OPD is 2.33 × 10 −11 A (at 0 V; Fig. 4b), which is comparable with the CP-OPDs reported so far 16,26 . On illumination at λ = 700 nm, R5011 blend devices show a linear response from 0.84 nW to 5 µW, which corresponds to a linear dynamic range (LDR) of ~76 dB (at 0 V; Fig. 4c). To evaluate the lowest detectable limit of our CP-OPD, we performed the dark-current noise measurement and calculated the noise-equivalent power. At 0 V bias, the R5011 device exhibits a low noise current level, ranging from ~2.02 to ~6.48 fA Hz −1/2 , as measured from 1 to 10 Hz. Based on the external quantum efficiency spectra ( Supplementary Fig. 24), spectra responsivity ( Supplementary Fig. 25) and measured noise (Fig. 4e), we evaluate the noise-equivalent power to be ~2.0 × 10 -14 W Hz −1/2 . We achieved a 700 nm D* of 1.04 (±0.03) × 10 13 Jones and peak D* of 1.07 (±0.05) × 10 13 Jones at 730 nm, which is among the highest reported value for CP-OPDs (Supplementary Table 9).
Under a 700 nm square-wave-modulated illumination, the R5011 device demonstrates an average rise/fall time of ~3.48 µs   Fig. 26). The decay of the device response driven by sinusoidal signals with different frequencies was then investigated, and the bandwidth calculated is comparable, reaching as high as ~120 kHz (Fig. 4d). This result is consistent with the fast-Fourier-transformed transient photocurrent curve (Supplementary Fig. 27), demonstrating the fastest response time among the reported CP-OPDs (Supplementary Table 9 lists a comparison). In terms of CPL sensitivity, the R5011 blend device exhibits photocurrents I L and I R of 0.55 and 0.60 mA (at 0 V; Fig. 4e) under the illumination of LH and RH CPL, respectively. The extracted g ph value is slightly amplified up to −0.113, when increasing the bias up to −2.0 V. The average g ph value at 0 V for R5011 and S5011 devices is −0.094 (±0.005) and +0.092 (±0.007), respectively, consistent with real-time measurements, which is almost one order of magnitude larger than the very first example of NIR CP photodetector reported using an intrinsically chiral active layer (Supplementary Figs. 28 and 29) 4 . These results indicate that chiral additives can effectively functionalize ITIC-based active layers for OPDs showing great sensitivity towards CPL signals at 700 nm (Supplementary Table 9 lists the CPL detection at different wavelengths), without imposing any detrimental effect on device performance as an unpolarized OPD.
We are also able to demonstrate that our strategy is generally applicable to other types of state-of-the-art NFAs, such as o-IDTBR 6 , IEICO 46 Fig. 30 shows the spectra and Supplementary Figs. 31 and 32 show the selective device performance). The CD bands of these systems can be tuned from 600 to 1,000 nm via modifications of the chemical structures. As such, our approach provides an effective way to fill the gap in the detection wavelength of existing chiral systems (Fig. 1).  In summary, we have described an effective strategy to induce strong CD in thin films based on NFAs by making use of strong excitonic coupling between adjacent NFA molecules. Owing to the strong π-π interactions among chromophores in NFA stacks, notable bisignate CD bands can appear, even though the individual NFA molecules are intrinsically achiral. CP-OPDs whose active layers incorporate these chiral assemblies display a large chiroptical response in the NIR region (g abs > 0.1), a region that is traditionally inaccessible to other organic semiconductors, which enables efficient CPL sensing in this range. We believe that such strong CP selectivity in the NIR regime, coupled with state-of-the-art response time and specific detectivity, has great potential to transform many real-world sensing and data transmission applications. In addition, our combined experimental and theoretical investigation provides a powerful tool to describe the molecular assemblies critical to chiroptical activities, from the chromophore interactions in dimers to the macroscopic scale.

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41566-023-01230-z.

Materials
Chiral additives R/S5011 were purchased from Nanjing Sanjiang Advanced Materials R&D. All the NFAs and PBDBT were purchased from Solarmer Materials. All the materials were used without further purification.

Solution preparation and thin-film deposition
ITIC and R5011 were dissolved in neat CF and CF/CB mixed solvent to a concentration of 20 mg ml -1 and blend to form a 10 wt% R5011 solution. Thin films were dynamically spin coated at a speed of 2,000 r.p.m. to give a ~120 nm film and deposited on clean fused silica substrates. The cleaning process for all the substrates involved rinsing in an ultrasonic bath with acetone, isopropyl alcohol, Hellmanex III (Sigma-Aldrich) and deionized water for 15 min. These were followed by an ultraviolet-ozone treatment for 15 min before spin coating. All the samples are measured without annealing and other post-deposition treatments. The thickness of the thin films was controlled by the spin speed and monitored using a DektakXT surface profiler.

Photophysical and morphological characterizations
The CD spectra and absorption spectra were simultaneously collected using a Chirascan (Applied Photophysics) spectrophotometer. The dissymmetry factor g abs was calculated following the equation g = (A L - A R )/A, |g| ≤ 2. The 2D GIWAXS measurements were carried out at the 14BL beamline of the Shanghai Synchrotron Radiation Facility (λ = 1.24 Å). The sample stage was equipped with a seven-axis motorized stage for the fine alignment of the sample, and the incidence angle of the X-ray beam was set at 0.2 for the neat and blend films. The GIWAXS patterns were recorded with a 2D charge-coupled device detector and an X-ray irradiation time within 100 s, dependent on the saturation level of the detector.

OPD fabrication and characterization
Prepatterned indium tin oxide glass substrates were sequentially cleaned in an ultrasonic bath with water solution of the surfactant, acetone and isopropanol (15 min each) and dried under a N 2 flow before ultraviolet-ozone treatment. PEDOT:PSS (CLEVIOS P VP AI 4083) was spin coated to the indium tin oxide substrates at 2,000 r.p.m. for 60 s. Then, they were immediately transferred to a N 2 glovebox. Then, R5011, ITIC and PBDBT were blend in a weight ratio of 4:6:6 in a CB/CF mixed solvent (CB/CF = 10:90 vol%) with a total concentration of 26 mg ml -1 and deposited onto the PEDOT:PSS layer to form a ~190 nm film. The film was then annealed at 100 °C for 5 min for better charge transport. PDINO (Ossila) was spin coated onto the active layer at 3,000 r.p.m. from a 0.5 mg ml -1 methanol solution. Afterwards, Al was thermally evaporated onto the organic layer under a high vacuum (~1 × 10 −6 mbar).
The characterizations of the CP-OPD were carried out under ambient conditions at room temperature. The devices were encapsulated using a drop of ultraviolet adhesive (~5 µl; Norland Products, NOA73) on the metal electrode side of the device and then covered with a glass slide. It was then solidified by illumination with an ultraviolet lamp through the cover glass side for about 15 s. Current-voltage (J-V) measurements (device pixel area, 0.047 cm 2 ) were performed using a Keithley 2400 instrument. For measurements under illumination, a mounted light-emitting diode (LED) (Thorlabs, M700L4; 700 nm) was used with a Thorlabs DC2200 driver, which was collimated with an adaptor lens (Thorlabs, SM2F32-A). For measurements of g ph , the collimated LED was circularly polarized using an ultrabroadband wire grid polarizer (Thorlabs, WP25M-UB; extinction ratio of >1,000:1 at 700 nm) and a mounted quarter-wave Fresnel rhomb retarder (FR600QM; retardance variation of <2% for 600-1,550 nm). The light intensity was monitored using an ultraviolet-enhanced Si photodetector (Newport; 200-1,100 nm).
The external quantum efficiency and spectra responsivity of the device were measured using a solar-cell spectral response measurement system (QE-R3011, Enli Technology), which was calibrated with a standard Si crystalline solar cell. The dark-current noise spectra of the devices were measured using a lock-in amplifier (SR830, Stanford Research Systems) coupled with a low-noise preamplifier (SR570, Stanford Research Systems), and the device was fixed in a grounded metal box. For Y18 devices, a mounted LED (Thorlabs, M880L3; 880 nm) was used.
The noise-equivalent power NEP and D* (in Jones) were calculated using the following equations: where i n is the dark-current noise (in A Hz −1/2 ), R (in A W -1 ) is the spectra responsivity and A is the area of the device (in cm 2 ). The average D* is calculated using 16 identical devices. The weak light response and LDR of the photodetector were carried out by using up to two neutral density filters and a Thorlabs DC2200 LED driver to vary the light power. The output power was calibrated with the Si photodiode. The current signals of the devices were recorded by the Keithley 2400 device. The LDR was calculated using the equation LDR = 20log(P high /P low ), where the subscripts high and low refer to the high and low power limits within the linear-response region of the OPD.
The damping measurements of higher frequency are extracted from the transient photocurrent measurements. The transient photocurrent of the photodetector was measured using an oscilloscope with an input impedance of 50 Ω when the device was excited by a pulse laser (wavelength, 337 nm; pulse width, ~4 ns) from a nitrogen laser (NL100, Stanford Research Systems) after attenuation with a neutral density filter. The rise/fall time and damping (low frequency) measurements were carried out using the LED driver with a 2 kHz square-wave pulse and sinusoidal light with various desired frequencies (<1 MHz).

DFT calculation and analysis
The ground-state geometries of the monomer and dimer were optimized at the B3LYP/6-311G** level of theory with GD3BJ (D3 with Becke-Johnson damping) dispersion corrections. Time-dependent DFT calculations of the ITIC monomer were carried out at the same level of theory to determine the chiroptical properties of the excited states. In case of the dimer, the hexyl groups attached on the benzene rings in the ground-state-optimized structure were replaced with hydrogen atoms in the time-dependent DFT calculations at the B3LYP/6-31G** level to maintain reasonable computation times. To compare the magnitudes of the electric and magnetic transition dipole moments, the SI units were converted to CGS-Gaussian units, as shown in a previous work 48 . The g-factors were obtained from equation (2).

Data availability
The relevant data are available in the main text and the Supplementary Information. Source data are provided with this paper.