Incremental optimization in donor polymers for bulk heterojunction organic solar cells exhibiting high performance


  • Mayank Mayukh,

    1. Department of Chemistry and The James Franck Institute, The University of Chicago, 929 E 57th Street, Chicago, Illinois 60637
    Search for more papers by this author
  • In Hwan Jung,

    1. Department of Chemistry and The James Franck Institute, The University of Chicago, 929 E 57th Street, Chicago, Illinois 60637
    Search for more papers by this author
  • Feng He,

    1. Department of Chemistry and The James Franck Institute, The University of Chicago, 929 E 57th Street, Chicago, Illinois 60637
    Search for more papers by this author
  • Luping Yu

    Corresponding author
    1. Department of Chemistry and The James Franck Institute, The University of Chicago, 929 E 57th Street, Chicago, Illinois 60637
    • Department of Chemistry and The James Franck Institute, The University of Chicago, 929 E 57th Street, Chicago, Illinois 60637
    Search for more papers by this author


The power conversion efficiency of an organic solar cell has now exceeded the 10% mark, which is a significant improvement in the last decade. This has been made possible due to the development of low-band-gap polymers with tunable electron affinity, ionization potential, solubility, and miscibility with the fullerene acceptor, and the improved understanding of the factors affecting the critical device parameters such as the VOC and the JSC. This review examines the latest strategies, results, and trends that have evolved in the design of solar cells with better efficiency and durability. © 2012 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys, 2012


Increasing awareness and growing concerns on the future energy security drive the research effort in the development of effective technology to harvest solar energy, the largest untapped, renewable energy source. Among these techniques, photovoltaic solar cells are very attractive because they can convert solar energy directly into electricity in a rather high efficiency. However, the high cost for the solar cell devices based on inorganic semiconductors have limited their widespread application. In the last decade, solar cells based on organic semiconducting materials have shown potential as an alternative energy source for the future, which has invigorated extensive research in this area.1–24 Of particular interest are the bulk heterojunction (BHJ) polymer photovoltaic devices comprised of a solution-processed active-layer that can be fabricated via high throughput techniques such as reel-to-reel wet coating, ink-jet printing, or spin-coating, which can facilitate the formation of large area, light weight, and potentially flexible devices.25 The active-layer in a BHJ solar cell device typically consists of a blend of donor and acceptor materials that is separated from top and bottom electrodes by exciton blocking and hole selective layers, respectively. Fullerenes such as C-60 and C-70 and their soluble derivatives, PCnBM (n = 61, 71,…), are the most popular choice for acceptors. Devices based on polymer/fullerene have achieved power conversion efficiency (PCE) values exceeding 10.0%.19, 26–28 Despite these advancements, the efficiency of a polymer–fullerene solar cell device is limited by factors such as inferior charge-carrier mobility, low efficiency in large area devices, and unknown long-term stability. Therefore, the challenge remains to design better donor materials for improved device performance to lower the cost per kilowatt so that the organic solar cell technology can compete with the current technologies based on inorganic semiconductors. In this review, we discuss the latest developments of donor polymers for BHJ solar cell devices that have exceeded PCE values of 6%. We also examine the effect of molecular design on critical energy offsets, band gap, and charge-carrier mobility, and the effect of processing condition on morphology, phase segregation, and overall device performance.


A series of events take place during the photoconversion process: (a) absorption of light to generate excitons; (b) diffusion of excitons; (c) photoinduced charge transfer (PICT) at the donor–acceptor interface, involving the dissociation of excitons into free electrons and holes; and (d) charge transportation and collection at the electrodes.29, 30 Upon absorption of the incident light, an electron in the HOMO of a donor (or acceptor) gets excited to the LUMO to form an exciton. An exciton must diffuse to the donor–acceptor interface to generate an electron and a hole via the charge transfer. It is generally accepted that for an effective exciton diffusion, the domain size of either donor or acceptor should be smaller than twice the diffusion length (LD) of an exciton so it can reach the interface during its lifetime.30, 31 An exciton in the donor dissociates by transferring an electron to the acceptor, whereas a hole is generated in the donor. This traditional model of charge generation involving exciton diffusion to a donor–acceptor interface followed by electron-transfer from the donor to the acceptor has been a subject of discussion. For example, Rumbles et al. proposed a two-step process in which a singlet exciton generated in the donor is transferred by the Förster mechanism to the acceptor, and then it dissociates by hole-transfer back to the donor.32 Recently, Heeger et al. argued that the rate of exciton diffusion is too slow to account for the observed ultrafast (100 fs) electron transfer rate.33 The authors proposed that the carrier delocalization should happen before the formation of the exciton because of the high initial mobility and transport assisted by the quantum effects, implying that the exciton diffusion is not involved in the charge transfer mechanism in polymer/fullerene BHJ solar cells.33 However, neither of these arguments can be generalized at this point in time.

The separation between an electron and a hole at which the Coulombic attraction energy balances the thermal energy, the Onsager radius (rc),34 is a function of dielectric constant and thermal energy:35

equation image(1)

where e is electronic charge, εr is the dielectric constant of surrounding medium, ε0 is the permittivity of the vacuum, k is the Boltzmann constant, and T is absolute temperature. The rc is much larger for organic semiconductors owing to the low dielectric constant. The probability of dissociation of an exciton depends on the relative mobilities of electrons and holes, and the energy offset between the donor and the acceptor electron affinities, that is, ELUMOD and ELUMOA (Energy-Level Tuning of Donor Polymers: Route to High VOC and JSC section). It is generally accepted that if the electron mobility exceeds the hole mobility by a factor of 100 or higher, dissociation is favored.36

Once the charges become free carriers, they migrate toward their respective electrodes through the bulk polymer–fullerene composite, along the donor and/or acceptor percolation channels, via either intrachain motion or interchain hopping.37 The electron mobility of fullerenes are of the order of 10−3 cm2 V−1 s−1 as measured in the space-charge-limited regime38 or about 10−1 cm2 V−1 s−1 as measured in field-effect transistors, which is relatively higher than most polymer materials.39 Therefore, for a balanced charge transport, it is necessary to enhance the hole mobility of the donor materials. It has been suggested that a molecular assembly consisting of an interdigitated network of donor and acceptor phases, comprising of well-ordered donor and acceptor within those phases, is the key to facilitate this process.40–42

The overall efficiency of organic solar cells is a function of efficiency of individual steps:

equation image(2)

where ηA is the efficiency for optical absorption, ηED is the efficiency for exciton diffusion, ηCT is the efficiency of charge transfer, ηCC is the efficiency of charge collection, and ηIQE is the overall internal quantum efficiency (IQE).29

An illustrative energy-level diagram for an organic solar cell device depicting HOMO and LUMO energy levels of donor and acceptor components, and Fermi level of electrodes is shown in Figure 1(a). Further, an equivalent circuit for an organic solar cell device can be drawn as shown in Figure 1(b), which consist of a diode and series (RS) and parallel (RP) parasitic resistances. The current–voltage (J–V) plot is often used to determine the performance of an organic solar cell device. The ideal and the actual device characteristics, in dark and under illumination, is shown in Figure 1(c). The most widely accepted mathematical model for J–V characteristics of a solar cell is given by the Shockley diode equation:43

equation image(3)

where J is the total current, V is the applied potential, J0 is the reverse saturation current, and JPh is the photocurrent. RS and RP are series and parallel resistances, respectively; n is the ideality factor, k is Boltzmann constant, T is absolute temperature, q is electronic charge, and A is the area of the device. Both potential (V) and current (J) can be measured experimentally, while area of the device (A), Boltzmann constant (k), and temperature (T) are known. Also, the ideality factor n is often assumed unity. The J–V plot can be used to calculate the open-circuit photopotential (VOC) and the short-circuit current (JSC) values can be calculated.43 Further, eq 3 can be solved at J = 0 and V = VOC:

equation image(4)

Finally, the PCE can be calculated using the following equation:

equation image(5)

where Pmax is the incident power density and FF is the fill factor [Fig. 1(c)]. The FF the ratio of the output power (VOC × JSC) divided by the theoretical maxima (Vmax × Jmax).43

Figure 1.

(a) Energy-level diagram for device operation, (b) circuit-equivalent of an organic solar cell, and (c) characteristic J–V plot in dark (blue) and under illumination (red).


The common design strategy for high performance polymers involve (a) synthesis from electron-rich (donor) and electron-deficient (acceptor) comonomers via a cross-coupling reaction; (b) highly conjugated, planar backbone structures via fused rings or otherwise, for promoting the interchain π−π stacking and mobility; (c) straight or branched chain alkyl or alkoxy groups to impart the solubility in common organic solvents to enhance processability; and (d) the presence of electron withdrawing moieties on monomer units to manipulate the HOMO and LUMO energy levels of the donor polymer. In this context, several copolymers of benzo[1,2-b:4,5-b']dithiophene (BDT),1, 2 4H-cyclopenta[2,1-b:3,4-b']dithiophene (CPDT),9, 13 dithieno[3,2-b:2′,3′-d]silole (DTS),17, 18 dithieno[3,2-b:2′,3′-d]germole,17 2,7-carbazole (CBz),6, 8 bithiophene,14, 44 thieno[3,4-b]thiophene (TT),1–4, 21 [2,1,3]-benzothiadiazole (BT),6, 8 naphtho[1,2-c:5,6-c]bis[1,2,5]thiadiazole (NT),7 thieno[3,4-c]-pyrrole-4,6-dione (TPD),13, 17 [1,2,5]thiadiazolo[3,4-c]pyridine (PyT),11 benzo[d][1,2,3]triazole (TAZ),13 have been reported that exhibit PCE of 6% or above in BHJ organic solar cell devices (Fig. 2). Soluble fullerene derivatives such as PC61BM and PC71BM are common choices for acceptor materials in BHJ solar cells (Fig. 3).

Figure 2.

Chemical structures of the donor–acceptor polymers.

Figure 3.

Chemical structures of the commonly used acceptor molecules in BHJ solar cell devices.

We now know that the conjugated polymers should exhibit a low optical band gap (1.2–1.9 eV) to maximize the absorption, spanning across the visible and near-IR region of the electromagnetic spectrum (Band-Gap Engineering and Light Harvesting section).45 One of the most promising approaches to achieve a low band-gap polymer is to utilize the intramolecular charge transfer interactions through the alternating incorporation of donor and acceptor units into the polymer backbone.46, 47 This strategy provides tunability of HOMO and LUMO energy levels of a donor–acceptor polymer via varying the HOMO and LUMO energy levels of its donor and acceptor units. In this context, thiophene derivatives are the most commonly used building blocks because of the electron-rich sulfur atom and the rigid five-membered ring, which facilitates the extension of conjugation and increases the π–π intermolecular stacking. Several thiophene-based donor monomers have been reported, such as bithiophene, BDT,1, 2, 4, 5, 11–13, 15, 16, 21, 22 CPDT,9, 13, 24, 48, 49 and DTS.18, 19 In addition, CBz derivatives have also been explored as donor units because of the electron-rich nitrogen atom, which contributes toward the aromaticity of the tricyclic ring system (Fig. 4).50

Figure 4.

Electron rich monomer units: BDT, benzo[1,2-b:4,5-b′]dithiophene; CPDT, cyclopenta[2,1-b:3,4-b']dithiophene; DTS, dithieno [3,2-b:2′,3′-d]silole; DTG, dithieno [3,2-b:2′,3′-d]germole.

The most common design strategy for an acceptor comonomer is to incorporate rigid heteroaromatic rings containing electronegative nitrogen or fluorine atoms. The introduction of strong electron acceptors into the polymer backbones impart a low HOMO level and hence a lower band gap of the polymers.1, 2, 4 BT is one of the widely used electron-deficient units,6, 8, 9 and recently several other electron-deficient moieties such as PyT,11 thieno[3,4-c]pyrrole-4,6-dione (TPD),13, 17, 18 NT,7 and isoindigo5 have been reported. In addition, fluorine-substituted BT,12 TT,1–4, 21 and TAZ13 have also been explored (Fig. 5).

Figure 5.

Electron-deficient monomer units: TPD, thieno[3,4-c]pyrrole-4,6-dione; BT, [2,1,3]benzothiadiazole; NT, naphtho[1,2-c:5,6-c]bis[1,2,5]thiadiazole; PyT, thiadiazolo[3,4-c]pyridine; TAZ, benzo[d][1,2,3]triazoles; TT, thieno[3,4-b]thiophene.

The most common polymerization route to these high performance polymers is Stille coupling, which involves coupling of an organotin compound with an arylhalide in the presence of a palladium catalyst.51, 52 Yu et al. developed BDT-based PTB1–7 polymers using the Stille coupling of alkoxy-substituted BDTs and ester-substituted TTs, which promoted the quinoid population and effectively reduced the band gap of the conjugated polymers.1, 2, 4, 21 The solubility of the polymers were greatly enhanced using branched alkyl side chains, and as a result, PTB2–7 exhibited better solubility in common organic solvents compared to PTB1 bearing linear alkyl side chains.1, 21 Also, the authors demonstrated that the intermolecular ordering and, hence, the charge-carrier mobility could be significantly improved upon substitution with fluorine atoms.1, 2, 4, 21, 53 Similarly, several low band-gap polymers were reported using the electron-deficient TPD unit.13, 15, 17, 18, 54 For example, Fréchet et al. reported polymers (P1–3) bearing BDT and TPD units with variable alkyl side chains attached to the TPD unit.15 Among these, P3 containing a less bulky octyl side chain on the TPD unit, exhibited enhanced π–π ordering and crystallinity in the solid state, even in the presence of PCBM.15 Interestingly, the PDTSTPD polymer containing TPD and silicon-bridged DTS units exhibited increased solid-state ordering and a lower HOMO energy level compared to the corresponding carbon-based fused rings.18 Further enhancement in the π–π stacking in the film was achieved upon substitution of the Si atom in PDTSTPD with a Ge atom. By virtue of the long C[BOND]Ge bond length, the steric hindrance between the bulky side chains and the polymer backbones was significantly reduced.17 Consequently, several other donor–acceptor polymers were developed using similar strategies.16, 22, 11–13, 14

Suzuki coupling, which involves coupling of a diboronic ester-substituted aromatic ring and dibrominated aromatic unit in the presence of palladium catalyst and base, has also been used to synthesize donor–acceptor polymers. For example, PCDTBT was synthesized by Suzuki coupling of a dithiophenyl benzothiadiazole (DBT) and a heptadecanyl carbazole.50


For an efficient photoconversion, an active layer should exhibit a high absorptivity and a panchromatic absorption, spanning through both the visible and the near-IR region of the electromagnetic spectrum. The solar photons incident on the earth's surface exhibits highest flux in the visible region (400–700 nm), followed by the near-infrared region (700–1400 nm). Therefore, the focus the organic solar cell research has recently been shifted to the design of donor polymers exhibiting absorption in the near-IR, where 50% of the solar radiation is incident (Fig. 6).55 Also, theoretical calculations suggest an optimum band gap of about 1.5 eV for a donor polymer to maximize the PCE of a single cell solar cell device.45 Therefore, PCEs could be significantly improved if optical absorption is extended to the near-IR, while minimally affecting the VOC.29, 56 This realization has prompted research in developing low band-gap polymers that could extend absorption to the near-IR region.

Figure 6.

Solar photon flux at the earth's surface (1000 Wm2 and AM1.5G) as a function of wavelength. The cumulative short-circuit current density from for an absorber material is shown on the Y-axis. (Reprinted from Ref. 55, with permission from Elsevier.)

The band gap of a polymer could be engineered by increasing the extent of delocalization of the π-orbitals. The extent of conjugation appears to be a function of number of monomer units in conjugation and varies linearly to a certain degree before leveling off, as seen in polythiophenes.57 Yu et al. demonstrated that the electrochemical band gap of a polythiophene polymer can be sequentially reduced (2.22 → 1.11 eV) on introduction of increasing ratios of the TT unit.3 Subsequent development of BDT- and TT-based polymers led to a significant reduction in the band gap and improvement in the device efficiencies.1–4 Similar strategies have been used to tune the band gap of the other polymer systems. For example, PBDT-DTNT exhibited a red-shifted absorption with a narrower band gap (2.16 → 1.58 eV) than PBDT-DTBT due to the electron-deficient NT unit.7 Interestingly, the HOMO and LUMO energy levels can also be tuned without significantly affecting the band gap. For example, on replacing the hydrogen atoms on the DTBT unit of PBnDT–DTBT with fluorine atoms, the HOMO and LUMO energy levels of PBnDT–DTffBT were both lowered by about 0.14 and 0.20 eV, respectively, while maintaining the band gap at about 2.2 eV.58

The band gap can also be tuned by improving the π–π stacking between the polymer chains. For example, the thin-film absorption spectrum of PBnDT–DTPyT saw a remarkable red-shifting relative to the solution spectrum owing to the enhanced molecular stacking in the solid-state arising from the symmetric nature of the BnDT unit.11 Similarly, higher absorptivity and red-shifting of the absorption peak was observed in a PBTTPD:PC71BM films, when cast from CHCl3 containing 0.5 vol % of diiodooctane (DIO), due to the enhanced miscibility of the polymer with the fullerene, thereby promoting ordering within the PBTTPD chains (Tuning the Morphology: Effect of Additives section).59 Likewise, owing to the planar nature of the TPD unit, the PCPDTTPD polymer exhibited enhanced interchain interactions and high charge-carrier mobility. Apart from the λmax at 653 nm, a shoulder peak at 702 nm, arising from the interchain stacking decreased with increasing chain lengths on the TPD unit. Therefore, the fine-tuning of HOMO and LUMO energy levels, and optical band gap, coherently or independently of each other, could provide a route to developing materials with optimized absorption signature.


The photoexcitation of the active layer leads to the formation of excitons, which diffuse along the concentration gradient to the donor–acceptor interface, where a PICT involving the dissociation of excitons into free electrons and holes takes place. Now, we know that this process is facilitated by the ELUMODELUMOA offset, where ELUMOD and ELUMOA are the electron-transport energy level of donor and acceptor, respectively. However, the relationship between the rate of electron transfer and the ELUMODELUMOA offset is not linear. According to the Marcus theory, this process can be realized as the charge migration at the intersection of donor and acceptor potential wells,60 where the exciton must overcome an activation barrier (ΔG), owing to coulombic interactions before it can dissociate into a free electron and a hole. This barrier has been postulated to be a nonlinear function of the Gibbs free energy (Δ) and the reorganization energy (λ), and therefore, the rate of electron transfer reaches its maximum when ΔG° = λ

equation image(6)

Beyond this point, the inverted Marcus region is reached, which could be detrimental to the PICT (Fig. 7).61 Therefore, careful attention should be paid in choosing a donor–acceptor pair so that the ELUMODELUMOA offset is maximized within the Marcus region. It has been suggested in the literature that the ELUMODELUMOA offset should be between 0.3 and 0.8 eV for an efficient PICT. Also, important device parameters such as the open circuit photopotential, VOC, and the short-circuit photocurrent, JSC, are functions of fundamental properties of donor and acceptor materials such as their ionization energy and electron affinity, respectively. It is widely accepted that JSC is a function of ELUMODELUMOA, whereas VOC is a function of the EHOMODELUMOA, where EHOMOD is the hole-transport energy of the donor, and ELUMOA is the electron-transport energy of the acceptor.

Figure 7.

Potential well for a donor/acceptor (D/A) system. Upon photoexcitation, 1D*/A is generated that subsequently undergoes an electron transfer to generate D+/A. Where Δ is the energy barrier for the reaction, ΔG is the energy difference between the reactant's minimum and the point of intersection between the two surfaces, and λ is the reorganization energy. (Reprinted from Ref. 35, with permission from American Chemical Society.)

The tuning of the energy levels in a donor polymer and the matching of the energy levels with that of an acceptor can be addressed synthetically. It is now known that systematic introduction of electron withdrawing groups on donor and acceptor units of a donor–acceptor polymer may lower the HOMO and LUMO energy levels, respectively. For example, among two very similar polymers, PTB2 and PTB3, the later exhibited a 0.1-eV lower HOMO energy level due to the presence of n-octyl groups, which is less electron donating relative to the n-octyloxy groups in PTB2.1, 2, 4, 22 On the other hand, PTB4 exhibited a 0.07-eV lower LUMO energy levels relative to PTB5 due to the presence of a fluorine atom on thienothiophene unit of the PTB4.22 Further, PTB4 registered a 0.1-eV lower HOMO level than PTB5, reflecting that the variation in the LUMO energy level is not independent of the variations in the HOMO energy level in the present case.22 These changes resulted in the increase in the VOC in the order PTB2PTB3PTB4.1, 2, 4, 53 Further, upon introduction of fluorine atoms on the 3 and 7-positons of the benzothiophene unit in the polymer, the HOMO energy level was lowered by 0.47 eV with a concomitant increase in the VOC by 0.16 V, as observed for PTBF2.53 Similarly, fluorine substitution on the thienothiophene unit in PTBF2 led to a decrease in the HOMO level; however, the VOC did not change significantly.53

You et al. also investigated the effect of varying electronics of the acceptor unit for a series of copolymers with BDT as a donor unit.11, 58 The replacement of the DTBT unit containing a benzene ring, with a DTPyT unit containing an electron-deficient pyridine unit led to a lowered HOMO energy level (−5.33 → −5.47 eV) with simultaneous increase in the VOC (0.83 → 0.85 V) for devices using PC61BM as an acceptor.11, 58, 62, 63 Similarly, upon replacement of the two hydrogen atoms at the 5- and 6-positions in the DTBT unit with two fluorine atoms in DTffBT on going from PBnDT–DTBT to PBnDT–DTffBT, the HOMO energy level decreased (−5.40 → −5.54 V) owing to the high electronegativity of the fluorine atom.58 On the other hand, the replacement of the DTBT unit with the HTAZ unit, the latter containing an electron donating N-atom, led to an increased HOMO energy level (−5.33 eV → −5.29 V).13 Further, the replacement of the two hydrogen atoms with the fluorine atoms on the benzotriazole unit upon going from PBnDT–HTAZ to PnDT-FTAZ led to a lowered HOMO energy level (−5.29 → −5.36 V).13

The tuning of the JSC can be similarly envisioned by varying the ELUMOD to provide the necessary driving force for overcoming the Coulombic interactions that an exciton must overcome to dissociate into free charges. It is now understood that ELUMOD for a donor–acceptor polymer is primarily determined by the acceptor unit, and hence, it could be varied without significantly affecting the EHOMOD.64–66 Also, it has been argued in the literature that the efficiency of a BHJ device is more sensitive to variation in ELUMOD than the band gap.45 Therefore, careful optimization of the ELUMOD could significantly improve the device efficiency. For example, the LUMO energy level of the polymers bearing the same acceptor unit were similar (∼3.4 eV), whereas their HOMO energies decreased linearly for PNDT–DTPyT, PBnDT–DTPyT, and PQDT–DTPyT (−5.2 → −5.4 → −5.5 eV). Thus, the HOMO and LUMO energy levels of a donor–acceptor polymer could be tuned by first varying the acceptor unit to get the desired ELUMOD, and then varying the donor unit, while keeping the acceptor unit the same, to get the desired EHOMOD. Recently, the Yu group demonstrated that the JSCs can also be tuned by tuning the dipole moment of the donor–acceptor polymer. For example, the JSCs for polymer/fullerene solar cells increased in the order PBB3PTBF2PTB2PTB7, following the increase in the difference between the ground and excited state dipole moments of the polymers.67 This result was supported by the long-lived excitonic state in PTBF2 and PBB3 relative to the PTB polymers, which led to a faster rate of recombination of the charges.67


Rigid and planar conjugated backbones in a semiconducting polymer are desirable due to the increased π–π stacking in the solid-state and the narrow band gap.68 On the downside, these structural features render the polymers insoluble in organic solvents, making it difficult to process them into thin-films that are required for solar cell applications.69, 70 One of the most common way to counter solubility and processability issues is to incorporate suitable alkyl or alkoxy side chains into the polymer backbone.71 However, as the sides chains are not photoactive, careful attention must be played while choosing them to avoid significant loss of photoactive real estate. Further, the properties of the side chains can affect the properties of a polymer beyond solubility or processability, affecting other important parameters such as miscibility with the acceptor and the thin-film morphology (Effect of Polymer Side Chain section), and the π–π-stacking distances. In this context, the effect of alkyl side chains for PTB polymers has been extensively studied.1, 2, 4, 53 Alkyl side chains attached to the BDT unit plays an important role in determining the π–π-stacking distance of PTB units. As the BDT unit has a larger aromatic ring size with two alkyl side chains on the TT unit, the BDT unit dominates the π–π-stacking ordering of the polymer in the film.21 Linear octyl chains on the BDT unit reduced the π–π-stacking distance of the polymer compared to the branched ethylhexyl chains, resulting in a strong π–π stacking of the polymers. As a result, PTB1-3 (PCE = 4.76–5.53%), bearing linear octyl side chains, exhibited a higher PCE compared to PTB4-6 (PCE = 3.10–2.26%), bearing branched side chains, when the devices were fabricated using PC61BM as an acceptor, in the absence of any additives. It turned out that the branched ethylhexyl chains on the TT unit exhibited optimum sterics, resulting in a better PCE for PTB2 and PTB3, at about 5.10 and 5.53%, respectively, relative to PTB1 (PCE = 4.76%) for devices fabricated using PC61BM as an acceptor. However, when highly branched butyloctyl chain was attached to TT, π–π-stacking distance in PTB6 was increased, and therefore, the PCE decreased to 2.26%.1, 2, 4, 53 The effect of polymer side chain on their miscibility with the fullerenes and the thin-film morphology was systematically studied using grazing-incidence wide-angle X-ray scattering (GIWAXS) for a series of PTB polymers.21 Interestingly, closer π–π-stacking distances led to higher fill factors (FF) in BHJ devices with fullerenes, suggesting that a shorter π–π-stacking distance favor charge transport across the active layer to the electrode (Fig. 8). Further, the grazing-incidence small-angle X-ray scattering (GISAXS) results also showed that the PTB polymers were successful in eliminating the formation of large PCBM domains, a serious issue often seen in other polymer/fullerene films. Similarly, ethylhexyl substitution on the electron-rich monomers such as diethylhexyl BDT, dithienosilole, and dithienogermole have been utilized to enhance the π–π stacking in polymers. The linear octyl side chains on the TPD unit enhanced the π–π-stacking ordering compared to the branched ethylhexyl side chain by virtue of their lesser steric bulk, which resulted in higher Jsc (5.5 → 10.6 mA cm−2) and PCE values (2.7 → 6.3%).15

Figure 8.

Correlation between fill factor and the π–π-stacking distance for the PTB/PCxBM BHJ solar cells: x = PC71BM (blue) and x = PC61BM (red). The line has been added to show the trend. (Reprinted from Ref. 21, with permission from Wiley.)

Solubility plays an important role in determining the device performance, for example, when linear hexyl and octyl side chains were attached to CPDT and TPD units, respectively, in the PCPDTTPD polymers, the corresponding device with PC71BM as an acceptor exhibited the best PCE of 6.41%. Short, linear butyl and hexyl alkyl chains on the TPD unit of the PCPDTTPD polymer resulted in poor solubility and lower PCE of 2.30 and 5.70%, respectively, similar to the effect of bulky ethylhexyl chains on the TPD units of the PCPDTTPD-EH polymer, resulting in low PCE of 5.71%.13 Similarly, Cao et al. also reported highly soluble PBDT-DTNT (and PBDT-DTBT) polymers containing 2,3-didecylthiophene substituted BDT and hexyl substituted DTNT (and DBT) units. Because of the strong steric hindrance between the alkyl chains of BDT and those of DBT, and the absence of such interactions in DTNT unit, the PCE for devices based on PBDT-DTNT was three times higher than that of PBDT-DTBT-based devices (2.1 → 6.0%).7 In the absence of an alkyl side chain on the electron-deficient comonomer, the solubility was compensated by incorporation of highly branched alkyl chains on the electron-rich comonomer. For example, POD2T-DTBT and PCDTBT polymers bearing highly branched 2-octyldodecyl and 9-heptadecanyl side chains, respectively, exhibited good solubility, and high PCEs at 6.26 and 6.1%, respectively.14, 50 Thus, the balance between solution processability and interchain ordering is greatly important to achieve an optimized photovoltaic performance.7


It is now known that an ideal BHJ architecture is one in which the donor and the acceptor phases achieve a nanoscale phase segregation involving an interdigitated network of donor–acceptor lamellae.40, 41, 72 The immiscibility of the polymer–fullerene blend often leads to the formation large domains of diameter exceeding the typical exciton diffusion length, rendering the photoharvesting process less than ideal. In this context, different approaches have been used to achieve optimal morphology, such as thermal72 and solvent42 annealing and usage of additives.44 The ultimate goal of these efforts is to control the domain size leading to the formation of interpenetrating network for an effective charge transport. The effectiveness of these approaches largely depends on the nature of donor polymers, for example, while the thermal annealing process was effective in P3HT/PCBM system, it failed completely in PTBx polymer system, leading to a dramatic decrease in device performance.73 It is now known that finer morphology (smaller domain size) and smooth surface (lower surface roughness) that can be achieved by using certain solvent additives, is crucial to the device performance.59, 74, 75 It has been hypothesized that such a morphology can not only facilitate exciton dissociation and but can also provide an easier route to percolation of the charge-carriers to their respective electrodes,59 which can improve the JSC, without significantly affecting the VOC.5, 74 Moreover, due to the simplicity of the process compared to thermal or solvent annealing,42, 72 the incorporation of solvent additives is emerging as a popular route for improving the organic solar cell device performance.

Though the exact mechanism of stabilization of polymer and fullerene phases in the thin-film by additives is not well understood, the choice of additive appears to be largely dependent on the miscibility of polymer–fullerene. Therefore, a number of additives such as alkanethiols,44 dichlorobenzene,74 diethylene glycol,76 1,8-diiodoalkane,59 1-chloronaphthalene,77 and so forth have been investigated for different polymer–fullerene systems. Transmission electron microscopy (TEM) analysis on PTB7 polymer showed obvious and large phase separation when it was blended with PC71BM in chlorobenzene (CB) without any additives, but a very uniform morphology was observed upon addition of DIO additive (Fig. 9).

Figure 9.

Left: TEM images of PTB7/PC71BM blend films: (a) prepared from chlorobenzene, and (b) prepared from mixed solvents chlorobenzene/diiodooctance (97/3, v/v). (The scale bar is 200 nm). (Reprinted from Ref. 4, with permission from Wiley.) Right: Cross-sectional TEM images of PSi:PC71BM and PGe:PC71BM-based PV cells without any additives (top) and with 5% DIO (bottom). (Reprinted from Ref. 17, with permission from American Chemical Society.)

Recent studies using GISAXS/GIWAXS and TEM studies have suggested an increased crystallinity of the polymer phase and a decreased size of PC71BM clusters for a PBTTPD/PC71BM solar cell upon addition of diiodohexane (DIH) (Fig. 9),9 leading to an improved PCE (5 → 7.3%).59, 78 Similarly, devices based on the active layer from isoindigo-based polymer (P3TI) and PC71BM that was processed with 2.5% (v/v) DIO in DCB resulted in a much finer phase separation compared to devices without the additive. This resulted in an enhanced JSC (10.5 → 13.1 mA cm2) and FF (0.63 → 0.69) and an overall increase in the PCE (4.8 → 6.3%).5 Amb et al. reported BHJ devices based on dithienogermole-based polymers P-Si and P-Ge as donors and PC71BM as an acceptor.17 TEM images of the devices revealed a dramatic difference in the extent of polymer–fullerene phase segregation and grain boundaries for devices processed with and without additives, with large PC71BM domains of dimensions 100–350 nm (lateral) and 45–65 nm (vertical) clearly seen for devices processed in the absence of additives (Fig. 9). A significant reduction in phase segregation was observed upon addition of 5% DIO for both P-Si and P-Ge polymers, which led to a large increase in the PCEs.17 Similar enhancement of the PCE has been observed for devices based on HSX-1/PC71BM devices.23

The extent of improvement of morphology of the active layer depends on the choice of additive and it's compatibility with the polymer fullerene system. For example, Li et al. reported a significant improvement in PCE (4.86 → 6.41%) for BHJ devices based on PCPDTTPD-Oc/PC71BM upon addition of 4.2% of DIO in DCB.2 A very similar polymer, PBDTTPD, bearing an octyl group on the cyclopenta[2,1-b;3,4-b′]-dithiophene (CPDT) unit and a 2-ethylhexyl group on the thieno[3,4-c]-pyrrole-4,6-dione (TPD) unit, did not show any significant change in morphology of the films with PC71BM when processed with 1-chloronaphthalene as an additive.77 Thus, compatibility of the additive with the polymer/fullerene system appears to be crucial for the improvement of film morphology. Also, the impact of an additive on the PCE could be less significant for a polymer–fullerene thin-film blend exhibiting high degree of miscibility and order. For example, N-alkylthieno[3,4-c]pyrrole-4,6-dione-based polymers with straight chain side groups, P1, tend to form a more ordered structure than the corresponding branched chain polymers P2 and P3, and hence addition of DIO only marginally improved the PCE for devices based on P1 (6.4 → 6.8%) when compared to those based on P2 and P3 (2.8 → 4.0% and 3.9 → 5.7%).15

The importance of nanoscale phase segregation over other parameters such as VOC and hole mobility can be seen realized from the fact that among thienothiophene polymers, despite the highest VOC and hole mobility values exhibited by PTB4, it did not show highest PCE in the solar cell devices.1 This has been attributed to the formation of large domains, as observed in the TEM analysis, due to the poor miscibility of the polymer with the fullerene acceptor owing to the bulky side chains (Fig. 9).1 The morphology was fine-tuned using a mixed solvent system consisting of 3 vol % DIO in DCB, which eliminated the large domains and led to finer structures, and an improved PCE (3.10 → 5.90%). On the contrary, a higher degree of organization in polymer–fullerene phase does not guarantee a higher PCE. For example, among thin-films of PC71BM with PTB-5, PBDTT, and PBDTT-CF, while PTB-5/PC71BM and PBDTT-C/PC71BM showed a fibrillar features, PBDTT-CF/PC71BM exhibited domains of different shapes and a lower degree of organization in the AFM analysis, but still recorded a PCE of 7.73%, much higher than those from PTB-5/PC71BM and PBDTT-C/PC71BM devices.1, 2, 4, 21, 22

The effect of additive also depends on the choice of casting solvent. The most dramatic effect of additive was observed for devices based on PDTSTPD (P-Si)/PC71BM, in which addition of 3 vol % DIO to the CB solution resulted in increase in the PCE from <1 to 7.3%, while no such improvement was reported for devices processed from DCB.79 TEM analysis on active layer from DCB and CB with DIO were significantly different, whereas the former showed large domains (ca. 100–200 nm in diameter), the presence of uniform phase segregation and the absence of large domains explained the higher PCE observed for the latter.78 Similarly, when 3.0% DIO was used as an additive in PTB7/PC71BM devices, active layer processed from CB showed a huge increment in the PCE (3.92 → 7.40%) compared to DCB (3.92 → 7.40%).

The effect of additive also appears to be dependent on the chain length of the additive. For example, among diiodobutane, DIH, and DIO that were investigated as additives in PBTTPD/PC71BM solar cells, devices utilizing 0.5 vol % DIH outperformed others by virtue of the improved JSC (9.1 → 13.1 mA cm2) and FF (0.58 → 0.61).59, 78 Thus, several parameters simultaneously affect the phase segregation in the polymer–fullerene active layer.


The charge-injection barrier between ITO electrode and active layer arising from the low work function of the untreated ITO is detrimental to the device performance.80 A robust interlayer could minimize the series resistance and careful optimization of the Fermi level could lower the Schottky barrier.81 In this context, several interlayer materials such as poly(3,4-ethylenedioxylenethiophene):poly(styrene sulfonic acid) (PEDOT:PSS),82 carbon nanotubes,83 transition metal oxides,84 and self-assembled monolayers (SAMs)85 have been used with varying degrees of success. Further, to improve the durability of polymer solar cells, several new interlayers including graphene oxide (GO) and MoOx have recently emerged.86, 87 In PTB7/PC71BM solar cell devices, device fabricated with GO interlayer registered a 20-fold increase in lifetime under humid atmosphere relative to PEDOT:PSS-based devices, by the virtue of inertness of the GO film against the environment.86 Similarly, in PCDTBT:PC71BM solar cells, devices fabricated using MoOx interlayer exhibited a high degree of environmental stability relative to those based on PEDOT:PSS; while 50% of the original PCE for the MoOx-based devices was retained after exposure to air for 720 h, the PCE for PEDOT:PSS-based devices degraded by 50% of the original value in just 16 h.87 Other metal oxides such as NiO have also shown improvements in the solar cells stability. For example, P3HT/PCBM-based solar cells, those with a NiO interlayer retained 90% of the PCE after air exposure for 60 days.88

At the other active-layer/electrode interface, LiF and Ca continue to be the most popular choice.89 N-type semiconducting titanium suboxide (TiOx)8 and zinc oxide nanoparticles (ZnO NPs) assisted with a SAM layer90 that have been used as optical spacers and hole-blocking layers have significantly improved PCEs of the polymer solar cells. Optical spacers can enhance the absorption efficiency with fixed thickness of active layer and result in an increased photocurrent.91 For example, Heeger et al. optimized the performance of BHJ devices based on PCDTBT by introducing a TiOx layer between the aluminum and the active layer, which enhanced the optical absorption without requiring thicker active-layers. The PCE of the device increased to 6.2% (5.96% certified by NREL) upon introduction of the TiOX layer.8 The enhancement of the performance was warranted by an IQE approaching 100%, signifying better charge-carrier generation and collection. Further, as TiOX is usually electron deficient, they offer a high degree of stability toward air and moisture.92 Polymeric TiOx layer synthesized by hydrolysis of titanium (IV) isopropoxide followed by sol–gel reaction exhibited long-term stability even at temperatures as high as 150 °C.93 Similar to the effect of TiOx interlayer, CrOx also showed remarkable improvement in the air-stability of the devices.94 For P3HT/PCBM solar cells, devices based on CrOx retained 50% of the original PCE after 6 days of exposure to ambient air, in comparison, devices based on LiF interlayer lost about 90% of the original PCE in just 5 h.94

As a lesson from the organic light-emitting diodes, thin layer of a conjugated polyelectrolyte in the device was found to enhance the efficiency.95 For example, Seo et al. reported enhancement of PCE (5.0 →6.5%) upon introduction of an alcohol soluble ionic conjugated diblock copolymer PF2/6-b-P3TMAHT between the polymer/fullerene active layer and the metal cathode.6 Similarly, He et al. achieved very high PCE up to 8.37% based on our PTB7/PC71BM system by incorporation of another alcohol/water-soluble conjugated polymer, PFN, as an interlayer between the polymer/fullerene active layer and the electrode.19 It is believed that the interlayer enhances the built-in potential across the device due to the presence of an interface dipole, which would improve the charge-transport properties, eliminate the build-up of space-charges, and reduces the recombination losses due to the increase in built-in field and charge-carrier mobility (Fig. 10).19 Moreover, the facile fabrication of CPE layer by solution-processing also indicated its promising application in commercial solar cell devices.

Figure 10.

The energy band structure of the devices without (a) and with (b) the CPE interlayer. (Reprinted from Ref. 19, with permission from Wiley.)


Practical limitations have put a theoretical limit to the single device efficiency at 10%.27, 28, 45 In this context, tandem solar cell devices, which consists of two or more solar cells connected in series or parallel, could be a promising solution.28, 96–98 The connection between the two devices is achieved using a recombination layer, an optically transparent layer that facilitate the alignment of quasi-Fermi levels of the acceptor of the first cell with the donor of the second cell while offering minimal resistance (Fig. 11).99 These devices can be designed from individual cells that absorb different windows of the incident solar spectrum, which can significantly broaden the absorption window.100 Also, the order of stacking of the individual cells can be decided based on the penetration depth of the incident radiation and the polymer band gaps (Fig. 11).9, 100 The individual cells could be arranged in series to give a cumulative VOC or arranged in parallel to give a cumulative JSC,96–98 which can provide a tool for optimizing the overall device efficiency.

Figure 11.

Strategic design rule for a tandem device consisting of two cells with different band gaps.

Tandem cells were originally developed by Hiramoto et al.,101 and since then, both vacuum-deposited102 and solution-processed96 active layers have been used to fabricate tandem devices. However, semitransparent thin-layer of metal that were used as the recombination layer in these devices led to optical losses.103 Recently, Gilot et al. reported solution-processable ZnO NPs as the recombination layer, which paved the path for an all-solution processed tandem devices.96 Recently, Yang et al. reported a tandem solar cell with efficiency approaching 7.0% using modified PEDOT:PSS as an interconnecting layer.20 Interestingly, Heeger et al. reported tandem solar cells and evaluated the robustness of the system using PCPDTBT/PCBM and poly(3-hexylthiophene)/PCBM as active layers for the two cells, and a sol–gel processed TiOx as the recombination layer. The tandem device exhibited a VOC of 1.24 V and a PCE of 6.5%, a significant improvement over the single cell devices.9 The reliability of the technology can be accessed from the fact that the efficiency of the tandem devices dropped (6.5 → 5.5%) by 15% in 3500 h even under nitrogen atmosphere. When exposed AM1.5G radiation, the efficiency dropped by about 30% after 40 h and by over about 40% after 100 h of continuous operation. Though these results are encouraging, the challenge remains to develop more robust systems with higher stability and longer life times.


With PCE exceeding 10.0%,19, 26–28 the field of organic solar cells continues to evolve, and this technology is now transferring from research labs to industrial fabrications. In order for this technology to be commercially viable, the cost per kilowatt that it can provide must be competitive with the available technologies, which requires a significant improvement in the PCE and the long-term stability. However, the PCE is always less than ideal due to a number of simultaneously occurring unfavorable phenomena, some of which are unavoidable. For example, optical loss arising due to photons with energy insufficient for photoexcitation or those with excessive energy leading to thermal loss are detrimental to the device efficiency.104 Even excitons generated after photons are effectively absorbed can radiatively or nonradiatively decay.104 Also, while the offset (EHOMODELUMOA) sets the upper limit of the VOC, the observed values are always lower than that owing to the structural relaxation and the field needed to drive the charge separation, and the loss in the VOC arising from the dark-current.45 Overall, the thermodynamic limit to solar cell efficiency in inorganic semiconductors have been put forward as 31 and 44% by the Shockley–Queisser limit and the “impact-ionization” theories, respectively.105

On the bright side, we now know the optimum band gap of the polymer should be less than 1.50 eV, the LUMO level less than −3.92 eV to match with fullerenes, and the hole mobility should be above ∼10−3 cm2 V−1 s−1.45 It has been proved possible to systematically tune the energy level of polymers via introduction of electron withdrawing groups on the acceptor units of the donor–acceptor polymers. Also, as the ELUMOD for a donor–acceptor polymer is primarily determined by the acceptor unit, it is now possible to vary the ELUMOD without significantly affecting the EHOMOD.64–66 Moreover, the efficiency of a BHJ device appears to be more sensitive to variation in ELUMOD than the band gap.45 We know that several parameters simultaneously affect the phase segregation in the polymer–fullerene active layer including chain length of the additive, casting solvent, chain length on the copolymer, and the extent of miscibility of the polymer–fullerene blend. Therefore, more efficient devices could be developed by (a) designing better materials;2, 3, 106 (b) developing better understanding of the device physics;32, 33 (c) improving processing techniques; (d) engineering better device architectures, among others. Further, while designing materials for organic solar cells, careful attention must be paid to ensure that the materials (a) can be synthesized in high purity at a low cost and be scalable; (b) have flexibility to make structural changes to tune their ionization potentials and electron affinities to optimize both band gap and VOC, coherently and/or independently; (c) possess good solubility in commonly used organic solvents; (d) are processable into thin-films that are thermally, chemically, and photochemically stable over a long period of time; (e) exhibits a high absorptivity and a panchromatic absorption; (f) are miscible with the acceptor and result in a fine phase segregation comprising of a bicontinuous donor and acceptor phases in thin films,40, 41 (g) exhibit optimum π−π stacking and crystallinity to ensure high charge-carrier mobility, (h) have a low series resistance, to match the high electron mobility of fullerene materials.38, 39 These desired properties must be synergistically integrated into one polymer system so that the PCE can be seamlessly optimized.


The authors acknowledge the financial support from NSF, AFOSR, DOE, NSF-MRSEC (the University of Chicago), Intel, and Solarmer Energy Inc., and Samsung Electronics for the preparation of this article.

Biographical Information

original image

Luping Yu received his B.S. degree in chemistry in 1982 and an M.S. degree in polymer chemistry in 1984 from Zhejiang University, Hangzhou, People's Republic of China. He earned his Ph.D. degree in 1989 and continued his postdoctoral training in the research group of Prof. L. R. Dalton at the University of Southern California. In 1991, Yu joined the Department of Chemistry at the University of Chicago as an assistant professor and is currently a Professor of Chemistry.

Biographical Information

original image

Mayank Mayukh received his bachelor's degree in Chemistry from Hindu College, University of Delhi, and master's degree in Chemistry from Indian Institute of Technology Delhi in 2004. After earning his Ph.D. degree in the research group of Prof. Dominic V. McGrath at the University of Arizona in 2011, he moved to the University of Chicago, where he is currently working as a postdoctoral scholar in the research group of Prof. Luping Yu.

Biographical Information

original image

Feng He received his B.E. degree in 2002 and Ph.D. degree in polymer chemistry in 2007 from Jilin University, China, under the supervision of Prof. Yuguang Ma. Then he moved to the University of Toronto, Canada, for his postdoctoral research with Prof. Mitchell A. Winnik. In 2009, he joined Prof. Luping Yu's group as a postdoctoral scholar at the University of Chicago, where his research interests are in the synthesis of low-band-gap copolymers for organic solar cell applications.

Biographical Information

original image

In Hwan Jung received his B.S. degree in 2006 and Ph.D. degree in 2011 in Chemistry from Korea Advanced Institute of Science and Technology (KAIST), South Korea, under the supervision of Prof. Hong-Ku Shim. He is currently working as a postdoctoral researcher in Prof. Luping Yu's group at the University of Chicago from 2011.