Size-Tuning of WSe2 Flakes for High Efficiency Inverted Organic Solar Cells George Kakavelakis,†,⊥ Antonio Esau Del Rio Castillo,‡,⊥ Vittorio Pellegrini,‡ Alberto Ansaldo,‡ Pavlos Tzourmpakis,† Rosaria Brescia,§ Mirko Prato,§ Emmanuel Stratakis,∥ Emmanuel Kymakis,*,† and Francesco Bonaccorso*,‡ †
Center of Materials Technology and Photonics and Electrical Engineering Department, Technological Educational Institute (TEI) of Crete, Heraklion 71004, Crete, Greece ‡ Graphene Labs and §Nanochemistry Department, Istituto Italiano di Tecnologia, Via Morego 30, 16163 Genoa, Italy ∥ Foundation of Research and Technology (FORTH), Institute of Electronic Structure and Laser (IESL), Heraklion 71110, Crete, Greece S Supporting Information *
ABSTRACT: The development of large-scale production methods of two-dimensional (2D) crystals, with on-demand control of the area and thickness, is mandatory to fulfill the potential applications of such materials for photovoltaics. Inverted bulk heterojunction (BHJ) organic solar cell (OSC), which exploits a polymer−fullerene binary blend as the active material, is one potentially important application area for 2D crystals. A large ongoing effort is indeed currently devoted to the introduction of 2D crystals in the binary blend to improve the charge transport properties. While it is expected that the nanoscale domains size of the different components of the blend will significantly impact the performance of the OSC, to date, there is no evidence of quantitative information on the interplay between 2D crystals and fullerene domains size. Here, we demonstrate that by matching the size of WSe2 few-layer 2D crystals, produced by liquid-phase exfoliation, with that of the PC71BM fullerene domain in BHJ OSCs, we obtain power conversion efficiencies (PCEs) of ∼9.3%, reaching a 15% improvement with respect to standard binary devices (PCE = 8.10%), i.e., without the addition of WSe2 flakes. This is the highest ever reported PCE for 2D material-based OSCs, obtained thanks to the enhanced exciton generation and exciton dissociation at the WSe2-fullerene interface and also electron extraction to the back metal contact as a consequence of a balanced charge carriers mobility. These results push forward the implementation of transition-metal dichalcogenides to boost the performance of BHJ OSCs. KEYWORDS: tungsten diselenide, liquid-phase exfoliation, sedimentation-based separation, inverted organic solar cell, efficiency enhancement
T
and charge transfer through the formation of a bicontinuous interpenetrated charge transport network in the active layer.21,22 The introduction of electron and hole blocking (or transport) layers between the polymer donors and acceptors active layer and the anode and cathode, respectively, enables the extraction of photogenerated carriers to the respective electrodes.23 Bulk heterojunction OSCs are gaining increasing attention due to their low cost,24 lightweight,25 and versatility for large-scale fabrication on flexible substrates.26,27 However,
wo-dimensional (2D) crystals display unique chemical and physical properties.1−7 Among the wide range of available 2D crystals,8 transition-metal dichalcogenides (TMDs), formed by a metal (Mo, W, etc.) atom layer sandwiched between chalcogenide atoms (e.g., S, Se, or Te), are studied for a number of applications, ranging from electronics1,9,10 and optoelectronics1,11−13 to spintronics.14,15 In the context of photovoltaics, one promising application for TMDs16,17 is in organic solar cells (OSCs) and, more specifically, OSCs based on the bulk heterojunction (BHJ) concept.18,19 The BHJ structure, which exploits a p-type polymer donor and n-type fullerene acceptor materials (mainly composed by a polymer−fullerene matrix),20 maximizes the donor/acceptor interfacial area, providing exciton dissociation © 2017 American Chemical Society
Received: January 16, 2017 Accepted: February 27, 2017 Published: February 27, 2017 3517
DOI: 10.1021/acsnano.7b00323 ACS Nano 2017, 11, 3517−3531
Article
www.acsnano.org
Article
ACS Nano
bulk WSe2,84,85 as well as of layered crystals in general,5,11,73−75,86 offers a viable route to obtain stable dispersions of exfoliated flakes. A key advantage of LPE over other methods5 relies on the possibility of scaling up and thus being appealing for the industrial exploitation of 2D crystals. The LPE of layered crystals can be performed either in water87 or in organic solvents.84,85,88,89 However, considering applications in the optoelectronic field,1,8,9,11,90 both the presence of exfoliation agents, e.g., surfactants,85,91 or DNA,87 surrounding the WSe2 flakes and the use of high boiling point organic solvents84,85,88,89 could be detrimental for the final device performance. However, LPE of WSe2 leads to a broad distribution of flake sizes.84−87 Flakes with diverse area are expected to behave differently from the optoelectronic point of view, due to the edge and quantum confinement effects,92−94 as well as for the electronic coupling with the blend, which is key for the extraction of charge pairs.55 While it is known that the optimum morphology of the different components of the blend determines the BHJ OSCs performance,55,56 the broad distribution of 2D flakes obtained by LPE would make it not possible to get quantitative information on the interplay between WSe2 flakes, as well as 2D crystals in general, and fullerene domain sizes. To address this issue, in this work we exploit the LPE of bulk WSe2 crystals in ethanol, by combining an ultrasonication process with sedimentation-based separation (SBS) to produce three different WSe2 samples with fine-tuned area (∼70, ∼240, and ∼720 nm2) and thickness (∼2.6, ∼6.1, and ∼8.5 nm), i.e., named as samples 1, 2, and 3, respectively. The three samples above are chosen since their areas are about one-third, the same, and three times larger, respectively, than the average PC71BM domain size in our PTB7:PC71BM blend (see Figure S1 in Supporting Information). We use ethanol, a low boiling point organic solvent, since it is compatible with the PTB7:PC71BM blend preparation and can be easily removed by low-temperature annealing.95,96 The as-prepared WSe2 samples are mixed with PTB7 and PC71BM to obtain ternary (PTB7/WSe2/PC71BM) blends used as the active materials for the OSCs, and the resulting photovoltaic performance of the BHJ OSCs is measured to determine the impact of the different WSe2 flake morphologies on the PCE of inverted ternary blend BHJ OSCs. We show that the integration of WSe2 flakes of ∼240 nm2 in area (similar to the area of the PC71BM domain) and 6.1 nm in thickness gives the best result among the three different WSe2 samples used, yielding an increase by ∼15% of the PCE of the BHJ OSCs (PCE = 9.28%) with respect to standard devices, i.e., without WSe2 flakes (PCE = 8.10%). This is the highest ever reported PCE for 2D material-based OSCs. The PCE enhancement, due to the addition of the WSe2 with the above sizes, is ascribed to the optimization of the ternary blend morphology, i.e., area matching between the WSe2 flakes and the PC71BM domains, which determines (i) optimized charge transfer through the introduction of additional conducting bridges, i.e., conduction band of WSe2 flakes is in between the PTB7 and PC71BM ones, and (ii) exciton generation by the WSe2 absorbing flakes, enhanced exciton dissociation at the WSe2:PC71BM interface and electron extraction to the back metal contact due to balanced charge carriers mobility. The latter reduces charge accumulation at the interfaces, hindering the formation of a positive space charge build up at the
until now, the typical performance of OSCs, which is around 10%,28 does not match that of silicon (power conversion efficiency, PCE, ∼ 25.6%)29 and other inorganic materials, such as four junction GaInP/GaAs/GaInAsP/GaInAs (PCE ∼ 46%).30 The low PCE of OSCs is mainly due to the limited absorption width (approximately 100 nm),31 the low charge carrier mobility (10−3 to 10−5 cm2 V−1 s−1 for both electron and holes),32 and the low exciton diffusion length (∼10 nm)33 of the currently used polymer donors and acceptors active layer.34−36 An option to tackle the limitations of binary BHJ OSCs, and thus significantly improve their PCE, is the addition of a third component material into the active layer, thus forming a ternary BHJ OSC structure.37 Different materials have been explored yielding PCE enhancement with respect to the binary blend in different ways: by broadening the absorption bandwidth of the active layer,38 by means of either the addition of a second acceptor39 or second donor40 material, by contributing to energy or charge transfer,41 or, finally, by improving the photogenerated charge dissociation.42 In this context, the high charge carrier mobility (∼202 and ∼140 cm2 V−1 s−1 for electrons and holes, respectively)43 and the high absorption coefficient (105 to 106 cm−1, in the visible44−46 and near-infrared region47−49) of WSe2 fulfill the requirements of a third component in a BHJ OSC. Most importantly, the valence band position of single-layer WSe2 is calculated to be 4.9 eV, while its conduction band is positioned at 3.6 eV with respect to the vacuum level.50 The valence band maximum shifts from Γ (single-layer WSe2) to K (bulk WSe2), determining a shrink of the energy band gap with a crossover from direct to indirect.51 Therefore, WSe2 energy levels are placed in between the ones of the PTB7-donor (3.3 eV) and PC71BM-acceptor (4.3 eV), a widely used and highly performing donor−acceptor couple composing the active material,52−54 enabling an efficient electron-cascade transfer. Critical to the optimization of the aforementioned processes is the morphology of the final blend and, thus, how the third component mixes with the other two materials.55,56 For this purpose, WSe2 flakes with controlled morphological properties (i.e., area and thickness) should be homogeneously mixed in the thin layer (typically ∼100 nm) of polymer− fullerene composite53 to determine the optimal 2D flake morphology able to maximize the photovoltaic efficiency of the resulting device. This is a challenging step that requires the production of flakes with optimized morphology. To address this goal, several methods to produce flakes of TMD and control their morphology can be exploited, by using bottom-up1,57−64 or top-down65−75 approaches. While the bottom-up approach to obtain WSe2 flakes, for example, by using chemical vapor deposition (CVD),64,76 are costly, topdown approaches are much cheaper and affordable, as, for example, the micromechanical cleavage (MC).77 However, although MC provides high-quality WSe2 flakes in terms of crystallinity, lateral size (up to micron scale), and thickness (single-layer crystals),4,78 it is not a scalable technique. Exfoliation of bulk layered WSe2 can also be performed by intercalation of lithium compounds79,80 or Lewis base intercalates, such as sodium ethoxide and sodium hexanolate.81 However, these methods have several shortcomings, such as the small lateral size of the flakes (
