B ← N Coordination: From Chemistry to Organic Photovoltaic Materials

devices based on B ← N-embedded materials should be also addressed. Although recent studies suggested high thermal stability of devices using B ← N-embedded materials as the active layers, [214,215] it is worri-some whether these B ← N-embedded materials are durable in moisture air for a long time. In-depth studies on material and device stability involving the B ← N-embedded molecules should be undertaken.Theperformances of B ← N-embedded materials are inferior to the state-of-the-art OPV materials. For example, J SC and FF of B ← N-type acceptors are relative low. The highest J SC of B ← N-type acceptors has been only 16.9 mA/cm 2 , [192] which lags behind the champion values of over 20 mA/cm 2 afforded by other types of outstanding acceptors. [221 – 223] Also, enhanced FF as high as 70% has been revealed only in one work. [189] Accordingly, how to synergistically enhance J SC and FF without sacri ﬁ cing the V OC of these B ← N-embedded OPV materials is an urgent and challenging job.


Introduction
The fascinating chemistry of boron has been drawing attention of chemists for a long time and has evolved from the originally chemical applications, e.g., used as reagents or catalysts in organic chemistry, to multidisciplinary uses of functional materials, especially organic optoelectronic materials. [1][2][3][4][5][6][7][8][9][10] The boron element (B), neighboring the carbon element (C), has an atomic number of 5. The trivalent B adopts the sp 2 hybrid mode. It has a trigonal-planar geometry with an empty p z orbital, rendering it strong Lewis acidity and high propensity to coordinate with Lewis bases. The most popular B Lewis acids are boron trihalides (BX 3 , X ¼ F, Cl, or Br) and B(C 6 F 5 ) 3 , also named BCF. [11] In contrast, conjugated molecules containing aromatic nitrogens (N) are quintessential Lewis bases and highly reactive toward trivalent boron due to their strong Lewis basicity induced by the lone pairs of electrons. This classic Lewis acid-base (B←N) interaction, essentially a chemical issue, is effective to adjust the optoelectronic properties of aromatic nitrogen-based molecules ( Figure 1a). Otherwise, this intermolecular B←N coordination is fragile and reversible, which can be easily damaged by nucleophiles, e.g., H 2 O or pyridine. As such, further incorporating B←N bonds into the conjugated systems via electrophilic borylation can accomplish fused conjugated units with rich optoelectronic properties ( Figure 1b). These B←N-bridged units are potentially useful for constructing more complex materials by covalently linking with various π-bridges, giving rise to a kind of novel organic electronic material, namely, B←N-embedded materials (Figure 1c). The B←N-embedded materials usually have tunable optical properties and rich redox properties due to the strong electron-deficient property of B centers, producing intramolecular charge transfer (ICT) effects. Moreover, the B←N bond not only has high polarity but also locks the molecular conformation to enhance the coplanarity, both of which are beneficial to the ordered molecular packing and charge transport. In general, the B←N-embedded materials have a broad window of optical, redox, and self-assembly properties that can be tuned to fulfill various electronic applications, especially organic photovoltaic (OPV) application.
The bulk heterojunction (BHJ) organic solar cells (OSCs), schematically shown in Figure 1d, adopting organic semiconductors as the active layers were developed rapidly in the past two decades. [12][13][14][15][16] Conventionally, the active layer is a blend of one electron donor (D, p-type) and one electron acceptor (A, n-type) with nanophase-separated morphology. As a result of extensive dedication to the material update and device optimization, the efficiencies of OSCs have been promoted to over 17%. [17][18][19][20] Due to the versatile structures of organic molecules, inventing novel and efficient photovoltaic materials to provide material foundation for the bright future of OSCs is a continuously pursued purpose of chemists and material scientists. To facilitate the photoelectricity conversion, the p-type and n-type organic semiconductors should be equipped with the following features: 1) strong light-harvesting ability, e.g., wide absorption coverage from visible to infrared regions and high absorption coefficients; 2) suitable frontier orbital energy levels, i.e., highest occupied molecular orbitals (HOMOs) and lowest unoccupied molecular orbitals (LUMOs) to drive the charge dissociation and reduce energy losses; 3) good compromise between aggregation and solubility to prepare well-defined electron donor/acceptor blends with suitable domain sizes and ordered molecular aggregates; and 4) excellent hole and electron mobilities to ensure efficient charge transport and collection.
These requirements fit well with the properties of B←Nembedded molecules, as discussed in the aforementioned section. Actually, in 2009, Bazan and coworkers initially studied the effects of intermolecular B←N coordination on the absorption spectra, bandgaps, and energy levels of oligomers containing aromatic N atoms, which are enlightening to the design of OPV materials. [21] In the same year, Roncali and coworkers reported boron-dipyrromethene (BODIPY, a classic dye molecule contain B←N bonds)-based small molecules, as the electron donors for OSCs. [22,23] Liu and coworkers used a B←N-bridged unit to build conjugated copolymers and applied these B←N-embedded polymers as the electron acceptors to fabricate OSCs. [24] These targeted studies on the B←N-based molecules provide a new toolbox to construct OPV materials. To date, although numerous B←N-bridged units have been reported, application of B←Nembedded materials to OSCs are in infancy. Currently, the best efficiencies of B←N-embedded materials used for OPV devices have been approximately 10%, [25,26] lagging behind efficiencies of state-of-the-art OPV devices, which have been increased beyond 17%. [17][18][19] However, the excellent properties of B←Nembedded materials imply that potential for the OPV application has not been exhausted and great room is left for their development. Accordingly, at this critical moment, systematical summaries on these kinds of materials are urgent to facilitate in-depth studies and efficiency improvement. Undoubtedly, the development of B←N-embedded OPV materials is built on the pioneer work on intermolecular B←N chemistry and invention of basic building blocks, i.e., B←N-bridged units. Accordingly, herein, we intend to review these types of OPV materials starting from the chemistry of intermolecular B←N coordination. Certain chemical issues of intermolecular B←N coordination and the resultant effects on the optoelectronic properties will be outlined. Then, a series of fused and symmetric B←N-bridged molecules that are potentially useful for constructing OPV materials will be reviewed. Finally, the B←N-embedded OPV materials will be summarized. This Review is trying to highlight ways in which high-performance OPV materials can be designed using B←N chemistry and B←N-bridged units.

Intermolecular B←N Coordination
Essentially, the intermolecular B←N coordination is a typically chemical issue. The acidity of B Lewis acids readily determines strength and equilibrium of the coordination. Next, coordination would cause alteration of the chemical environment of the relevant atoms, which can be reflected in the nuclear magnetic resonance (NMR) signals. Then, this chemical interaction would impose significant effects on material properties, e.g., absorption, emission, energy levels, bandgaps, and device performance ( Figure 2).

Lewis Acidity of Trivalent B Lewis acids
The representative Lewis acids are boron trihalides, BX 3 (X ¼ F, Cl, or Br). Due to their strong Lewis acidity, they are reactive   toward water and other nucleophiles and hard to be handled in the moisture air. Lewis acidity is correlated with their reactivity. The reactivity of the organoboron can be effectively reduced by introducing three aromatic phenyl (Ph) groups onto the B center to form π-bonding (BPh 3 ). The bulky mesityl (Mes) groups can further quench the Lewis acidity of the B center. Introducing Mes groups on the B centers can protected the p z orbitals from nucleophiles, leading to sufficiently stabilized organoborons, e.g., B(Mes) 3 . [27] Another ubiquitous B Lewis acid is tris (pentafluorophenyl) borane (B(C 6 F 5 ) 3 , BCF), [11,28,29] which is commonly used as the cocatalyst of olefin polymerization. [30][31][32] Due to perfluorinated substitution, BCF also has strong Lewis acidity and reactivity toward water. The quantitative tests of Lewis acidity have been disclosed in experimental and theoretical methods by Childs, Laszlo, Beckett, et al., leading to a general sequence, as shown in Figure 3. [11,[33][34][35] 2.1.2. NMR Signals of 11 B, 1

H, and 19 F in the Intermolecular B←N Systems
B←N coordination is distinguishable in NMR spectra of 1 H, 11 B, and 19 F. 11 B NMR spectra directly report the chemical environment of B after coordinating with Lewis bases. For the BCF, upon reacting with water to form BCF•H 2 O, δ ( 11 B) moved from 59 ppm (br) to 6.65 ppm. [36] Similarly, when BCF coordinated with aromatic N atoms in polymers 6 and 7, the chemical shifts also moved to the higher field (Figure 4a), which was a straightforward evidence of the B←N coordination. [37] Structures of 6 and 7 are shown in Figure 5. In addition, 1 H NMR signals around the N atoms are also altered after B←N coordination. As shown in Figure 4b, chemical shifts of H 2 and H 3 on pyridine moved to the lower field after the BCF was bound to N, which can be ascribed to the decreased electron density on pyridine upon coordinating with strong electron-deficient BCF. However, NMR signal of H1 moved to the higher field upon the B←N coordination. This can be interpreted by the conformation distortion due to the coordination of bulky BCF, transforming the chemical environment of H1 from the deshielding to shielding region of C═O groups. [38] It is noteworthy that the intermolecular B←N coordination is a reversible reaction. The equilibrium between free molecules and B←N adducts can be tracked by the 1 H NMR spectra at different temperatures. As shown in Figure 4c, the 1 H NMR spectra of 4-BCF in CD 2 Cl 2 showed broad proton resonances at 300 K, indicating the existence of an exchange mechanism between free 4 and 4-BCF adducts. [39] Upon cooling to 280 K, the resonances sharpened and the signals of 4 and 4-BCF coexisted. Further cooling to 230 K moved the equilibrium to the 4-BCF adduct formation and sharp resonances of 4-BCF were observed. These results indicated that the low temperature was favorable for the formation of 4-BCF adducts. This can be further verified by the 19 F NMR spectra of BCF (Figure 4d). The 19 F NMR spectra of free BCF showed three resonances at À128.2, À143.8, and À160.9 ppm, corresponding to the ortho-, meta-, and pararesonances of BCF, respectively. Upon coordinating with 4 at 300 K, some new resonances emerged, indicating the exchange of free BCF and 4-BCF adducts. Upon cooling to 230 K, 15 new resonances appear, indicating the formation of 4-BCF, leading to the equivalent environment of 15 fluorines. [39]

Adjustment of Optoelectronic Properties via B←N Coordination
The intermolecular B←N coordination has been studied for various purposes, e.g., determining the Lewis basicity of N-based solvents, [40] conformational behavior, [41] and supramolecular self-assembly. [42] However, manipulating optoelectronic properties through intermolecular B←N complexation was not revealed until 2009, when Bazan and coworkers provided a bandgap control method for benzothiadiazole (BTZ)-based oligomers via coordinating with BCF. [21] The absorption spectrum of oligomer 1 ( Figure 5) remarkably shifted to lower energy upon coordinating with different equivalents of BCF (Figure 6a), indicating narrowed optical bandgaps. Later, they extended the research to analogous oligomers and polymers containing cyclopentadithiophene (CDT) and benzothiadiazole (BTZ) or pyridalthiadiazole (PT) (2-5 in Figure 5). [39] In addition, we studied the energy levels of N-based polymers (6 and 7) and their BCF adducts by cyclic voltammetry. [37] It is found that upon coordinating with BCF, the HOMOs slightly changed whereas the LUMOs significantly lowered, leading to narrowed bandgaps ( Figure 6b).
These studies demonstrated an effective strategy to extend the absorption, narrow the bandgaps, and decrease the energy levels, especially LUMOs of conjugated molecules containing aromatic N atoms via coordinating with trivalent B Lewis acids. This strategy is generally valid for conjugated molecules containing aromatic N atoms, e.g., pyridine-capped diketopyrrolopyrrole (DPP) (8), [38] azaisoindigo (9), [43] diazopentacene (10), [44] and pyrazinoquinoxaline (11). [45] It is noteworthy that Bousquet and coworkers disclosed a diazapentalene-dithienosilole copolymer (PDAT-DTS, 12), whose absorption bands shifted bathochromically from the first-infrared window (peaked at 840 nm) to second-infrared window (peaked at 1070 nm) upon coordinating with BF 3 (Figure 6c). [46] This is the lowest energy of absorbance that can be reached through B←N coordination, which is instructive for designing ultranarrow-bandgap molecules.
Theoretical simulations were conducted to gain insight into the reasons behind the red-shifted absorption, narrowed bandgaps, and lowered LUMOs upon B←N complexation. It was found by Bazan et al. that the electron-deficient nature of the N-based molecules would be enhanced upon coordinating with B Lewis acids due to the fractional charge transfer from Lewis basic molecules to the Lewis acids. Thus, the enhanced electron affinity of N-based molecules is responsible for the significantly lowered LUMOs, narrowed bandgaps, and lower energy of absorbance. [47,48] Calculations using ab initio and density functional theory (DFT) methods carried out by Karamanis et al. suggested that the intermolecular B←N coordination would trigger extrastrong one-and two-photon quantum transitions followed by intense transfer of charge, which will then significantly alter the absorption profiles of the molecule systems. [49] Along with absorption, the independence of fluorescence properties on B←N coordination was also studied. The absorption and fluorescence spectra of 13 were synergistically red shifted after it was bound to BCF. Moreover, the fluorescence quantum yields of 13 were enhanced in both solution (Φ s ¼ 0.86 vs. 0.99) and solid state (Φ f ¼ 0.21 vs. 0.50) upon B←N coordination. [50] B←N complexation was also observed between the alkynyl aryl-conjugated imine (14) and various B Lewis acids, e.g., BX 3 (X ¼ F, Cl, and Br), BCF, and BPh 3 to shifted the absorption and fluorescence to lower energy. [51] Besides solid fluorescence, the crystallization fluorescence enhancement was also surveyed in the crystal of 15-BCF complex. Crystallization-induced emission enhancement in 15-BCF was ascribed to the intermolecular C-H•••F interactions in the crystals, which may restrict the molecular motions and deactivate the nonradiative decay pathway. [52] As most of the intermolecular B←N reactions were conducted in solutions, Hayashi et al. conducted the coordination in solid films by exposing the polymer (16 and 17) films to HCl or BF 3 vapor. The coordinations were also verified by the deepened film color in both visible and ultraviolet light. [53,54] However, they reported an exceptional situation of N-based polymers, whose absorption spectra blue shifted upon reacting with BF 3 . For comparison, two polymers based on pyridine (18) and N-alkyldiarylamine (19) were synthesized separately and both films were exposed to BF 3 vapor. Interestingly, the film absorption of 18 was red shifted, whereas the film absorption of 19 was blue shifted (Figure 6d). It is explained that the N in 19 transformed from sp 2 to sp 3 hybrid upon coordinating with BF 3 , resulting in a reduced mean conjugated length and thus a blueshifted absorption band. [55] Based on the B←N coordination, the B Lewis acids are also used as dopants for organic semiconductors containing aromatic N atoms. In contrast to the conventional doping modes, e.g., oxidative doping and Bronsted acid doping, the mechanism of Lewis acid doping is not very clear. [56][57][58][59] However, the B Lewis acid doping has been sporadically reported to increase carrier densities, fill traps, and enhance p-type transport of nitrogen-based conjugated molecules. Poverenov et al. disclosed the nonoxidative doping of alternating copolymers based on 3,4-ethylenedioxythiophene (EDOT) and BTZ (20) using BF 3 . [60] Upon coordinating with BF 3 , the BTZ-based polymers switched from neutral to doping state, causing notably increased conductivity. This doping state can be reversibly transformed to neutral by adding external bases, leading to the recovery of conductivity. Nguyen and coworkers found that the mobile carrier concentration in polymer 21 can be remarkably increased and the hole mobilities were improved by two orders of magnitude through Lewis acid doping using suitable equivalents of BCF. [61] Another study discovered that upon doping with suitable equivalents of BCF, the hole mobilities of indenopyrazine copolymers (22) increased due to introducing extra holes, which effectively filled the traps in the films. However, excessive equivalents of BCF would cause negative effects on the mobilities because of defect formation and structural disorder. [62] Improved understanding of Lewis acid doping drives the further applications of B Lewis acids to tune the performance of various optoelectronic devices, e.g., organic light-emitting diodes (OLEDs), [63,64] organic field effect transistors (OFETs), [62,65,66] OPVs, [67,68] and perovskite solar cells (PSCs). [69,70] For example, upon coordinating with BCF, OLEDs based on polymer 16-BCF showed red-shifted emission and increased tune-on voltage due to the lowered HOMO of 16-BCF adduct, which elevated the hole injection barrier. [63] The intermolecular B←N coordination has been proven powerful to adjust the optical properties and energy levels of organic semiconductors over a broad window. However, it is required to add equivalent or excess B Lewis acids, which are highly moisture sensitive, making this operation hard to handle in air. Moreover, the B←N adducts are hydrolytically sensitive, limiting their applications to various devices. For the doping effects of B Lewis acids toward N-based semiconductors, more studies are required to gain an in-depth insight into the doping mechanism. Alternatively, to obtain stable semiconductor materials, especially OPV materials, incorporating the B←N bonds into conjugated systems is a sensible strategy to design B←N-bridged units with tunable properties.

B←N-Bridged Conjugated Units
Fusing the B←N bonds symmetrically into the conjugated systems through electrophilic borylation creates B←N-bridged units Figure 6. a) Absorption adjustment of oligomer 1 through coordinating with BCF; b) energy-level control of polymers 6 and 7 via coordinating with BCF; c) absorption shift of polymer 12 (PDAP-DTS) from first-infrared to second-infrared window upon coordinating with BF 3 ; and d) absorption (solid lines) and fluorescence (dash lines) shifts of 18 (PPF) and 19 (PAF) films before (black and gray) and after (red and blue) exposing to BF 3 vapor. Reproduced with permission. [21,37,46,55] Copyright 2018, Centre National de la Recherche Scientifique (CNRS) and The Royal Society of Chemistry; Copyright 2009 and 2019, American Chemical Society.
www.advancedsciencenews.com www.advenergysustres.com with ladder-type and coplanar backbones. Several synthetic methodologies have been developed for electrophilic borylation, [71][72][73][74][75][76] which tends to occur on aromatic cycles, hydroxyl, or amine groups, producing C-B←N, O-B←N, or N-B←N-bridged units, respectively ( Figure 7). The substituents (X) on the B centers can be halos, Ph, Mes, or C 6 F 5 . Similar to their precursors of intermolecular B←N adducts, these merging units also possess rich optical and redox properties. More importantly, in contrast to the unstable intermolecular B←N adducts, the chemical stability greatly improves for these units. In addition, the coplanar and symmetric backbones are beneficial to the self-assembly in solid state. These features are desirable for organic electronic applications. Unfortunately, a large number of B←N-bridged units disclosed in literature were subjected to physiochemical characterizations without further attempt toward organic electronic applications. Herein, we summarize some representative B←N-bridged units in terms of their absorption, fluorescence, energy levels, and crystal structures (Table S1, Supporting Information), aiming to excavate the potential for electronic, especially for OPV application.

Conjugated Molecules Containing C-B←N Bonds
Intermolecular B←N coordination followed by electrophilic C-H borylation on adjacent aromatic cycles leads to C-B←N-bridged ladder-type molecules. Figure 8 shows representative C-B←N-bridged molecules with symmetric and ladder-type backbones. Yamaguchi and coworkers presented an unprecedented C-B←N-bridged molecule (23). [75] Electrostatic potential surface (EPS) maps indicated obviously an enhanced electron affinity of 23 in contrast to the B-free precursor ( Figure 9a). The single-crystal structure of 23 showed offset face-to-face π-stacking array with a π-π distance of 3.60 Å (Figure 9b). The electron mobility of 23 determined by time-of-flight (TOF) measurements reached 1.5 Â 10 À4 cm 2 /Vs and no p-type charge transport was observed. This seminal work opens the window of electronic materials based on B←N-bridged units.
Recently, a series of A-π-A molecules using a C-B←N-bridged unit as the π-linker were reported (24a-c). [77] Depressed LUMOs (%À4.31-À4.03 eV) were observed for these molecules, indicative of strong electron affinity. Organic thin-film transistors (OTFTs) were prepared using these molecules and typical n-type semiconductor properties were found with electron mobilities reaching 1.2 Â 10 À2 cm 2 /Vs. Otherwise, similar to  www.advancedsciencenews.com www.advenergysustres.com the intermolecular B←N adducts, the intramolecular B←N coordination was found to show reversible formation depending on the external conditions. Only one N atom in the BTZ unit (25) can complex with the adjacent B, and the B←N coordination was reversibly transformed between 25 and 25' under different solvents and temperatures. [78] This can be partially ascribed to the low Lewis acidity of Mes-substituted B. When two BTZ units were linked with π-bridges, e.g., fluorene or carbazole, two six-membered B←N heterocycles were formed peripherally in 26 and 27, respectively. [71,79] As the bulky Mes groups on the B centers were replaced by Ph and C 6 F 5 groups, Lewis acidity of B was enhanced and the B←N bindings were strengthened in 26 and 27. In contrast to B-free precursors, the HOMOs and LUMOs of 26 and 27 significantly decreased due to the introduction of B electron-deficient centers.
Using fluorene as the linker, Wakamiya and coworkers compared inter-and intramolecular B←N coordination in terms of thermal stability, optical properties, and electron affinity. [80] Compared with the intermolecular B←N adducts (28), the B←N-embedded 29 showed higher thermal decomposition temperature (225 vs. 387 C), red-shifted absorption (352 vs. 399 nm), fluorescence (413 vs. 476 nm), and enhanced electron affinity with a more positive reduction potential (À2.30 vs. À1.81 V). Pyridine was also linked to fluorene bilaterally to form the B←N-bridged 30, which displayed strong blue emission for both 30a (λ fl ¼ 434 nm, Φ fl ¼ 0.42) and 30b (λ fl ¼ 437 nm, Φ fl ¼ 0.68). [81] Fang and coworkers reported two B←N-bridged ladder units, 31 and 32, whose LUMOs were significantly lowered to À4.40 and À4.21 eV, respectively, due to the introduction of the electron-deficient B and substituting Br atoms on the B centers. [82] They also found the solvatochromism effect for these two molecules, that is, the absorption edges and peaks gradually blue shifted as the Lewis basicities of solvents increased. Later, based on the same backbone, we reported four units substituted by different numbers of Br atoms (33a-d) to study the effects of Br number on the absorption and energy levels. [83] In addition, the B←N-free unit, that is, the C-C analogue, namely, indacenodithiophene (IDT, 34), was also adopted for comparison. From IDT to B←N-bridged unit without Br substituents and further increasing the Br substituents, the absorption spectra red shifted and the energy levels decreased in a regular trend (Figure 9c). However, due to the strong electron-deficient property of B, these moieties are hydrolytically sensitive and cannot be purified by archetypical column chromatography. To enhance the stability of these units, we replaced the Br groups on the B centers by www.advancedsciencenews.com www.advenergysustres.com aromatic groups, [84] obtaining a chemically stable unit with excellent photovoltaic properties, which will be discussed in the following section.
Apart from the five-membered B←N heterocycles, ladder-type molecules containing six-membered B←N heterocycles were also reported. Jakle and coworkers presented two B←N-bridged polycyclic aromatic hydrocarbons by forming B←N Lewis pairs at the periphery of anthracene (39a-b). [87,88] Due to the polarization effect of B←N bonds, 39a exhibited dramatically lowered LUMO and displayed an enhanced quinoidal character relative to the all-carbon congener and the B-free ligand ( Figure 9d). Interestingly, this molecule shows excellent O 2 -sensitized property in the presence of light, resulting in selective and revisable formation of endoperoxides (Figure 9e). Using a similar synthetic strategy, they also synthesized two regioisomeric structures containing B←N bonds based on pyrene (40 and 41). [89] For both Ph and Et substitutions, the six-member-contained units (41) show largely red-shifted absorption and emission maxima relative to the five-member-contained units (40).

Conjugated Molecules Containing O─B←N Bonds
In contrast to C-B←N groups, the O-B←N groups are more endurable to nucleophiles due to the stronger electron-donating ability of oxygen, effectively quenching the Lewis acidity of B. Li and coworkers disclosed a fluorescent dye by fusing O─B←N bonds into the imide fragments of the archetypical perylenetetracarboxylic diimide (PDI) dye molecule, resulting in 42 ( Figure 10), which exhibited a wider absorption covering the UV-vis region, bathochromic emission peaks, and stronger electron-accepting ability than those properties of PDI. [90] Zhang and coworkers reported a series of ladder-type O-B←N-embedded molecules substituted by the phenyl groups (43a-d and 44a-b) for OLED applications as electron-transporting emitters. [91,92] Single crystals of these structures determined by X-ray crystallography implied that the coplanar backbones were departed by the bulky phenyl groups, rendering these molecules high-fluorescence quantum yields in the solid state. Moreover, the electron mobilities tested by TOF measurements were on the magnitude of 10 À4 cm 2 /Vs, which are remarkably higher than that of the commonly used electron-transporting materials Alq 3 (10 À5 cm 2 /Vs). To this end, intense green-to-yellow emission for 43 and red-to-near-infrared (NIR) electroluminescence for 44 are verified by preparing OLED devices using these molecules as the electron-transporting emitters.
Kubota et al. reported a quinoid-type bisboron complex (45) with a forbidden S 0 -S 1 transition at 800 nm and an allowed transition at %620 nm. [93] As proposed by the authors, the forbidden S 0 -S 1 transition may originate from the highly symmetrical structure. However, upon negative scanning, using cyclic voltammograms (CV), bisboron complexes undergo a two-electron reduction forming aromatic dianions, which shows an allowed S 0 -S 1 transition at %410 nm.
Recently, Wakamiya and coworkers reported a series of BF 2 -bridged dyes containing O-B←N groups (46a-c). [94] For 46a and 46b, visible absorption bands appear at 625 and 457 nm, respectively. Interestingly, 46c exhibits an intense NIR absorption at 922 nm with a large absorption edge reaching 1150 nm. Moreover, the LUMOs of these molecules are lowered to À4.6-À4.4 eV, indicating strong electron-accepting ability.

Conjugated Molecules Containing N─B←N Bonds
The most representative and classic N-B←N-bridged molecules are BODIPY-based dyes, which were first invented in 1968. [95] As several critical reviews have focused on these molecules, [96,97] herein, we are not going to repeatedly review BODIPY and its derivatives. However, inspired by BODIYP, ligands containing aromatic N atoms and amines at proper sites are readily reactive with trivalent B Lewis acids, e.g., BF 3 , BPh 3 , and BCF, to obtain various N-B←N-bridged molecules ( Figure 11). For example, chelate 47a was subjected to BPh 3 to accomplish one B-complexed (47b) and two B-complexed (47c) molecules. [98] As the B number increased from 47a to 47b and 47c, the   Figure 12a) and the LUMOs lowered (À2.58 vs. À3.00 vs. À3.22 eV), whereas the HOMOs (À5.06 vs. À5.05 vs. À4.98 eV) were nearly immobile. Similarly, syn (48a)-and anti (49a)-ladder-type anilido-pyridine ligands were subjected to BF 3 to synthesize boron difouorides-bridged compounds 48b-c and 49b-c, respectively. [99] As expected, the increased B number caused bathochromic absorption and emission spectra, lowered LUMOs, and narrowed bandgaps for both syn-and anti-series. Nakamura et al. conducted borylation reactions on benzodipyrrole-quinoxaline (50a) and benzodipyrrole-benzothiadiazole (51a) chelates, respectively, leading to a series of anti-ladder-type boron complexes 50b-d and 51b-d. [100] Introducing boron into the parent precursors shifted the absorption from visible region to NIR region up to 1100 nm. Moreover, the absorption and energy levels were tuned by changing B numbers and installing fluorines on the diarylboron. Fang and coworkers synthesized two ladder-type N-B←Nbridged molecules (52a-b) with highly reversible and multistage redox activities and multicolor electrochromism properties. [101] As proposed by the authors, the hyperconjugation between the sp 3 hybrid B and N can stabilize the radical intermediates in the oxidation and reduction processes ( Figure 12b). Thus, two separated and revisable anodic peaks in oxidation and other two revisable cathodic peaks in reduction processes were observed (Figure 12c), corresponding to four accessible redox states, i.e., radical anions (À1), dianions (À2), radical cations (þ1), and dications (þ2), which were highly stable upon Zhang and coworkers reported a dual-boron-cored luminogen (53) with a large Stokes shift, aggregation-induced emission activity, and reversible piezochromism, which were further applied to imaging living cells and sensing of fluoride anions. [102] They also disclosed a class of disk-like molecules containing triple-boron centers (54a-b). [103] Single crystals of these diskshape molecules showed slightly deformed skeletons. In contrast to the phenyl-substituted 54b, the F-substituted 54a showed red-shifted absorption (368 vs. 394 nm) and decreased LUMOs (À2.93 vs. À3.17 eV). Later, this molecule was used as the negative photoresist in photolithography because it can be cross linked via exposure to white and UV light. [104] Liu and coworkers uncovered two N-B←N-bridged azaacenes composed of six or eight linearly annelated rings (55a-b). [105] These fused units exhibited low-lying LUMOs (%À3.87-3.70 eV). Solution-processed OTFTs prepared from these azaacenes showed n-type characteristic with the maximum electron mobility reaching 0.21 cm 2 /Vs. Replacing the fluorines in 55a and 55b by propynyl groups led to 56a and 56b, respectively. [106] Due to the electron-donating ability of propynyl, 56a and 56b displayed significantly increased HOMOs (À5.38 vs. À5.87 eV and À5.43 vs. À5.87 eV) in contrast to their F-substituted congeners 55a and 55b, leading to narrowed HOMO-LUMO gaps. Moreover, both the absorption and emission spectra moved to longer wavelengths for propynyl-substituted 56a and 56b in comparison with these properties of F-substituted congeners. Recently, based on the similar ligands, they introduced four B centers by one-pot multifold borylation cyclization reaction to obtain two quadruply N-B←N fused molecules (57a-b) with fascinating properties. [107] Due to the crowd gathering of electron-deficient B centers, these 2D molecules exhibited strong electron affinity with extremely low-lying LUMOs of %À4.5 eV. 2D conjugation rendered these molecules strong propensity to form micrometer-sized wires (Figure 13a-c). OFETs based on single-crystal microwires of 57b afforded the highest electron mobility of 1.6 cm 2 /Vs (Figure 13d-e). Moreover, excellent ambient stability was found www.advancedsciencenews.com www.advenergysustres.com for the OFET devices ( Figure 13f ). The excellent OFET performance and stability were ascribed to strong electron affinity and high crystallinity of 57b, beneficial to the electron injection and transport and air stability. They also reported another disklike π-framework containing four N-B←N units (58). [108] A depressed LUMO of À4.07 eV was tested and an average electron mobility of 3.4 Â 10 À4 cm 2 /Vs was measured by preparing solution-processable OTFT using 58. Their inspiring studies unambiguously clarified that introducing multiple B electrondeficient centers into the conjugated backbones can readily decrease the LUMOs, enhance the electron affinity, and lead to n-type semiconductor characteristic. DPP is a pigment that has been broadly applied to OPVs, OFETs, and fluorescence probes due to its high molar extinction coefficient and fluorescence quantum yield, excellent charge carrier mobility, and good light and thermal stability. [109][110][111][112][113][114] It consists of a condensed lactam core bearing aromatic rings, e.g., phenyl, thienyl, and furanyl groups. Considering the already existed amino group in the lactam, chemical modifications at the pending aromatic rings or the carbonyl groups to introduce aromatic N atoms would feasibly construct N,N' ligands, acting as the precursors toward N-B←N structures. One of the readymade N,N' ligands is the pyridine pended DPP unit (59), which was reactive in the presence of BPh 3 to obtain B-complexed DPP (60, Route I in Figure 14), as discovered by Kanbara and coworkers. [115] In comparison with the DPP precursor 59, B-complexed 60 gave red-shifted absorption (504 vs. 617 nm) and emission (514 vs. 638 nm) and remarkably decreased LUMOs (À3.47 vs. À3.79 eV). Unfortunately, the limited solubility of nonalkylating DPP precursors resulted in very poor isolation yields of borylation. Accordingly, few follow-up reports on this reaction strategy were disclosed. [116] Daltrozzo and coworkers conducted the condensation reactions using DPP and 2-heteroaromatic acetonitrile under acidic conditions using POCl 3 as the activating agent ( Figure 14, Route II). [117] This modification at the carbonyl groups built N,N' ligands with NIR absorption but without fluorescence ( Figure 14, Route II, a). Further borylation with BF 3 or ClBPh 2 led to N-B←N-bridged chromophores with selective NIR absorption and intense fluorescence ( Figure 14, Route II, b). Using this strategy, heterocyclic peripheral groups were tailored to obtain 61a-f, which displayed gradually bathochromic absorption and fluorescence, [118] indicative of strong perturbation of electronic structures of these units by the heteroaromatic rings. Based on the same backbone, more structural cuttings on the heterocyclic peripheral groups and substituents were conducted in their later work, indicating effectiveness of this strategy. [119][120][121] Interestingly, they extended the stiff π-electronic system using a bifunctional bridging heteroaromatic acetonitrile and two equivalent DPP, forming quadruply B-complexed molecules (62a-b). [122] These rigid π-conjugation systems had very narrow absorption bands, high absorption coefficients, and excellent photochemical stability. Furthermore, they selectively absorb photons in the NIR region (710-900 nm) without any strong absorption bands in the visible region (380-700 nm), which are of great interest for various technical applications.    Regretfully, although their optical properties were sufficiently explored, extending studies on electrochemistry, energy levels, and application to electronic devices have been blank for these N-B←N-bridged DPP chromophores.
There is a structural drawback in these pyrrolopyrrole cyanines due to the cyano-substituents at the meso-positions, leading to intensified distortion between aromatic rings at the α-positions and the pyrrolopyrrole (Figure 14, BNDPP1). [118] The poor coplanarity suggests weak conjugation of aromatic rings toward the pyrrolopyrrole and small effect on the electronic properties of the chromophoric system. Shimizu et al. developed an alternative strategy to synthesize B-complexed DPP dyes based on the titanium tetrachloride (TiCl 4 )-mediated Schiff base forming reaction using heteroaromatic amine ( Figure 14, Route III). [123] This one-pot method resulted in B-complexed DPP dyes with lower steric hindrance between α-aromatic rings and the pyrrolopyrrole due to the unsubstituted meso-imine ( Figure 14, BNDPP2). Daltrozzo and coworkers also conducted this reaction using POCl 3 as the activator at a high temperature over 220 C. [124] As such, the electronic properties of the conjugated systems can be effectively tuned by tailoring not only the fused peripheral heterocycles but also the aromatic substituents at the α-positions. Tailoring the fused heterocycles from pyridine (63), benzothiazole (64), quinolone (65), to benzoindole (66) resulted in gradually red-shifted absorption (638 vs. 655 vs. 671 vs. 747 nm). When the phenyl substituents in 64 (638 nm) were replaced by p-piperidinophenyl groups (67: 733 nm) or thienyl groups (68: 699 nm), red-shifted absorption was also observed. Extending the thienyl substituents in 68 to bithienyl (69: 756 nm) and terthienyl groups (70: 803 nm) also gave rise to red-shifted absorption by 50 nm per thienyl group. [123,125] The fluorescence spectra varied in the same order as the absorption spectra. Further elongating the thienyl numbers at the αpositions to red-shift absorption was demonstrated in their later report. [126] Moreover, upon cutting the fused heterocycles and the α-aromatic substituents, significantly changed energy levels and bandgaps were observed. It worth to note that these B-complexed DPP molecules (63)(64)(65)(66)(67)(68)(69)(70) showed low-lying LUMOs of À3.8-À3.5 eV, suggesting strong electron affinity and potential electron-transport properties.
Dimerization of the B-complexed DPP dyes via a bithienyl bridge led to a black dye (71) with panchromatic absorption covering the UV-vis-NIR region, which was significantly extended in contrast to that of the monomer (72). [127] However, the biphenyl-linked dimer 73 had the similar absorption profiles to that of the corresponding monomer 74. This phenomenon was interpreted by the better backbone coplanarity of the bithienyl-bridged dimer (71) than that of the biphenyl-linked dimer (73), causing strengthened skeleton conjugation of 71.
Except for DPP, other electron-deficient bislactams, e.g., isoindigo and benzodipyrrolidone, were also applied to the Schiff base forming reaction and subjected to borylation using BF 3 . [128,129] In contrast to the isoindigo (483 nm) and benzodipyrrolidone (453 nm) precursors, the B-complexed 75 (668 nm) and 76 (672 nm) led to remarkably red-shifted absorption. [128] Nonfluorescence was observed for these B-complexed bislactams. Interestingly, these B-complexed bislactams generally had depressed LUMOs of %À4.00 eV, even lower than LUMOs of DPP-based B complexations, indicating their potential ability for n-type semiconductors.
These bislactam-based B complexations prepared via Schiff base forming reactions have been demonstrated as a promising class of chromophores with good backbone coplanarity, extended absorption covering the vis/NIR region, and low-lying energy levels, which are desirable for n-type semiconductors. Unfortunately, their applications to organic electronics are in infancy and only a few studies on the OPV applications have been reported sporadically (vide infra). [125,127,130]

Conjugated Polymers Containing B←N Bonds
Using the B←N-bridged units as building blocks, certain polymers consisting of B←N bonds with intriguing optoelectronic properties have also been revealed through transition-metal-catalyzed cross-coupling reactions ( Figure 15). Jakle and coworkers prepared two B←N-embedded polymers (77 and 78) based on a B-complexed fluorene-pyridine unit. [131] Polymers 77 and 78 showed strong fluorescence in solutions with quantum yields of 0.55 and 0.78, respectively, and bright yellow luminescence in solid states. Liu et al. reported two N-B←N-embedded polymers (79a-b) based on a B-bridged aniline-pyridine unit. [132] Due to strong ICT effects, 79a and 79b showed extended absorption spectra to NIR at %700 nm, corresponding to narrowed optical bandgaps of %1.50 eV.  Recently, an O-B←N-embedded polymer (80b) analogous to poly(p-phenylene vinylene) with NIR absorption and were reported. [133] In comparison with the B-free polymer 80a, B-complexed 80b showed significantly a bathochromic absorption maximum (510 vs. 702 nm), lowered HOMO (À5.30 vs. À5.38 eV) and LUMO (À3.23 vs. À3.98 eV), and narrowed bandgap (2.07 vs. 1.60 eV). Moreover, nonfluorescence was observed for 80a, whereas NIR luminescence was found for B-complexed 80b. Hole and electron mobilities determined by TOF measurements were in the magnitude of 10 À3 -10 À2 cm 2 /Vs for both 80a and 80b. In addition to the strategy of copolymerizing B←N-bridged units, postborylations at the suitable sites of N-containing conjugated polymers have also been demonstrated to prepare B←N-embedded polymers. Ingleson and coworkers conducted this strategy to accomplish a series of B←N-embedded polymers (81) with NIR emission and applied them to OLED devices and biological imaging. [134]

B←N-Embedded OPV Materials
The OPV materials essentially emphasize solution processability, optical absorption ability, energy levels, charge transport ability, crystallization, and aggregation. Specifically, the electron donor/ acceptor synergism in terms of complementary D/A absorption, matchable D/A energy-level alignments, balanced hole/electron transport, and proper D/A miscibility is the well-established avenue toward high-performance devices. These properties are critically determined by the device technics and material structures of the active layers. As discussed in the aforementioned sections, the B←N chemistry has derived various B←N-bridged electronic units and B←N-embedded materials with extended absorption covering the vis/NIR regions, depressed energy levels, high charge transport ability, and excellent molecular crystallization. When suitable partners (D or A) are selected, these B←N-embedded materials will act as A or D to convert the solar energy to electrical energy. In the past 10 years, B←N-embedded OPV materials have been triggered and given more attention.
Herein, we summarize the known B←N-embedded OPV materials to offer an insight into the design motif and development of these materials. Figure S1, Supporting Information, shows chemical structures of acceptors or donors that have been paired with B←N-embedded OPV materials in this Review.

BODIPY-Based Donors
BODIPY and its derivatives are desirable candidates for OPV donors due to their broad absorption to NIR region, high molar extinct coefficient (%10 5 M À1 cm À1 ), and suitable HOMOs (%À5.5 eV) and LUMO (À3.5 eV) to match with the fullerene acceptors, e.g., PC61BM and PC71BM. [135] Chemical modification at the α-, β-, and meso-positions ( Figure 16) with various electron-rich or -deficient units are readily effective to tailor the absorption, energy levels, aggregation, and charge carrier mobilities. According to the molecular geometry, three categories of BODIPY-based donors, e.g., modified BODIPY single cores, BODIPY dimers, and BODIPY polymers, can be classified.
BODIPY single cores: αand meso-substitutions of BODIPY are synthetically accessible. Table 1 shows the properties and performance of modified single BODIPY donors. Roncali reported two BODIPY derivatives (82a-b) with meso-positions substituted by iodobenzene and α-positions modified by styryl. [22] The symmetric 82b showed red-shifted absorption and a narrower bandgap in contrast to that of asymmetric 82a. Using PC61BM as the acceptor, OPV devices based on 82a and 82b gave efficiencies of 1.17% and 1.34%, respectively, which were amazing during the report time. This pioneering work triggered the application of B←N-embedded materials to OPV donor materials. Ternary devices using blend of 82a/82b/PC61BM as the active layer showed an enhanced efficiency of 1.70%. [23] Later, they further modified the mesoposition with oligothiphene, leading to 82c with similar absorption and energy levels to 82b but a significantly improved power conversion efficiency (PCE) of 2.17%, which was mainly ascribed to the promoted hole mobilities of oligothiphene-modified 82c. [136] Four BODIPY molecules, i.e., 83a-d substituted at the α-positions with vinylthiophene groups, were uncovered to afford PCEs of 1.40%, 4.70%, 0.90%, and 1.50%, respectively. [137] The record efficiency of 4.70% for 83b was interpreted by the lowered HOMO levels, broad and strong external quantum response, and high hole mobility.
Triphenylamine (TPA) is a typical electron-donating unit that has been frequently utilized to modify BODIPY. Ziessel and coworkers substituted the α-positions of BODIPY using vinylthiophene-linked TPA, leading to strong push-pull molecular systems (84a and 84b). [138] Single-crystal structures suggested strong π-π and S-S interactions with layer-to-layer distance of %3.5 Å. Using PC71BM as the acceptor, 84a and 84b gave efficiencies of 1.20% and 1.50%, respectively. Substitution at the αand meso-positions using vinyl-linked TPA resulted into two molecules (85a-b) with strong absorbance from 500 to 800 nm. [139] However, presumably due to the limited hole mobilities and nonoptimum device configurations, poor short-circuit current density ( J SC ), low fill factor (FF), and medium efficiencies of %1.0% were tested for these two molecules. Connecting TPA units to β-positions of BODIPY and tailoring the mesosubstituents with different fat chains gave three molecules (86a-c) with panchromatic absorption from 300 to 800 nm. [140] Poor photovoltaic performance with PCEs lower than 1.0% were observed, which was attributed to the weak charge transfer between the donor and PC61BM along with the device deterioration imposed by moisture and oxygen. Similar to TPA, another electron-donating unit carbazole was used to tailor the mesopositions of BODIPY obtaining 87a and 87b, which afforded high efficiencies of 5.05% and 4.80%, respectively, by thermal annealing and solvent vapor annealing, refreshing the efficiency records of BODIPY-based devices. [141] More importantly, high V OC s of over 1.0 V were obtained for these carbazole-functioned BODIPY due to their deep HOMOs. Xu and coworkers reported a series of BODIPY derivatives by substituting at the α-positions with thienyl, fluorene, and carbazole groups. [142] Selecting the thienyl-functioned 88 to prepare OPV devices afforded a preliminary efficiency of 2.12% without any thermal annealing or high-boiling temperature additive.
Except for the electron-donating units, various electrondeficient moieties were also introduced to BODIPY core.  The typically electron-deficient unit benzothiadiazole was linked to α-positions of BODIPY bridged with thienyl to accomplish a push-pull-push molecule (89). [143] Due to the strong intramolecular interaction, three broad absorption bands covering the UV-vis-NIR regions with absorption onset reaching 900 nm and narrowed bandgap of 1.36 eV were observed for 89. It also exhibited ambipolar transport properties with hole and electron mobilities on the magnitude of 10 À4 cm 2 /Vs, as demonstrated by OFET devices due to its deep LUMO (À4.08 eV). Despite its excellent absorption and charge transport ability, unexpected mediocre efficiency of 1.26% was found for 89/fullerene-based devices. The poor performance mainly stemmed from the inferior J SC and FF, which were attributed to insufficient LUMO offset (< 0.3 eV) between 89 and fullerene derivatives and high recombination rates in the films. Sobral and coworkers disclosed a push-pull system by modifying the meso-positions with electron-pull perfluorophenyl and tailoring the α-positions with different electron-push groups, e.g., methyl, phenyl, naphthyl, and anthracyl. [144] As the electron-donating ability of α-substituents increased from 90a to 90d, the bandgaps gradually narrowed with LUMOs lowered and HOMOs raised. By fabricating the OPV devices, 90c exhibited champion performance (PCE ¼ 2.79%) with a high V OC of 1.0 V and J SC of 7.72 mA/cm 2 . Another push-pull system by substituting the αand meso-positions with TPA, phenothiazine, and nitrobenzene was reported with strong NIR absorption coefficients (91a-c). [145] Preliminary device characterization showed efficiencies of 1.3%-1.6% using PC71BM as acceptor and the efficiencies can be further improved to 1.71% via thermal annealing. Recently, a series of simple BODIPY molecules with different replacements at meso-positions were reported (92a-f ). [146] Although poor OPV performance was tested for these molecules, the authors proposed a view that the presence of electronwithdrawing groups at the meso-position had a detrimental effect on the photovoltaic performance, whereas the meso-free structures showed improved efficiencies.
In contrast to the early reported αand meso-substitutions, the β-modification of BODIPY was not revealed until Lin et al. reported the first β-functioned molecular system (93a-d). [147] In comparison with 93a and 93b, inserting alkyne entities (93c and 93d) released steric congestion and resulted in red-shifted absorption. The alkyne-inserted 93c and 93d gave www.advancedsciencenews.com www.advenergysustres.com better OPV performance than that of 93a and 93b due to the stronger light-harvesting ability, balanced hole/electron mobilities and favorable film morphology. Other typical electrondonating units, e.g., thienyl, bithienyl, fluroene, and carbazole, were also covalently connected to β-positions of BODIPY leading to 94a-d. [148] Moderate efficiencies of %1.0%-2.0% were tested for these molecules and 94b exhibited the highest efficiency of 2.15% due to its prominent hole mobility. Interestingly, β-function with alkyne-linked BDT attained a panchromatic absorptive BODIPY derivative 94e with a narrow bandgap and high molar extinct coefficient. [149] Through thermal annealing and solvent vapor annealing, the OPV devices based on 94e/PC71BM displayed a J SC of 12.98 mA/cm 2 , an outstanding FF of 62%, and a high efficiency of 5.61%. A fused carbazole trimer, namely, triazatruxene (TAT), was also used to modify BODIPY to tune the absorption and molecular packing. [150] Leclerc and coworkers reported the β-functionalized BODIPY using vinyl-and alkyne-linked TPA, respectively, to obtain 95a and 95b. [151] Remarkably, 95b/PC71BM-based device showed a prominent FF of 65% and a high efficiency of 5.8%. The high FF was ascribed to the high hole mobility and weak charge recombination, which critically stemmed from the enhanced solid-state packing due to introduction of the rigid and largely conjugated TAT unit. Excitingly, β-substituted by furan-linked tetrathiafulvalene produced a high-performance BODIPY derivative 96. [152] Thermally annealed OPV devices based on 96/PC71BM afforded a high efficiency of 7.2% with a J SC of 13.79 mA/cm 2 and an FF of 67%. Recently, two star-shaped BODIPY molecules substituted by DPP and porphyrin were reported. [153] Devices based on 97a and 97b demonstrated low energy losses of 0.63 and 0.50 eV, respectively. Along with their strong light absorption, efficient exciton dissociation, and charge transport, recorded efficiencies of 6.67% and 8.98% were tested for 97a-and 97b-based devices.
Apart from the modification on side groups, fusing other heterocycles to the BODIPY core is also an important strategy to regulate the photovoltaic performance of BODIPY. Ma and coworkers reported an aza-BODIPY with B-, O-fused benzene rings (98b). [154] Compared with the BF 2 -substituted aza-BODIPY 98a, highly fused 98b exhibited bathochromic absorption up to 800 nm in solution. Planar heterojunction (PHJ) solar cells based on 98b/C60 gave a V OC of 0.8 V, which was outstanding among the low-bandgap donors of E g < 1.5 eV. Benzannulation on aza-BODIPY cores is also effective to extend absorption. [155] Kraner et al. studied the influence of side groups on the photovoltaic performance of benzannulated aza-BODIPYs (99a-c). [156] They found that upon increasing the size of attached chains, grown trap density was observed for 99b and 99c. To this end, 99a without methyl or methoxy side groups had higher charge carrier mobilities and OPV performance than 99b and 99c. Another BODIPY derivative with benzannulation at β-positions and B-, O-fused benzene rings (100) was synthesized and used as the ternary component of P3HT/IC70BM-based devices. [157] From binary device to P3HT/IC70BM/100-based device, the efficiency was promoted from 3.7% to 4.3%. A series of fluorene-substituted aza-BODIPYs with improved thermal stability were reported (101a-c). [158] Based on these molecules, BHJ devices prepared by vacuum evaporation using C60 afforded an optimal efficiency of 4.5%. PHJ devices prepared from 102 and C60 exhibited a 13 nm-mixed layer at the 102/C60 interface, which was enlightening to understand the extent of spontaneous mixing between the donor and acceptor materials. [159] The strong NIR absorption of BODIPY molecules endows great potential for the tandem solar cells. Ma and coworkers synthesized three furan-fused BODIPYs with strong NIR absorption from 700 to 900 nm. [160] Single-junction devices based on 103a, 103b, and 103c gave PCEs of 2.5%, 4.6%, and 6.1%, respectively. Furthermore, the tandem solar cells consisting of 103c-based NIR subcell and a matching "green" absorber subcell displayed a greatly enhanced efficiency of 9.9% due to the complementary absorption. Recently, they also reported two furan-fused BODIPYs and adjusted the aggregation mode via the endsubstituents. [161] Replacing the end H (104a) by F atoms (104b), the aggregation style was changed from H-to J-mode. Finally, the H-aggregated device was found with reduced nonradiative loss (0.35 vs. 0.49) and higher efficiency (5.5% vs. 4.2%) than the J-aggregated device.
BODIPY dimers: Dimerization is a useful strategy to enlarge the conjugation, extend absorption, decrease the bandgap, and thus enhance the OPV performance of BODIPY derivatives. In general, α-, β-, and meso-positions are available for the dimer linkage ( Figure 17). Zhan and coworkers reported a meso-linked BODIPY dimer bridged by terthiophene (105a), which exhibited significantly enhanced J SC (11.28 vs. 6-7 mA/cm 2 ) in comparison with the corresponding single BODIPY core due to improved packing order and hole mobilities of dimer. [162] Using the same strategy, they further synthesized BDT (105b)-and DPP (105c)-bridged BODIPY. [163] Connection to dye molecule DPP conferred the BODIPY dimers enhanced light-harvesting ability. Accordingly, 105c-based device produced a high J SC of 13.4 mA/ cm 2 , which was 1.7-fold of 105b's J SC value. Dimerization using dye molecules as the linkers not only extends the conjugation but also strengthens the visible absorption greatly. Two similar DPPbridged BODIPY dimers 106a and 106b with high molar extinct coefficients across the visible and NIR regions were also reported. [164] However, poor PCEs of lower than 1.0% were tested due to unfavorable film morphology. Recently, two meso-linked BODIPY dimers bridged with dithienosilole (107a) and dithienogermole (107b) were also synthesized, affording desirable efficiencies of 4.58% and 4.12%, respectively ( Table 2). [165] Peng et al. reported a series of α-connected BODIPY dimers using BDT (108a), fluorene (108b), and thieno[3,2-b]thiophene (108c) as the bridges. [166] The BDT-bridged 108a showed the highest efficiency of 4.75% due to the largest hole mobility and the best nanophase separation with the smallest domain sizes among the three blend films. Similarly, α-connected BODIPY dimers bridged with fluorene (109a), carbazole (109b), BDT (109c), and phenothiazine (109d) were also synthesized. [167] Among these dimers, the BDT-bridged 109c also yielded the highest efficiency of 4.65% due to its superior charge transport property and favorable film morphology. Porphyrin was linked to a BODIPY dimer, producing 110, which exhibited a record efficiency of 5.29%. [168] The high efficiency stemmed from a high V OC of 0.90 V, a J SC of 10.48 mA/cm 2 , and an FF of 56%. Recently, corrole, a tetrapyrrolic congener of porphyrin, was also used as the bridge to link two BODIPYs, leading to 111. [169] Using PC71BM as the acceptor, 111-based device gave    [183] www.advancedsciencenews.com www.advenergysustres.com an even higher PCE of 6.6% with a V OC of 0.90 V, a J SC of 11.46 mA/cm 2 , and an FF of 61%. BODIPY polymers: Although alternating copolymerization is a well-established strategy to design various OPV polymers, only a small amount of examples on the BODIPY-based OPV polymers has been sporadically disclosed to now ( Figure 17). The first report on BODIPY-based OPV polymers can be traced to 2010, when Frechet and coworkers synthesized two alternating copolymers containing BODIPY and ethynyl (112a) or thienyl units (112b). [170] Broad absorption bands extending to 800 nm and low-lying LUMOs of %3.70 eV were measured for the two polymers. Selecting PC61BM as the acceptor, OPV devices based on 112a and 112b yielded moderate efficiencies of 1.30% and 2.00%, respectively. The unsatisfying PCEs mainly originated from the poor J SC s (%4.80 mA/cm 2 ), which were ascribed to nonideal nanoscale phase separation and low hole mobilities (10 À7 -10 À6 cm 2 /Vs). A BODIPY-Pt-conjugated polymer (113) was reported and used as the donor to be paired with PC61BM. [171] Inferior efficiencies of lower than 1.0% were tested for these devices. One of the critical factors damaging the performance was recognized as the amorphous nature with poor molecular packing order stemming from the bulky Bu 3 P ligand. Alternating copolymerization using BODIPY and vinylthienyl units produced an ultralow-bandgap polymer (114, E g opt ¼ 1.15 eV) with panchromatic absorption spectrum ranging from 300 nm to 1100 nm. [172] However, low J SC of 3.39 mA/cm 2 was tested using 114 as donor and PC71BM as acceptor. This unexpected low J SC was attributed to 114's extremely low-lying LUMO of À4.02 eV, causing insufficient LUMO offset between donor and acceptor.
Sharma and coworkers reported a BODIPY-thienyl-ethynyl copolymer (115a) and its porphyrin-enriched analogue (115b). [173] Upon attaching porphyrin to the polymer sides, the absorption bands were greatly extended, corresponding to decreased optical bandgaps from 1.74 eV (115a) to 1.59 eV (115b). Devices treated by solvent vapor annealing gave PCEs of 7.40% and 8.79%, respectively, for 115a and 115b. The better performance of 115b was ascribed to its broader absorption spectra, lower bandgap, and higher charge transport ability. Later, they reduced the side chains of thiophene units in 115a leading to 115c. [174] Using a DPP derivative, namely, SMDPP as the acceptor ( Figure S1, Supporting Information), 115c-based nonfullerene polymer solar cells (PSCs) gave an amazing high efficiency of 9.29%. Based on 115c, better performance of SMDPP than PC71BM was interpreted by the stronger light absorption at NIR region and higher LUMO of SMDPP, leading to reduced energy loss and synergistically improved J SC and V OC . Thiophene and thieno [3,2-b]thiophene were directly copolymerized with BODIPY obtaining 116a and 116b, respectively. [175] Mobilities tested by OFETs reached 5 Â 10 À3 and 2 Â 10 À4 cm 2/ Vs for 116a and 116b, respectively. Inverted BHJ solar cells prepared from 116a/PC71BM showed a PCE of 6.16%. Well-balanced hole/electron mobilities of 3.6 Â 10 À5 / 2.7 Â 10 À5 cm 2/ Vs were demonstrated by space-charge-limited current (SCLC) method.
Recently, Li and coworkers reported the alternating copolymers of BODIPY and BDT (117a and 117b). [176] The α-methyl groups of BODIPY were found to heavily affect the performance. In contrast to 117a, the methyl-anchored 117b showed increased energy levels, improved crystallinity, and dramatically enhanced hole mobilities (2.0 Â 10 À7 vs. 1.9 Â 10 À3 cm 2 /Vs). Selecting N2200 as the polymer acceptor, all-PSCs were fabricated using 117a and 117b as the donors, affording a poor PCE of 0.32% and a greatly enhanced PCE of 5.8%, respectively. Very recently, using 117b as the donor, they selected two small-molecule acceptors of ITIC-2Cl and BTP-2Cl to prepare fullerene-free PSCs, yielding high efficiencies of 7.56% and 9.86%, respectively. [25] 9.86% was the highest efficiency of the single-junction device based on BODIPY donors.
Generally, BODIPY has excellent optoelectronic properties, especially its strong absorption in NIR and tunable energy levels through α-, β-, or meso-modification, which are desirable for OPV donor application. Otherwise, performance of BODIPYbased donors lags behind the state-of-the-art OPV donors, which gave optimal efficiencies over 13% for small-molecule donors and beyond 17% for polymer donors. [12,[177][178][179] From our perspective, the following aspects can be responsible for the limited performance of BODIPY-based donors. First, for most BDOIPYbased small molecules, although absorption bands have been extended to the NIR region, a conspicuous valley emerges in the range of 450-550 nm, [22,136,137,141,162] limiting visible absorption. Second, modification on the BODIPY core usually results in nonlinear molecular geometry and inferior backbone coplanarity, inhibiting the ordered molecular packing and deteriorating the hole mobilities. Moreover, non-optimal film morphologies of the devices resulting from the immature device technics and nonideal device structures in the early stage also severely limited the performance. In addition, most of devices adopted the fullerene derivatives as the acceptors rather than the excellent successor, fused-ring nonfullerene acceptors, [180] e.g., ITIC, IT-4 F, and Y6, which have been demonstrated the most outstanding acceptors to now. It is worth noticing that the recent efficiency record of BDOIPY-based donors was created by adopting BTP-Cl as acceptor. [25] Accordingly, extensive efforts should be devoted to rational material design, optimal device technics, and acceptor screening to further promote the performance of these B←N-embedded OPV donors.

Other B←N-embedded donors
BODIPY analogues, B-complexed DPP dyes invented by Shimizu et al., as discussed in the previous section, were also applied to OPVs as the donor materials. The initial attempt was conducted using 63 as the donor and PC61BM as the acceptor to prepare BHJ devices. [125] Poor performance (PCE ¼ 0.20%) was tested due to over-aggregation of 63. By bearing branched chains onto the backbone, obtaining 118 with released aggregation, and through adopting PC71BM as the acceptor, the efficiency was promoted to 1.27%. Dimerization of B-complexed DPP (71) with panchromatic absorption covering the UV-vis-NIR region was also selected to fabricate OPV devices. [127] However, an unsatisfactory efficiency of 0.74% was obtained using 71/PC71BM as the active layer, which was ascribed to the low solubility of 71, leading to unfavorable film morphology. Recently, they reported three A-D-A-type molecules (119a-c) with extended absorption and high molar extinct coefficients using CDT as the π-linker to connect DPP or B-complexed DPP units. [130] As the B number increased from 119a to 119b, and 119c, the absorption peaks red shifted and the efficiencies significantly increased from 0.18% to 1.49, and 3.88%. Besides the N-B←N-based molecules, three O-B←N-embedded molecules (120a-c) were also prepared and the PHJ solar cells using C60 as the acceptor showed low efficiencies of %1.00%. [181] B←N-bridged fluorene was used as the linker of DPP dimers to obtain 121, which exhibited a high open-circuit voltage (V OC ) of 0.91 V upon blending with PC71BM as the photosensitive layer due to the low-lying HOMO of the donor. [182] Otherwise, moderate J SC of 4.79 mA/cm 2 and poor FF of 32% were produced for the devices due to the nonoptimal phase separation and unbalanced hole/electron mobilities. Duan et al. suspended two B←N-bridged fluorene units to a DPP moiety symmetrically to adjust the aggregation style. [183] In the solid films, both the DPP-bearing fluorene (122a) and B←N-bridged fluorene (122b and 122c) displayed H-aggregation. Upon thermal treatment to the films, DPP-bearing fluorene units (122a) showed enhanced H-aggregation, whereas flipped aggregation from J-to H-style was observed for DPP-bearing B←N-bridged fluorene (122b and 122c). OPV devices based on the three molecules gave low moderate performance as a consequence of suboptimal film morphology, bimolecular recombination, and triplet-state formation. B←N-bridged thienylthiazole was also connected to DPP dimers (123a-c), showing broad spectral coverage from 300 to 800 nm. [184] OPV device based on 123b yielded highest V OC of 0.92 V due to its deepest HOMO (À5.62 eV). In contrast to 123b, 123c showed higher J SC (6.87 vs. 8.00 mA/cm 2 ) but significantly inferior FF (51% vs. 38%) and V OC (0.92 vs. 0.84 V), leading to a lower PCE (3.21% vs. 2.56%). Polymers containing B←N-bridged thienylthiazole (124b) were synthesized to compare with thienylthiazole-based homopolymers (124a). [185] The B←N-embedded 98b showed a much lower HOMO/LUMO (À5.20/À3.23 vs. À4.98/À2.71 eV) and narrower optical bandgap (1.73 vs. 1.85 eV) than these values of 124a. To this end, from 124a to 124b, V OC s were remarkably increased from 0.45 to 0.82 V and the PCEs were obviously enhanced from 0.63% to 3.74%.

B←N-Embedded Acceptor Materials
The depressed energy levels confer the B←N-embedded materials great potential toward OPV acceptor materials. B←Nembedded acceptors have been flourishing in the past 5 years, and plenty of high-performance acceptors consisting of B←N bonds have been developed (Figure 18).
Liu and coworkers reported an unprecedented strategy by introducing B←N bonds into conjugated polymers to enhance the electron affinity. [24] Through replacing the C-C bridges in 125 by B←N bonds, the resultant 126 based on the B←N-bridged unit BNTT showed a synergistically decreased HOMO (À5.37 vs. À5.90 eV) and LUMO (À3.11 vs. À3.76 eV) by 0.53 and 0.65 eV, respectively. Apparent fluorescence quenching was observed in P3HT/126 mixed solution, suggesting charge transfer between P3HT and 126. OPV device preparation using P3HT/126 as the active layer afforded a preliminary efficiency of 0.14%. Although the primitive performance is very low, this strategy is rather enlightening for the design of novel OPV electron acceptors. Their later studies found that the poor performance of 126 originates from the bulky pendant phenyl groups inhibiting the compact π-π stacking and limiting electron mobilities. By selecting a longer spacer isoindigo as the comonomer, the steric hindrance effect that arises by the phenyl groups along the backbone can be alleviated. [186] Thus, 127a displayed significantly enhanced electron mobilities by two orders of magnitude in contrast to that value of 126. Moreover, by selecting PTB7-Th as the donor, all-PSCs based on 127a yielded a respectful efficiency of 5.04%, which was among the highest ranked of all-PSCs at that moment. Replacing the comonomer of isoindigo by another dye moiety of pyridine-flanked DPP led to an amorphous polymer acceptor 127b. [187] Screening the donor polymers among PTB7-Th, J71, and PffBT4T-2OD, all-PSCs based on 127b produced enhanced efficiencies of 6.60%, 6.42%, and 5.25%, respectively. Engineering the fluorine-substituting numbers of benzodithiophene (BDT)-BTZ-based polymer donors to couple with 127b afforded an optimal efficiency of 6.45% (Table 3). [188] BNTT is an asymmetric unit and copolymers based on BNTT are in regiorandom modes, which are unfavorable for the order aggregation in blend films of OPV devices. To avoid this defect, we reported a symmetric, rigid, and coplanar unit, namely, BNIDT. As discussed in the aforementioned section, Fang and coworkers initially reported the synthesis of 31, which was moisture sensitive and could not be handled with column chromatography. [82] We modified the units by replacing Br groups on the B centers with aromatic units, e.g., thienyl and phenyl units, which greatly improved the stability in air. [84] Furthermore, we first applied this unit to construct polymers   [209] www.advancedsciencenews.com www.advenergysustres.com by copolymerizing with thiophene and 3,4-difluorothiophene, resulting in 128a and 128b, respectively. [189] In contrast to BNTT, BNIDT possesses a symmetric structure, extended conjugation, and decreased bandgap. Consequently, 128a and 128b exhibited high electron mobilities of 10 À3 -10 À2 cm 2 /Vs, as determined by OFETs. Selecting PBDB-T as the donor, all-PSCs based on 128a gave an all-right efficiency of 4.03%. Interestingly, 128b-based devices optimized by thermal annealing without any additives produced a high PCE of 8.78%, refreshing the efficiency record of all-PSCs based on B←N-embedded polymer acceptors at that time point. Particularly, the high FF of 70% is comparable with the state-of-the-art values of all-PSCs based on the benchmark polymer acceptor N2200. The high performance of 128b was ascribed to the strong light-harvesting ability, high and balanced hole/electron mobilities, suitable donor/acceptor phase separation, and weak charge recombination. One of the most important features of the polymer is the coplanar and electron-deficient backbone flanked with aromatic side groups, which can not only stack in a relative ordered mode but also avoid over-aggregation. We also copolymerized BNIDT with halogenated BDT units and systematically explored the effects of nonhalogenation, fluorination, and chlorination on the performance of polymer acceptors. [190] From 129a to 129b and 129c, efficiencies were gradually promoted from 1.60% to 3.71% and 4.23%, indicating the positive effects of halogenation.
Copolymerizing BNIDT with DPP led to 129d with a broad absorption and narrowed bandgap of 1.48 eV. [191] Poor device performance was observed for 129d using PBDB-T as the donor, mainly due to the unmatchable energy levels. Recently, we selected vinyl as comonomer and tailored the side groups of BNIDT with methyl (Me) to obtain 130a and 130b. [192] These two polymers had a low bandgap of %1.40 eV with absorption edge reaching 900 nm. By selecting PTB1 as the donor, all-PSCs based on 130a and 130b gave high efficiencies of 7.25% and 6.20%, respectively. The J SC of 130a-based devices was as high as 16.90 mA/cm 2 , which was the champion value among the reported B←N-embedded acceptors.
Besides the B←N-embedded polymers, B←N-bridged units were directly utilized as nonfullerene acceptors for PSCs. Piers et al. synthesized a series of B←N-bridged units (131a-d) by electrophilic borylation on the 2, 5-diarylpyrazines. [193] Absorption and energy levels were feasibly tuned via tailoring the substituents on aromatic cycles or the B centers using different electron-push or -pull groups. Through coupling with PTB7-Th to prepare fullerene-free PSCs, 131a, 131b, 131c, and 131d afforded PCEs of 0.40%, 1.90,%, 0.20%, and 0.30%, respectively. Good V OC s of 0.79-0.96 V made these types of molecules attractive for further structural modification toward high-performance nonfullerene acceptors.
Liu and coworkers reported an unprecedented N-B←Nbridged building block named BNBP with a low-lying LUMO, strong electron affinity, high backbone coplanarity, and compact π-π stacking distance. [194] Using this novel electron-deficient unit, they constructed plenty of OPV acceptors and conducted systematic studies on the structure-property relationship and device technics, [195] which refreshed the efficiency record of B←N-type acceptors and motivated the development of B←N-embedded OPV materials greatly.
For D-A-alternating copolymers, tailoring comonomers and side groups are regular strategies to manipulate the performance. The BNBP-based homopolymer (132) had a narrow absorption band peaked at 626 nm in solution and an optical bandgap of 1.93 eV. [196] It is low-lying LUMO of À3.59 eV and moderate electron mobility of 4.4 Â 10 À6 cm 2 /Vs suggested that it qualifies for the OPV acceptor. Selecting PTB7-P, BDB-T, and J61 as the donors, all-PSCs based on 132 gave PCEs of 2.44%, 2.63%, and 3.04%, respectively. Copolymerizing with other units gave rise to a series of D-A polymers with tunable energy levels and absorption. BNBP-thiophene copolymer (133a) exhibited a LUMO of À3.50 eV and a bandgap of 1.92 eV. [197] Changing the counit from thiophene to selenophene, the resultant 133b showed a decreased LUMO of À3.66 eV and a narrowed bandgap of 1.87 eV. Thus, from 133a to 133b, all-PSCs produced lowered V OC (1.12 vs. 1.03 V) but remarkably increased J SC (5.24 vs. 10.02 mA/cm 2 ) and PCE (2.27% vs. 4.26%). [197] Interestingly, selecting suitable donors to pair with 133a can accomplish commonly high V OC s beyond 1.0 V due to the optimal LUMO level of the acceptor. [198,199] For example, adopting PCPTBT as the donor and 133a as acceptor yielded an extraordinarily high V OC of 1.3 V. [198] Using 3,3'-difluoro-2,2'-bithiophene as the comonomer, the resultant 134 displayed slightly red-shifted absorption and a lowered bandgap of 1.86 eV. [200] More importantly, devices based on PTB7-Th/134 exhibited a low energy loss of 0.51 eV and a further enhanced PCE of 6.26%. Besides all-PSCs, they also paired 134 with a series of small-molecule donors to prepare molecular donor/polymer acceptor (M D /P A ) OSCs. The small molecules usually have strong aggregation property and lead to unfavorable film morphology when blended with polymers. To alleviate the aggregation of small molecules, they proposed several effective strategies, for example, utilizing small molecules with bulky side chains, [201] preparing ternary devices using two small molecules as donors, [202] or increasing the molecular weights of polymer acceptors. [203] These attempts promoted the efficiencies of M D /P A OSCs based on 134 from 2.91% to 6.40%.
CDT unit was also copolymerized with BNBP obtaining 135a with HOMO/LUMO of À5.20/À3.20 eV. [204] Considering the relative elevated energy levels of 135a, P3HT was selected to prepare all-PSCs. It is found that these devices based on P3HT/135a exhibited efficiencies over 1.0% at a broad window of D/A ratios from 0.5:1 to 9:1. Among them, D/A ¼ 5:1 gave the best efficiency of 1.76%. To reduce the LUMO of 135a, they modified the CDT units by replacing the alkyl chains with fluorine, leading to 135b. [205] In comparison with the methyl-substituted 135c, fluorinated 135b showed downshifted LUMO by 0.1 eV and increased electron mobility by one order of magnitude. Using PTB7-Th as the donor, all-PSCs based on 135b gave a high efficiency of 3.76%, whereas 135c afforded poor a PCE of 0.04%.
Recently, they synthesized a series of oligothiophene-BNBP copolymers. [206] By extending the length of oligothiophene, polymers displayed gradually increased HOMOs/LUMOs and enhanced crystallinity. Thus, 136a-c can work as acceptors, whereas 136d-e were readily suitable for donors to pair with fullerene acceptor. When worked with J61, 136c gave the best efficiency of 6.51%. In contrast, 136e yielded the efficiency of 5.79% when paired with PC71BM. To extend the absorption, they connected BNBP with the dye molecule DPP alternatingly, leading to 137 with a small optical bandgap of 1.56 eV. [207] All-PSCs based on PTB7/137 gave a moderate efficiency of 2.69% Parallel to the backbone design, side-group tailoring is also powerful to tune the performance. Instead of alkyl-substitution on the N sites of BNBP, they anchored conjugated alkoxyphenyl groups on the BNBP, leading to 138a and 138b. [208] In contrast to their alkyl-substituted congeners, the alkoxyphenyl-attached polymers had low-lying LUMO energy levels, enhanced π-π stacking, and high electron mobilities. Finally, the devices based on 138a and 138b gave low energy losses of 0.47 and 0.51 eV along with efficiencies of 3.77% and 4.46%, respectively.
In another work, they designed a series of BNBP-bithiophene backbone polymers with different side chains on the N sites of BNBP and the β-sites of thiophene. [209] The LUMOs gradually decreased by cutting the side groups from 139a-h, indicating that the LUMOs of these BNBP-based polymers are feasibly controllable by tailoring side chains. Moreover, the substituents on the B sites are also powerful in regulating the properties. By transforming the substituents on B sites from four phenyl groups (140a) and two phenyl/two fluorine groups (140b) to four fluorine groups (140c), the BNBP-based polymers show blue-shifted absorption spectra, decreased LUMO/HOMO energy levels, and enhanced electron affinities, as well as increased electron mobilities. [210] In their previous studies, electron-rich or -deficient units were directly connected to BNBP. Recently, they developed a series of high-performance BNBP-based polymer acceptors by selecting thiophene as the π-bridge to link BNBP with electron-deficient units. Using difluorophenylene and perfluorophenylene, respectively, as the electron-deficient counits, two copolymers 141a and 141b were obtained. [211] In contrast to 141b, 141a exhibited a small π-π distance of 3.60 Å, leading to an enhanced electron mobility of 5.4 Â 10 À4 cm 2 /Vs. All-PSCs based on J61/141a gave an improved efficiency of 5.46%. Using a special but simple method, that is, dissolving polymer J61 and 141a individually and then blending them immediately before spin coating, can effectively improve the film morphology and promote the efficiency to 7.09%. [212] Furthermore, 141a was applied to indoor photovoltaic (IPV) cells to convert the relatively weaker indoor illumination into electrical power. [213] The IPV cells based on CD1/141a gave a PCE as high as 27.4% under fluorescent lamp illumination of 2000 lx. Simultaneously, 141b was applied to M D / P A OSCs by pairing with a small-molecule donor DR3TBDTC with carbazole side groups. [214] The bulky side groups alleviated the donor aggregation and the device performance was enhanced to 8.01%. More importantly, these M D /P A OSCs exhibited amazing thermal stability, i.e., maintaining 89% of its initial efficiency after thermal annealing of the active layer at 180 C for 7 days.
Using benzothiadiazole as counit, they developed other two high-performance polymers 141d and 141e. [215] Using BD3T as the donor, M D /P A OSCs were fabricated with 141d and 141e, affording PCEs of 5.06% and 9.51%, respectively, creating an efficiency record of M D /P A OSCs. Moreover, 84% of initial PCE can be retained after thermal treatment at 150 C for 72 h, suggesting high thermal stability of the devices. Recently, they also prepared all-PSCs adopting 141c as the acceptor and CD1 as the donor. Through device optimization, high efficiencies of %10% can be obtained, which are the highest levels of all-PSCs based on B←N-embedded polymer acceptors. [26,216,217] Except for the polymers, BNBP was also used to build smallmolecule acceptors by attaching strong electron-deficient units (142a-c). [218,219] These molecules possessed excellent absorption ability in the visible and NIR regions. By tailoring the side groups, the energy levels, absorption, and photovoltaic performance can be adjusted. Fullerene-free PSCs based on these B←N-embedded molecules gave the best efficiency of 7.06%. These systematic studies indicated that BNBP is an excellent building block with comparable or even better properties than the benchmark dye units of NDI and PDI for OPV acceptors.
Consequently, B←N-embedded donor materials represented by BODIPY derivatives were developed for 10 years, whereas B←N-embedded acceptor materials featured with BNTT, BNBP, and BNIDT units have been flourishing in the past 5 years. For the BODIPY-based donors, researchers initially focused on the chemical modification of the BODIPY single cores. Along with the immature device configurations and technics, performance of these BODIPY-based donors developed slowly. Later, as more attention was paid to the BODIPY dimers or polymers, accompanied by optimized device structures and technics and emergence of several outstanding nonfullerene acceptors, performance of BODIPY-based OPV donors was promoted rapidly. For the B←N-embedded acceptors, BNBP and BNIDT have been demonstrated as promising building blocks that can be comparable or even better than these classic units of NDI and PDI. However, based on these units, attention is mainly focused on the polymer acceptors. Only few examples of B←N-embedded small-molecule acceptors have been reported although they also possess excellent properties for OPV acceptors.

Prospects
The initial contributors to the boron chemistry may be gratified to witness the prosperous situation of boron-based materials, especially the B←N-based OPV materials. From the chemical issues to OPV materials, this classic B←N coordination has experienced several decades and gradually developed into a representative class of OPV materials. Concerning the application of B←N coordination to OPV materials and devices, we propose some potential research interests as well as challenges in the future from our point of view. B Lewis acids have been used as dopants for organic semiconductors containing aromatic N atoms to fill traps and improve mobilities. [61,62] The film morphology and mobilities of the OPV devices are also critical to device performance. To this end, taking advantage of intermolecular B←N interactions to manipulate performance of OPV devices deserves in-depth exploration. However, using B Lewis acids to adjust the performance of devices adopting aromatic N-based semiconductors as the active layers has not been reported although this strategy may be reasonably effective. Recent studies on performance improvement via adding BCF to PTB7-Th and adding Pt Lewis acid to tetrazine-based polymer are instructive. [68,220] Developing novel or modifying the known B←N-bridged units to construct high-performance OPV materials will be a long-term interest. Till now, several synthetic methods toward various B←N-bridged units have been established. On the one hand, inventing novel units is the continuous pursuit of chemical and material scientists. On the other hand, some known units have excellent properties, that can be promising for OPV materials if proper modification can be applied. Unfortunately, most of these units have not been sufficiently applied to electronic devices, especially OPV devices. BODIPY, BNTT, BNIDT, and BNBP are the most successful B←N-bridged units that have been applied to OPV materials. In addition, known units with regular and coplanar backbones, broad absorption, lowered energy levels, and ideal packing modes, typically such as DPP-based B←N molecules, may be promising for OPV materials if effective structure tailor can be undertaken. Actually, for OPV application, attention should be paid to not only the optical properties and energy levels but also the molecular aggregation, mobilities, and miscibility with their partners through judicious molecular design.
In parallel, challenges can also be recognized in several aspects. The strong Lewis acidity of B makes some B←N-bridged units sensitive to moisture, which may limit their application to OPV materials despite their outstanding optoelectronic properties. In parallel, the long-term stability of devices based on B←N-embedded materials should be also addressed. Although recent studies suggested high thermal stability of devices using B←N-embedded materials as the active layers, [214,215] it is worrisome whether these B←N-embedded materials are durable in moisture air for a long time. In-depth studies on material and device stability involving the B←N-embedded molecules should be undertaken.
The performances of B←N-embedded materials are inferior to the state-of-the-art OPV materials. For example, J SC and FF of B←N-type acceptors are relative low. The highest J SC of B←Ntype acceptors has been only 16.9 mA/cm 2 , [192] which lags behind the champion values of over 20 mA/cm 2 afforded by other types of outstanding acceptors. [221][222][223] Also, enhanced FF as high as 70% has been revealed only in one work. [189] Accordingly, how to synergistically enhance J SC and FF without sacrificing the V OC of these B←N-embedded OPV materials is an urgent and challenging job.

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