Solar Paint: from Synthesis to Printing

Water-based polymer nanoparticle dispersions (solar paint) offer the prospect of addressing two of the main challenges associated with printing large area organic photovoltaic devices; namely, how to control the nanoscale architecture of the active layer and eliminate the need for hazardous organic solvents during device fabrication. In this paper, we review progress in the field of nanoparticulate organic photovoltaic (NPOPV) devices and future prospects for large-scale manufacturing of solar cells based on this technology.


Introduction
Organic photovoltaics (OPVs) offer enormous potential as inexpensive coatings capable of generating electricity directly from sunlight [1,2].These polymer blend materials can be printed at high speeds across large areas using roll-to-roll processing techniques, creating the tantalising vision of coating every roof and other suitable building surface with low-cost photovoltaics [3].
Conventionally, OPV devices are fabricated from mixtures of donor and acceptor materials dissolved in organic solvents, which are deposited to produce an interpenetrating network [4].Current fabrication methodologies rely on the thermodynamics of demixing to produce phase segregated regions with the required optimum size of 20-50 nm [5].However, two key aspects of current OPV OPEN ACCESS fabrication are not well-suited to building large-area PV modules using high-speed printing techniques.First, using current fabrication approaches to control phase segregation across large areas is problematic [6].Second, current OPV technology typically uses chlorinated solvents (e.g., chloroform), which are under continual regulatory pressure due to their hazardous and toxic nature [7,8].As such, the increasingly harsh technical requirement for using these solvents means that their implementation in high-speed printing lines will be highly problematic, if not economically impractical [9].
The relatively low efficiencies reported for OPVs derived from nanoparticulate films spin coated from water are perhaps unsurprising.To disperse these colloidal particles in water, a surfactant is added to lower the interfacial energy and to provide colloidal stability [23].Upon spin coating, these surfactant molecules might be expected to form an insulating barrier between neighbouring particles that would tend to disrupt charge transport within the photovoltaic device [24,25].In addition, semi-conducting polymer particles are usually highly sensitive to oxygen and especially water contaminants, which tend to rapidly degrade OPV performance [26].

Synthesis of NPs via the Miniemulsion Process
The nanoparticle fabrication method utilises ultrasound energy delivered via an ultrasound horn inserted into the reaction mixture to generate a miniemulsion (Figure 1).The ultrasound horn makes the formation of sub-micrometre droplets possible by applying high shear force.A liquid aqueous surfactant-containing phase (polar) is combined with an organic phase of polymer dissolved in chloroform (non-polar) to generate a macroemulsion, then ultrasonicated to form a miniemulsion.The polymer chloroform droplets constitute the dispersed phase with an aqueous continuous phase.This is a modification of the usual method for generating polymer nanoparticles where the dispersed phase was liquid monomer; this alternative method was first reported by Landfester and Kietzke [17,18].Immediately after miniemulsification, the solvent is removed from the dispersed droplets via evaporation, leaving polymer nanoparticles.The final nanoparticle size can be varied by changing the initial concentration of surfactant in the aqueous phase.

Synthesis of NPs via Precipitation Methods
As an alternative to the miniemulsion approach, precipitation techniques offer a simple route to the production of semiconducting polymer nanoparticles via the injection of a solution of active material into a second solvent of poor solubility [27,28].As such, the synthesis is quick, does not use surfactant, requires no heating (and therefore, no prefabrication annealing of the nanoparticles) in the nanoparticle synthesis phase and can readily be scaled up for the large-scale synthesis of material.In general, the dispersions have been shown to have lower stability and exhibit a compositional change upon standing due to preferential precipitation of particles of differing composition.However, the precipitation approach does offer the opportunity for inclusion of the nanoparticle synthesis as part of an active printing process, with particles being generated as and when required.Furthermore, Hirsch et al. have shown that by successive solvent displacement, it is possible to synthesise inverted core-shell particles where the structural arrangement is counter to the inherent surface energies of the materials [29].
In 2012, Stapleton et al. reported multilayered photovoltaic devices fabricated from PFB:F8BT nanoparticles, which demonstrated the highest power conversion efficiencies observed for these polyfluorene nanoparticle materials [30].This increased performance was achieved through the control of the surface energies of the individual components in the polymer nanoparticle and the post-deposition processing of the polymer nanoparticle layers.Significantly, this work showed that the fabricated nanoparticulate organic photovoltaic (NPOPV) devices were more efficient than the standard blend devices (Figure 2).b work [30]) and ~0.8% for Ca/Al, and that there is a distinct optimized thickness for the NP devices (Figure 3).The optimal thickness is a consequence of the competing physical effects of the repair and filling of defects for thin films [32,33] and the development of stress cracking in thick films [34].Indeed, Stapleton's work showed that the optimal layer thickness in these devices corresponds to the critical cracking thickness (CCT) above which stress cracking occurs, resulting in low shunt resistance and a reduction in device performance [30].Overall, the work by Vaughan et al. provided further confirmation that the intrinsic morphology of NPOPV PFB:F8BT devices enhances the exciton dissociation relative to the corresponding BHJ structure.Moreover, the use of a Ca/Al cathode results in the creation of interfacial gap states (Figure 4), which reduce the recombination of charges generated by the PFB in these devices and restores open circuit voltage to the level obtained for an optimized BHJ device, resulting in a PCE approaching 1%.

The P3HT:PCBM NPOPV Material System
The first NPOPV devices based on the more common P3HT:PCBM system were reported by Lee et al., who synthesised P3HT:PCBM NPs using the miniemulsion method and then demonstrated their PV effect using conducting atomic force microscopy (AFM) under illumination [35].At almost the same time, Larsen-Olsen et al. reported fully-printed reel-to-reel (R2R) devices using an inverted geometry also based on P3HT:PCBM NPs that exhibited a best PCE of only 0.3% [22].Power Conversion Efficiency (%) Reproduced with permission from [31], published by American Institute of Physics, 2012.

Number of Layers
In a subsequent paper, Ulum et al. demonstrated NP-OPV devices fabricated from water-dispersed P3HT:PCBM nanoparticles that exhibited power conversion efficiencies (PCEs) of 1.30% and peak external quantum efficiencies (EQE) of 35% [36].However, unlike the PFB:F8BT NPOPV system, the P3HT:PCBM NPOPV devices were less efficient than their bulk heterojunction counterparts.Scanning transmission X-ray microscopy (STXM) revealed that the active layer retains a highly structured NP morphology and comprises core-shell NPs consisting of a relatively pure PCBM core and a blended P3HT:PCBM shell (Figure 5).However, upon annealing, these NPOPV devices undergo extensive phase segregation and a corresponding decrease in device performance.Indeed, this work provided an explanation for the lower efficiency of the annealed P3HT:PCBM OPV devices, since thermal processing of the NP film results in an effectively "over-annealed" structure with gross phase segregation occurring, thus disrupting charge generation and transport.Holmes et al. reported both the intra-and inter-particle morphology of P3HT:PCBM nanoparticles before and after thermal annealing treatment, as a function of P3HT molecular weight [37].The morphological changes resulting from thermal annealing were shown to be highly dependent upon the molecular weight of the polymer (Figure 6).Diffusion of PCBM in the polymer matrix was found to govern the morphology of annealed NP films, with the diffusion of molecular PCBM dominating for high molecular weight P3HT and the diffusion of larger PCBM cores being significant for low molecular weight P3HT.Using the Stokes-Einstein continuum model for diffusion, the authors were able to predict a threshold P3HT molecular weight of between 25 and 38 kg mol −1 for the PCBM core diffusion in P3HT:PCBM NP films, which was in good agreement with the experimental range of 16-44 kg mol −1 .In a follow-up paper, Holmes et al. demonstrated that polymer molecular weight not only affects film morphology, but also the electronic and compositional structure of the nanoparticulate film.They showed that it is the domain composition that is most highly correlated with device performance and that this composition is driven by the PCBM mobility and aggregation within the nanoparticulate structure [38].Their work shows that it is possible to generate a range of NP structures with quite small changes in formulation, and their work suggests that a uniform distribution of conjugated polymer and fullerene throughout the NP volume should deliver the best device performance [39].The precipitation method has also been used to fabricate P3HT:PCBM NPOPVs, with Darwis et al. demonstrating comparable or improved power conversion efficiencies with respect to the corresponding surfactant-based NPOPV devices [40].Interestingly, the precipitated NPs are initially fully blended and do not exhibit the core-shell morphology characteristic of the surfactant-containing P3HT:PCBM NPs.

The P3HT:ICBA NPOPV Material System
In 2013, Ulum et al. reported the performance of nanoparticulate organic photovoltaic (NPOPV) devices fabricated from poly(3-hexylthiophene) (P3HT):indene-C60-bisadduct (ICBA) blends.These devices exhibit power conversion efficiencies of 2.5%, which is the highest so far reported for NPOPV cells [41].STXM mapping revealed that the unannealed NPs have a core shell structure with an enhanced ICBA concentration in the core region.Upon annealing, although the films maintained their NP structure, their core shell morphology was lost, and a mixed composition was observed (Figure 7).
Using the Flory-Huggins model, Ulum et al. showed that ICBA is miscible in P3HT at all weight fractions, which was supported by the STXM observations.Thus, the improved performance of the annealed P3HT:ICBA NPOPV devices is driven by the enhanced miscibility of ICBA in P3HT, which results in a more efficient intermixed structure.[42].As is the case with previous work on OPV devices based on NPs synthesized using free precipitation [40], many depositions are required to form the continuous film needed for a working device.However, most encouragingly, this work does show that high efficiencies are possible using the NPOPV approach.

Summary of NPOPV Performance
A summary of the performance of NPOPV devices reported over the past few years is presented in Table 1.It is clear from the table that the performance of NPOPV devices has increased dramatically, with a rise of three orders of magnitude reported from 2003 to 2014.

Conclusions and Future Outlook
The recent development of water-based NPOPV coatings represents a paradigm shift in the development of low-cost OPV devices.This approach simultaneously provides control of morphology and eliminates the need for volatile flammable solvents in device production; two key challenges of current OPV device research.Indeed, the development of a water-based solar paint offers the tantalising prospect of printing large-area OPV devices using any existing printing facility.Moreover, it is increasingly recognised that the development of a water-based printable OPV system would be highly advantageous and that the current material systems based on chlorinated solvents are not suitable for commercial scale production.
The work described in this review shows that the new NPOPV methodology is generally applicable and that NPOPV device PCEs can be competitive with devices built from organic solvents.However, these studies also reveal that, from a materials point of view, NPs behave completely differently from polymer blends spun from organic solvents.Effectively, the NPs are a completely new material system, and as such, the old rules for OPV device fabrication that have been learnt for organic-based OPV devices no longer apply.
In the case of NPOPVs based on polyfluorene blends, the NP morphology results in a doubling of the device efficiency.However, for polymer:fullerene blends (e.g., P3HT:PCBM and P3HT:ICBA), morphology formation in the NP films is highly complex, and other factors (such as core diffusion) can dominate, resulting in unoptimised device structures and efficiencies.
The future outlook for these materials is extremely promising, with device efficiencies having increased from 0.004% to 4% in less than five years.The next stage of development will involve understanding the mechanisms that determine NP structure and NP film morphology and how these can be controlled and optimised.To date, the ability to control the morphology of OPV active layers on the nanoscale has yet to be realized.However, recent work demonstrates that the application of NP materials may allow this goal to be achieved.

Figure 1 .
Figure 1.Schematic of the NP synthesis process.

Figure 3 .
Figure 3. Variation of power conversion efficiency (PCE) with the number of deposited layers for PFB:F8BT nanoparticulate organic photovoltaic (NPOPV) devices fabricated with an Al cathode (filled circles) and a Ca/Al cathode (open circles).Dotted and dashed lines have been added to guide the eye.An average error has been determined based upon the variance for a minimum of ten devices for each number of layers.Reproduced with permission from [31], published by American Institute of Physics, 2012.

Figure 4 .
Figure 4. Energy level diagrams for PFB:F8BT nanoparticles in the presence of calcium.(a) Calcium diffuses through the nanoparticle surface; (b) Calcium dopes the PFB-rich shell, producing gap states.Electron transfer occurs from calcium producing filled gap states; (c) An exciton generated on PFB approaches the doped PFB material (PFB*), and a hole transfers to the filled gap state, producing a more energetic electron; (d) Electron transfer from an exciton generated on F8BT to either the higher energy PFB lowest unoccupied molecular orbital (LUMO) or the filled lower energy PFB* LUMO is hindered.Reproduced with permission from [31], published by American Institute of Physics, 2012.

Figure 5 .
Figure 5. (a) Transmission electron microscopy (TEM) image of the unannealed P3HT:PCBM NP film.Corresponding scanning transmission X-ray microscopy (STXM) maps of: (b) P3HT composition and (c) PCBM composition for the unannealed P3HT:PCBM NP film.(d) TEM image of the annealed P3HT:PCBM NP film.Corresponding STXM maps of: (e) P3HT composition and (f) PCBM composition for the annealed P3HT:PCBM NP film.The horizontal line in the composition images is produced by a small (<5%) beam intensity variation occurring in the P3HT mass plot.A 1-μm scale bar is shown.Reproduced with permission from [36], published by Elsevier B.V., 2013.

Table 1 .
[30]ary of NPOPV device characteristics reported in the literature.The asterisked values are estimated based on a best estimate fill factor (FF) of 0.28 from bulk blend devices[30].