2.1.

Polymeric Donors

2.1.1.

Poly (2,7-carbazoles)

The electron-donating nitrogen unit of the central fused pyrrole ring makes carbazoles electron-rich compounds. The solubility of the polymer is ensured by functionalization of the central nitrogen with an alkyl chain (see molecular structure in Fig. 2). Since carbazole derivatives present good thermal and photochemical stability and high charge mobility, they are promising candidates to be incorporated in polymers for photovoltaic applications.^22,34,35

Fig. 2

Molecular structure of 2,7-carbazole.

The conjugation of the carbazole unit to a benzothiadiazole moiety through a thiophene bridge gives rise to a material known as PCDTBT, see Fig. 3. In combination with ${\mathrm{PC}}_{70}\mathrm{BM}$ as an acceptor, organic solar cells with power conversion efficiencies $>6\%$ have been achieved by several groups.^36–38 These numbers can be improved to 7.2 to 7.5% by using advanced interface materials and antireflection coatings.^27,39

Fig. 3

Highest reported efficiencies for different polymers in single devices.

A slightly different approach has been attempted by conjugating the benzothiadiazole moiety to a dithiophene unit instead of a carbazole one. The result is a polymer known as PCPDTBT (Fig. 3). Organic solar cells with efficiencies up to 5.5% are usually reported for this polymer when it is blended with ${\mathrm{PC}}_{70}\mathrm{BM}$.⁴⁰ Moreover, by introducing a fluor atom into this molecule (Fig. 3), which lowers the increase of the polymer HOMO level and, thus, the $V_{\mathrm{oc}}$, efficiencies of 6.16% have been also achieved.⁴¹

A further modification of this copolymer with two fluor atoms at the benzothiadiazole unit gives rise to a difluorobenzothiadiazole (DFBT) moiety that is further conjugated to a dithienopyran (DTP) segment instead of a single dithiophene. The result is a copolymer known as GPDTP-DFBT (Fig. 3) with efficiencies in single devices of $>8\%$ (Ref. 42) and also an outstanding performance in tandem configuration, as we will see in Sec. 4.3. A more developed semicrystalline version known as PPDT2FBT (Fig. 3) forms a well-distributed nanofibrillar networked morphology with the fullerene, which results in balanced hole and electron mobility, and tight interchain packing. Relatively thick films of $\sim 300\mathrm{nm}$ yield record efficiencies of 9.4%.⁴³

2.2.

New Fullerene Derivatives

Nowadays, almost all reported cells with remarkable efficiencies employ soluble fullerene derivatives as an electron acceptor. Among them, ${\mathrm{PC}}_{60}\mathrm{BM}$ and ${\mathrm{PC}}_{70}\mathrm{BM}$ are the most widely used. Both of them have the same LUMO values, which considerably limit the $V_{\mathrm{oc}}$ of operating devices.

A new n-type fullerene derivative, indene-${\mathrm{C}}_{60}$ bis-adduct (ICBA) (see Fig. 4 for the molecular structure) was demonstrated to be a good alternative acceptor. ICBA has a higher LUMO level ($-3.74\mathrm{eV}$) in comparison to either ${\mathrm{PC}}_{60}\mathrm{BM}$ or ${\mathrm{PC}}_{70}\mathrm{BM}$ ($-4.2\mathrm{eV}$), which leads to a higher $V_{\mathrm{oc}}$. In this way, P3HT based devices with ICBA as an acceptor resulted in cells with $V_{\mathrm{oc}}$ values of 0.84 V and enhanced PCE of 6.5%.^66,67

Fig. 4

Molecular structure of indene-${\mathrm{C}}_{60}$ bis-adduct.

Cheng et al. also reported P3HT based inverted solar cells using ICBA as an acceptor with a $V_{\mathrm{oc}}$ of 0.82 V and a PCE of 4.8%. Further improvement was carried out by the incorporation of a cross-linked fullerene derivative interlayer, C-PCBSD, resulting in the following configuration: ITO/ZnO/C-PCBSD/ICBA:P3HT/PEDOT:PSS/Ag. The incorporation of C-PCBSD increased the photocurrent from 10.6 to $12.4\mathrm{mA}{\mathrm{cm}}^{-2}$ and the FF from 55 to 60%, and, thus, pushed up the PCE to 6.2%.⁶⁸

Many other fullerene derivatives such as bisadducts,⁵ diphenylmethano fullerenes,⁶⁹ dimethylphenylmethano fullerene bisadducts,⁷⁰ and endohedral fullerenes^71,72 are also being synthesized and tested in the search for large $V_{\mathrm{oc}}$ devices based on internal bulk polymer:fullerene heterojunction systems. In spite of these advances in the synthesis of new fullerenes, and even though soluble ${\mathrm{C}}_{70}$ derivatives, on average, yield a 10% increase in photocurrent in comparison to ${\mathrm{C}}_{60}$ ones thanks to a slightly increased absorption in the visible, the lack of a stronger absorption in the visible and in the infrared still hinders further developments in achieving higher conversion efficiencies. Non fullerene containing heterojunction thin films can overcome these shortcomings, since the energy levels (and, hence, absorption) can be modified through the choice of co-couple units due to the well-distributed frontier molecular orbitals.^73,74

3.1.

Use of Additives

Additives are used for the generation of a more favorable nanomorphology for the transport of electrons and holes, thereby enhancing the final efficiency of the device. Additives do not react with the polymer or with the fullerene; the only aim is to favor the bicontinuous percolation pathways for exciton dissociation and charge transport. Two basic requirements for the correct choice of an additive are (1) the boiling point of the additive has to be much higher than the primary solvent and (2) only one component will be selectively dissolved in the additive. For example, considering polymer:fullerene blends, only the fullerene is selectively dissolved in the selected additive.¹⁷

Some of the most studied additives are based on alkanedithiols. Due to the ability of alkanedithiols to selectively dissolve the fullerene component while the polymer is less soluble, an optimum nanomorphology is formed.¹⁷ Peet et al. compared the performance of P3HT and PCPDTBT devices processed from pristine solvents and with blends of the same primary solvent and different alkanedithiols. They reported improvements in the PCE from 2.8 to 5.5% when 1,8-octanedithiol was added to chlorobenzene (CB).⁴⁰

1,8-di(R)octanes with different functional groups (R) are also often used as additives in order to favor the nanomorphology of the film, which improves device efficiency. Figure 6 represents the influence of DIO on the final morphology of the film. When no additives are used, there are big aggregates of fullerene and the penetration of the acceptor molecules in the polymer network is more difficult, resulting in large fullerene and polymer domains. On the contrary, DIO is a much better solvent for the fullerene than CB while the polymer has limited solubility in DIO and is very well dissolved in CB. With the addition of DIO in this case, the fullerene is selectively dissolved, facilitating the percolation of acceptor molecules in the polymer network, resulting in a more favorable nanomorphology of the film due to the optimization of the domain size and polymer:fullerene interface.¹⁸

Fig. 6

Graphical sketch of polymer and fullerene when using only chlorobenzene as a solvent (a) and with diiodooctane as an additive (b).¹⁸

The work carried out by Liang et al. for devices based on PTB7 show the influence of DIO on device performance when using it as additive.⁵⁵ Preliminary studies showed that $\mathrm{PTB}7:{\mathrm{PC}}_{70}\mathrm{BM}$ ($1:1.5$) films prepared from ODCB and DIO (97%:3% in volume) increased the FF from 60.25 to 68.9% and the PCE from 6.22 to 7.18%, in comparison to devices processed from pristine ODCB. In this case, the photocurrent remained constant. However, when CB was used as the primary solvent, a considerable increment was also observed in the photocurrent (from 10.2 to $14.5\mathrm{mA}{\mathrm{cm}}^{-2}$). Also, the FF rose from 50.52 to $\sim 69\%$. In this way, it was possible to improve the efficiency from 3.92% (when CB was used as solvent) up to 7.4% when DIO was used as an additive. All these data are collected in Table 1. They also studied the influence of the additive on the morphology of the active layer by transmission electron microscopy. Images revealed that DIO promotes the formation of smaller domains. The film also showed a higher uniformity due to the good miscibility between PTB7 and ${\mathrm{PC}}_{70}\mathrm{BM}$ and the formation of an interpenetrating network. In this way, higher FF and photocurrent values were achieved, thus increasing the PCE.

Some other works have also shown the potential of DIO as an additive to improve the performance of devices made from different active materials. Lee et al. demonstrated an enhancement from 3.4 up to 5.1% by using DIO in BHJ organic solar cells based on $\mathrm{PCPDTBT}:{\mathrm{PC}}_{70}\mathrm{BM}$ systems.¹⁶ Zhang et al. reported a PCE increment from 1.4 to 4.8% when DIO was used as the additive to process devices from an LBG polymer consisting of thieno-thiophene substituted BDT (PTTBDT-C8).⁷⁶ Finally, Bijleveld et al. also documented the use of DIO for processing devices with DPPs.²⁹ They also observed an enhancement in the PCE from 2 to 5.6% and substantial changes in the film morphology when DIO was added to chloroform.

3.2.

Solvent Mixtures

Similarly to additives, the combination of two or more solvents with different boiling points and limited solubility for one of the components can help control the nanomorphology of the polymeric blend. Janssen et al. reported the influence of adding ODCB to chloroform in a series of DPP based device performance.^44,77 The analysis was based on a narrow bandgap polymer known as pBBTDPP2, in combination with ${\mathrm{PC}}_{60}\mathrm{BM}$ or ${\mathrm{PC}}_{70}\mathrm{BM}$ diluted in chloroform (${\mathrm{CHCl}}_3$), ODCB, or a mixture of both solvents. When the polymer:fullerene was dissolved in chloroform, just 1.1% of PCE was achieved. This low performance was attributed to the amorphous nature of the deposited layer. When using ODCB instead, due to the limited solubility of pBBTDPP2 in this solvent, after spin-casting the solution, films with larger degrees of crystallinity were obtained, which resulted in devices with higher photocurrent values and, thus, higher PCEs, 2.9%. The PCE of pBBTDPP2 based devices was significantly improved by combining chloroform and ODCB. With the combination of these two solvents in a ratio of $4:1$, a large amount of semicrystalline polymer film was obtained. In this way, PCE values of 3.2 and 4% were achieved when using ${\mathrm{PC}}_{60}\mathrm{BM}$ and ${\mathrm{PC}}_{70}\mathrm{BM}$ as electron acceptors, respectively. Due to the large difference in vapor pressure, the final film morphology was essentially determined by the slow evaporation of ODCB, leaving the polymer sufficient time to partly crystallize before precipitation. Cells from chloroform:ODCB, apart from higher photocurrent values, also provided higher FFs with respect to the other two options (see Table 2).

Table 2

Photovoltaic parameters of pBBTDPP2:PCBM solar cells.44

Zhang et al. also observed a significant enhancement in the photocurrent density for a polyfluorene copolymer/fullerene blend. When introducing a small amount of CB into the chloroform solvent, a uniform domain distribution was reached, resulting in more efficient devices.⁷⁸ In further studies, 1-chloronaphthalene in ODCB or small amounts of nitrobenzene in CB have also been used as additives for P3HT:PCBM based devices in order to improve device performance with the aim of achieving a higher degree of crystallinity.^79,80

3.3.

Postprocessing Treatments

The possibility to act on film morphology during and after the deposition process has also been thoughtfully studied. First, it is possible to quite accurately control the drying time of the films by using alternative depositing techniques to spin-coating, as, for example, Dr. Blading. Based on the work carried out by Li et al. about the slow drying for film nanomorphology control in conventional architecture devices,¹¹ Ajuria also slowed down the drying process of the P3HT:PCBM photoactive film in inverted configuration devices [600-nm-thick photoactive layer, see Fig. 7(a)].⁸¹ Film drying times were delayed from 1 to 2 s to $\approx 10$ and $\approx 30\mathrm{s}$ for films dried directly in air, protecting the film with a Petri dish as a lid in order to limit the contact of the film with air and create a solvent atmosphere below the Petri plate to minimize the presence of air, respectively.

Fig. 7

(a) Field emission scanning electron microscopy cross-sectional view of an inverted ITO/ZnO/P3HT:PCBM/PEDOT:PSS/Au solar cell. Different layers and their thickness are depicted. (b) $JV$ curves of different photoactive layer drying techniques.⁸¹

Results revealed that the slow drying of the film assists the growth of a photoactive self-ordered nanostructure, improving the PCE from 1.44 to 3.57% [see Fig. 7(b) and Table 3]. The self-organization of the polymer has been shown to improve field effect carrier mobilities by more than a factor of 100 in P3HT.^82,83 On the contrary, the destruction of self-organized structures during fast drying (unordered growth) present unbalanced charge carriers and lower charge mobilities.¹¹ Under these precisely controlled drying conditions, it is feasible to increase the thickness of the active film up to 600 nm, thus maximizing the light absorption without negatively affecting charge transport. Raising the Dr. Blade plate temperature from ambient conditions to 70°C had a similar effect.

Table 3

JV characteristics of different photoactive layer drying techniques.81

Alternatively, some other approaches to later manipulate the bulk morphology once the film has been already deposited from solution have been attempted with successful results. Zhou et al. spin-coated methanol on top of an already dry active layer comprising $\mathrm{PTB}7:{\mathrm{PC}}_{70}\mathrm{BM}$.⁸⁴ They reported simultaneous improvements on device series resistance, charge mobility, charge recombination, and charge extraction mainly by means of surface modification. All these improvements gave rise to enhanced $J_{\mathrm{sc}}$ and FF, which lead to 7.9% PCE devices in comparison to 7.1% for nontreated films.

Finally, a novel approach explores the viability of vapor printing as a fast postprocessing technique.^85,86 A carrier gas transporting the vapor solvent is delivered through a nozzle promoting local self-assembly of polymer chains. This enables finding an optimal nanostructure in promisingly short times. Changes in the degree of crystallinity led to a twofold increase in PCE with respect to as-cast samples.

4.1.

Alternative Interlayers

An efficient charge extraction process following light absorption and charge generation requires the use of conducting electrodes wisely chosen in order to have energetic levels alignment and, hence, ideally ohmic contacts between the metal and the semiconductor, thus avoiding charge barriers and undesired losses. Depositing an extra buffer layer between the photoactive film and the metallic electrode has been demonstrated as an efficient way to improve the contact properties.

Indium tin oxide (ITO) is typically used as the bottom contact (anode) in conventional configuration organic solar cells. ITO has a high work function around 4.8 eV, which is very dependent on washing treatments.^91,92 However, these variations and the roughness of the resulting films usually result in important contact losses.⁹³ In order to improve the quality of this contact and favor the formation of an ohmic contact, a p-type PEDOT:PSS interlayer, which has a work function of 5 eV, has been traditionally used.⁹⁴ Due to the acidic nature of PEDOT:PSS and its tendency of being easily degraded in contact with either air or moisture and negatively affecting the stability of the whole device, different transition metal oxides, such as ${\mathrm{V}}_2{\mathrm{O}}_5$, ${\mathrm{MoO}}_3$, ${\mathrm{WO}}_3$, and NiO, which are considered more stable, were introduced as alternative p-type interlayers in relatively highly efficient and stable devices.^27,95,96 In the cathode side, instead, calcium was initially inherited from the organic light emitting diode (OLED) development as the hole blocking layer due to its low work function. In spite of the good results obtained for photovoltaic devices in combination with silver, it is, however, very reactive to oxygen and moisture and, thus, device stability is further jeopardized. Therefore, alternative inorganic low work function interlayers were introduced. On one side, inorganic salts, such as LiF,^97–99 CsF,^100,101 and MgF,¹⁰² have been vacuum evaporated in combination with aluminum in order to enhance the contact selectivity and, hence, the PCE. It is believed that due to their high internal dipole moment, thin layers of these materials are able to alter the electrode work function by inducing a shift of the vacuum energy level. Regardless of the increase in efficiency obtained with the use of these materials, they are principally vacuum deposited, which hinders the way toward lower-cost devices based on R2R processing. Solution processable alternatives make use of salts like ZnS,¹⁰³ LiAc,¹⁰⁴ or ${\mathrm{Cs}}_2{\mathrm{CO}}_3$.^105–107 Furthermore, polymeric salt compounds, known as polyelectrolytes, have been used with promising results.¹⁰⁸ Also, small molecules with opposing internal charges, zwitterions, have been demonstrated as interfacial layers in organic solar cells.^109,110 Khan et al. demonstrated efficient conventional configuration P3HT:ICBA based solar cells substituting the Ca layer with a thin film of a hydrophilic polymer processed from a polyethylenimine, 80% ethoxylated solution, PEIE.¹¹¹ As a result, devices with a $V_{\mathrm{oc}}$ of 0.78 V, $J_{\mathrm{sc}}$ of $9.1\mathrm{mA}{\mathrm{cm}}^{-2}$, FF of 65%, and PCE of 4.6% were achieved.

On the other side, as previously commented on for the anode, solution processable films of n-type inorganic semiconductors, such as ${\mathrm{TiO}}_x$ and ZnO, processed from nanoparticle dispersions are typically used as n-type interlayers.^112–115 Nowadays, many research groups, mainly motivated by the improvement observed in the FF and the parallel resistance, have focused their investigation on alternative interlayers that can further enhance the device’s overall performance.

As it was shown in Sec. 3.1, Liang et al. improved the PCE of $\mathrm{ITO}/\mathrm{PEDOT}:\mathrm{PSS}/\mathrm{PTB}7:{\mathrm{PC}}_{70}\mathrm{BM}/\mathrm{Ca}/\mathrm{Al}$ based organic solar cells from 3.92 to 7.40% by adding 3% of DIO to the primary solvent CB.⁵⁵ However, more specifically related to the use of effective interlayers, further improvement was carried out by He et al. when a thin polymeric film of poly [(9,9-bis(3´-(N , N—dimethylamino) propyl)-2,7-fluorene)- alt -2,7-(9,9–dioctylfluorene)] (PFN) was incorporated as the cathode interlayer, increasing the PCE to 8.22%. Significant and simultaneous enhancements in $J_{\mathrm{SC}}$, $V_{\mathrm{oc}}$, and FF were observed.¹¹⁶ PFN is an alcohol/water soluble conjugated polymer that creates an interfacial dipole between the PFN and the polymeric $\mathrm{PTB}7:{\mathrm{PC}}_{70}\mathrm{BM}$ blend. Moreover, the incorporation of PFN is believed to exert a strong electric field at the active layer/cathode interface, which may strongly influence charge transport and extraction. Thus, the effects of the interlayer on the improvement of device performance were due to improved charge-transport properties, reduction of any possible space charge effect at the interface, and reduced surface recombination losses.¹¹⁷ Regarding dipole formation at this interface and related to morphology issues, which were previously addressed in Sec. 3.3, it has been recently suggested that vertical segregation of the fullerene content toward the cathode can also result in interfacial dipoles of varying strength, which affect the selectivity of the contact in a similar way as buffer interlayers.^118,119

Alternatively, Martínez-Otero et al. incorporated bathocuproine (BCP) as the interlayer of the cathode to achieve an optimal optical interference for PBDTTT-C and PTB7 based devices. As a result, for devices processed with BCP as the interlayer instead of Ca, an increment in PCE from 6.3 to 7.5% was observed for PBDTTT-C based cells and from 7.4 to 8.1% for the PTB7 based ones.⁸⁸ In the case of PTB7, the increment in PCE was related to the increase in photocurrent due to the enhanced reflectivity of the buffer layer/electrode back contact. For the PBDTTT-C instead, apart from the photocurrent improvement, the FF was also enhanced from 60.3 to 67.9%, which was related to the reduction of recombination at the interfaces.

4.2.

Inverted Configuration

As previously commented on at the beginning of Sec. 4, inverting the polarity of the device and processing the electron collecting electrode onto the ITO and the hole collection electrode on top of the active layer has also been successfully used to increase the efficiency of devices made with the same active materials. Ajuria et al. documented this effect for P3HT:PCBM devices for which impressive FFs $>70\%$ were reported for inverted designs.¹²⁰

In the case of PTB7 devices, further improvement in PCE was observed by He et al. when inverting the polarity of the device, reaching a very remarkable 9.2%.⁸⁷ As it can be seen in Fig. 8 and Table 4, the efficiency improvement for inverted structure devices is mainly due to the larger $J_{\mathrm{sc}}$ with respect to the conventional one, $17.2\mathrm{mA}{\mathrm{cm}}^{-2}$ in comparison to $15.4\mathrm{mA}{\mathrm{cm}}^{-2}$, respectively. The drastically enhanced $J_{\mathrm{sc}}$ for inverted devices was assigned to an improved ohmic contact, which eases photogenerated charge carrier collection and an optimum photon harvesting of these devices.

Fig. 8

Performance for inverted configuration (red filled symbols) ($\mathrm{ITO}/\mathrm{PFN}/\mathrm{PTB}7:{\mathrm{PC}}_{70}\mathrm{BM}/{\mathrm{MoO}}_3/\mathrm{Al}$) and conventional configuration (black open symbols) ($\mathrm{ITO}/\mathrm{PEDOT}:\mathrm{PSS}/\mathrm{PTB}7:{\mathrm{PC}}_{70}\mathrm{BM}/\mathrm{PFN}/\mathrm{Ca}$) devices. (a) Current density-voltage measurements and (b) external quantum efficiency (EQE) graph.⁸⁷

Table 4

Best device performance/parameters for PTB7:PC70BM based conventional and inverted solar cells with PFN as interlayer.87

The interlayers used for inverted configurations can also be exposed to different treatments in order to modify their intrinsic properties and try to improve device efficiency. Adhikary et al. improved $\mathrm{PBDTTT}\text{-}\mathrm{CT}:{\mathrm{PC}}_{70}\mathrm{BM}$ based inverted configuration devices from 6.46 to 8.34% by exposing the cell to UV-ozone.⁵⁸ This treatment modifies the wurtzite phase crystallinity of the ZnO films, leading to faster film mobilities and improved charge extraction properties. While exposure times around 5 min resulted in an ideal crystalline structure, longer exposure times induced the formation of p-type defects, pushing the ZnO Fermi-level further away from the vacuum level and decreasing the wurtzite crystallinity.

The work function of the materials can also be manipulated by surface modifiers. Zhou et al. demonstrated that compounds based on polymers containing simple aliphatic amine groups can substantially reduce the work function of conductors, including metals, transparent conductive metal oxides, conducting polymers, and graphene.¹²¹ Kyaw et al. incorporated PEIE on top of ZnO to enhance the efficiency of the cell by lowering the work function of ZnO (from 4.5 to 3.8 eV).¹²² With this surface modification, the PCE was enhanced from 6.29 to 7.88% thanks to simultaneous enhancements in $J_{\mathrm{sc}}$, $V_{\mathrm{oc}}$, and FF.

4.3.

Tandem Cells

The main losses that occur in a solar cell are transmission and thermalization losses [see Fig. 9(a)]. Photons with lower energy than the energy bandgap will not be able to generate excited states, resulting in transmission losses. On the contrary, photons with energies larger than the energy bandgap will generate hot charge carriers that will relax down to the LUMO of the polymer, giving rise to thermalization losses.^7,123–125 By stacking together in the same device two cells with polymers that absorb at different wavelengths, i.e., wide- and low-bandgap polymers, these losses can be reduced. The reduction in thermalization losses can be carried out by conversion in the subcell with a wide-bandgap polymer. On the other hand, transmission losses are lowered by absorption of the low energy photons in the subcell with a small-bandgap polymer. This results in polymer solar cells with ideally enhanced power conversion efficiencies.

Fig. 9

(a) Thermodynamic losses related to light absorption. (b) Layout of an organic tandem cell with two devices stacked one on top of another.

Tandem solar cells basically consist of stacking two or more devices one on top the each other with an adequate interlayer between them [see Fig. 9(b)]. The first deposited subcell is usually referred to as the front cell, while the one that is deposited on top is referred as the back cell. The connection among the subcells can be performed either in series (two terminals) or in parallel (three terminals).^125–127 The front layer should ideally absorb only most of the light corresponding to wavelengths of its maximum absorption coefficient. On the contrary, the light that is not absorbed by the front cell, belonging to wavelengths of maximum absorption of the second layer, will be more efficiently absorbed by the polymer of the back cell. Thus, the solar irradiance is absorbed in a more efficient manner without the need of thick layers that can induce charge transport losses.

Simple models have been developed for organic tandem devices in order to predict the efficiency increase when going from single to tandem cells. These models are based on the energy bandgaps of the absorbers used in each subcell and efficiencies up to 15% are estimated with the appropriate combination of materials.^128,129

One of the first reports about organic tandem solar cells with two identical small molecule subcells stacked together separated by a thin gold layer was published in 1990 by Hiramoto et al.¹³⁰ However, due to the limited choice of small molecule materials with significantly different absorption spectra, they started exploring hybrid tandem cells combining polymer and small molecule subcells.^131–135 In 2005, Kawano et al. reported the first tandem cells composed of two identical polymer BHJ subcells stacked together with an interlayer comprising sputtered ITO and spin-coated PEDOT:PSS.¹³⁶ One year later, Hadipour et al. showed a tandem cell consisting of subcells based on low- and wide-bandgap polymers with complementary absorptions.¹³⁷ However, all these attempts resulted in efficiencies $<5\%$ and many times even below those efficiencies reported for single cells made with one of the same comprising materials. A major breakthrough in the area of polymeric tandem cells was the demonstration of all solution processable polymer tandem cells with efficiency $>6\%$ by Kim et al.¹³⁸ More recently, in 2010, Gilot et al. studied the effect of current matching on the tandem device performance, which provides more insight in order to achieve high-performance tandem cells.¹³⁹ It is worth saying that the performance of polymeric tandem cells during the last four years has been limited to $\sim 7\%$ mainly due to the lack of high-performing low-bandgap polymers.^139–142 Nevertheless, in 2012, Dou et al. designed a new LBG polymer and achieved an inverted configuration tandem device with 8.6% PCE.^143,144 Recently, Li et at. demonstrated tandem and triple-junction polymer solar cells with power conversion efficiencies of 8.9 and 9.6%, respectively, by combining an LBG polymer known as PMDPP3T (with absorption in the near-infrared region up to 960 nm) and the wide-bandgap polymer PCDTBT.²⁰

Single cells based on $\mathrm{PMDPP}3\mathrm{T}:{\mathrm{PC}}_{60}\mathrm{BM}$ and $\mathrm{PCDTBT}:{\mathrm{PC}}_{70}\mathrm{BM}$ result in devices of 6 and 4.7% PCE, respectively. However, using the same layer thicknesses and stacking the subcells in a tandem cell [see Fig. 10(a)], the PCE was enhanced to 8.9%. Further improvement in the efficiency of the organic solar cell was carried out by adding an additional layer of the same small-bandgap material, creating a triple-junction device, as the one depicted in Fig. 10(d). In a tandem cell, the $J_{\mathrm{sc}}$ is limited by the wide-bangap front cell. However, this limitation is circumvented by splitting the small-bandgap subcell into two separate cells (middle and back cell) with different thicknesses to ensure that both absorb the same number of photons. The increase in $V_{\mathrm{oc}}$ of the triple junction (2.09 V with respect to 1.49 V) compensates the loss in $J_{\mathrm{sc}}$ [compare Figs. 10(b) and 10(c) with Figs. 10(e) and 10(f)]. In this way, the efficiency was enhanced up to 9.6%.²⁰

Fig. 10

Tandem and triple junction device layout [(a) and (d)], $JV$ characteristics [(b) and (e)], and EQE graph [(c) and (f)].²⁰

Efficiencies up to 10.2% have been published by You et al. by stacking together in a tandem device two identical subcells based on $\mathrm{PDTP}\text{-}\mathrm{DFBT}:{\mathrm{PC}}_{70}\mathrm{BM}$.¹⁴⁵ It is worth mentioning that single $\mathrm{PDTP}\text{-}\mathrm{DFBT}:{\mathrm{PC}}_{70}\mathrm{BM}$ based devices provide efficiencies of 8.1%. The absorption in the visible part of the tandem cell was significantly increased from 70 to 90% with respect to that of the single cell, suggesting that the performance of single cells for this material is a difficult compromise between light absorption and charge transport. Films of 80 nm are not able to absorb all the light; thicker films of 120 nm yield higher photocurrents ($19\mathrm{mA}{\mathrm{cm}}^{-2}$) but smaller FF (58%) than those for thinner films of 80 nm ($17\mathrm{mA}{\mathrm{cm}}^{-2}$ and 65%, respectively). When implemented in a tandem configuration, however, it is preferred to have efficient charge transport since the lack of absorption of thinner films can be compensated by the extra absorption of the second cell that additionally offers a gain in $V_{\mathrm{oc}}$.

The best polymer-fullerene tandem solar cells have a reported certified efficiency of 10.6% [see Fig. 11(b)] and feature a response up to 900 nm.¹⁹ A wide-bandgap polymer, P3HT, was blended with ICBA in the front cell. As commented previously, this combination allows for a larger $V_{\mathrm{oc}}$ in comparison to more standard fullerene derivatives. The LBP PDTP-DFBT was used with either ${\mathrm{PC}}_{60}\mathrm{BM}$ or ${\mathrm{PC}}_{70}\mathrm{BM}$ as the back cell [see Fig. 11(a)]. In Table 5, the detailed parameters of single and tandem cells are summarized. As can be seen, single cells provide efficiencies of 6.1 and 7.1% for P3HT:ICBA and $\mathrm{PDTP}\text{-}\mathrm{DFBT}:{\mathrm{PC}}_{60}\mathrm{BM}$, respectively. It is remarkable that although the use of ${\mathrm{PC}}_{70}\mathrm{BM}$ with the LBG material shows higher efficiencies than ${\mathrm{PC}}_{60}\mathrm{BM}$ in single cells, thanks to the extended absorption of the former in comparison to the latter, this is not the case when implemented in tandem configuration. Either the front active film already absorbs at those wavelengths at which ${\mathrm{PC}}_{70}\mathrm{BM}$ can represent an improvement with respect to ${\mathrm{PC}}_{60}\mathrm{BM}$ and/or the extra current of the back cell breaks the current matching between the front and the back cell. The photocurrent of the tandem is obviously limited by the subcell with the lowest value. This is clearly the case for the tandem that makes use of ${\mathrm{PC}}_{60}\mathrm{BM}$ in the back cell, whose photocurrent is very similar to that measured for an equivalent single cell of the front cell. However, when ${\mathrm{PC}}_{70}\mathrm{BM}$ is used, the currents of both cells are unbalanced and penalize the overall performance of the tandem with additional losses that result in a photocurrent even lower than that of the limiting subcell. The $V_{\mathrm{oc}}$ is the perfect summation of each subcell, resulting in the most efficient reported tandem device to date.

Fig. 11

(a) Device structure of the tandem solar cell. (b) $JV$ characteristics of the tandem cell as measured by NREL.¹⁹

Table 5

Poly(3-hexylthiophene) (P3HT) and PDTP-difluorobenzothiadiazole (DFBT) single junction cell and tandem solar cell performance.19

Even though, as commented before, triple-junction devices can in some cases minimally improve the performance of tandem cells,²⁰ they are very difficult and costly to implement. The enhancement obtained in terms of efficiency does not always compensate these two issues and one has to carefully analyze whether this approach is worth taking into account for a potential mass production. In any case, and considering only efficiency issues, an efficient triple-junction tandem organic solar cell with a record conversion efficiency of 11.5% has been recently published.¹⁴⁶ Three complementary absorbers with bandgap energies from 1.4 to 1.9 eV are used to obtain balanced absorption rates and matched photocurrents among the three subcells. In similar terms, Yusoff et al. made use of a combination of a wide-bandgap, a medium-bandgap, and low-bandgap materials to report double-junction devices of 10.4% and a record efficiency of 11.83% for a triple-junction cell.¹⁴⁷ The wide-bandgap material was a copolymer based on the alternation of dithienolsilole, thiophene, and thiazolothiazole segments (PSEHTT), whose bandgap lies at 1.82 eV, while the medium- and low-bandgap materials were the already introduced PTB7 (1.6 eV) and PMDPP3T (1.3 eV), respectively. Apart from the active materials, this work also makes use of all the other concepts mentioned in the other sections of this review, introducing, for example, new interlayers based on lithium zinc oxide and C60 self-assembled monolayers. The resulting devices show an impressive $J_{\mathrm{SC}}$ of 10.30 and $7.83\mathrm{mA}{\mathrm{cm}}^{-2}$, FF of 65.5 and 67.5, and $V_{\mathrm{oc}}$ of 1.54 and 2.24 V for the tandem ($\mathrm{PSEHTT}:\mathrm{ICBA}//\mathrm{PTB}7:{\mathrm{PC}}_{70}\mathrm{BM}$) and triple device ($\mathrm{PSEHTT}:\mathrm{ICBA}//\mathrm{PTB}7:{\mathrm{PC}}_{70}\mathrm{BM}//\mathrm{PMDPP}3\mathrm{T}:{\mathrm{PC}}_{70}\mathrm{BM}$), respectively.