Organic photovoltaics: The current challenges

Organic photovoltaics are remarkably close to reaching a landmark power conversion efficiency of 20%. Given the current urgent concerns regarding climate change, research into renewable energy solutions is crucially important. In this perspective article, we highlight several key aspects of organic photovoltaics, ranging from fundamental understanding to implementation, that need to be addressed to ensure the success of this promising technology. We cover the intriguing ability of some acceptors to undergo efficient charge photogeneration in the absence of an energetic driving force and the effects of the resulting state hybridization. We explore one of the primary loss mechanisms of organic photovoltaics—non-radiative voltage losses—and the influence of the energy gap law. Triplet states are becoming increasingly relevant owing to their presence in even the most efficient non-fullerene blends, and we assess their role as both a loss mechanism and a potential strategy to enhance efficiency. Finally, two ways in which the implementation of organic photovoltaics can be simplified are addressed. The standard bulk heterojunction architecture could be superseded by either single material photovoltaics or sequentially deposited heterojunctions, and the attributes of both are considered. While several important challenges still lie ahead for organic photovoltaics, their future is, indeed, bright. ©


INTRODUCTION
The performance of organic photovoltaic (OPV) devices has drastically improved over the last few decades, but with climate change at the forefront of the world's attention, the development of this sustainable energy solution is now more important than ever.Commercially available photovoltaics are traditionally fabricated from inorganic semiconductors such as silicon; however, "next-generation" materials such as perovskites and organics are rapidly gaining in feasibility.Although still lower in efficiency, OPV cells have considerable advantages: short energy payback times, the ability to be printed in high-throughput manufacturing processes, and high red-NIR absorptivity in very thin films.Combined, these characteristics could enable their incorporation into visibly transparent windows, significantly enhancing the commercial feasibility of building-integrated photovoltaics and providing another crucial tool in the arsenal to fight climate change.
After years of stagnant OPV device efficiencies, the OPV field is currently enjoying a renaissance, with power conversion efficiencies now approaching 20%.This is due to the advent of nonfullerene electron acceptors (NFAs), which have largely replaced the more conventional fullerene (C 60 ) acceptors.Non-fullerenes offer multiple benefits over fullerenes, including substantially higher absorptivities and greater synthetic tuning for specific matching to a donor, with their typically planar structures often enabling improved electron mobilities.In 2017, Zhao et al. reported on the first OPV making use of a non-fullerene acceptor that exceeded the power conversion efficiencies of fullerene-based cells. 1 They made use of a fluorinated PBDB-T donor and a fluorinated non-fullerene ITIC [3,9-  1), achieving a power conversion efficiency (PCE) of 13%.Since then, NFA-based OPV devices have been steadily climbing in efficiency, recently reaching 19% in 2022. 2 These recent advances in device efficiencies have spurred more research into the fundamental properties of these NFA materials, uncovering numerous intriguing results, a few of which will be discussed in this perspective article.4][5][6][7][8] Instead, this perspective article will focus on selected key challenges-by no means an exhaustive list-that remain to first understand and then implement OPVs.We will cover topics that influence both fullerenes and NFAs, encompassing energy offsets and hybridization, nonradiative recombination, the effects of triplet states, and progress toward simpler device architectures.

ENERGY OFFSETS AND HYBRIDIZATION
Organic solar cells require both an electron donor and acceptor to separate excitons into the free charges that produce electricity.After the photoexcitation of a donor or acceptor molecule to form an exciton, the exciton must first diffuse to the donor/acceptor interface.There, the exciton can dissociate to form a charge transfer (CT) state via electron or hole transfer.Due to the typically low dielectric constants of organic materials, the exciton has a high binding energy.This energy barrier must be overcome for the exciton to separate into an electron and a hole and produce photocurrent.The magnitude of the Coulombic attraction between an electron and a hole can be calculated using Coulomb's law [Eq.(1)] and is significantly greater than the thermal energy at room temperature (k B T = 0.025 eV), where e is the charge of an electron, εr is the dielectric constant of the organic material in question, ε 0 is the vacuum permittivity, and r is the electron-hole separation.
The traditional viewpoint of OPV charge generation mechanisms requires a driving force-an "energy offset"-to overcome the exciton binding energy (Fig. 2).In a typical organic solar cell, the frontier energy levels of the donor and acceptor must have an energetic offset that provides this driving force to split the exciton efficiently.The driving force then originates from the energetic difference between the singlet exciton (S 1 state) and CT state energies, ∆E S1−CT .Here, we define the concept of "energy offset" as the LUMO level offset when referring to electron transfer or the HOMO level offset when referring to hole transfer, although it should be pointed out that ∆E S1−CT is an equally valid definition.The CT state must then dissociate into free charges, forming the charge separated (CS) state.The magnitude of the driving force required for efficient charge photogeneration has been the source of significant research with numerous viewpoints offered.One school of thought, for example, is that excess driving force enables a vibrationally excited ("hot") CT state that then provides an impetus for ballistic electron transport, enabling the CT state to overcome its own binding energy and form free charges. 9Lane et al. showed a link between excitation energy and photocarrier mobility, suggesting that "hot" excitons will form "hot" CT states at the interface. 10Additionally, a study by Tamura and Burghardt using the density functional theory (DFT) concluded that charge separation occurs via "hot" CT states. 11They also noted that the excess vibrational energy was particularly beneficial to charge separation when the excess energy was near that of the exciton binding energy.][14] However, the importance of a hot CT state has faced disagreement from other groups.In one study, Vandewal et al. found that the internal quantum efficiency of selected polymer:fullerene systems remained high even if the CT state manifold was directly accessed. 15hey attribute this observation to the ultrafast rate of vibrational relaxation of hot CT states, which outcompetes charge separation, thereby leading to charge photogeneration from relaxed CT states.Some of the disagreement-but not all-arises from semantics and the definition of the term "CT state."Here, we define it as the intermolecular donor/acceptor CT state intermediate formed during charge photogeneration, and it is distinct from a local singlet exciton.
Altering the excitation wavelength to access different vibrational levels within the local singlet exciton manifold, for example, will not necessarily probe the presence or effects of hot CT states, particularly if vibrational relaxation within the singlet excition manifold is very rapid.
To address the ambiguity in the role of "hot" states, Sosorev et al. considered a two-step charge generation model assuming the presence of "hot" CT states. 16An analysis using this model led them to propose two categories of OPVs: "CT state dissociation limited" and "CT state generation limited."They suggest that charge generation in "CT state dissociation limited" systems is enhanced by "hot" CT states providing energy to overcome a significant charge separation barrier, while the latter possess a smaller barrier and are therefore limited by the rate of CT state formation.
The open-circuit voltage (V OC ) and short-circuit current (J SC ) of an OPV device are in delicate balance, with the optimization of one often being at the cost of the other.For example, increasing the interfacial energy offset between the donor and acceptor may increase the photocurrent but will simultaneously limit the V OC .Therefore, a great deal of research is currently focused on minimizing the energetic offset at OPV heterojunctions.However, if the reduced energy offset is achieved by widening the bandgap of one component, this can also reduce the J SC .How then do we achieve both high current and high voltage in an organic solar cell?
In 2015, Takimiya et al. demonstrated remarkably low voltage losses of ∼0.52 eV in a polymer:fullerene blend with an ∼0.1 eV LUMO offset between the donor and acceptor materials, showing that the small offset did not limit charge separation in this case. 17he following year, Liu et al. 18 showed efficient charge generation in a polymer:NFA solar cell with a negligible energetic offset.Their blend exhibited a photoinduced electron transfer of 3 ps and a high electroluminescence quantum yield, enabling much higher solar cell efficiencies despite the observed minimal difference in singlet exciton and CT state energies.Since then, several more systems, including either fullerene or NFAs, have been observed to undergo efficient charge photogeneration with a minimal energetic offset, sparking a flurry of research activity to uncover the fundamental reasons behind these observations.
A central issue in understanding the effects of minimal energy offsets lies in the method used to determine an accurate numeric value for the energy offset itself.No standardized method for determining ionization potential (IP) or electron affinity (EA) exists.][21][22] The results from current methods, including cyclic voltammetry and photoelectron spectroscopy, are typically only compared within the same method.Yet, even for cyclic voltammetry, differences arise from using peak potentials vs onsets, and even variations in the position of the natural hydrogen electrode relative to vacuum to provide the conversion factor between volts and electron volts. 23Furthermore, EA and IP values can alter from the pristine material to the blend, 24 adding an additional element of complexity.Given the large uncertainties in EA and IP values, this creates an inherent difficulty in quantifying energy offsets, particularly in the minimal offset regime, where the singlet and CT emission spectra overlap.
A key aspect of NFAs is that they absorb light strongly, in stark contrast to traditional fullerenes.This opens up the probability of both the donor and acceptor contributing to charge photogeneration by hole transfer from the acceptor in addition to the standard electron transfer from the donor polymer.Zhong et al. 25 demonstrated that this hole transfer can occur on sub-picosecond timescales, even in the absence of an energetic offset.They used planar heterojunctions and dilute blends in addition to device-optimized blends to decouple charge transfer and exciton diffusion, enabling intrinsic charge transfer rates to be determined.Intriguingly, while the hole transfer rate was, indeed, correlated with a driving force, all rates were sub-picoseconds, even when the driving force approached zero.Zhong et al. attributed this to the low reorganization energies associated with the hole transfer process.

PERSPECTIVE scitation.org/journal/jcp
Indeed, it has been shown that the electron transfer channel in non-fullerene blends can be inoperable due to the ultrafast singlet energy transfer from the polymer to the NFA. 26As such, the LUMO level offset may be less relevant in these cases.However, Karuthedath et al. also suggest a minimum ionization potential offset of 0.5 eV to ensure an efficient charge transfer.They relate this to band-bending at the donor/acceptor interface (Fig. 2), which essentially increases the ionization energy of the donor and decreases the electron affinity of the acceptor, effectively increasing the CT state energy.This band-bending inhibits exciton dissociation into CT states at low ionization energy offsets by creating a negative driving force for exciton dissociation under these conditions.Natsuda et al. suggested that energy and charge transfer are in competition in contemporary OPV blends, with smaller domains favoring the latter. 27he band bending observed in non-fullerene blends is caused by the electrostatic interaction of the charges with the large quadrupole moments of the NFA molecules.Quadrupole moments are large in NFA materials because their typical A-D ′ -A ′′ -D ′ -A molecular architecture induces an alternating charge density "checkerboard pattern," which is enhanced due to the aggregation properties of NFAs. 28While Karuthedath et al. 26 show that the band-bending is detrimental to the conversion of excitons to CT states, they also highlighted that it is favorable for the CT state dissociation into free charges.This was beautifully demonstrated by Perdigón-Toro et al. one year earlier for the high-performing PM6:Y6, 29 showing that its barrierless CT state dissociation was possible due to Y6's large quadrupole moment (Fig. 2) and the resulting electrostatic interfacial field that compensates for the Coulomb dissociation barrier.Minimizing the energetic offsets has a natural consequence that the CT state is energetically very close to the singlet state [local exciton (LE)].Importantly, this can lead to a hybridization between the two states (Fig. 3). 30One of the most important ramifications of this is that nonradiative voltage losses can be suppressed due to the resulting increase in the CT state luminescence, 31 and this will be discussed in the section titled "NON-RADIATIVE RECOMBINA-TION AND THE ENERGY GAP LAW."It has also been suggested by Hinrichsen et al. 32 that this hybridization allows an equilibrium to form between singlet, CT, and CS states and that the dissociation of interfacial charge-transfer states is thermally activated rather than barrierless.In this case, it was suggested that the long lifetime of the disorder-free CT state enables no energetic offset requirement.
In a similar vein, Classen et al. 33 suggested that the key to achieving these high organic solar cell efficiencies at minimal energy offsets is a long exciton lifetime.The minimal energy offset inhibits the rate of exciton splitting (electron or hole transfer) at the donor/acceptor interface, and, therefore, a long exciton lifetime enables this slow exciton splitting to take place.For blends with Y6, one of the most efficient non-fullerene acceptors currently known, the exciton splitting efficiency is almost unity even at minimal energy offsets.The authors relate this to Y6's exceptionally and unusually long exciton lifetime of 1 ns.Classen's experimental findings are corroborated by including hybridization via a twostate model with local excitons in equilibrium with CT states.The researchers show that the exciton splitting efficiency is dictated by both the energetic offset and, crucially, the exciton lifetime.
However, an important point to note is that the singlet exciton/CT state hybridization and equilibria observed by multiple groups-and invoked to rationalize high-performing NFA systems-are not actually exclusive to NFAs.If a fullerene is matched with a very low bandgap polymer (thereby also yielding a negligible energy offset), the exciton and CT states will also hybridize and form an equilibrium.This minimizes non-radiative voltage losses in much the same way as non-fullerene acceptors can. 34espite this, such low offset fullerene devices still show substantially smaller power conversion efficiencies compared to recent low offset NFA systems.An important future step, then, is to decouple the magnitude of the energy offset from the identity of the acceptor to truly understand what makes NFAs so special.It appears that the large quadrupoles present for NFAs play a crucial role in this, but the opposing nature of the consequent band-bending-promoting CT state dissociation but inhibiting exciton dissociation at small energetic offsets-requires further investigation and new strategies to address this.

NON-RADIATIVE RECOMBINATION AND THE ENERGY GAP LAW
To enable organic photovoltaics to be competitive on the world stage, one of the most critical factors still to be solved is non-radiative decay back to the ground state.Within the context of OPVs, nonradiative decay usually refers to the relaxation of the interfacial donor/acceptor charge transfer state back to the ground state without photon emission, thereby losing the potential energy as heat.Even with the introduction of highly efficient non-fullerene acceptors pushing efficiencies close to 20%, non-radiative decay remains a major loss mechanism.To maximize light absorption across the solar spectrum, optical bandgaps are being pushed further into the near-infrared (NIR) and most NFAs absorb strongly in this region.Indeed, Y6, one of the highest performing NFAs so far, has an absorption maximum of around 800 nm.However, as the bandgap and thus the singlet exciton energy decrease, this typically increases the efficiency of nonradiative internal conversion (illustrated in Fig. 4).In general, the non-radiative decay involved in the deactivation of an excited state is typically internal conversion followed by vibrational relaxation.The internal conversion step involves a horizontal transition between two isoenergetic vibrational levels of the excited state and the ground state, the rate of which depends largely on the vibrational wave function overlap.This, therefore, gives rise to the energy gap law: within the weak coupling limit, as the energetic spacing between two electronic states decreases, the efficiency of internal conversion between them is exponentially increased. 35Non-radiative relaxation can, therefore, become very fast in low bandgap materials, which typically precludes long exciton lifetimes and high photoluminescence yields.
Collado-Fregoso et al. demonstrated the presence of the energy gap law in polymer:fullerene blends using transient absorption spectroscopy.They investigated the effect of changing the CT state energy, thus altering ∆E S1−CT . 36They found that, indeed, non-radiative geminate recombination from the CT state obeys the energy-gap law.Consequently, lowering the CT state energy enhances non-radiative geminate recombination, which the authors suggested is a result of increased vibronic coupling from the CT state to the ground state.
Non-radiative decay limits the efficiency of OPV devices by reducing the attainable V OC .The V OC of an OPV device, one of the primary parameters controlling device efficiency, is dictated by the non-radiative voltage losses subtracted from the radiative opencircuit voltage: V OC = V OC,rad − V OC,nr .In turn, V OC,nr is dictated primarily by ln[kr/(kr + knr)], where kr and knr are the radiative and non-radiative decay rate coefficients, respectively.A zero knr leads to a zero V OC,nr .For NIR-absorbing materials, therefore, the ultimate goal is to maximize kr at the expense of knr, producing low bandgap materials with long excited state lifetimes and high photoluminescence (PL) quantum yields.However, there is an inherent dichotomy between this ideal and the energy gap law that needs to be addressed, first by tackling the fundamental understanding of nonradiative decay in complex organic photovoltaic blends, which is still very much in progress.
The V OC of an OPV device is intrinsically related to the energy gap between the HOMO of the donor and the LUMO of the acceptor and is thus closely connected to the energy of the CT state.One of the main impacts of minimal energy offset systems, therefore, is to maximize the achievable V OC of a device.As mentioned in the section titled "ENERGY OFFSETS AND HYBRIDIZATION," this has an additional effect of also reducing the energetic separation between the local singlet exciton state and the charge transfer state, enabling the two states to couple together (hybridization).It has been shown for both fullerene and NFA blends that this hybridization allows intensity borrowing: the typically dark CT state gains oscillator strength by coupling with the bright singlet exciton.Eisner et al. 31 demonstrated this using a variety of both fullerene and non-fullerene blends with both experimental and simulation methods with a three-state model, showing a trend of lower non-radiative voltage loss as ∆E S1−CT is reduced.The lower non-radiative voltage losses arise from the enhanced oscillator strength of the CT to ground transition when the electronic coupling between the S 1 and CT states is strong.Notably, it is the lower bandgap material that should have a high S 1 → S 0 oscillator strength: in NFA blends, this is typically the NFA rather than the polymer.
The importance of high photoluminescence and electroluminescence quantum yields to maximize V OC was also reinforced by the design rules reported by Qian et al. 30 Furthermore, as ∆E S1−CT decreases, the reverse transition from the CT state back to the singlet state becomes more viable, thus enabling an additional radiative relaxation pathway and providing another avenue for the reduction in non-radiative decay losses.Indeed, Chen et al. 37 demonstrated that when ∆E S1−CT is small and hybridization is in play, the energy gap law is no longer adhered to due to the additional coupling effects of the bright local exciton.This conclusion further verifies that the PERSPECTIVE scitation.org/journal/jcpselection of donor and acceptor materials with high photoluminescence yields promotes high power conversion efficiencies.For example, it has been shown that relatively low non-radiative voltage losses of 0.15 eV are achievable in an NFA-based OPV device consisting of high photoluminescence yield materials. 38However, Liu et al. noted that hybridization does not always lead to improved CT state photoluminescence and reduced voltage loss. 39They showed that hybridization in one blend led to faster non-radiative recombination than predicted by the Marcus-Levich-Jortner model, making the case for consideration of both the radiative (kr) and non-radiative (knr) rate coefficients of the neat components rather than just their photoluminescence quantum yields.Interestingly, the recent computational and spectroscopic work by Azzouzi et al. 40 reports that reducing ∆E S1−CT in NFA blends decreases the rate coefficients of both singlet exciton and CT state dissociation, alongside an increase in free charge carrier recombination back to CT states.Furthermore, they note that a deeper donor ionization potential not only decreases ∆E S1−CT but also reduces the free energy offset between the CT state and the free charge carriers.These intriguing results provide a broader context for the effects of marginal offset systems, showcasing the trade-off between maximizing V OC and maintaining reasonable J SC and fill factor by demonstrating that the poorer performance in some low-offset systems can be caused by accelerated recombination of charges back to the CT state.

TRIPLETS
Triplet states commonly form during the operation of OPV cells. 41There are three main mechanisms by which donor or acceptor triplet states form in an organic solar cell, depicted in Fig. 5: intersystem crossing (ISC), back electron transfer from a spin-mixing charge transfer (CT) state, 42 and bimolecular recombination of free carriers. 43The latter two mechanisms, therefore, both proceed via a CT state and are dependent on the energetics of the system, kinetic competition, spin statistics, and the nature of the spin-mixing.The rate of ISC is controlled by the magnitude of the spin-orbit coupling and the energy gap between the singlet and triplet states involved, ΔE ST .The spin-orbit coupling depends on either the presence of heavy atoms and/or changes in orbital angular momentum, whereas ΔE ST depends on the exchange energy.
Interest in triplet states has been mounting owing to their recently discovered prevalence in highly efficient non-fullerene acceptor OPV blends. 44Considering that triplet formation is traditionally seen as a loss mechanism in OPVs, this is a very surprising result.Since the T 1 state is often one of the lowest energy states of the system, for example, its formation and subsequent relaxation back to the ground state entail a non-radiative voltage loss.Gillett et al. 44 used a combination of transient absorption spectroscopy and electron paramagnetic resonance spectroscopy to show that a series of NFA-based OPV blends typically exhibited non-geminate NFA triplet formation in preference to geminate triplet formation, including the high-efficiency blend PM6:Y6.It was also noted that in the few blends without evidence of triplet formation via charges, high quantum yields of ISC were observed instead.Indeed, it is estimated that for certain blends, 90% of charge recombination occurred via the NFA triplet and this had a negative impact on the V OC .However, the authors noted that hybridization between CT and local FIG. 5. Formation pathways for triplet states in organic solar cells.In intersystem crossing (ISC, blue), optical excitation to the singlet manifold (such as the S 1 state) on either the donor or acceptor is directly followed by relaxation to the triplet manifold (represented here for simplicity as only the T 1 state).Geminate triplet formation (green): the initial excitation is followed by exciton dissociation (1), resulting in a spin-mixing (2) CT state; back electron transfer (BET) from the 3 CT state to the triplet manifold (T 1 ) then occurs rather than CT state dissociation into free charges (3, FC).In the non-geminate triplet formation pathway (red), the free charges recombine to reform the CT state (4), from which back electron transfer to the T 1 state on the donor or acceptor (depending on which has the lowest energy T 1 state) occurs.
excitons could destabilize the 3 CT state at close donor-acceptor separations such that the back charge transfer to create the NFA triplet is suppressed.
Triplet formation is prevalent in polymer:fullerene blends as well.The generation of triplets rather than charge carriers necessitates that the polymer or fullerene triplet is lower in energy than the charge separated state.This is particularly likely for polymers with high ionization potentials (deep HOMO), as the high IP increases the energy of the CS state above either triplet, as has been reported for both polythiophene and polyfluorene systems. 45,46In 2015, Dimitrov et al. 43 reported one of the first demonstrations of a spinmixing CT state that underwent geminate recombination to form triplet states.Importantly, they linked this to consequently lower free charge carrier yields and, thus, a lower device performance.More recently, Privitera et al. 47 suggest this geminate triplet formation to be a general phenomenon for fullerene-based blends.
Triplet formation can also depend on the morphology of the active layer.Rao et al. 48,49 have shown that ordered fullerene domains enable wave function delocalization of the CT state.The greater effective separation of hole and electron promotes CT state separation over the competing process of back electron transfer to create triplets.In polymer:NFA blend systems, the triplet formation is also influenced by morphology.Gillet et al. have shown, for example, that the relative energies of the 1 CT and 3 CT states depend on the donor/acceptor spatial separation, where the 3 CT state is destabilized and the 1 CT stabilized at short distances due to hybridization between the CT and local excitons in PTB7-Th:IEICO-2F. 44As such, 3 CT states will favor interfacial sites where donor-acceptor spatial separations are larger.Since the donor-acceptor electronic coupling-and, hence, the back electron transfer rate-decreases with increasing spatial separation, this consequently reduces non-geminate triplet formation.
Fluorination has also been shown with both polymers 50 and acceptors 51 to be an effective way to suppress triplet formation.

PERSPECTIVE scitation.org/journal/jcp
Wang et al. showed that the triplet formation via bimolecular recombination of charges was reduced upon fluorination of an IDIC-based NFA, resulting in an increased charge carrier lifetime and improved device performance. 51The authors attribute this to a modified energetic landscape, where fluorination causes an interfacial mixed 3 CT/ 3 LE state that is lower in energy than the T 1 state in the bulk NFA, thereby suppressing triplet generation via the non-geminate recombination pathway.It should be noted that fluorination can also alter other aspects such as morphology and charge carrier mobility.For example, Chen et al. 50showed that non-geminate triplet formation was slower in the PM6:IXIC-4Cl blend compared to the PBDB-T:IXIC-4Cl blend, noting that PM6 contains fluorine atoms while its analog PBDB-T does not.The former blend shows a better device performance.Chen et al. ascribe their observation to the better π-π stacking and charge carrier mobility in the PM6:IXIC-4Cl film, which facilitate CT state delocalization and, thus, dissociation into free charge carriers.
An additional triplet pathway that can have a large negative influence on the OPV device performance is triplet-charge annihilation (TCA). 49,52It has not been frequently considered in the OPV literature, and its precise mechanism is poorly understood because TCA is typically inferred rather than directly observed.Our group recently reported a small molecule:fullerene system in which TCA was clearly demonstrated. 53DRCN5T:fullerene blends-among the fullerene-based OPVs of highest efficiencies-showed a pronounced DRCN5T triplet formation.The TCA was demonstrated in both DRCN5T:fullerene and DRCN5T:NFA blends using microsecond-TAS, where the presence of triplet-quenching oxygen substantially enhanced the charge carrier population and lifetime.Although DRCN5T, with its high oxygen stability, offers an excellent opportunity to showcase TCA, it is reasonable to presume that this mechanism is prevalent in other blends, particularly in cases where charge densities are high.
Indeed, illuminated organic conjugated materials are well known to produce singlet oxygen ( 1 O 2 ) via Dexter energy transfer between the triplet and ambient 3 O 2 .This photochemistry is broadly exploited in the field of photodynamic therapy, where reactive 1 O 2 species can be used to attack cancerous cells. 54Despite this, remarkably few studies have focused on the role of 1 O 2 in the degradation of OPV cells.Soon et al. reported that polymer films with a shorter triplet lifetime were generally more stable. 55,56They proposed that in these cases, the formation of 1 O 2 was outcompeted by the thermal relaxation of the triplet state.However, the significance of this degradation pathway for working OPVs is debated, given that the concentration of O 2 in devices is typically low.
Research on the link between triplet states and stability is now largely focused on the intrinsic photo-instabilities in the bulk heterojunction (BHJ) itself, which cannot be mitigated by better encapsulation barriers.8][59][60][61] Distler et al. proposed that this dimerization degrades performance by reducing the fullerene mobility. 59urthermore, Che et al. showed that triplet states can also catalyze degradative reactions in donor:NFA systems. 62Using IT-4F as an example, they showed that the six-electron photoisomerization between the dicyanomethylene unit and the thiophene ring was mediated by the acceptor triplet state.Recently, Luke et al. conducted a study to investigate the improved stability of ITIC-2F as compared to ITIC and ITIC-DM.The authors proposed that the higher crystallinity of ITIC-2F compared to other IT-core derivatives reduces the likelihood of triplet formation, which, in turn, minimizes the possibility of photoisomerization. 63lthough traditionally seen as a loss mechanism, the unique properties of triplets are increasingly being manipulated to enhance device efficiencies through strategies such as singlet fission [64][65][66] and up-conversion. 67,683][24][25][26] Despite the relatively low efficiency of these triplet solar cells, the point is clear: it is possible to extract energy from triplet states in a solar cell.Importantly, Laquai et al. showed that triplets could create polarons through triplet-triplet annihilation (TTA), enhancing the maximum theoretical efficiency. 27New high energy singlets were created via TTA, where these singlet excitons had enough energy to undergo delayed charge separation on the nanosecond timescale.
There are a number of potential strategies for utilizing triplets in OPVs.The ΔE ST in conjugated polymers is typically large, on the order of 0.7 eV. 41However, reducing this exchange energy has the potential advantage of allowing triplet recycling via reverse intersystem crossing (RISC).Reducing ΔE ST is also necessary for thermally activated delayed fluorescence (TADF), a key strategy for enhancing organic light emitting diode (OLED) efficiencies.As such, the same orbital decoupling strategies employed in the TADF-OLED field could also be applicable for OPVs.
One of the key reasons that the influence of triplet states in OPVs is arguably less well understood than that of singlet excitons and charge carriers is because triplets are an intermediate species.Neither directly photogenerated nor extracted from a device, triplet parameters such as their absorption cross-section, actual populations, and quantum yields of creation/decay are often difficult to measure accurately.We took an important step in this regard recently by using triplet sensitization to measure the molar extinction coefficients of several key NFA triplets, 69 which should allow more facile analysis of triplets in NFA-based OPV blends in the future.
However, it is critical to quantify the effects triplets have on device performance.In this context, one of the most promising techniques is transient electrically detected magnetic resonance, 70 which can be performed on complete OPV devices.This technique, when combined with standard transient electron paramagnetic resonance, enables crucial links to be made between triplet formation/decay and the photocurrent produced by the device.

BEYOND BULK HETEROJUNCTIONS: SINGLE-MATERIAL AND SEQUENTIAL DEPOSITION DEVICES
One of the great paradigm shifts in the OPV field was the introduction of the BHJ architecture, which superseded the bilayer geometry.The BHJ architecture greatly increases the donor/acceptor interfacial area and, therefore, significantly improves exciton dissociation and charge generation.However, BHJs suffer from a major disadvantage in that the enhanced donor/acceptor interfacial area also considerably increases the probability of charge recombination.Despite this, BHJ architectures offer much better efficiencies The Journal of Chemical Physics PERSPECTIVE scitation.org/journal/jcpthan the original strictly planar architectures.][73] However, the advent of new materials, particularly NFAs with their high mobilities and longer exciton diffusion lengths (20-50 nm), 33,74 is promoting a resurgence in the bilayer concept.These have been called several names in the literature: planar-mixed heterojunctions, pseudo-bilayers, or planar heterojunctions.Such structures are typically achieved by sequential deposition (SD) of the donor and acceptor layers from solution via blade-coating or spin-coating, utilizing solvent orthogonality to form a bilayer with a highly textured interface (Fig. 6).The extent to which the two layers mix-and, thus, the donor/acceptor interfacial area-depends on factors such as solvent and donor/acceptor miscibility.6][77][78] Another advantage of planar geometries is the greater reproducibility because of a reduced dependence on microstructure and nanomorphology than BHJs.It is now possible to obtain high efficiency SD solar cells, 79,80 with efficiencies of over 19% reported. 81,82][85][86][87][88] A major point of differentiation between SD devices and BHJs is that the SD strategy allows the processing and optimization of each individual layer.donor and the acceptor materials, respectively. 89This fine control of device fabrication would not be possible with a BHJ architecture.This control of the individual layers has also enabled efficient SD devices with thick layers, 90,91 an important requirement for OPV commercialization feasibility.Cai et al., for example, used bladecoatings of thick sequential layers to obtain 16.9% efficiencies with a 300 nm thick active layer. 92This result was of particular interest because even thicker devices (500 nm) almost maintained that optimum efficiency (15.7%).
In line with this, it is possible to tune the molecular orientation in a planar heterojunction donor/acceptor interface. 93The relative orientation at the donor/acceptor interface has been shown in the literature to be an important factor dictating the efficiency of key photophysical pathways. 77,94,95One of these pathways is the long-range energy transfer by Forster resonance, which can be enhanced in layer-based geometries.Park et al. distinguished energy transfer from charge transfer by incorporating an insulating layer between the donor and acceptor layers 96 and demonstrated that layer geometries facilitate considerably longer-range energy transfer than between isolated molecules or small domains.Energy transfer is known to be an important process in OPV blends and materials. 26,97,98 remarkable point is that, in many cases, SD devices show better stabilities than their BHJ counterparts. 84,85,87,89,90,93,99For example, it is possible to adjust the layer ordering such that the more stable component receives the most light through the transparent electrode.Another origin for this larger stability is the intrinsically separated and purer phases of SD devices in comparison with BHJs.
Both BHJ and SD organic photovoltaic devices require two separate components blended together to form the active layer: the electron donor and acceptor.To achieve an efficient photovoltaic performance in organic materials requires both components to overcome the binding energy of the initial photogenerated exciton to create free charges.This is the case even for the highest efficiency OPV devices, which incorporate novel non-fullerene acceptors. 100,101Although efficient charge photogeneration can be achieved using this two-component configuration, it is accompanied by a host of other issues: energy losses, 34 morphology variations, 102 miscibility, 103 stability, 104 and added fabrication complexity.It is, therefore, desirable to target OPVs with a single-material active layer. 105However, organic small molecules and polymers on their own do not possess the high levels of charge photogeneration 106 or long charge carrier lifetimes required. 107Indeed, while many pristine organic materials can generate charge carriers, these typically recombine on ultrafast timescales due to their inability to escape their mutual Coulombic attraction.Furthermore, many pristine conjugated polymers exhibit solely triplet formation. 108,109he possibility of single-material OPVs has been examined in recent years with research into charge photogeneration in pristine, polycrystalline α-sexithiophene. 110Grain boundaries between crystalline domains possessing different molecular orientations generate the required interfacial energy offset for efficient exciton dissociation into free charges.Remarkably, an EQE of 44% and a V OC of 1.61 V were achieved, highlighting the potential of single-material OPVs.A follow-up work 111 showed that while the exciton dissociation probability at a grain boundary was low (0.5%), the extraordinarily long exciton diffusion length of ∼45 nm was able to compensate for this.Intrinsic charge photogeneration in the high-efficiency nonfullerene acceptor Y6 has also been observed. 107High populations of charge carriers were observed at very early times, attributed to an intermolecular polarization pattern that promotes exciton dissociation.However, these charge carriers decayed back to the ground state rapidly, leading to poor-performing OPV devices.The desirable properties of Y6 as an electron acceptor have also led it to be incorporated into intramolecular donor/acceptor block copolymer structures for single material OPVs, but again, high levels of charge carrier recombination were reported. 112,113ather than a block copolymer strategy to incorporate both the donor and acceptor components, double-cable polymers have also been trialed for single-material OPVs.Typically, these materials have a polymer donor backbone with acceptor moieties as side chains that are required to stack together to create the necessary electron channels.One of the most successful was reported by Jiang et al., who achieved a remarkable 8.4% OPV device efficiency. 114his was possible due to the fine-tuning of the miscibility by altering the position of Cl atoms such that exciton diffusion and dissociation were promoted.
Recently, we reported a conjugated polymer with a donor/orthogonal acceptor motif.The orthogonal acceptor spatially isolated the LUMO from the HOMO, allowing for intramolecular CT states to form.These CT states were able to dissociate into free charges in both the pristine solution and the film.The charge carriers possessed remarkably and unusually long millisecond lifetimes due to the stabilization imparted by the spatial separation of the polymer's donor and orthogonal acceptor motifs.Although this strategy mitigated the fast recombination observed in other pristine materials, another issue arose: the donor/orthogonal acceptor polymer was also able to efficiently produce triplet states on ultrafast timescales.While evidence existed that the triplet and charge carrier generation were linked, the excessive triplet formation still limited charge photogeneration yields.
Single-material OPVs remain an extremely worthwhile goal, but there remain significant questions to be answered.Charge carriers need to be produced with high yields and long lifetimes.Furthermore, these carriers must be able to be extracted in a device.Finally, in order to be commercially viable, the synthesis must be scalable in an industrial setting, which may place limitations on the complex donor-acceptor copolymer design.

CONCLUDING REMARKS
Extensive research has been conducted over the last few decades with the aim of increasing the power conversion efficiencies of OPV devices.In particular, a thorough understanding of the photophysics and device physics involved will be required for future organic devices to match the efficiencies displayed by the current state-of-the-art inorganic cells.After the relatively recent emergence of high efficiency NFA blends, it has become apparent that NFAs possess several key characteristics that have led to significant efficiency increases in the last few years: high absorption coefficients, strong packing motifs, synthetic tunability, and the ability to generate charges with low energy offsets.The trade-off between V OC and J SC that once hindered the performance of fullerene devices can be overcome.Newfound importance is now placed on long exciton lifetimes, favorable interfacial electrostatics, and wave function delocalization at the interface to reduce the effective exciton binding energy in lieu of a significant energy offset to provide a driving force.These properties have been demonstrated in non-fullerene acceptors such as Y6, which has been successfully used in several state-of-the-art organic devices reaching efficiencies of 19%.However, there are still several aspects of OPV materials and devices that need to be tackled, including the role of triplet states, the current dependence on BHJ architectures, and non-radiative voltage losses.It is likely that further improvement in the OPV device performance and enhancement in their commercial feasibility will heavily rely on delving more deeply into these aspects, along with the careful design of acceptor molecules to optimize desirable properties.

FIG. 1 .
FIG. 1. Structures of selected small molecule donor and acceptors and polymers referred to in the text.

FIG. 3 .
FIG. 3. Potential energy surfaces illustrating the hybridization between the bright LE/S 1 and dark CT states at small energy offsets (small ΔG LE-CT ), enabling an increase in the oscillator strength of the CT state via intensity borrowing.Pertinent rate constants are shown, including the radiative (r) and nonradiative (nr) decays back to the ground state and the facile interconversion between LE and CT states (k S1-CT and k CT-S1 ) enabled by their enhanced coupling.

FIG. 6 .
FIG. 6.An illustration of the differences between a coating of a BHJ (a) and a sequentially deposited (SD) device (b) via blade-coating; a representation of the morphologies formed in each case (c).Figures (a) and (b) are reproduced with permission from Sun et al., Energy Environ.Sci. 12, 3118 (2019).Copyright 2019 Royal Society of Chemistry.