Nanoporous Polymer Re ﬂ ectors for Organic Solar Cells

harvesting and hence the PCE of OSCs. To mitigate these transmission losses and to harvest those unabsorbed photons, (KSOP). nancial


Introduction
Among the emerging photovoltaic technologies, organic solar cells (OSCs) stand out with their design flexibility in color and shape, their lightweight, their mechanical flexibility, their potential sustainability all along the value chain, and their optional semitransparency. [1][2][3][4][5][6][7] Recently, power conversion efficiencies (PCEs) of OSCs have surpassed 18%, enabled by the advent of nonfullerene acceptors, by tuning of the microstructure of the bulk-heterojunction, by tailoring the spectral absorption, and by smart device design. [8] Printing technologies, including inkjet printing, doctor blading, and roll-to-roll printing, are widely considered toward large-scale fabrication and commercialization of OSCs. [1] Other than the lightharvesting layers, charge carrier transport layers and electrodes must also be solution processable for any fully integrated printing process. While the front electrode can be printed from conductive polymers, the back electrode is envisaged to be printed from silver inks, in order to use its back reflections for enhanced light harvesting inside the device. The feasibility of printing and the sustainability of materials would gain from the implementation of back electrodes also comprising conductive polymers instead of silver or using only small amounts thereof. Earlier works have shown that the substitution of rear metal electrodes by highly conductive polymers is feasible, [9,10] and that the limited conductivity of polymer electrodes can be mitigated by incorporating silver nanowires while preserving the mechanical robustness of the solar cell. [10] Although this concept is of great use to realize semitransparent solar cells, e.g., for window or façade integration, the much smaller reflectance of the polymer electrode versus the commonly used metal electrode reduces light harvesting and hence the PCE of OSCs. To mitigate these transmission losses and to harvest those unabsorbed photons, Due to their high transparency, electrodes fabricated from conductive polymers are often implemented in semitransparent organic solar cells. Opaque solar cells usually employ metal back electrodes with high reflectivity for best photon confinement in the light-harvesting layer. Herein, a bilayer back electrode comprising conductive polymers and nanofoamed poly(methyl methacrylate) (PMMA) is investigated, the latter of which creates diffuse reflection of the incoming light. By tuning the thickness of the nanofoamed PMMA layer, absorption and transmission of the solar cells can be tailored from opaque to vastly transparent. Due to its diffusive character, this versatile electrode enhances the light absorption in the wavelength regimes with lower absorption coefficient. The solar cells are particularly suited for deployment in frosted window applications.
various methods have been proposed, including multilayer distributed Bragg reflectors, [11][12][13][14][15][16] pyramidal rear reflectors, [17] and cholesteric liquid crystal reflectors. [18] Light harvesting can also be enhanced by prolonging the way of the photons through the device by light scattering. Tang et al. proposed a pigmented dielectric reflector consisting of titania nanoparticles embedded within a polydimethylsiloxane matrix to be implemented as a cost-effective light-trapping structure for OSCs. [19] The use of pigmented dielectric reflectors for PV was first explored 20 years ago by Cotter et al. [20] in thin silicon solar cells and further used to enhance the PCE of thin-film silicon devices. [21] In order to increase the light scattering efficiency of dielectric reflectors and to increase the refractive index contrast, the host material was further removed using snow globe coating. [22] The spacing between aggregates of titania nanoparticles filled with air that exhibits dimensions comparable to the wavelength of light, thus resulted in longer optical paths and some PCE enhancement. [23] Such nanoporous reflectors can also be realized from polymer networks, where light scattering depends on the morphology of the porous network in the polymer matrix, which can be tuned through different processes, including electrospinning or phaseseparation methods. [24][25][26][27] In this study, we demonstrate that the reflective rear metal electrode in an OSC can be replaced by a combination of a conductive polymer electrode and a reflecting (light scattering) nanoporous polymer film, the latter of which is fabricated by supercritical CO 2 foaming. The process is versatile, cost-effective, reproducible, rapid, and industrially viable. It can be used with various polymers and generates mechanically stable nanoporous films. In a previous study, we reported that foamed polymer films are highly reflecting over the visible and near-infrared spectrum. [28] Upon variation of the thickness of the foamed films, the ratio of reflected/transmitted light can be tuned easily. We take advantage of this effect to tailor light management in OSCs for their light harvesting properties and their transmittance, depending on the envisaged application.

Device Design
To explore the photovoltaic performance enhancement brought by the nanoporous polymer reflectors, we have chosen the solar cell architecture that is depicted in Figure 1a, comprising indium tin oxide (ITO), zinc oxide (ZnO), a blend layer of the conjugated polymer donor poly[ [4,8- (PBDB-T) and the molecular acceptor 3,9-bis(2-methylene-(3-(1,1-dicyanomethylene)-indanone))-5,5,11,11-tetrakis(4-hexylphenyl)dithieno [2,3- (ITIC) as well as a poly (3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) counter electrode. PBDB-T:ITIC was chosen for light harvesting due to its high efficiency and its excellent thermal stability. [29] The surface resistivity of the 70 nm-thick polymer electrode was reduced to 35 Ω sq À1 by the incorporation of silver nanowires (AgNW) that were readily dispersed in the aspurchased PEDOT:PSS formulation (HYE). [10] The rear reflector comprises a volumetric network of light scattering nanopores and was mechanically attached to the PEDOT:PSS:AgNW electrode. The incoming light propagates through the glass substrate first. After passing through the device, the unabsorbed photons are retro-reflected into the light-harvesting layer, eventually enhancing the overall absorption and the PCE of the solar cell. For reference, we mechanically attached an electrically isolated silver (Ag) layer to the PEDOT:PSS:AgNW electrode, mimicking the reflectance of a metal electrode while not changing the conductivity of the electrode, hence separating the optical effects from the electronic effects.

Optical Properties of the Nanoporous Polymer Reflectors
At the outset of our study, we foamed a commercially available poly(methyl methacrylate) (PMMA) sample with a thickness of  500 μm. The foamed sample exhibits an almost 100% broadband reflection by light scattering (close to 0% transmission) over the wavelength range of interest and hence is an effective back reflector for opaque solar cells. Figure 2 depicts the light reflectance of this 500 μm-thick sample with the optical properties of a typical opaque silver electrode. We note that the decrease of reflectance below 400 nm can be attributed to UV absorption of additives in certain types of commercial PMMA. In order to tailor the translucency of the solar cells, we fabricated a series of nanoporous PMMA reflectors following the process described in our previous work. [30] The nanoporous reflectors were produced individually by foaming of PMMA with supercritical CO 2 at 50 MPa for 1 min. This facile process enables the rapid and homogeneous introduction of nanopores (here with an average diameter of 320 AE 40 nm) into the PMMA films while maintaining closed and planar film surfaces (i.e., free from pores), hence facilitating their mechanical and optical coupling to the solar cells. The coupling of the PMMA reflectors was further facilitated by their mechanical flexibility due to their low thickness between 10 and 50 μm. To tailor the translucency of the nanoporous reflectors, we deliberately maintained the foaming parameters but adjusted the thickness of the PMMA film prior to foaming. Notably, similar effects can be achieved by varying the foaming parameters and hence the morphology of the porous network (including the density and the average diameter of the nanopores) while maintaining the film thickness.
Four nanoporous reflectors were fabricated by gradually increasing the thickness of the PMMA film. The target thicknesses of the pristine PMMA films (i.e., before foaming) were 10, 20, 30, and 50 μm (thicknesses achieved: 8.9, 19.2, 31.8, and 52 μm). Upon foaming, i.e., the incorporation of CO 2 , the above listed samples inflated to 9.3, 21.5, 35.5, and 61.1 μm. The relative increase of the sample thickness toward thicker samples upon foaming can be attributed to the formation of a skin layer without pores where the CO 2 diffuses out of the sample rather quickly. Hence, thin samples hardly gain in thickness during the foaming process, whereas thick samples inflate significantly (relative thickness gain: 4%, 12%, 12%, and 18%). We note that the CO 2 diffusion and skin-layer formation produce a nonhomogenous vertical gas inclusion profile throughout the sample, which is difficult to describe quantitatively. Figure 2 depicts the reflectance spectra of the foamed samples in comparison to the Ag electrode (thickness: 100 nm) and the 500 μm foamed commercial reference. The spectra of the nanoporous reflectors are essentially featureless, yet display a spectral dependency with a higher reflectance toward smaller wavelengths, comparable to the size of the nanopores, which induce complex multiscattering effects.
The left side of Figure 3 depicts the series of nanofoamed reflectors with different thicknesses, demonstrating the increasing diffuse reflection of light toward larger film thicknesses. At a nominal thickness of 50 μm, the nanofoamed reflector is effectively opaque, and all incoming light is scattered (diffuse reflection).

Solar Cells with Nanoporous Reflectors
Finally, we attached the nanoporous reflectors to the solar cells, behind the rear PEDOT:PSS:AgNW electrodes. While the PEDOT:PSS:AgNW electrodes alone would be a perfect choice for semitransparent solar cells, the additional nanoporous reflectors warrant enhanced PCEs due to (partly) backscattered light. The thickness of the nanoporous reflector allows to tailor the PCE or, likewise, the translucency of the device as depicted on the right side of Figure 3. Notably, if used as a semitransparent solar cell, the nanoporous reflectors produce a certain haze on the transmitted light, depending on the actual thickness of the nanoporous layer, which renders this device concept particularly interesting for frosted window applications.
The solar cells and the thickness of the bulk-heterojunction were optimized for best PCEs in the absence of any reflector, i.e., a standard semitransparent OSC comprising a PEDOT: PSS:AgNW top electrode. At a PBDB-T:ITIC layer thickness of 150 nm, the devices achieved a PCE of 5.9 AE 0.1%. Figure 4a-c depicts the transmittance, the absorptance, and the external quantum efficiency (EQE) of this and all other solar cells discussed in this work. In addition, Table 1 summarizes the average visible transmission (AVT) of the solar cells with respect to the photopic response of the human eye. [31] Each of the three characteristics is vastly dominated by the absorption of the PBDB-T:ITIC blend. All characteristics show severe flattening when nanoporous reflectors are employed, with the effect being most pronounced toward thicker reflectors which we attribute to the strong backscattering of light: the absorptance of rather weakly absorbed spectral regimes particularly benefits from the light-path elongation within the absorber layer by scattering. At the maximum of the OSC transmittance at 460 nm in Figure 4a, the transmittance of the full device can thus be varied from 21.3% at a reflector thickness of 10 μm down to 5.7% at a thickness of 50 μm. Notably, the absorption shows only negligible influence from thin-film interferences as it is often observed when using metal electrodes. Hence, the transparency color perception of the devices is almost invariant against electrodes with different reflection strength. Figure 4d     demonstrate the excellent and almost lossless retro-reflection of light from the nanoporous reflectors.

Conclusions
We have demonstrated that the combination of polymer electrodes and reflective nanoporous polymer films from foamed PMMA can replace the common metal electrodes in thin-film (organic) solar cells. Using a rapid and cost-effective supercritical CO 2 foaming process, we have fabricated PMMA reflectors of different thicknesses in order to tailor the degree of reflectivity and translucency toward the requirements of different applications, including fully opaque solar cells.
In addition, the scattering effects enhance light harvesting in spectral regions where the bulk-heterojunctions absorb only weakly. It produces a certain transparency color stability and a distinct frosted-window effect. While our approach was illustrated on a PBDB-T:ITIC bulk-heterojunction solar cell, it can be readily translated to other organic blends and even other thin-film PV technologies. In organic photovoltaics, this electrode configuration can be an interesting asset toward a sustainable and all-organic solar cell without incorporation of heavy metals or rare elements.

Experimental Section
Fabrication of Nanoporous Polymer Reflectors: Glass substrates (2 Â 2 cm 2 ) were cleaned with acetone and isopropanol in an ultrasonic bath (5 min), and then exposed to oxygen plasma (2 min). Immediately after, PMMA resist (AR-P 672.11, ALLRESIST GmbH, M w ¼ 0.95 MDa) was spin-coated, and thermally annealed on a hot plate (110 C, 5 min). The thickness of the PMMA films was tuned from nominally 10 to 50 μm by repeatedly spin coating the PMMA solution. The 10 μm layer was obtained by spin coating twice (1200 rpm, 30 s; 1700 rpm, 30 s). The other samples (20, 30, 50 μm) were spin-coated repeatedly by three, five, and eight times, respectively (1000 rpm, 30 s). Then the PMMA films on glass were put in an autoclave vessel fed with pure CO 2 (80 C, saturation pressure 50 MPa, 1 min). Fast pressure quenching was initiated by quickly opening the outlet valve of the foaming apparatus. More details about the process and the growth mechanism of the pores are described in our previous work. [30] Finally, the nanofoamed films were peeled off from their glass substrates. The 500 μm-thick PMMA reference sample (M w ¼ 1.86 MDa) was purchased from Topacryl AG and subdued to microcellular foaming without further treatment.
Device Fabrication: Patterned indium tin oxide (ITO)-coated glass substrates (R □ ¼ 13 Ω sq À1 ) were cleaned by sequential ultrasonication in acetone and isopropanol (10 min each). The ZnO electron extraction layers (30 nm) were doctor bladed in air (gap 150 μm, speed 10 mm s À1 , 35 C) from nanoparticle dispersion by using a universal thin-film applicator (ZUA 2000, Zehntner). The ZnO nanoparticle dispersion was synthesized according to previous literature reports. [32] Then the samples were thermally annealed on a hotplate (120 C, 10 min) and transferred into a nitrogen glove box for the remaining fabrication and characterization process. The light-harvesting polymer donor PBDB-T (synthesized according to the literature [33] ) and the molecular acceptor ITIC (1-Material) were dissolved (1:1 w/w, 20 mg mL À1 ) in chlorobenzene (Sigma-Aldrich). The solution was stirred on a hotplate overnight (60 C), and 1,8-diiodooctane (0.25 vol%, DIO, Alfa Aesar) was added to the stock solution half an hour before deposition. Then the solution was doctor bladed (60 C, applicator gap 200 μm, speed 20 mm s À1 ) to form a 150 nm-thick bulkheterojunction layer. The samples were thermally annealed on a hotplate (160 C, 10 min). For the top electrode, a poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PEDOT:PSS) formulation including dispersed silver nanowires (Clevios HYE, Heraeus Deutschland GmbH & Co. KG) was doctor bladed (60 C, applicator gap 200 μm, speed 5 mm s À1 ) and thermally annealed (120 C, 5 min) to yield a thickness of 70 nm. Then the PEDOT:PSS:AgNW layers were structured using dicing tape. The photoactive area of 0.105 cm 2 was defined by the cross section of the two electrodes. Finally, the reflectors were attached on the rear side of the device deliberately omitting refractive index matching oil since the oil would have damaged the HYE electrode. The reflectors were selfadhering. Only the 500 μm sample required additional fixing with tape. For reference, we fabricated a mock-silver-electrode (Ag, 100 nm, 99.9%, ChemPur) on glass that was sublimed in a vacuum chamber (<0.1 mPa). When attached to the semitransparent solar cell in order to imitate a rear silver electrode, it was electrically isolated from the device by a 300 μm thin and highly transparent PMMA layer and fixed with tape.
Characterization: The cross-sectional micrograph of the nanoporous PMMA film shown in Figure 1b was obtained at an accelerating voltage of 1 kV on a scanning electron microscope (SEM; SUPRA 55 VP, Zeiss), after the sample was immersed into liquid nitrogen, broken to expose its nanoporous network, and covered with a thin gold layer. To determine the typical dimensions of the foamed nanopores, 100 pores were analyzed using ImageJ (software version 1.52a).
The reflectance spectra of the nanofoamed reflectors, before and after coupling to the OSC, were measured under close-to-normal (8 ) light incidence employing an integrating sphere (Lambda 1050 UV/vis/NIR spectrometer, PerkinElmer).
The optoelectronic properties of the OSCs were measured using a shadow mask (0.105 cm 2 ). Current density-voltage ( J-V ) curves were measured with a source-measure unit (Keithley 2400, Keithley Instruments Corporation) under illumination from a Class AAA solar simulator (Sciencetech AX LightLine with LA200 Homogenizing Optics, ASTM AM1.5G), and using a KG5 filtered silicon reference cell (91 150-KG5, Newport). For the EQE measurements, chopped monochromatic probe light was generated by a high-pressure xenon plasma lamp (450 W LSH601, LOT Oriel) and a monochromator (Omni-λ300, LOT Oriel), including a set of wavelength filters (MSZ3122, LOT Oriel) and optics. The respective current densities were measured using a lock-in amplifier (DLPCA-S, Femto Messtechnik GmbH and eLockIn 203, Anfatec Instruments AG) synchronized with the chopped monochromatic probe light.