1.

Introduction

In the last decades, organic bulk heterojunction (BHJ) solar cells have gradually evolved from a purely academic research topic to a promising emerging technology with world-wide R&D activities. With the first commercial applications on the market (e.g., the solar bags of Neuber ¹ ) and efficiencies reaching $>10\%$ , ² the lifetime of the devices (predicted to be several years ³ ) becomes important. In order to test the durability of solar cells during operation, a variety of stress factors must be considered, such as humidity and oxidizing agents in ambient atmosphere, incident ultraviolet (UV), and visible radiation, and the high temperatures reached after prolonged exposure to sunlight. Since gathering fundamental knowledge on these degradation processes represents an important step toward the realization of more durable devices, various laboratories are investigating them and have agreed on testing protocols that allow for an easier comparison of results among different groups. ⁴ Regarding the failure mechanisms in BHJ organic photovoltaics, several review papers can be found. ^{5 – 7} Herein, we restrict ourselves to a brief overview of some of the most reported causes of losses in power conversion efficiencies (Table 1).

Table 1

Extrinsic and intrinsic degradation mechanisms in organic solar cells.

Reversible degradation can occur due to oxygen doping of some metal oxides, widely employed as interfacial layers between the BHJ and the cathode. ⁷ On the other hand, exposure to light accelerates the reversible p-doping of the polymer as it absorbs oxygen from the atmosphere, with a resulting decrease in short-circuit current density ( $J_{\mathrm{sc}}$ ). ^{6 , 10} Alongside these effects, other mechanisms lead to irreversible degradation and fatal device failure. The ingress of water into the device can cause fast oxidation of the low-work function metal cathode (Yb, Ca, Al). ⁸ Moreover, oxygen was found to promote the rupture of the $\pi$ –conjugated backbone in the electron donor polymer within the active blend under UV illumination, with a subsequent drop in absorbance. This phenomenon is less effective in the presence of a fullerene derivative, ⁶ such as the most widely used acceptor material, [6,6]-phenyl- ${\mathrm{C}}_{61}$ -butyric acid methyl ester (PCBM).

Some of these issues can be prevented by appropriate encapsulation ¹⁴ and can, thus, be considered as extrinsic degradation mechanisms. ^{15 , 16} Next to extrinsic failures, a variety of intrinsic failure mechanisms, linked directly to the properties of the materials used (as further defined in Ref. 16), may also occur. Table 1 provides a selection of the important extrinsic and intrinsic degradation phenomena observed in organic solar cells.

In this contribution, we focus on one particular intrinsic failure mechanism: the thermally induced disruption of the active layer blend (nano) morphology of polymer:fullerene BHJ organic solar cells (Fig. 1). During thermal annealing, the BHJ system gains energy, which allows it to order itself in a more thermodynamically stable way. This leads to a thermally induced separation of the polymer and fullerene phases. As a result, small PCBM crystals tend to group into larger crystalline domains, a phenomenon called Ostwald ripening. ^{14 , 17} The formation of these PCBM-rich features leads to a decrease of the interfacial area between donor and acceptor and, thus, to a diminution of exciton dissociation and of the number of percolation paths to the electrodes. ⁶ In this phase separation phenomenon, the glass transition temperature ( $T_{\mathrm{g}}$ ) of the blend plays an important role. ¹⁸ Above $T_{\mathrm{g}}$ , the blend component molecules become mobile, which allows diffusion and clustering of the fullerene derivatives.

Fig. 1

Schematics of the evolution of fullerene aggregation in thermally annealed bulk heterojunction organic solar cells.

Better insights regarding the nature of the phase separation process have been obtained through a variety of characterization methods. In the next paragraph, we provide a short overview of the analytical, thermal analysis and electro-optical techniques used to investigate the morphology of BHJ active layers at their initial stage as well as during and after thermal stress. Subsequently, the effects of thermally induced morphological changes on the photovoltaic properties are described. Models to predict the impact of structural degradation are introduced, together with analyses of its impact on the durability of the devices. Finally, we introduce various routes proposed to improve the thermal stability of the active layer blend morphology, such as the use of high- $T_{\mathrm{g}}$ polymers, the addition of functionalized/thermo-cleavable side-chains, cross-linking, or the photo-induced dimerization of the fullerene component.

2.

Characterization of the BHJ Blend Stabil

2.1.

Investigation of the Active Layer Morphology

The stability of the active layer blend morphology is crucial for the long-term operation of the solar cells. Optimal intermixing requires that the distance between the domain interfaces of the polymer and fullerene materials does not exceed the exciton diffusion length, typically in the order of 10 nm. ¹⁵ A variety of techniques have been employed to determine on which length scale the intermixing occurs. Optical microscopy can be used to study the presence of large structures in the active layer film. Information on their height is obtained via stylus profilometry or atomic force microscopy (AFM) for micro- and nanoscale features, respectively. ^{19 , 20} AFM represents a valuable tool to analyze the film topography, ²¹ and it also enables one to distinguish structures from the different components in the blend through peak force tapping ²² and conductive-AFM. ^{22 , 23} Scanning electron microscopy (SEM) ²⁴ and transmission electron microscopy (TEM) ²⁰ techniques are used to image thin layers of organic semiconductors as well. Electron tomography reconstructions of three-dimensional volumes, ²⁵ together with cross-sectional SEM and TEM, ⁶ allow the investigation of the vertical disposition of donor-acceptor domains throughout a layer ^{2 , 25} or throughout a cross-section of a device. ²¹ Figure 2 (top row) shows an example of the evolution of the morphology of a poly[2-methoxy-5-( $3^{\prime }$ , $7^{\prime }$ -dimethyloctyloxy)-1,4-phenylene vinylene] (MDMO-PPV): PCBM ( $1:4$ wt%) blend after various durations of thermal annealing at 110°C (above the $T_{\mathrm{g}}$ of the blend), as visualized by TEM. It is clearly observed that the initially finely dispersed phases rapidly evolve toward the formation of aggregated structures, identified as PCBM domains. Due to the thermal treatment, the PCBM molecules diffuse out of the originally intermixed photoactive layer and aggregate into large domains. This Ostwald ripening (Fig. 1) is observed for numerous polymer:fullerene blends. ^{20 , 21 , 24}

Fig. 2

Transmission electron microscopy (TEM) images of poly[2-methoxy-5-( $3^{\prime }$ , $7^{\prime }$ -dimethyloctyloxy)-1,4-phenylene vinylene] (MDMO-PPV): [6,6]-phenyl- ${\mathrm{C}}_{61}$ -butyric acid methyl ester (PCBM) ( $1:4$ ) (top) and high- ${\mathrm{T}}_{\mathrm{g}}$ PPV: PCBM (bottom) blend films after various annealing times at 110°C. The selected-area electron diffraction (SAED) insets show the crystalline nature of the aggregates. Reprinted from Ref. 26, with permission from Elsevier.

Diffraction studies can also provide information on ordered structures in BHJ films. Selected-area electron diffraction (SAED) allows attributing the observed structures to specific blend components. ^{20 , 27} X-ray diffraction measurements enable determination of the nature of the crystalline structures. ²³ In grazing-incidence geometry, information is obtained on the ordering and orientation of crystallites and crystalline planes in a thin film ^{2 , 27} or on the stacking of the $\pi$ -chains in largely amorphous materials. ²⁸ As an illustration of the use of diffraction techniques in the study of thermally induced Ostwald ripening in BHJ organic solar cells, the insets of Fig. 2 (top row) show the crystalline nature of the PCBM aggregates formed during thermal annealing, as identified via SAED.

Together with UV-visible absorption ^{29 , 30} and Raman spectroscopy, ²⁹ these microscopy and diffraction techniques can be used to determine the initial morphology of BHJ active layers ²¹ as well as the changes in the blend structures due to annealing treatments ^{23 , 27} or aging of the film. ^{6 , 14}

3.

Effect of Active Layer Morphology Degradation on Photovoltaic Properties

As illustrated above, annealing at high temperatures promotes morphological rearrangement in BHJ blends because of the available thermal energy. The new phase distribution can have specific effects on the final photovoltaic performance of the devices, depending on the nature of the components. When a partially crystalline polymer (such as P3HT) is mixed with PCBM, an increase in hole mobility is reached by thermal annealing, as the latter promotes a strengthening of the crystalline character of the polymer. ²³ Concurrently, high temperatures induce PCBM aggregation. As a consequence, systems of this sort show an optimal annealing approach, which allows for high mobility in the crystalline polymer, but beyond which the formation of large PCBM domains causes a decrease in performance. ²³ On the other hand, when there is no ideal crystalline structure to be reached, as in the case of largely amorphous polymers, thermal annealing above the $T_{\mathrm{g}}$ of the blend leads to the formation of fullerene clusters only. ^{17 , 37} In the following section, it will be shown that these thermally induced morphological changes influence the short-circuit current density, the fill factor (FF), and the open-circuit voltage ( $V_{\mathrm{oc}}$ ), thus altering the final photovoltaic energy conversion efficiency.

3.1.

Effect of Morphology Degradation on ${\mathsf{J}}_{\mathsf{s}\mathsf{c}}$

The formation of aggregates of fullerene derivatives in BHJ films leads to a deterioration of the $J_{\mathrm{sc}}$ of the devices, partially due to a reduction of the interfacial area between the donor and acceptor domains, and partially due to the diminution of the photoactive area. ³⁸ For organic BHJ solar cells, it has been reported by several authors that the short-circuit current density ( $J_{\mathrm{sc}}$ ) decreases upon prolonged annealing and that the degradation occurs at a faster rate when higher annealing temperatures are applied. ^{9 , 26} In blends of MDMO-PPV with PCBM, the degradation curves for $J_{\mathrm{sc}}$ have been fitted with Eq. (1) to reflect the Ostwald ripening occurring in the active layer during annealing ¹⁷

Eq. (1)

$$J_{\mathrm{sc}}(t)=J_{\mathrm{sc}}(\infty )+J_{\mathrm{sc}}(0)\exp (-k_{\mathrm{deg}}\sqrt{t}).$$

In Eq. (1), $t$ represents the annealing time and $k_{\mathrm{deg}}=A\exp (-E_a/k_{B}T)$ is a rate constant that characterizes how fast the degradation evolves, with $E_a$ the activation energy in eV, $k_B$ the Boltzmann constant ( $8.62*{10}^{-5}\mathrm{eV}{\mathrm{K}}^{-1}$ ), and $A$ , a constant that depends on the degradation mechanisms and the experimental conditions. Figure 3 shows the fitted curves (solid lines) and the measured decay in short circuit current at different annealing temperatures.

Fig. 3

Simulated curves matching well with the drop in short-circuit current ( $I_{\mathrm{sc}}$ ) of MDMO-PPV: PCBM solar cells upon thermal annealing. Reproduced with permission from Ref. 17.

A similar approach was proposed to predict the lifetime of devices using the Lipshitz-Slyozov theory for spinodal decomposition, which predicts domain-size ( $R$ ) evolution following $R\sim t^n$ , with $n$ a material-dependent exponent, ³⁹ stressing the strong relationship between the employed polymer:fullerene system and its inherent rate of morphological degradation.

The link between the rearrangement of the active blend’s structure on the nanoscale and the drop in $J_{\mathrm{sc}}$ has also been investigated by Schaffer et al. ³⁸ through the acquisition of $I–V$ characteristics and in situ micro-focused grazing incidence small-angle x-ray scattering ( $\mu$ GISAXS) patterns of P3HT:PCBM devices during operation. They documented the change in P3HT domain size and density within the active layer blend matrix and introduced a model to quantitatively determine the $J_{\mathrm{sc}}$ of a device from the recorded morphology data by correlating the active interfacial area between the two species in the blend with the dimension of the structures, as measured with μGISAXS.

4.

Methods to Stabilize the BHJ Active Layer Morphology

Thermally induced structural changes in BHJ active layers at operating temperatures represent an intrinsic limit to the employability of certain material systems in outdoor conditions, ¹⁵ in which temperatures $>100°\mathrm{C}$ can be reached (temperatures up to 175°C were measured for silicon modules in Arizona ⁵⁰ ). For this reason, we dedicate this section to some of the methods that have been proposed to improve the morphological stability of photoactive material blends. ⁶

A careful choice of the donor-acceptor ratio ⁵¹ and the casting solvent can result in not only more efficient, ⁵² but also more durable organic solar cells. ³⁰ Depositing photoactive layers via different solvents was demonstrated to dictate the $T_{\mathrm{g}}$ and the ordering of the $\pi \text{-}\pi$ stacking in amorphous poly[ $N$ -9’’-heptadecanyl-2,7-carbazole-alt-5,5-(4’,7’-di-2-thienyl-2’,1’,3’-benzothiadiazole)] (PCDTBT):[6,6]-phenyl- ${\mathrm{C}}_{71}$ -butyric acid methyl ester ( ${\mathrm{PC}}_{70}\mathrm{BM}$ ) blends. ^{37 , 53} This system, stable at outdoor solar cell operating temperatures, thanks to its high $T_{\mathrm{g}}$ ( $\sim 130°\mathrm{C}$ ), ³⁷ needs to be optimized through processing conditions ⁵² or solvent annealing ⁵⁴ and has been determined to reach an extrapolated lifetime of seven years. ³ Removing residual solvents by treating the layers with methanol also proved to reduce thermally induced fullerene aggregation. ⁵⁵ Additionally, a lower regioregularity and higher purity of the donor polymers, as well as the addition of block copolymer compatibilizers, has also been demonstrated to result in more stable films. ^{5 , 56 – 58}

It was shown that employing polymers with a $T_{\mathrm{g}}$ higher than the temperature agreed on for standard testing [85°C (Ref. 4)] helps to slow down PCBM clustering over the lifetime of the solar cells, due to the high rigidity of the polymer chains in said conditions. ^{6 , 26 , 59} An example is shown in Fig. 2 (bottom row), where TEM analysis indicates that the employment of a high- $T_{\mathrm{g}}$ PPV polymer results in a more thermally stable blend as compared to standard MDMO-PPV. A similar comparison was reported by Vandenbergh et al. (poly[2-methoxy-5-(2’-phenylethoxy)-1,4-phenylene vinylene] versus MDMO-PPV). ⁶⁰ A high $T_{\mathrm{g}}$ also ensures stability of the electronic properties of (semicrystalline) polymers. ⁵³

A number of research groups have investigated the effect of polymer side-chain substitution on the blend stability. ^{61 – 65} Kesters et al. ⁶⁶ showed that the introduction of a small amount (5 to 15%) of functionalized side-chains in poly(3-alkylthiophene) (P3AT) copolymers results in a considerably increased blend stability upon prolonged thermal stress. Devices employing functionalized P3AT copolymers in blends with PCBM showed initial efficiencies comparable to regular P3HT:PCBM solar cells and drastically reduced fullerene aggregation at the ISOS-3 standard testing temperature of 85°C for a timescale of 700 h. ⁴ Figure 4 illustrates TEM images of functionalized P3AT copolymer:PCBM blends after different thermal annealing times (in comparison with standard P3HT:PCBM). SAED allowed identifying the formed clusters as PCBM crystals.

Fig. 4

TEM images and SAED patterns of polymer (P2 to P4): PCBM ( $1:1$ ) blends degraded at high temperature. Reprinted with permission from Ref. 66.

The use of block copolymers containing both donor and acceptor domains has also been proposed as a strategy to improve photoactive layer stability, ⁶⁷ although their synthesis remains challenging. ⁷ In these compounds, the donor and acceptor phases are covalently linked, allowing the material to be employed in single-component organic solar cells with stable morphology. Furthermore, rigid polymers can also be obtained by thermal cleaving of the side-chains, introduced to ensure good solubility but, in principle, not needed for optoelectronical functionality. This strategy has also been proven to lead to morphologically stable BHJ films. ⁵⁹

Another popular strategy uses cross-linking of the active materials, either via polymer or fullerene component, to freeze in the optimal morphology. This can, for instance, be achieved by adding cross-linkable groups to the polymer side-chains ⁶⁸ and curing the film with UV light. The employment of a cross-linked fullerene derivative, aligned in vertical nanorods, achieving a (quasi)ordered organic/organic nanostructure, allowed for an improved thermal stability, as well. ⁶⁹

Modifying the nature or the number of side-chains on the fullerene results in more amorphous compounds, ^{55 , 70} which can also lead to an improved morphological stability. Liao et al. were able to exploit supramolecular interactions to impose a given order to the BHJ film and ensure structural firmness. ⁷¹ They introduced a fullerene derivative, [6,6]-phenyl-C 61-butyric acid pentafluorophenyl ester ( ${\mathrm{PCBP}}^{\mathrm{F}}$ ), in which the methyl group of PCBM is substituted by a pentafluorophenyl ring, and then exploited the known interaction between ${\mathrm{C}}_{60}$ and the pentafluorophenyl moiety to produce solar cells with enhanced high-temperature stability.

Piersimoni et al. ²⁰ and Li et al. ⁷² demonstrated the stabilizing effect of fullerene photodimerization on the active layer blend morphology. Light soaking of polymer:PCBM films before or during annealing above the $T_{\mathrm{g}}$ of the blend resulted in reduced fullerene aggregation, although the effect becomes less appreciable with increasing temperatures ²⁰ due to thermally induced dissociation of the dimers. These experiments were conducted using either PCBM or ${\mathrm{PC}}_{70}\mathrm{BM}$ , in combination with MDMO-PPV, P3HT, and PCDTBT as donor polymers (annealing the latter at higher temperature because of the elevated $T_{\mathrm{g}}$ of the polymer). Optical microscopy images, shown in Fig. 5, confirmed that thermal aging in the dark promotes the formation of clusters, attributed to the PCBM, while the same treatment under illumination does not significantly affect the morphology of the BHJ layers. While these images prove the stabilizing effect of this photoreaction on the blends’ morphology, its impact on the performance of devices resulted strongly system-dependent. Indeed, it can contribute to raising the $J_{\mathrm{sc}}$ in PCDTBT-based solar cells, ⁷² but it was also reported by other authors to be responsible for performance degradation when combined with a bithiophene-co-thiazolothiazole push-pull copolymer. ⁷³

Fig. 5

Comparison between optical reflection microscopy images of as-cast films (top row), films annealed at 180°C (b), and 110°C [(e) and (h)] in the dark, and films annealed at the same temperatures under solar illumination. Reproduced with permission from Ref. 20.

5.

Conclusions and Outlook

Prolonged storage or operation of polymer:fullerene BHJ organic solar cells can result in athermally induced disruption of the optimal photoactive layer morphology via phase separation and Ostwald ripening of the fullerene domains, leading to a significant reduction of the photovoltaic performance. Improving the morphological stability of the photoactive blends upon thermal stress is, therefore, an important challenge for future commercialization of these devices. We briefly reviewed the degradation mechanisms resulting in performance failure, as well as the main techniques employed to characterize the thermally induced changes in the active layer (nano)morphology. The most obvious indication for BHJ structural damage is the formation of large crystalline domains of the fullerene derivative, with a consequent reduction of the interfacial area between the donor and acceptor domains, and a consequent negative effect on the photovoltaic parameters. Insights on the underlying mechanisms provide the foundation to model the effect of this structural reorganization on the photovoltaic parameters, allowing the prediction of device lifetimes and degradation rates.

Toward more intrinsically stable BHJ organic solar cells, several roads have been proposed and are still under investigation, regarding the choice of materials, the exploitation of observed phenomena, and the modification of compounds. The reviewed solutions will undoubtedly continue to find their way toward the development of stable organic photovoltaic devices. On the other hand, dedicated thermal analysis of the blends, as well as accurate investigation of the morphological changes in degraded photoactive layers, can be regarded as valuable tools in the direction of intrinsically stable organic solar cells.