Scholarly article on topic 'Graphene and transition metal dichalcogenide nanosheets as charge transport layers for solution processed solar cells'

Graphene and transition metal dichalcogenide nanosheets as charge transport layers for solution processed solar cells Academic research paper on "Materials engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Materials Today
OECD Field of science
Keywords
{}

Abstract of research paper on Materials engineering, author of scientific article — Nikolaos Balis, Emmanuel Stratakis, Emmanuel Kymakis

This review focuses on the recent progress and critical aspects of the utilization of solution processable graphene oxide (GO) and graphene-like transition metal dichalcogenides (TMDs) nanosheets as either the hole or the electron transport extraction layer in solution processed solar cells (SPSCs), including organic, dye-sensitized, quantum dot and perovskite solar cells. GO and TMDs are becoming very attractive due to their solution processability and more importantly due to their tunable electronic structure via proper functionalization routes, which in effect can optimize the charge transport process to the collection electrodes. Challenges and future applications of these materials in large area SPSCs, as well as the transition of the related technology from the laboratory scale to the industry will be analyzed and discussed.

Academic research paper on topic "Graphene and transition metal dichalcogenide nanosheets as charge transport layers for solution processed solar cells"

Materials Today • Volume 00, Number 00• April 2016

RESEARCH

.„JSitrfxmmg.

ELSEVIER

Graphene and transition metal dichalcogenide nanosheets as charge transport layers for solution processed solar cells

Nikolaos Balis1, Emmanuel Stratakis2'* and Emmanuel Kymakis1'*

1 Center of Materials Technology & Photonics and Electrical Engineering Department, School of Applied Technology, Technological Educational Institute (TEI) of Crete, Heraklion 71004, Greece

2 Institute of Electronic Structure and Laser (IESL), Foundation for Research and Technology-Hellas (FORTH), Heraklion 71110, Greece

This review focuses on the recent progress and critical aspects of the utilization of solution processable graphene oxide (GO) and graphene-like transition metal dichalcogenides (TMDs) nanosheets as either the hole or the electron transport extraction layer in solution processed solar cells (SPSCs), including organic, dye-sensitized, quantum dot and perovskite solar cells. GO and TMDs are becoming very attractive due to their solution processability and more importantly due to their tunable electronic structure via proper functionalization routes, which in effect can optimize the charge transport process to the collection electrodes. Challenges and future applications of these materials in large area SPSCs, as well as the transition of the related technology from the laboratory scale to the industry will be analyzed and discussed.

Introduction

Energy related issues and environmental problems such as pollution and global warming are ringing the alarm bell to humanity. In the last decade, the large increase in world population has led to a significant increase of the emission of greenhouse gases from fossil fuels. In this context, clean and renewable energy, and in particular solar energy through photovoltaics (PV) is the most promising solution, with the overall European PV contribution to amount to 3% of Europe's electricity demand, while the PV contribution to the global electricity demand reached 0.87% in 2013 [1]. Although, crystalline silicon based solar cells currently dominate the market', the so called 3rd generation solar cells, which are generally referred to solution processable solar cells (SPSCs), in which molecules or inorganic clusters are the primary absorbers, have been significantly emerging due to their low energy payback time [2].

The low cost manufacturing of SPSCs is based on printing technologies of flexible solar panels [3], which can be easily integrated on the surfaces of interior or exterior building spaces. Moreover, in contrast to Si counterparts, their attributes such as

*Corresponding authors: Stratakis, E. (stratak@iesl.forth.gr), Kymakis, E. (kymakis@staff.teicrete.gr)

transparency, flexibility and high weight-specific power can open futuristic applications of solar energy. As shown in Fig. 1, these emerging thin-film solar cells can be categorized as (a) organic solar cells (OSCs), (b) dye sensitized solar cells (DSSCs), (c) quantum dot solar cells (QDSSCs) and (d) perovskite solar cells (PSCs). The corresponding power conversion efficiency (PCE) exceeds 10%, 14%, 9% and 21%, respectively [4,5]. Despite the progress, further breakthroughs concerning the performance as well as the stability of the SPSCs devices are required for potential commercialization.

DSSCs are photoelectrochemical cells and can be considered the most mature 3rd generation PV technology [6]. They typically consist of a nanoporous TiO2 scaffold anode, sensitized with a light absorbing dye molecule [7]. A liquid electrolyte is used to transport ions towards a counter electrode, while solid-state mesoscopic devices have also been developed [8]. A critical drawback concerning the DSSCs is their long-term stability under prolonged illumination and high temperatures, although significant progress has been recently accomplished [9].

Colloidal QDSSCs utilize solution-processed nanocrystals (e.g. PbS), also known as quantum dots (QDs), to absorb light [10]. Their success lies on the size-effect tunability of the QDs bandgap, enabling light absorption from the visible to the near IR and in

1369-7021/© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). http://dx.doi.org/10.1016/ j.mattod.2016.03.018 1

MATTOD-744; No of Pages 15 ARTICLE IN PRESS

RESEARCH Materials Today • Volume 00, Number 00• April 2016

FIGURE 1

Schematic illustration of the device architecture of the most important emerging SPSCs technologies.

particular, present the MX2 structure, where M is a transition metal of groups 4-10 (typically Mo, Nb, W, Ni, V, or Re) and X is a chalcogen (typically Se, Te or S) acting as semiconductors (MoS2, WS2), metals or semi-metals (TaS2, NbS2) [36]. Similar to graphene, TMDs are characterized by strong covalent bonds providing the stability of 2D crystals, as well as weak van-der-Waals (40-70 meV) forces between the planes. Such forces are prone to be broken, facilitating controllable liquid exfoliation in a range of common solvents [37]. Once exfoliated, thin TMDs nanosheets possess a high surface area and aspect ratio, comprising a combination of excellent electronic, mechanical and photophysical properties [38,39].

This review focuses on the utilization of solution processed graphene and TMDs as the charge transport layers on SPSCs, including hole and electron extraction layers of single junction devices, as well as recombination interlayers of tandem SPSCs.

Electron extraction layers (EEL) and hole extraction layers (HEL) are core elements of all SPSCs, facilitating a similar function in most cases. Specifically, their primary operations concern (Fig. 2):

turn, the development of multijunction solar cells using a single material system [11]. QDSSCs are improving rapidly, with a record certified cell efficiency of 9.2% [12], utilizing micron-scale pyramid features as part of a hierarchical structured device. Besides this, QDSSCs exhibit remarkable stability in air [13]. Key challenges include the understanding of the strong effect of the QD surface ligands on the performance, as well as the poor charge transport properties of the QD-based active layer due to the presence of mid-gap states or inherent disorder [14].

PSCs evolved from solid-state DSSCs, with leading certified efficiencies reaching 20.1% [15]. Perovskite [16] refers to a ABX3 crystal structure, e.g. hybrid organic-inorganic lead halide CH3NH3Pb(I,Cl,Br)3, and is characterized by wide optical bandgap and strong absorption, long carrier diffusion lengths, low recombination losses, low material cost, and bandgap tunability [17]. Future challenges include addressing with PSCs' high sensitivity to moisture, low device stability and incorporation of toxic lead [18].

Finally, OSCs employ a different architecture, as the active layer comprises a conjugated, electron donor, polymer and an, electron acceptor, fullerene derivative mixed in a bulk heterojunction (BHJ) blend [19]. The conjugated polymer absorbs light and the excitons created diffuse at the donor-acceptor interfaces [20]. The architecture is completed by a hole and a electron extraction layer, which are contributing to the unhampered flow of charge carriers through the device. The PCE of OSCs recently exceeded 11% [21]. Future challenges of OSC technology include addressing with OSCs' high sensitivity to oxygen, poor absorption and low device stability.

In general, all types of SPSCs require significant advances on several important issues including stability against degradation, cost reduction and performance enhancement. Owing to extraordinary physical and mechanical properties graphene [22-25] and its derivatives, graphene oxide (GO) and reduced graphene oxide (rGO), have been employed in various ways [26-28] as well as in almost every structural unit of SPSCs [29-32]. Following the discovery of graphene, scientists expanded their research interest in other 2D materials [33] including metal nitrides, metal oxides (MO) and TMDs, taking advantage of the impressive properties that such materials exhibit in monolayer form [34,35]. TMDs, in

FIGURE 2

(a) Energy level diagram and charge transport procedures within an OSC before and after the insertion of charge extraction layers. The main functions of such layers are the minimization of the energy barriers demanded for charge extraction and the selective extraction of a single carrier type and blocking of the opposite one; (b) spatial distribution of the squared optical electric field strength within OSCs with and without an optical spacer. The optical spacer addition leads to the maximization of the electric field strength into the active layer; (c) Atomic Force Microscopy (AFM) images of bare ITO, PEDOT:PSS and graphene-based HTLs. The corresponding mean roughness values are 2.26, 1.21, 2.52 and 2.50 nm. A charge transport layer should be as smooth as possible to facilitate the carrier transport to the respective electrodes. Reprinted with permission from Ref. [120] © Royal Society of Chemistry 2015.

Materials Today • Volume 00, Number 00• April 2016

(i) The minimization of the energy barriers demanded for charge extraction (Fig. 2a);Effective charge extraction is determined by the energy barrier height at the interface between the photoactive layer and the electrode. This difference can be lowered via engineering the interface contacts, thus creating high-quality ohmic junctions to facilitate charge transport and maximize the open circuit voltage (Voc) attained.

(ii) The selective extraction of a carrier type and blocking of the opposite one (Fig. 2a). Interfacial layers are usually employed into the aforementioned solar cells in oeder to selectively extract the one charge carrier type and block the other respectively. Charge extraction/transport interlayers are wide band gap materials to cope with the undesirable exciton quenching. Besides this, interlayers should dispose suitable energy levels to effectively extract the respective charge carriers.

(iii) The optical properties matching (optical spacer) between the various device layers (Fig. 2b) [40]; Insertion of an interlayer with suitable optical properties between the active layer and the anode, especially in the case of OSCs, leads to a beneficial light harvesting effect. Indeed, such interlayer could modulate the optical field inside the device and give rise to the electric field strength maximization into the active layer and thus enhancement of light absorption.

(iv) The smoothness of the interfaces between the collection electrodes and the photoactive layer (Fig. 2c). A successful interlayer should preserve the minimal roughness of the respective electrodes and, in turn, the uniformity of the contacts with the active layer. Besides this, the interlayer roughness could influence the morphology of the photoactive layer itself, since during the drying process, the blend components rearrange according to the surface energy of the attached interlayers.

In this context, the efficiency enhancement of SPSCs can be potentially realized through the optimization or replacement of

the currently state-of-the-art materials used as the charge transport and/or extraction layers. Hence, this review will only concentrate on the utilization of graphene and graphene-like TMDs as alternative charge transport/extraction layers for high-performance SPSCs, since it is by far the most promising and mature application with respect to both efficiency and stability. Moreover, the review deals for the first time with solution processable graphene and TMDs charge transport layers of all the types of solution processed PV devices, giving a benchmark outlook of the technology and contribute to the further development and scalability of 3rd generation PV technology.

The current state of the art of this technology is presented in Table 1. The future prospects and potential developments of graphene- and TMDs-based materials in this exciting field will be additionally delineated, providing emphasis to the smooth transition of the technology from the laboratory scale to the industry.

SPSCs incorporating graphene-based layers

Dye sensitized solar cells

The archetypical structure of a DSSC (Fig. 3a) comprises the photoanode, the cathode and the electrolyte. The photoanode consists of a transparent conductive electrode (TCE), usually a glass substrate coated either with fluorine tin oxide (FTO) or indium tin oxide (ITO), the semiconductor, usually a sintered titania (anatase phase) nanoparticle scaffold and the (organome-tallic or organic) dye, adsorbed on the semiconductor surface. The electrolyte primarily contains an iodide/triiodide redox couple dissolved in acetonitrile, while the cathode consists of a platinized FTO electrode. The photons enter the solar cell through the anode backside and get absorbed by the sensitizer, exciting electrons. The excited electrons are injected into the conduction band of the adjacent semiconductor and subsequently flow to the cathode, where there are transferred to the electrolyte. Finally, electrons are transferred to the sensitizer to regenerate it and complete the circuit.

TABLE 1

Record efficiencies of various SPSCs containing various 2D materials as charge transport layers.

Charge transport interlayer 2D material Modification SPSC PCE (%) Ref.

Electron extraction layer rGO TiO2 DSSC 8.67 [51]

rGO TiO2-CdS/CdSe/ZnS QDSSC 4.20 [102]

Graphene TiO2 PSC 15.60 [158]

rGO PCBM P3HT-based OSC 3.89 [144]

rGO ZnO PCDTBT-based OSC 6.72 [140]

rGO TiO2 P-based OSC 5.33 [143]

MoS2a P3HT-based OSC 2.73 [168]

Hole extraction layer GOa PSC 12.40 [164]

rGO Fluorine P3HT-based OSC 4.57 [129]

GO PEDOT:PSS PTB7-based OSC 8.21 [113]

GQDs DR3TBDT-based OSC 6.82 [126]

GO Ag NPs PBDTTT-CF-based OSC 7.54 [121]

GO VO(bilayer) PTh4FBT-based OSC 6.70 [114]

GO Cl PCDTBT-based OSC 6.46 [133]

GO Cl PTB7-based OSC 8.52 [147

MoS2 P3HT-based OSC 4.02 [167]

MoS2 PTB7-based OSC 8.11 [167]

Transport interlayer GO TiO2(bilayer) PSBTBT/PSEHTT-based tandem OSC 8.40 [152]

a Inverted structure.

MATTOD-744; No of Pages 15 ARTICLE IN PRESS

RESEARCH Materials Today • Volume 00, Number 00• April 2016

FIGURE 3

(a) Schematic representation of the structure of a DSSC. (b) Overview of processes and typical time constants under working conditions in a Ru-based DSSC with iodide/triiodide electrolyte. The red arrows indicate recombination processes. (c) Schematic illustration of DSSC devices and respective electron transport path. On the left, the reference device with TiO2 as anode semiconductor and N719 dye as sensitiser, is depicted. On the right, a DSSC with graphene-decorated TiO2 nanoparticle network is illustrated. (a) Reprinted with permission from Ref. [43] © 2009 American Chemical Society, (b) Reprinted with permission from Ref. [42] © 2010 American Chemical Society, (c) Reprinted with permission from Ref. [60] © 2015 Elsevier B.V.

Towards optimization of these devices, researchers have mainly concentrated on maximizing light harvesting, eliminating losses due to alternative electron transfer pathways and minimizing the overpotentials that govern the electron transfer towards the electrodes [41]. The main undesirable electronic processes in DSSCs are charge transfer between the photoanode and either the sensitizer (recombination) or the oxidized species of the electrolyte that have not been reduced by the platinized cathode (back transfer). Taking an insight into DSSC kinetics (Fig. 3b), one can realize that

extraction of electron towards anode encounters with back transfer as the time constants of both procedures are comparable [42,43]. As concerns the recombination problems between FTO and triodide, the insertion of a thin, dense electron blocking layer (20-100 nm) between FTO and titania scaffold, has been widely used. On the other hand, several approaches have been also proposed towards improving electron transport into the semiconducting scaffold, such as using one-dimensional [44] or core-shell [45] and vertically oriented nanostructures [46] or integrating carbon derivatives into the photoanode [47]. In this context, the incorporation of graphene-based materials into the photoanode has been widely explored [48,49]. The main goal is to leverage graphene's high conductivity as well as the ability to tune the work function (Wf) of graphene-based derivatives to match it with the TiO2 conduction band (Fig. 3c). Both approaches could improve the electron extraction efficiency and mobility in DSSCs. Besides that, a wide range of interventions has been also realized in the cathode part of the devices. As mentioned before, the most competitive material employed as the catalyst is platinum. Nevertheless, research efforts are focusing on replacing it, mainly due to its scarcity and high-cost.

As concerns the anode, the first report regarding the insertion of graphene into the photoanode was by Kim et al. [50] who introduced a TiO2-graphene nanocomposite as a blocking layer in order to inhibit electron back transfer from FTO to triodide, while the same route was followed by other research groups [51,52]. More extensively explored, is the utilization of titania-graphene composites as the semiconductor scaffold. Yang et al. [53] were the first to realize TiO2-graphene and TiO2-carbon nanotubes composites onto FTO substrates, using a simple doctor blade technique. In particular, the graphene-based photoanodes showed superior performance by transporting the electrons through a versatile ultra conductive network, while suppressing recombination and enhancing light utilization.

More recently, other group [54] published similar results and attributed the remarkable PCE increase to graphene's beneficial incorporation into the semiconducting scaffold. Cheng et al [55] prepared graphene-TiO2 hybrid photoanodes following a two-step autoclave-hydrothermal route while Anjsusree and his coworkers [56] tested a 1-pyrene carboxylic acid stabilized gra-phene-titania composite in photocatalysis and DSSCs as well. They reported that there is an optimum graphene content, beyond which the performance of the devices becomes deteriorated. This is due to the shielded light-harvesting of dye molecules by graphene, which impairs the light adsorption of the dye-sensitizer and reduces the number of photogenerated electrons. Zhao [57] and his group prepared a mixed colloid of graphene and TiO2 using a one-step hydrothermal process and a graphene/TiO2 photoanode was successfully prepared. The measurements showed that the graphene/TiO2 film show high porosity and large specific surface area, allowing the increased adsorption of dye molecules. Zhu and co-workers [58] prepared graphene/TiO2 composites by following a one-step solvothermal route, while Sacco and co-workers [59] also reported improved performance for DSSCs attributed to graphene's beneficial contribution. Zhi et al. [60] presented an easy prepared hybrid composite film of graphene-titania on plastic substrate. While, using C28H16Br2 as the precursor, Liu et al. [61] presented a facile novel process to coat

Materials Today • Volume 00, Number 00• April 2016

TiO2 particles with few-layer graphene. Finally, Xu et al. [62] replaced the semiconductor titania scaffold with ZnO structures, giving rise to prolonged electron lifetime and reduced electron recombination losses. In this case, the DSSCs incorporating an optimal amount (1.2 wt%) of graphene showed a markedly PCE enhancement compared with the devices without graphene.

In a different concept, Xiang and his co-workers [63] used nitrogen doped rGO (n-rGO) as the graphene derivative. What is noteworthy in this work is that Voc was also improved, as the value of Voc is affected by the electron potential inside TiO2. After doping of TiO2 with n-rGO nanosheets, a carbon impurity level is formed inside the band gap and the electron density in the conduction band of TiO2 is increased. As a result the Fermi level is slightly shifted giving rise to Voc increase. Following a different route, Fang et al. [64] reported a remarkable PCE enhancement via incorporating graphene quantum dots (GQDs) to exploit hot electrons as well as multi electron photogeneration by a single photon. The maximum PCE reported was 6.10%, which was 19.6% higher than that of the non-graphene based reference device. Finally, following an entirely different approach, Neo et al. [65] used GO as an auxiliary binder in TiO2 pastes. In this case, contrary to previous reports, GO is acting as a structural coherent agent enabling the fabrication of TiO2 films with improved structure.

As far as the mechanism for improved performance is concerned, Fang et al. [66] reported that the Jsc and PCE enhancement of the graphene-based DSSCs were preliminary attributed to the increased dye absorption due probably to the increase of the films' specific surface area, as well as the improved conductivity of graphene. While, using conductive AFM measurements, He et al. [67] revealed a size-dependent electron transport property of the TiO2-graphene composite photoanodes. They showed that, as the titania particle size is decreasing, a more continuous electron transport network is formed that contributes to higher electron mobility.

On the other hand, the incorporation of graphene into the cathode is a widely explored option, since platinum has to be replaced due to highly-cost and scarcity reasons. The ideal counter electrode should present high conductivity and catalytic activity, as well as low cost. Graphene derivatives are competitive alternatives although their conductivity and catalytic properties have counter proportional values, since the oxygen-defect sites of gra-phene lattice increase their catalytic attitude, but simultaneously reduce their conductivity. However, the most efficient DSSCs proposed till now, utilize a graphene-based counter electrode. Specifically Kakiage et al. [68] reported a highly efficient device consisted of an Au/graphene cathode resulting to a 14.3% PCE. The electrolyte used was based on a cobalt (III/II) complex while the titania scaffold was sensitized through the synergistic effect of a silyl-anchor and a carboxy-anchor organic dye. A similar approach was followed by Yang et al. [69] who also employed a metal-graphene complex as the catalyst. In an analogous way, Mathew et al. [70] presented a DSSC based also on a cobalt (III/II) complex and on a graphene-based electrocatalyst as the cathode. The PCE achieved (13%) proves the highly potential of graphene even as a sole material in the cathode.

Amid the alternative graphene-based approaches that have been proposed towards replacing platinum, including graphene-poly-mer [71], graphene-metal [72,73], graphene-carbon [74,75] or heteroatom doped-graphene electrodes [76,77], the most effective

attempts include sulfur, nitrogen of boron doped-graphene cathodes. Yu et al. [78] synthesized a N-doped graphene electrode presenting an improved catalytic activity in comparison to the reference materials due to the high charge polarization stemming from the difference in electronegativity between carbon and nitrogen. Following a similar route Jeon et al. [79] reported an F-doped graphene electrode, while Jung et al. [80] introduced a B-doped graphene material as electrocatalyst. In both cases, a cobalt complex was utilized as the electrolyte along with an organic compound as the dye, further indicating that graphene-based materials overwhelm the platinum counterparts at least concerning the DSSCs containing a Co(II)/Co(III) complex.

Quantum dot sensitized solar cells

In principle, QDSSCs adopt a similar design to that of DSSCs and follow the same operation principles (Fig. 4). The main differentiation is the replacement of the organic dye by inorganic QDs, such as CdTe [81], CdS [82], CdSe [83], PbS [84] and ZnS [85]. QDs are inorganic semiconductor nanocrystals with a size comparable to the Bohr radius of the corresponding exciton. Owing to such unique structure, QDs are characterized by a strong quantum confinement effect [86], high extinction coefficient [87], tunable band gap [88] and large intrinsic dipole moment [89]. The most substantial consequence of the quantum confinement effect is that the QD energy band gap, Eg, is dependent on the particle size, i.e. Eg ~ 1/r2, where r is the radius of the QD [90].

FIGURE 4

(a) The architecture and (b) the energy band diagrams of a QDSSC, incorporating a ZnO-graphene/CdS/CdSe nanostructure as the photoanode. Reprinted with permission from Ref. [101] © 2014 Elsevier B.V.

RESEARCH

What is considerably attractive in QDs is their potential in multiple electron generation [91] and hot electron utilization [92]. As a result the theoretically predicted conversion efficiency of QDSSCs is elevated to 44%, which is far beyond the value of 31% for silicon-based solar cells due to the Schockley-Queisser limit. Nonetheless, the record efficiency measured for such devices still lags behind that of DSSCs due to electron-hole recombination at the QDs' surface states, carrier recombination at the QD/semicon-ductor scafold interfaces and electron recombination (scavenging) of free electrons by the electrolyte [93].

Towards addressing those hurdles, a series of alternative routes [94,95] have been proposed and in this context, graphene has attracted major interest as a candidate for improving QDSCs' performance. This is mainly due to the property of ballistic electron transport within graphene nanosheets. The first work on graphene-based QDSSCs was carried out by Guo et al. [96], who fabricated a photoanode composed of successive layers of gra-phene and CdS. Graphene was electrophoretically deposited on FTO while CdS was successfully applied by chemical bath deposition (CBD). Recently, Barpuzary and his co-workers [97] reported an enhanced PCE of 2.82% by fabricating a solar cell with a ZnO-CdS-GO photoanode, using a simple doctor blade technique. The ternary system formed demonstrated an enhanced electron transfer due to the band alignments of ZnO, CdS, and GO within the composite. Besides this, GO is functioning as electron acceptor, giving rise to enhanced separation of charges generated under photoexcitation.

Zhao et al. [98] followed a different synthetic route, using an autoclave-hydrothermal technique to synthesize the GO-TiO2 composite while CdS QDs were incorporated by successive ionic layer adsorption and reaction method (SILAR). It is shown that the electron transport was improved and the total PCE was increased from 1.07% to 1.70%, while a reduced charge transfer resistance was demonstrated by electrochemical impedance spectroscopy (EIS) measurements. Kumar et al. [99] incorporated few layered graphene platelets into the photoanode, noncovalently function-alized with pyrenecarboxylic acid. Graphene platelets were electrophoretically deposited on TiO2 while CdS and PbS QDs were adsorbed using the SILAR method. Graphene served as an efficient electron acceptor due to its high electrical conductivity (2.55 Scm-1) and suitable Wf, both of which facilitate electron propagation. These results were supported by emission decay measurements.

Several other groups reported on the beneficial effect of gra-phene in QDSSCs. He and his group [100] adsorbed CdS/CdSe/ZnS by CBD method achieving a total PCE of 4.02%. According to the authors, graphene is a high rate electron diffusion medium and thus it leads to improved collection efficiency of photoelectrons. In addition to this, the graphene-titania composites enhanced the light harvesting efficiency via exploiting the light scattering effect. As shown in Fig. 4, Ghoreishi et al. [101] used ZnO as the semiconductor scaffold and CdS/CdSe as the sensitizers. Using EIS measurements, they confirmed an improvement in electron lifetime and resistance recombination decrease, under the presence of graphene. While, Zhu et al. [102], achieved a total 4.20% PCE via spin coating a layer of graphene-TiO2 slurry onto FTO, followed by CdS and ZnS adsorption by CBD and SILAR methods respectively. They concluded that graphene frameworks acted beneficially

compared to 2D-rGO sheets, acting as more efficient electron transfer mediums into TiO2 scaffolds. Nevertheless, undesirable charge recombination was revealed by EIS measurements, indicating that graphene frameworks may additionally act as recombination traps.

Organic solar cells

OSCs exhibit a different architecture that DSSCs and QDSSCs. The photoactive layer comprises a conjugated polymer and a fullerene-based material, blended into a bulk heterojunction (BHJ) morphology, consisting of donor and acceptor domains extended to a lengthscale of approximately 20 nm. The planar bi-layer structure is inefficient in this case, considering that the exciton diffusion length in conjugated polymers is limited to ~10nm. The light excited excitons in the polymer donor are tightly bound (~1 eV) and can dissociate only due to the energy band offset between the lowest unoccupied molecular orbitals (LUMO) of the blended materials. Besides this, dissociation takes place as long as the interface band gap between the LUMO of the fullerene and the highest occupied molecular orbital (HOMO) of the polymer is less than the exciton energy. Following exciton dissociation, a free electron and a free hole are created which are transporting towards the low work function cathode and the high work function anode respectively. The configuration is completed with the insertion of an EEL and a HEL, as energy intermediates, between the electrodes and the blend. A HEL should have a suitable Wf to allow for the built-in electrical field across the active layer and for holes to transport towards the anode. Similarly, an EEL should exhibit a low Wf for electrons to efficiently transport to the cathode.

The most common material used as HEL is PEDOT:PSS. It is highly conductive and exhibits the suitable Wf for efficient holes transport. However, its acidic nature is a major drawback and detrimental for devices' stability. Other materials commonly used are inorganic metal oxides (e.g., MoO3, V2O5, NiO) [103-105]. Concerning the EELs, the most efficient materials are n-type inorganic semiconductors, including ZnO, TiO2 and TiOx, n-type organic semiconductors [106] as well as metals and salts [107,108]. Nonetheless, the range of candidate HEL and EEL materials is restricted due to the organic nature of the polymer active layer, considering that high temperature annealing and vacuum deposition techniques may be incompatible with the sensitive photoactive polymer blend. In this context, graphene and its derivatives are extensively explored as extraction layers in order to exploit their great advantages including solution processability, facile deposition techniques, low cost fabrication and versatility in functionalization processes [109,110].

Graphene-based hole extraction layers

Li et al [111] first reported the successful replacement of PED-OT:PSS by a 2 nm GO layer. Similar to PEDOT:PSS, GO adheres to fundamental characteristics of a HEL including smoothing of ITO surface, adequate transparency and a suitable Wf of 4.9 eV. Besides this, as transient open-circuit voltage decay measurements showed, in the presence of GO HEL, the recombination rate in poly(3-hexylthiophene-2,5-diyl) (P3HT)-based OSCs, was significantly reduced, establishing GO as an effective hole transport material. While, Kim et al [112 fabricated solar cells based on poly({4,8-di(2-ethylhexyloxyl)benzo[1,2-b:4,

Materials Today • Volume 00, Number 00• April 2016

5-b0]dithiophene}-2,6-diyl)-alt-({5-octythieno[3,4-c]pyrrole-4,6-dione}-1,3-diyl):fullerene (PBDTTPD:PCBM) blends as the photoactive layer. The presence of GO benefited the stability of solar cells while the device with GO as the HEL yielded a record PCE of 5.5%. Following another route, Yu and his co-workers [113] attempted to combine PEDOT:PSS with GO and via an easy solution-mixing method, to synthesize an effective HEL. The final PCE yielded for the PEDOT:PSS-GO device was superior than the one based on PEDOT:PSS. Finally, in a different approach, other groups [114,115] utilized composites of GO with transition metal oxides as HEL in OSCs. Such composites enabled inverted polymer solar cells to exhibit high Voc and fill factor (FF) values, as well as an improved PCE of 6.7%.

One of the crucial parameters of a HEL is its thickness and since GO is an insulator it has to be quite low in order for the charge transfer to take place through a quantum mechanical tunneling process. In this context, the development of stable and homogenous GO films of a few nm with simple coating techniques is desirable. Indeed, in Li's report the thickness of GO layer played a crucial role in devices' performance as the fill factor (FF) dramatically decreased as the GO layer thickness was increased up to 10 nm. However, the most serious drawback of GO is its low conductivity. Therefore, following Li's work, many reports [116,117] focused on addressing this issue via utilization of rGO instead of GO, incorporation of conductive agents into the GO layer and functionalization of GO. Nevertheless, in situ thermal reduction of GO restricts its application on plastic substrates, while rGO utilization is hampered by its low solubility in most common solvents. All these issues have to be taken under consideration and actually define the future research directions on graphene-based HELs.

It should be also noted that the GO HEL is actually modified to a partially reduced GO (prGO), during either the subsequent fabrication steps by thermal annealing or during illumination by photoreduction [118]. Therefore, GO refers to the starting material and not the actual HEL, which is a prGO.

In this context, Murray et al. [119] treated a GO HEL with UV irradiation in order to reduce it and at the same time to increase its Wf to match the energy levels of the poly [[4,8-bis[(2-ethylhexyl) oxy]benzo[1,2-b:4,5-b0]dithiophene-2,6-diyl][3-fluoro-2-[(2 ethyl-hexyl) carbonyl]-thieno[3,4-b] thiophenediyl]] (PTB7) donor. It is shown that the resulting prGO layer templates more effectively the PTB7 p-stacking orientation, which is favorable for charge extraction. In addition, although the performance of prGO-based device was comparable to the one with PEDOT:PSS, the improved stability seems to remain prGO's key advantage. Following a different route, Jun-Seok Yeo et al. [120] used rGO as the HEL by employing p-hydrazinobenzene sulfonic acid hemihydrate as the reduction agent. The resulting srGO presented a very high dispersion concentration (20 mg ml-1) in pure water, without the assistance of insulating surfactants, and a high conductivity of 3.18 Scm-1. Besides this, the Wf of srGO exhibited a higher value (5.04 eV) compared to other graphene derivatives, rendering it compatible with the HOMO level of several donor polymers. The srGO was successfully employed in solar cells with P3HT, PTB7 and poly[4,8-bis(2-ethyl-hexyloxy)benzo[1,2-b:4,5-b0]dithiophene-2,6-diyl-alt-(4-octanoyl-5-uoro-thieno[3,4-b]thiophene-2-carboxylate)- 2,6-diyl] (PBDTTT-CF) as polymer donors along with. In the presence of srGO the PCE

was improved as concerns the P3HT-based device from the initial 3.56% to 3.64% while the PTB7 and PBDTTT-CF based OSCs exhibited PCEs of 7.18% in both cases, values highly comparable to those of PEDOT:PSS-based OSCs. An even improved efficiency regarding to PBDTTT-based OSCs was demonstrated by Yuan et al. [121] who inserted a GO-Ag nanoparticles layer between ITO and PEDOT:PSS. By leveraging the strong coupling between the plasmon resonance effect, of GO-Ag nanoparticles and incident light and the subsequent improved light absorption and exciton generation rate, the final PCE yielded was increased by 15% resulting in a 7.54% for the PBDTTT-based device.

Apart from reducing GO there are other routes to cope with GO's insulating nature. Jaemyung Kim et al. [122] leveraged single wall carbon nanotubes in an effort to improve GO's conductivity and at the same time to allow the fabrication of thicker films. Another approach was developed by Liu et al. [123], who developed a well-soluble sulfated reduced GO, with -OSO3H groups attached to the carbon basal plane of rGO (i.e., rGO-OSO3H). The OSC device using the rGO-OSO3H HEL exhibited a PCE of 4.37%, which was comparable to the one based on PEDOT:PSS (4.39%). Our group worked on the replacement of PEDOT:PSS [124] via incorporating Au nano-particles into GO and depositing a GO-Au HEL in OSCs based on P3HT and PC60BM. The resulting OSCs exhibited enhanced stability, while the PCE was increased from 2.90% for the GO-based device to 3.37% for the device containing Au nanoparticles, outperforming in both occasions the OSCs based on PEDOT:PSS. A similar work was published by Chuang et al. [125] using a P3HT:1',1",4',4" Tetra-hydrodi[1,4]methanonaphthaleno[1,2:2',3',56,60:2",3"] [5,6]ful-lerene-C60, C60 derivative, indene-C60 bisadduct (ICBA) pair in the active layer, yielding a total PCE of 5.05%. Li and his co-workers [126] developed a HEL based on GQDs and tested it in polymer-based and small molecule-based solar cells. GQDs presented superior conductivity and more homogeneous morphology compared to GO and PEDOT:PSS, as well as improved stability. Besides this, the PCEs achieved were similar to the reference devices. Finally, an interesting approach was carried out by other groups [127,128], who employed 1D graphene derivatives as HEL in P3HT-based and PTB7-based devices. Especially in the case of graphene nanoribbons-carbon nanotubes the film coating was realizedusing inkjet printing technology proving the versatility of graphene and the potential of fabricating highly efficient devices, since the solar cells prepared yielded a total PCE of 7.60%.

Another, well-reported, approach to enhance the performance of GO-based HELs is via tuning of their energy levels and thus Wf values, by proper functionalization routes. Using a phenylhydra-zine-based reductant containing fluorine atoms, Kim et al. [129] reported on a novel functionalized rGO material, taking advantage of the higher Wf (4.97 eV) of fluorinated GO (FrGO), compared to GO and rGO. As a result, the devices fabricated with FrGO exhibited similar performance to that of PEDOT:PSS-based ones, however they outperformed those in stability. Another route was proposed by Li et al. [130] who developed a simple modified Hummers' method for synthesizing a series of graphene oxides with precise oxidation (pr-GO) via strictly controlling pre-oxidation steps, oxidant content and oxidation time, resulting in different values of the Wf, ranging from 4.74 to 5.06 eV. The larger electronegativity of oxygen atoms in pr-GO, creates surface Cd+-Od- dipoles via extraction of p electrons from graphene lattice which leads to stronger interfacial

RESEARCH

dipole moments at the interface with ITO. As a result, the ITO levels shift away from the vacuum level, giving rise to a higher Wf value. The final PCE yielded was 3.74% for a P3HT:PCBM-based device superior to the 3.60% achieved with the PEDOT:PSS-based counterpart. An alternative method towards controlling the reduction, thus the Wf of rGO, was carried out by Noh et al. [131], via g-ray irradiation of a GO solution in ethanol and water. While, Chen et al. [132] tackled with the dilemma between GO and rGO by simply employing rGO into a GO solution. Finally, our group [133] demonstrated a different approach based on GO Wf tuning via a laser-induced doping technique (Fig. 5a). In particular, following pulsed laser irradiation of ultrathin GO films in the presence of a dopant chloride precursor gas, a simultaneous reduction and Cl-doping of GO lattice was realized. It is shown that the chloride atoms are bonded to both the edges and the basal plane of the GO lattice, leading to a tunable Wf for the irradiated layers, as a function of the laser exposure time. The reason for the increased Wf is attributed to

the formation of C-Cl dipoles (Fig. 5b) with different electronegativity, resulting in downward shift of the Fermi level and the consequent increase of Wf. PCDTBT: PC71BM-based OSCs devices using photochlorinated GO as HEL exhibited a remarkable increase in PCE, compared to reference PEDOT:PSS-based devices (Fig. 5c).

Graphene-based electron extraction layers

The first effort to employ a graphene-based EEL in OSCs was carried out by Liu et al. [134] who neutralized GO with Cs2CO3 to yield a GO-Cs derivative. Compared to GO, the Wf of GO-Cs was significantly lowered (3.9-4.1 eV) to energetically match well with PC61BM's LUMO, thus facilitating electron extraction and transportation towards the cathode. As a result, the performance of GO-Cs-based OSCs was comparable to the reference, which used a LiF EEL. At the same time, when GO was used as HEL as well, the PCE yielded was higher than the reference one. A similar approach was followed by Yang and his co-workers [135] who investigated a

FIGURE 5

(a) Wf of GO-Cl films as a function of the exposure at different number of laser pulses, Np. The inset shows the dependence of Cl-doping (/(Cl2p)//(CqS)) with the reduction degree ((/(C-C)//(C-O)); (b) 3D chemical structure of the photochlorination of the GO, illustrating the formation of dipoles by Cl atoms in the form of Cl-C covalent bonds at the edges and/or Cl-C=0 groups located outside the graphene basal plane; (c) J-V characteristics of PCDTBT:PC7qBM photovoltaic devices with different HELs under simulated AM 1.5, 100 mW cm~2 solar irradiation; (d) Schematic energy level diagram showing the effect of polar C-Cl bonds of the work function. Reprinted with permission from Ref. [133] © Royal Society of Chemistry 2014.

Materials Today • Volume 00, Number 00• April 2016

Cs2CO3 modified GQDs (i.e GQDs-Cs2CO3) layer as EEL in inverted OSCs based on P3HT:PCBM. GQDs-Cs2CO3 presented a suitable Wf value of 4.18 eV in order to effectively act as EEL. Moreover, the respective OSCs fabricated presented superior stability against devices with Cs2CO3 alone as EEL. Besides this, the presence of Cs+ ions on GQDs in GQDs-Cs2CO3 composites further retarded Cs+ ion diffusion into the polymer layer, leading to improved device stability.

Jayawardena et al. [136] reported highly efficient OSCs based on PTB7, incorporating rGO-ZnO and rGO-TiO2 nanocomposites as EEL. The PCE yielded was superior in both cases (7.50% for the former and 7.46% for the latter) against devices containing individual ZnO (7.39%) and TiO2 (7.22%) as EELs. Moreover the researchers compared the OSCs containing rGO-MO composites with devices containing bathocuproine as EEL. In the latter case, the PCE obtained (7.47%) was quite comparable to the rGO-MO-based OSCs, as the presence of rGO played a key role in balancing the hole and electron mobilities. In this context,other groups have also reported the utilization of ZnO-rGO nanocomposites as cathode layers in inverted OSCs. Chen et al. [137] utilized a complex amphi-philic fullerene-end-capped poly(ethyleneglycol) (C60-PEG) agent

as to further modify the ZnO nanorods-rGO sublayer, aiming to improve the compatibility between the inorganic part and the organic PTB7-based active layer. As elsewhere reported [138] PVP favors homogeneous distribution of the RGO due to the strong p-p interactions between graphene and PVP molecules and also acts as a stabilizer that controls in situ growth of sol-gel-derived ZnO nano-particles on the graphene. Hu et al. [139] also reported a high efficient (8.4%) PTB7-based device utilizing a novel ZnO-graphe-ne:ethyl cellulose nanocomposite by in situ formation of ZnO nanocrystals in a highly uniform graphene sheets via liquid ultrasound exfoliation. Ethyl cellulose not only increases the dispersion of graphene but also maintains its high-conductivity.

The beneficial presence of rGO-MO composites as EEL was also attested in PCDTBT-based OSCs by Beliatis and his co-workers [140], in P3HT-based OSCs by Zhang et al. [141,142] and in devices incorporating a low band gap quinoxaline based donor-acceptor copolymer by Sharma and his co-workers [143].

Qu et al. [144] utilized as EEL an rGO-fullerene composite, by attaching PCBM onto GO lattice via a pyridine moiety. The rGO-PCBM nanocomposite presented higher solubility compared to rGO, while the Wf value, measured via kelvin probe microscopy

0 2 4 6 S 13 1ft IT 1» 1» » 06 aj 01 OA

Binding Energy <oV) Binding Energy (oV) v (Voluge)

FIGURE 6

(a) The energy level diagram depicting the energy levels of all materials used in OSC; (b) UPS valence band region (the inset represents the expanded view of the shallow valence features) and secondary cut-off region (c) for GO and GO-Li (expanded view of the secondary cut-off region features); (d) J-V characteristics of PCDTBT:PC71BM photovoltaic devices with different EELs. (b)-(d) Reprinted with permission from Ref. [145] © 2014 American Chemical Society.

RESEARCH

measurements, was low enough as to match well with the LUMO of the electron acceptor. The PCE yielded was 3.89% for an OSC based on P3HT, quite improved compared to OSCs without an EEL (3.39%).

A quite effective graphene-based derivative was synthesized and inserted as an EEL into OSCs by our group [145]. Specifically, a GO-Li layer was inserted between TiOx and the photoactive blend acting not only as an interfacialengineering material that leads to improved EEL-cathode ohmic contact and enhanced electric field amplitude into the active layer, but also as a protection layer against oxygen and humidity, providing better device stability (Fig. 6). In this case, the PCE of 6.29% measured for GO-Li based device (Fig. 6d) was superior to that of devices using GO (4.89%) or no interfacial layer (5.51%). The improved performance was attributed to the lower, compared to GO, Wf value of GO-Li (4.3 eV), estimated by ultraviolet photoelectron spectroscopy (Fig. 7b and c). As a result the Wf of GO-Li displays a perfect match with the LUMO level of PC71BM (4.3 eV). The Wf lowering was attributed to the presence of the less electronegative Li atoms. In particular, when bonded to GO, Li atoms lose their valence electrons to the GO plane, and the resulting Li+ ions induce dipoles [146]. This charge transfer from Li

to GO plane shifts the Fermi level towards the vacuum, yielding a Fermi level difference between the two materials of 0.67 eV, which is responsible for the decrease in Wf.

Furthermore, it was recently demonstrated by our group that the simultaneous utilization of Wf tuned functionalized GO derivatives as both the HEL and EEL can lead to a significant increase in the PCE of high efficient OSC, leading to a record PCE of 9.14% for graphene based transport layers [147]. The Wf tuning of GO took place by either photochlorination for Wf increase [133], or lithium neutralization for Wf decrease [145]. In this way, the Wf of the GO-Cl layer can be perfectly matched with the HOMO level of the polymer donor, enabling excellent hole transport. While the Wf of the GO-Li can be perfectly matched with the LUMO level of the fullerene acceptor, enabling excellent electron transport. The PCE enhancement was attributed to reduced recombination at the interfaces, balanced charge mobilities, and enhanced electric field in the OSC.

Graphene-based transport interlayers in tandem OSCs

Tandem solar cells are multijunction photovoltaic devices comprised of two or more subcells stacked together in order to achieve

FIGURE 7

(a) Cross-sectional SEM image of an ITO/rGO/Perovskite/PCBM/BCP/Ag device. The corresponding energy-level diagram for each layer shown in (b); (c) UV-vis absorption spectra of CH3NH3Pbl3 films on glass/ITO/PEDOT:PSS, glass/ITO/GO, and glass/ITO/rGO; (d) J-V curves for PSCs with various HELs incorporating rGO. Reprinted with permission from Ref. [165] © 2015 Elsevier B.V.

Materials Today • Volume 00, Number 00• April 2016

a higher overall efficiency. Tandem devices architecture has attracted great interest but unlike devices with inorganic materials, the progress in tandem OSCs, especially in solution-processed ones, is hindered by difficulties during multiple solution deposition steps. Among the key layers of a tandem OSC is the inter-conection layer (IL) that separates and simultaneously connects the successive devices. This layer has to be sufficiently continuous, compact, conductive, highly transparent, of low surface roughness, and high mechanical and chemical stable. In this context, many materials including TiOx/PEDOT:PSS [148], ZnO/PEDOT:PSS [149] and Al/TiO2/PEDOT:PSS [150] have been proposed as interconnection layers.

The first graphene-based IL in tandem OSCs was reported by Tung et al. [151], who employed a GO/PEDOT:PSS nanocomposite as the IL in a two-cell configuration, based on P3HT as the donor material. The two separate subcells were also measured, together with control devices with PEDOT and GO ILs respectively. The subcells were connected in series, thus the researchers obtained an increased Voc of 0.94 V and a final PCE of 4.14% for the tandem device, higher than the respective values of Voc (0.59 V and 0.53 V) and PCE (2.92% and 3.75%) for the subcells. The presence of GO in the composite was crucial as it increased the conductivity of PEDOT:PSS leading to a dispersion with increased solution viscosity and subsequently highly adhesive properties.

In another report, Yusoff et al. [152] incorporated a GO/TiO2 recombination layer, into a tandem OSC device comprising two subcells in series. The total Voc (1.62 V) was measured to be the sum of the Voc values obtained for the subcells (0.94 V and 0.68 V), indicating that the interconnection realized was ideal, offering a resistance-free adhesion for the front and the rear cell. The tandem OSCs fabricated were all-solution processed and stable, while the TiO2/GO IL exhibited a large LUMO and HOMO contrast between its two interfaces with the bottom and top active layers, leading to low absorption losses. As a result the PCE attained reached 8.40%.

Finally, another interesting approach was carried out by Chen et al. [153], who introduced a GO-Cs/Al/GO/MoO3 IL between two PCDTBT-based subcells. The results obtained, indicated that the GO-based IL provided an efficient recombination region for electrons and holes generated from the front and rear cells, due to excellent energy level alignment of the materials involved. The PCE obtained was 3.91% with a Voc of 1.69 V, which was almost the sum of the Voc values of the respective subcells, indicating the beneficial role of the graphene-based recombination layer.

Perovskite solar cells

A common characteristic of both DSSCs and QDSCs is the presence of a liquid part, namely the electrolyte, in the structure of the devices. Considering the low PCE attained for both technologies, alternative strategies proposed towards all-solid devices [154,155]. A great advancement towards this direction was the introduction of organometal trihalide perovskites as light-harvesting materials. The strong and broad absorption of those materials led to highly efficient PSC photovoltaic devices, which are currently demonstrated to have the highest ever reported PCE values in SPSCs [156]. The archetypical structure of PSC is analogous to DSSCs with the perovskite layer substituting the dye and a hole transport material, commonly 2,29,7,79-tetrakis-(N,N-di-p-methoxyphenyl-amine)-9,99-spirobifluorene (SpiroMeOTAD), replacing the electrolyte.

Moreover, the perovskite layer is coated on the top of a mesos-tructured layer of a MO, Al2O3 or TiO2, while the structure is completed with a compact TiO2 blocking layer between perov-skite/MO and FTO (Fig. 1c).

Graphene-based electron extraction layers in PSCs

The EEL seems to be of major significance in PSCs, as it facilitates the selective charge collection by the anode. MO materials are commonly used for the anode, however their use restrains the potential of plastic substrates utilization, considering the high temperature sintering required to improve the crystallinity and conductivity of such oxides.

Graphene composites may offer the conductivity that is missed when the MO EEL is processed in too low temperatures [157]. Wang et al. [158] proposed a TiO2-graphene composite as EEL, achieving a 15,6% PCE, superior to devices without EEL, with titania-only or graphene-only layers. The layer was processed in low temperatures, up to 150 °C. Different graphene contents were tested and similar to the case of DSSCs and QDSCs, there is an optimum content beyond which, graphene is acting rather as a recombination network than a conductive pathway. As graphene's Wf matches that of TiO2 and FTO, graphene is functioning as an energy intermediate, thus it is reducing the energy overpotential behaves as an electron funnel towards the anode.

Zhu and co-workers [159] reported an improved PCE in PSCs via inserting a graphene-QDs (GQDs) EEL between the perovskite and TiO2. The fabricated cells are characterized by strong difference among the electron and hole injection times (0.4 and 0.6 ns) on one side and hot electrons cooling time (fs scale) on the other. According to that work and based on transient absorption measurements, the devices containing graphene presented improved performance, based on achieving high electron transport times from perovskite to TiO2 conduction band. This result led to a 15.2% improvement for the GQDs-containing device attributed to energy coupling enhancement as well as improved electron transfer.

Lastly, Arresti and coworkers [160] utilized GO-Li as the EEL in mesoporous PSCs, by inserting it between the sensitizer and the m-TiO2 layer. In this way, the electron injection from the sensitizer to the TiO2 layer was remarkably improved leading to an increase in the Jsc. Moreover, the GO-Li insertion resulted in a reduction of about 50% of the cells J-V curves hysteresis, stemming from the passivation of m-TiO2 oxygen vacancies that are the main reaction centers of moisture attack [161].

Graphene based hole extraction layers in PSCs

Less explored is the use of graphene derivatives as HELs, since the materials already used are quite competitive. Initially, Li and co-workers [162] inserted a GO buffer layer between perovskite and Spiro-MeOTAD. As a result, the GO-treated perovskite's surface became highly hydrophilic and thus enabled the smooth coverage of Spiro-MeOTAD. Accordingly the holes injection to the hole collector was improved, giving rise to an enhanced solar cell performance. This was clarified by EIS measurements indicating that GO not only enhances hole selectivity of HEL but also inhibits electron-hole recombination.

Another approach was proposed by utilizing graphene derivatives as HELs in inverted PSCs. In those configurations, fullerene

RESEARCH

was usually layered above perovskite, as fullerene's high electron affinity facilitates energy transfer processes towards the cathode. Perovskite was actually sandwiched between fullerene and a hole transport material, usually poly(3,4-ethylenedioxythiophene) polystyrene sulfonate (PEDOT:PSS). Contrary to oxide substrates that require high temperature annealing, those configurations allow solar cell fabrication through solution, simple and low-temperature processes, i.e. compatible with flexible formats. The inverted devices based on fullerene and PEDOT:PSS achieved a 10% PCE [163], however still lag behind PSCs based on mesopor-ous structures. Moreover, the presence of PEDOT:PSS is accompanied by well-known drawbacks derived from its acidic and aqueous nature. In this context, Wu et al. [164] worked on introducing GO as a HEL, instead of PEDOT:PSS. The PCE was clearly improved from the initial 9.26% for the reference device, to 12.40% for the GO-containing champion device. A similar approach was conducted by Yeo and co-workers [165], who employed rGO as a HEL (Fig. 7) in an inverted PSC structure, resulting in a remarkable PCE improvement. As time resolved photoluminescence measurements showed, the fast decay process, originated from the quenching of charge carriers and the slow decay, attributed to the radiative recombination of free charge carriers, were both decreased, indicating that the majority of charge carriers generated under illumination were transported to the respective contacts with less recombinations.

Similarly, a GO/PEDOT:PSS composite fabricated by sequential spin coating, was used as the HEL in planar PSCs, showing a superior PCE and life-time compared with the reference PED-OT:PSS cell mainly due to decreased series resistance (RS) and increased shunt resistance (RSh) [166].

SPSCs containing transition metal dichalcogenide-based interlayers

Most of the research efforts made towards introducing MX2 mono-layered structures inside solar cells, concern their utilization as HELs in OSCs. In a first effort, Gu et al. [167] employed a MoS2 nanosheet layer as the HEL in P3HT and PTB7-based OSCs and achieved PCEs of 4.02% and 8.11% respectively (Fig. 8). Kelvin probe measurements revealed an 80 meV energy offset between P3HT and MoS2, equal to the one measured between P3HT and MoO3, indicating that MoS2 is an efficient HEL. Besides this, transient electrical measurements showed that the solution processed layered MoS2 EEL exhibited low trap density.

In a different approach, Yun et al. [168] used HAuCl4.3H2O to p-type dope a MoS2 layer and increase its Wf to 4.76 eV, compared with the 4.52 eV of its un-doped counterpart. As a result, the devices containing the doped MoS2 HEL showed a better performance reaching a total PCE of 3.38%. Following an opposite route, the same group treated MoS2 with NaBH4 to induce n-type doping and used it to form an EEL in inverted-structured devices. The improvement of the respective devices was sound and outperformed the reference ones based on undoped MoS2 EEL.

Le et al. [169] increased the Wf of MoS2 via UV/Ozone (UVO) treatment of MoS2 layers. As a result, the Wf increased to 4.9 eV that matches energetically with the HOMO level of P3HT. The PCE yielded was 2.31%, slightly lower compared to PEDOT:PSS-based devices. On the contrary, the use of MoS2 along with PEDOT:PSS led to a lower PCE than PEDOT:PSS alone-based solar cells;nevertheless

FIGURE 8

(a) Schematic of the inverted-type organic solar cells. An active layer was sandwiched between a ZnO-modified ITO cathode and a MoS2 buffer layer modified Ag-based top anode. (b) Schematic drawing of the structureof the thin layer MoS2 buffer layer (side view). (c) Schematic drawing of the monolayer flake of MoS2 along the 0001] (top view). Electrical output characteristics of devices based on MoS2 as HEL: (d) J-V curves and (e) EQE spectra of the - PTB7:PC71BM OSCs with MoS2 as the HEL Reprinted with permission from Ref. [167] © 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim.

the presence of MoS2 benefitted the performance of the OSCs regarding stability issues. The UVO treatment technique was also applied by Yang et al. [170], aiming at the introduction of MoS2 as HEL in PTB7-based OSCs. Subsequently, a comparison among devices incorporating PEDOT:PSS, undoped MoS2 and UVO treated MoS2 HELs was carried out. UVO treatment increased the Wf of MoS2, while the respective OSCs showed superior performance compared to PEDOT:PSS and undoped MoS2 cases.

In an effort to address stability issues, Liu and his co-workers [171], treated MoS2 sheets in solution with a hydrophilic surfactant. In particular, MoS2 sheets were modified with positively charged hexadecyltrimethylammonium chloride molecules that inhibit aggregation phenomena. The resulting OSCs exhibited long-term stability, as their performance slightly deteriorated, only after 100 days of storage.

An alternative approach was carried out by Yang et al. [172], who embedded Au nanoparticles into MoS2 HEL, in order to take advantage of plasmon enhancement effects. The simulation and experimental measurements conducted, showed that the MoS2-Au composite HEL can more efficiently utilize the enhanced near-field, particularly along the horizontal direction. As a result, the plasmonic device exhibited a PCE of 7.25%, which was enhanced by 17.3%, compared to the OSC with the MoS2 HTL alone.

Other 2D materials apart from MoS2 were also used as HELs leading to an enhanced OSC performance. Le et al. [173] utilized UVO treated WS2 as HEL in P3HT-based devices. As expected, the Wf value was increased but the PCE of the device was improved, compared to PEDOT:PSS-devices, only when WS2 sheets were

Materials Today • Volume 00, Number 00• April 2016

incorporated into PEDOT:PSS to provide a composite HEL. The same group reported [174] on the application of TaS2 in both HEL and EEL. In particular, the Wf of TaS2 is 4.4 eV, establishing it as a suitable EEL, while following UVO treatment, the Wf increased to 5.1 eV and the material obtained was introduced as HEL. The PCE yielded for the devices using TaS2 as HEL was comparable to the reference ones, based on PEDOT:PSS, while the inverted structures with TaS2 as EEL, instead of TiOx, presented superior performance.

Finally, Gu and his co-workers [175] introduced NbSe2 as a HEL in inverted devices based on PTB7 as the donor material. The PCE attained was 8.10%, outperforming the devices based on vacuum-deposited MoO3. Based on scanning kelvin probe microscopy measurements complemented with transient electrical output characteristics, the authors attributed this superior performance to the existence of a surface dipole, as well as to the unique flakelike 2D structure giving rise to lower trap density.

Conclusions and future outlook

In this review, we provided an overview the potential of using graphene materials and derivatives as well as TMDs as electron and hole transport layers in several types of SPSCs, including DSSCs, QDSSCs, OSCs and PSCs. It is presented that graphene offers great advantages towards all solution processed photovoltaic devices, especially if we take into account the easy, solution-compatible and low-cost fabrication methods pursued. Indeed, graphene can be practically embedded into every layer of photovoltaic devices, while it enables alternative and facile fabrication routes even when it cannot outperform its counterparts. On the other hand, TMDs were also attested as efficient HELs in OSCs, though the research progress on these materials is still lagging behind graphene.

The key requirements towards large scalability and commercialization of the 3rd generation solar cell technology are high efficiency, solution processability, long-term stability, low-cost manufacturing and compatibility with flexible substrates. Although efficiency and stability issues are still unresolved, the main advantage that 2D materials provide, when used as charge transport interlayers, is their compatibility with large-scale solution-processable printing techniques, along with work function tunability and ambipolar transport ability. For instance, concerning DSSCs and QDSCs, despite their remarkable stability and the imponderable properties of QDs, the presence of liquid mediators is the main handicap of both technologies towards commercialization. In both technologies, the presence of graphene within the photoanode, as a highly conductive host backbone, is confirmed to be a promising alternative for improved charge transport. Indeed, graphene and its derivatives favorably increase the semiconductor's surface area and porosity, leading to additional dye adsorption that facilitate charge extraction. Besides this, such materials can promote electron transport, retard charge recombination, function as intermediate energy cascade materials and finally increase Jsc due to the optical scattering effect. It is noteworthy that research has not expanded yet to the field of TMDs, since TMDs have only been used as Pt-free counter electrodes in DSSCs. Incorporation of highly conductive TMDs into the photoanode, should offer a promising approach towards improving charge carrier mobilities of DSSCs and QDSSCs. In a first step, this can be realized by following the synthesis and fabrication routes already used in the case of graphene. With regard to PSCs, graphene materials have been slightly explored but can provide several advantages

like faster charge transport and higher transparency; besides this, graphene interlayers may allow a better crystallinity of the perov-skite component. Nonetheless, PSCs suffer from low stability, which is the major but still unresolved issue for this technology. In this context, the adoption of techniques and protocols developed to design interlayers for OSCs may be desirable. Particularly, due to the fact that the employment of low temperature-processable interlayers is an essential step towards highly-efficient flexible PSCs.

On the other hand, the application of graphene and other 2D materials in OSCs has been widely explored, particularly for hole extraction and transport, where graphene materials are proved to be quite competitive against PEDOT:PSS and MOs. This can be attributed to their excellent solution processability, work function tunability, ambipolar transport and improved stability. Indeed, contrary to PEDOT:PSS that suffers from high acidity and hygro-scopicity, graphene-based materials are not acidic and quite stable. Moreover, in contrast to graphene and other 2D materials, transition MOs exhibit poor mechanical flexibility and usually demand high temperature annealing or vacuum deposition processes, both incompatible with plastic and flexible substrates. Nonetheless, the simultaneous requirement of ultra thin films (only a few nm thick) as well as of printing techniques providing full-coverage, homogeneous layers remains a challenge. In this context, the development of suitable surface modification techniques, to render 2D materials compatible with large area solution processable OSCs, is required. Another challenge is the proper functionalization of graphene to match well with the energy levels of newly designed high efficient polymers. Finally, research in graphene films should cope with roughness, which is still an unresolved issue, although extensive research has been carried out. Besides this, although GO presents high solubility, its insulating nature inhibits the formation of good ohmic contacts. At the same time, high quality pristine graphene is achievable only with techniques incompatible with solution processability.

Concerning the electron extraction procedures, graphene derivatives were mostly combined with state-of the art MOs to improve charge mobility. Nonetheless, graphene-MO composites still have to compete other highly efficient solution-processable materials, like poly [(9,9-bis(3'-(N,N-dimethylamino)propyl)-2,7-fluorene)-alt-2,7-(9,9 dioctylfluorene)], Ca and LiF. In the field of tandem OSCs, graphene materials offer high transparency and low electrical and optical losses across the sub-cells.

Finally, the employment of other 2D materials as OSCs inter-layers is quite satisfactory, especially in the case of MoS2, where the efficiency obtained was quite competitive compared to PED-OT:PSS. At the same time, the dispersions prepared were stable for long periods, a property that addresses the PEDOT:PSS instability, attributed to its acidic nature. Besides this, TaS2, WS2 and NbSe2 have been also attested as effective interlayers, given that in most cases tuning of Wf was necessary to form good ohmic contacts with the active layer. In this context, materials like WSe2, MoTe2 and MoSe2 should be also tested as HELs within OSCs, taking always into account the need of Wf tuning, according to the HOMO level of the respective donor material. On the other hand, a lower-lying Wf is required for EELs, to energetically match the LUMO level of fullerene-based acceptors. The only TMD tested as EEL to date, is TaS2, while similar efforts could be also made to insert, e.g. MoS2, WS2 or other TMDs with suitably located energy

RESEARCH

levels. Regarding PSCs, TMDs have not yet been used as charge transport interlayers, although the standards for HELs and EELs are quite similar to OSCs. In the case of DSSCs and QDSSCs, as mentioned before, research has mostly been focused on doping of the MOx scaffold with highly conductive graphene derivatives. The same principle could be investigated with the graphene-like TMDs. Finally, another approach could be the use of multilayered TMDs considering that the optoelectronic properties of such heterostructures are different compared to those of pristine TMD layers. In this context, band gap engineering of TMDs heterobilayers may provide further opportunities to realize efficient carrier extraction layers. Whatever the case, research in TMDs interlayers is at an infant stage and taking into account the plethora of 2D materials, it is highly expected that such materials should play a crucial role in the near future.

In conclusion, further research is demanded towards incorporating 2D materials in SPSCs. Advancements can be realized by developing properly modified 2D materials and/or by further optimizing device architecture. No doubt, advances in this topic will be crucial for the future of graphene and graphene-like materials in energy harvest technologies and related applications.

Acknowledgements

The research leading to these results was funded by the European Union Seventh Framework Programme under Grant Agreement No. 604391 Graphene Flagship.

References

[1 [2 [3 [4 [5 [6 [7 [8 [9 [10 [11 [12 [13 [14 [15 [16 [17 [18 [19 [20 [21 [22 [23 [24 [25 [26 [27 [28 [29 [30 [31 [32 [33 [34 [35 [36 [37

Trends in Photovoltaic Applications, IEA-PVPS (2014). J. Jean, et al. Energy Environ. Sci. 8 (2015) 1200. X. Wang, et al. Energy Environ. Sci. 8 (2015) 790.

http://www.nrel.gov/ncpv/images/efficiency_chart.jpg (accessed March 2016). M.A. Green, et al. Prog. Photovolt.: Res. Appl. 23 (2015) 1. M. Gratzel, Inorg. Chem. 44 (2005) 6841.

H.J. Snaith, L. Schmidt-Mende, Adv. Mater. 19 (2007) 3187.

I. Chung, et al. Nature 485 (2012) 486.

D. Perganti, et al. Electrochim. Acta 179 (2015) 241. X. Lan, et al. Nat. Mater. 13 (2014) 233.

J.J. Choi, et al. Adv. Mater. 23 (2011) 3144.

A.J. Labelle, et al. Nano Lett. 15 (2015) 1101.

C.H.M. Chuang, Nat. Mater. 13 (2014) 796.

W. Tress, et al. Adv. Energy Mater. 3 (2013) 873.

N.-G. Park, Nat. Mater. 14 (2015) 140.

N.J. Jeon, et al. Nature 517 (2015) 476.

J.M. Ball, et al. Energy Environ. Sci. 6 (2013) 1739.

F.T.F. O'Mahony, et al. J. Mater. Chem. A 3 (2015) 7219.

E. Stratakis, E. Kymakis, Mater. Today 16 (2013) 133. H. Hoppe, N.S. Sariciftci, J. Mater. Res. 19 (2004) 1924.

C.C. Chen, et al. Adv. Mater. 26 (2014) 5670. K. Novoselov, et al. Science 306 (2004) 666. S. Das, et al. J. Mater. Res. 29 (2014) 299. A.C. Ferrari, et al. Nanoscale 7 (2015) 4598.

A.K. Geim, K.S. Novoselov, Nat. Mater. 6 (2007) 183. Z. Liu, et al. Chem. Soc. Rev. 44 (2015) 5638. Z. Yin, et al. Adv. Energy Mater. 4 (2013) 1300574.

D. Chen, et al. Energy Environ. Sci. 6 (2013) 1362. Yuan, et al. Chem. Eur. J. 20 (2014) 1.

D. Konios, et al. Adv. Funct. Mater. 25 (2015) 2213.

F. Bonaccorso, et al. Adv. Funct. Mater. 25 (2015) 3870. N. Balis, et al. ChemNanoMat 1 (2015) 346.

F. Bonaccorso, et al. Science 347 (2015) 1246501. M. Chhowalla, et al. Nat. Chem. 5 (2013) 263. M. Chhowalla, et al. Chem. Soc. Rev. 44 (2015) 2584. A. Gupta, et al. Prog. Mater. Sci. 73 (2015) 44. H. Wang, et al. Small 10 (2014) 2165.

[38 [39 [40 [41 [42 [43 [44 [45 [46 [47 [48 [49 [50 [51 [52 [53 [54 [55 [56 [57 [58 [59 [60 [61 [62 [63 [64 [65 [66 [67 [68 [69 [70 [71 [72 [73 [74 [75 [76 [77 [78 [79 [80 [81 [82 [83 [84 [85 [86 [87 [88 [89 [90 [91 [92 [93 [94 [95 [96 [97 [98 [99 [100 [101 [102 [103 [104 [105 [106 [107 [108 [109 [110

G. Eda, et al. ACS Nano 6 (2012) 7311. M. Xu, et al. Chem. Rev. 113 (2013) 3766. J.Y. Kim, et al. Adv. Mater. 18 (2006) 572.

H.J. Snaith, Adv. Funct. Mater. 20 (2010) 13.

A. Hagfeldt, et al. Chem. Rev. 110 (2010) 6595. G. Boschloo, A. Hagfeldt, Acc. Chem. Res. 42 (2009) 1819. A.B.F. Martinson, et al. J. Phys. Chem. A 113 (2009) 4015. S. Lee, et al. Langmuir 23 (2007) 11907.

D. Barpuzary, et al. ACS Appl. Mater. Interfaces 15 (2014) 12629.

A. Kongkanand, et al. Nano Lett 7 (2007) 676.

J. Roy-Mayhew, et al. Chem. Rev. 114 (2014) 6323. D.W. Chang, et al. J. Mater. Chem. A 2 (2014) 12136. S.R. Kim, et al. Chem. Phys. Lett. 483 (2009) 124.

B. Tang, G. Hu, J. Power Sources 220 (2012) 95.

T. Chen, et al. Adv. Funct. Mater. 22 (2012) 5245. N. Yang, et al. ACS Nano 4 (2010) 887. Y. Kusumawati, et al. J. Phys. Chem. C 118 (2014) 9974. G. Cheng, et al. ACS Appl. Mater. Interfaces 5 (2013) 6635.

G.S. Anjusree, et al. RSC Adv. 3 (2013) 12933.

J. Zhao, et al. Electrochim. Acta 156 (2015) 261. M. Zhu, et al. J. Power Sources 262 (2014) 349. A. Saccoa, et al. Electrochim. Acta 131 (2014) 154. J. Zhi, et al. J. Power Sources 281 (2015) 404. Y. Liu, et al. Nanoscale 6 (2014) 6755. F. Xu, et al. J. Phys. Chem. C 117 (2013) 8619. F. Xiang, et al. ACS Sustain. Chem. Eng. 2 (2014) 1234. X. Fang, et al. Electrochim. Acta 137 (2014) 634.

C.Y. Neo, J. Ouyang, J. Power Sources 222 (2013) 161. X. Fang, et al. Sol. Energy 101 (2014) 176.

Z. He, et al. Adv. Mater. 25 (2013) 6900. K. Kakiage, et al. Chem. Commun. 51 (2015) 15894. J. Yang, et al. J. Am. Chem. Soc. 15 (2014) 5722. S. Mathew, et al. Nat. Chem. 6 (2014) 242. W. Sun, et al. J. Mater. Chem. A 1 (2013) 2762. V.-D. Dao, et al. Carbon 81 (2015) 710. Y. Li, et al. J. Mater. Chem. A 1 (2013) 6342. J. Ma, et al. ChemSusChem 7 (2014) 3304.

H. Zheng, et al. ACS Appl. Mater. Interfaces 5 (2013) 6657. X. Meng, et al. Adv. Energy Mater. 5 (2015) 1500180.

Y. Zhang, et al. RSC Adv. 5 (2015) 10430. M.J. Ju, et al. Adv. Mater. 26 (2014) 3055.

I.-Y. Jeon, et al. Adv. Funct. Mater. 25 (2015) 11170. S.-M. Jung, et al. Chem. Mater. 26 (2014) 3586.

X. Shen, et al. J. Power Sources 277 (2015) 215. L. Yang, et al. Nanoscale 7 (2015) 3173. A. Rasin, et al. J. Phys. Chem. C 119 (2015) 2297.

D. Ding, et al. RSC Adv. 5 (2015) 6462.

N. Balis, et al. Electrochim. Acta 91 (2013) 246.

S. Baskoutas, A.F. Terzis, J. Appl. Phys. 99 (2006) 013708.

W.W. Yu, et al. Chem. Mater. 15 (2003) 2854.

D.J. Norris, M.G. Bawendi, Phys. Rev. B 53 (1996) 16338.

M. Shim, P. Guyot-Sionnest, J. Chem. Phys. 111 (1999) 6955.

Q. Zhang, et al. Chem. Soc. Rev. 42 (2013) 3127.

O.E. Semonin, et al. Science 334 (2011) 1530.

W.A. Tisdale, et al. Science 328 (2010) 1543.

K. Tvrdy, et al. Proc. Natl. Acad. Sci. U. S. A. 108 (2011) 29.

M.S. De la Fuente, et al. J. Phys. Chem. Lett. 4 (2013) 1519.

P.K. Santra, P.V. Kamat, J. Am. Chem. Soc. 134 (2012) 2508.

C.X. Guo, et al. Angew. Chem. Int. Ed. 49 (2010) 3014.

D. Barpuzary, M. Qureshi, ACS Appl. Mater. Interfaces 5 (2013) 11673. J. Zhao, et al. Electrochim. Acta 96 (2013) 110.

P.N. Kumar, et al. J. Phys. Chem. C 118 (2014) 18924.

J. He, et al. RSC Adv. 4 (2014) 2068.

F.S. Ghoreishi, et al. J. Power Sources 271 (2014) 195.

Y. Zhu, et al. ACS Appl. Mater. Interfaces 6 (2014) 13833.

J. Liu, et al. Adv. Mater. 24 (2012) 2774.

S.-P. Cho, et al. Sol. Energy Mater. Sol. Cells 132 (2015) 196.

F. Jiang, et al. Adv. Mater. 27 (2015) 2930.

Z. He, et al. Nat. Photon. 6 (2012) 591.

D.W. Zhao, et al. Sol. Energy Mater. Sol. Cells 94 (2010) 985.

T. Yanagidate, et al. Jpn. J. Appl. Phys. 53 (2014) 02BE05.

J. Liu, et al. Energy Environ. Sci. 7 (2014) 1297.

C.C. Chueh, et al. Energy Environ. Sci. 8 (2015) 1160.

Materials Today • Volume 00, Number 00• April 2016

111] S.-S. Li, et al. ACS Nano 4 (2010) 3169. [144] S. Qu, et al. ACS Nano 7 (2013) 4070.

112] H.P. Kim, et al. Sol. Energy Mater. Sol. Cells 110 (2013) 87. [145] G. Kakavelakis, et al. Chem. Mater. 26 (2014) 5988.

113] J.C. Yu, et al. ACS Appl. Mater. Interfaces 6 (2014) 2067. [146] N.D. Lang, Phys. Rev. B 4 (1971) 4234.

114] Y.-H. Chao, et al. Adv. Energy Mater. 3 (2013) 1279. [147] D. Konios, et al. J. Mater. Chem. A 4 (2016) 1612.

115] J.J. Intemann, et al. Adv. Funct. Mater. 24 (2014) 1465. [148] J.Y. Kim, et al. Science 317 (2007) 22.

116] Y.-J. Jeon, et al. Sol. Energy Mater. Sol. Cells 105 (2012) 96. [149] V. Shrotriya, et al. Appl. Phys. Lett. 88 (2006) 073508.

117] J.-M. Yun, et al. Adv. Mater. 23 (2011) 4923. [150] J. Gilot, et al. Appl. Phys. Lett. 90 (2007) 14351.

118] B. Paci, et al. RSC Adv. 5 (2015) 106930. [151] V.C. Tung, et al. J. Am. Chem. Soc. 133 (2011) 9262.

119] I.P. Murray, et al. J. Phys. Chem. Lett. 2 (2011) 3006. [152] A.R.B.M. Yusoff, et al. Nanoscale 5 (2013) 11051.

120] J.-S. Yeo, et al. J. Mater. Chem. A 2 (2014) 292. [153] Y. Chen, et al. Nano Lett. 14 (2014) 1467.

121] K. Yuan, et al. ACS Appl. Mater. Interfaces 6 (23) (2014) 21069. [154] E. Stathatos, et al. Chem. Mater. 15 (2003) 1825.

122] L. Kim, et al. Adv. Energy Mater. 1 (2011) 1052. [155] N. Balis, et al. J. Photochem. Photobiol. A 214 (2010) 69.

123] J. Liu, et al. J. Phys. Chem. Lett. 3 (2012) 1928. [156] A. Kojima, et al. J. Am. Chem. Soc. 131 (2009) 6050.

124] E. Stratakis, et al. Nanoscale 5 (2013) 4144. [157] M. Batmunkh, et al. J. Mater. Chem. A 3 (2015) 9020.

125] M.-K. Chuang, et al. Nanoscale 6 (2014) 1573. [158] J.T.-W. Wang, et al. Nano Lett. 14 (2014) 724.

126] M. Li, et al. Phys. Chem. Chem. Phys. 15 (2013) 18973. [159] Z. Zhu, et al. J. Am. Chem. Soc. 136 (2014) 3760.

127] J. Liu, et al. Adv. Mater. 26 (2014) 786. [160] A. Agresti, et al. Adv. Funct. Mater. (2016), http://dx.doi.org/10.1002/

128] S.J. Lee, et al. Chem. Commun. 51 (2015) 9185. adfm.201504949.

129] S.-H. Kim, et al. Nanoscale 6 (2014) 7183. [161] W.-J. Yin, et al. J. Mater. Chem. A 3 (2015) 8926.

130] C. Li, et al. Org. Electron. 15 (2014) 2868. [162] W. Li, et al. J. Mater. Chem. A 2 (2014) 20105.

131] Y.-J. Noh, et al. Carbon 79 (2014) 321. [163] J. You, et al. ACS Nano 8 (2014) 1674.

132] L. Chen, et al. ACS Appl. Mater. Interfaces 6 (2014) 22334. [164] Z. Wu, et al. Nanoscale 6 (2014) 10505.

133] E. Stratakis, et al. Nanoscale 6 (2014) 9925. [165] J.-S. Yeo, et al. Nano Energy 12 (2015) 96.

134] J. Liu, et al. Adv. Mater. 24 (2012) 2228. [166] D.-Y. Lee, et al. J. Mater. Chem. A 8 (2016) 1513.

135] H.B. Yang, et al. ACS Appl. Mater. Interfaces 6 (2014) 1092. [167] X. Gu, et al. Adv. Energy Mater. 3 (2013) 1262.

136] K.D.G.I. Jayawardena, et al. J. Mater. Chem. A 1 (2013) 9922. [168] J.-M. Yun, et al. J. Mater. Chem. C 1 (2013) 3777.

137] T. Hu, et al. J. Mater. Chem. A 3 (2015) 10890. [169] Q.V. Le, et al. Phys. Chem. Chem. Phys. 16 (2014) 13123.

138] T. Hu, et al. Chem. Eur. J. 20 (2014) 17178. [170] X. Yang, et al. J. Mater. Chem. A 2 (2014) 7727.

139] A. Hu, et al. ACS Appl. Mater. Interfaces 7 (2015) 16078. [171] W. Liu, et al. RSC Adv. 4 (2014) 32744.

140] M.J. Beliatis, et al. Adv. Mater. 26 (2014) 2078. [172] X. Yang, et al. J. Mater. Chem. A 2 (2014) 14798.

141] Y. Zhang, et al. Electrochim. Acta 117 (2014) 438. [173] Q. Van Le, et al. Phys. Status Solidi RRL 8 (2014) 560-565.

142] Y. Zhang, et al. Electrochim. Acta 143 (2014) 18. [174] Q. Van Le, et al. Phys. Chem. Chem. Phys. 16 (2014) 25468.

143] G.D. Sharma, et al. Org. Electron. 15 (2014) 348. [175] X. Gu, et al. ChemSusChem 7 (2014) 416.