One-Step Synthesis of TiO2/Graphene Nanocomposites by Laser Pyrolysis with Well-Controlled Properties and Application in Perovskite Solar Cells.

This work presents an original synthesis of TiO2/graphene nanocomposites using laser pyrolysis for the demonstration of efficient and improved perovskite solar cells. This is a one-step and continuous process known for nanoparticle production, and it enables here the elaboration of TiO2 nanoparticles with controlled properties (stoichiometry, morphology, and crystallinity) directly grown on graphene materials. Using this process, a high quality of the TiO2/graphene interface is achieved, leading to an intimate electronic contact between the two materials. This effect is exploited for the photovoltaic application, where TiO2/graphene is used as an electron-extracting layer in n-i-p mesoscopic perovskite solar cells based on the reference CH3NH3PbI3-x Cl x halide perovskite active layer. A significant and reproducible improvement of power conversion efficiencies under standard illumination is demonstrated, reaching 15.3% in average compared to 13.8% with a pure TiO2 electrode, mainly due to a drastic improvement in fill factor. This beneficial effect of graphene incorporation is revealed through pronounced photoluminescence quenching in the presence of graphene, which indicates better electron injection from the perovskite active layer. Considering that a reduction of device hysteresis is also observed by graphene addition, the laser pyrolysis technique, which is compatible with large-scale industrial developments, is therefore a powerful tool for the production of efficient optoelectronic devices based on a broad range of carbon nano-objects.

Introduction

The use of titanium dioxide (TiO2) is nowadays widespread in many fields such as waste-water purification or photocatalysis due to its physical properties and chemical stability. In the field of sustainable energy, titanium dioxide has attracted more and more interest for rechargeable batteries, supercapacitors, photocatalytic hydrogen generation, and solar cells.[1,2] More particularly, at the nanometric scale, due to the critical contribution of active surfaces, TiO2 offers interesting properties for many applications, and in particular for third-generation solar cells, such as dye-sensitized, organic, or perovskite solar cells.[3,4] In this field, a main limitation to current generation is associated with the recombination of excitons and/or charge carriers at many different levels. Therefore, regarding photogenerated electrons, the scientific community put many efforts to enhance the charge extraction layers of photovoltaic devices by synthesizing TiO2-based thin films modified with metallic nanoparticles or carbon nanostructures.[5] In particular, since the development of new synthesis methods of graphene materials, TiO2/graphene nanocomposites showed a strong potential to replace pure TiO2 to reach optimal material properties for the photovoltaic application. Graphene materials are indeed now well known for their excellent conductive properties and high specific area, attracting strong interest for electronic applications.[6]

In the field of perovskite solar cells, high efficiencies of more than 23% have been recently reached,[7,8] but efforts remain to be performed to improve charge extraction, especially concerning the development of efficient and reliable charge transporting electrodes and selective contacts. The motivation for using TiO2/graphene nanocomposites as a mesoporous electron transport layer aims at improving charge transfer and electron collection, thus leading to reduced charge trapping and recombination that can occur at the surface of titanium dioxide.[9−13] These benefits rely on the possibility for efficient electron transfer mechanisms from TiO2 to graphene, as reported in the field of dye-sensitized solar cells.[14]

Among the numerous works reported over the last few years, the study carried by Nicholas et al. clearly demonstrates the benefit of TiO2/graphene nanocomposites as an electron transport material in perovskite solar cells by reaching 15.6% power conversion efficiency (PCE) for the best cell using a reference n–i–p device configuration.[15] One main benefit of the strategy is the possibility to achieve efficient TiO2 electrode at low temperatures (<150 °C), although it requires several independent processing steps for the preparation of graphene, TiO2, and the final nanocomposite.

Different pathways have been reported for the elaboration of TiO2/graphene nanocomposites, such as the sonication of graphene and titanium dioxide dispersions. In this case, TiO2 nanoparticles anchor onto the graphene layers, leading to partial covering and weak bonding between both materials.[16] Other alternatives reported in the literature include sol–gel and hydrothermal methods, where graphene oxide is reduced in presence of a titanium-based precursor or a solution containing dispersed TiO2 nanoparticles.[16] To this date, these procedures cannot enable the elaboration of TiO2–graphene composites with controlled properties for both TiO2 and graphene materials while ensuring a strong interface between them and also leading to a high production yield.

In this work, we present a one-step and continuous synthesis method of TiO2/graphene nanocomposites by laser pyrolysis. Indeed, the synthesis by laser pyrolysis of such nanocomposites is a continuous process that leads to a production rate of several grams per hour. Moreover, laser pyrolysis neither use any solvent nor require cleaning or washing steps, which are inevitable in most of the other way of production mentioned in the literature.[17,18] This singular technique enables the direct synthesis of well-controlled TiO2 nanoparticles on graphene sheets. Relevance of this process has been previously demonstrated for TiO2/carbon nanotube nanocomposite production, exhibiting an improved efficiency when used as active materials in solid-state dye-sensitized solar cells.[19] In this context, the present study focuses on the morphological and structural properties of the TiO2/graphene nanocomposites synthesized by laser pyrolysis and their integration into mesoporous (or mesoscopic) n–i–p perovskite solar cells. A clear improvement of photovoltaic performance is related to the intimate contact between graphene and TiO2 associated with the synthesis method, which is clearly beneficial to charge extraction.

Results and Discussion

Physical Properties of the TiO2 and TiO2 Graphene Nanopowders

The synthesis of TiO2/graphene nanocomposite has been performed in a one-step method by laser pyrolysis, as described in Section . This very fine technique enables a continuous production of nanoparticles that are obtained as powders.[20] Besides the excellent purity of the materials, laser pyrolysis can also provide high production rate (up to 3 g/h at lab scale) in the case of pure TiO2 synthesis, with no graphene addition.[21] Characteristics of the obtained nanopowders (size, crystallinity) are tuned by the choice of experimental conditions (including laser power, reactor pressure, and gas flow rate).[22] We aim at producing particles as small as possible so that the specific surface, hence interfacial interactions, will be maximized. Experimental parameters have been carefully chosen to favor the production of anatase phase.[22] Pure TiO2 nanoparticles were synthesized by laser pyrolysis as a reference material from pure titanium (IV) isopropoxide (TTIP). To produce TiO2/graphene nanocomposites, 0.04 wt % reduced graphene oxide (rGO) was dispersed into TTIP using ultrasounds and the suspension was carried out to the laser beam. The obtained nanocomposite-based powder is then labeled as 0.04G/TiO2.

These as-formed powders obtained from the laser pyrolysis are of gray color. This gray color is attributed to a significant amount of residual carbon phases (25%) due to the decomposition of the precursors (TTIP and C2H4). Therefore, annealing under air (6 h at 430 °C) was systematically performed to remove this carbon as it is known to alter the electronic properties of the obtained materials. The annealing temperature was carefully determined from thermogravimetric analysis, which is presented in the Supporting Information (Figure S1), to avoid graphene combustion while ensuring amorphous carbon elimination. After annealing, the color of powders turn to white, as depicted in Figure S2 in the Supporting Information.

The annealed powders were then thoroughly analyzed, and we especially focus here on two samples referred as TiO2 and 0.04G/TiO2 samples, whose main characteristics are presented in Table .

Table 1

Main Features of TiO2 and 0.04G/TiO2 Annealed Powders Elaborated by Laser Pyrolysis

				crystallinity (normalized intensity)
	production rate (g/h)	specific surface (g/m2)	particle size (BET) (nm)	anatase	rutile	srilankite	crystallite size (nm)
TiO2	2.8	63	25	1	0.17	0	15
0.04G/TiO2	0.18	97	16	1	0.20	0.1	10

The morphologies of the powders have been first studied by scanning electron microscopy (SEM). Figure a shows an overview of the general aspect of the powders and confirms good homogeneity of the samples. Moreover, through meticulous analysis, we were able to distinguish graphene sheets within the TiO2 matrix for sample 0.04G/TiO2. Figure b shows a SEM image centered on a graphene sheet at a higher magnification. Within the powder, graphene is present in a small amount (considering that 0.04 wt % of graphene was introduced in the precursor mixture prior to the laser pyrolysis process) and is surrounded by TiO2 nanoparticles; therefore, most of the TiO2/graphene layers appear covered by a high amount of TiO2 nanoparticles and cannot be easily seen through SEM characterization. Besides, TiO2 particles seem to be very small as they reach the resolution limit of the microscope. Transmission electron microscopy (TEM) was therefore performed to get into more details.

Figure 1

(a, b) SEM images of the TiO2/graphene nanocomposite within the TiO2 nanoparticle powder at different magnitudes. (c) TEM image of as-received graphene, before laser pyrolysis. (d) TEM images of the TiO2/graphene nanocomposite obtained by laser pyrolysis (0.04G/TiO2).

Figure c shows the TEM image of as-received graphene, before laser pyrolysis. Its transparency emphasizes that only few carbon layers are involved, leading to very thin graphene sheets. This morphology seems to be preserved after the nanocomposite synthesis and after annealing, as shown in Figure d. As we can see, graphene is completely covered by TiO2 nanoparticles, as one can only distinguish the edges of the graphene layer. For TEM observations, as mentioned in Section , the preparation of the microscope grids involves dispersion of the powders into ethanol using a high-power ultrasonic probe to separate the particles. Despite this treatment, a good coverage of graphene by titania particles is observed, which suggests a good quality of interface between the two materials. We highlight here that the quality of the interface between TiO2 nanoparticles and carbon nanotubes synthesized by laser pyrolysis was already pointed out by our previous work,[19] which shows an intimate electronic contact between the two constituents. However, while carbon nanotubes are highly robust in terms of mechanical and thermal properties, graphene is much more sensitive to such stress. Achieving an intimate interface with TiO2 without significantly altering the aspect of the graphene sheets is therefore an additional benefit of the laser pyrolysis process in our opinion, which remains difficult to achieve using other preparation methods.

In addition, complementary X-ray diffraction (XRD) measurements on the two powders are presented in Figure a, as well as their corresponding diffraction patterns.

Figure 2

(a) XRD images of TiO2 (pure TiO2) and 0.04G/TiO2 (TiO2/graphene) nanopowders and (b), (c) their respective diffraction patterns (JCPDS card nos. 21-1272, 21-1276, and 84-1750). (d) Raman spectra of TiO2 and 0.04G/TiO2 focusing on the specific TiO2 signatures.

For both materials, the major phase present is anatase, although a small fraction of rutile is also present (see Table ). The average size of the crystallites is evaluated to be between 10 and 15 nm from Scherrer’s equation,[23,24] showing that TiO2 particles can be polycrystalline as the mean particle size evaluated by the Brunauer, Emmet and Teller (BET) method is around 20 nm (as reported in Table ). A comparison of the two diagrams emphasizes the presence of several additional peaks in the 0.04G/TiO2 nanocomposite. This phase can be attributed to the orthorhombic TiO2-II phase, also named srilankite. This result is confirmed by Raman spectra recorded for pure TiO2 and 0.04G/TiO2 samples, as shown in Figure d. Besides the anatase signatures at 395, 516, and 634 cm–1, clear signatures of srilankite TiO2-II are evidenced at 282, 314, 354, and 424 cm–1 in the case of the 0.04G/TiO2 sample. This phase is observed in all TiO2/graphene nanocomposites produced using our laser pyrolysis process. Usually, the TiO2-II is formed under high pressures and its structure is of the α-PbO2 type. In this work, this phase is observed only when graphene is present in the reaction. We therefore assume that the presence of graphene, although in less amount, does change the synthesis conditions. Graphene would induce the creation of a reductive atmosphere that reduces a small fraction of anatase or rutile phase formed within the reactor during the laser pyrolysis. This TiO2-II phase was also reported in another work[25] that uses a flame synthesis method to synthesize TiO2 particles and produces a small fraction of TiO2-II (srilankite) under oxygen-lean conditions. In addition, as graphene is not detectable by XRD, further information about the composite properties has been investigated by Raman scattering and is presented in the Supporting Information (Figure S3). The investigations performed by Raman spectroscopy confirm the presence of graphene within the nanopowders. They also show a significant increase of the D-to-G band ratio in the composite compared to that in pristine graphene, which suggests that defects have been introduced through the pyrolysis process or through close interactions of the graphene sheets with TiO2 nanoparticles. This fact is consistent with the observations made by TEM.

We highlight here that these characterizations confirm the good crystallization of TiO2 in the anatase form and the presence of graphene in our nanocomposites powder. Moreover, especially through TEM analysis, the interface between TiO2 and graphene seems to be intimate, which is thus very promising for our application, as it can be associated with a good electronic contact.

From Powders to Mesoporous Electrodes

Mesoporous TiO2 and TiO2/graphene electron transport layers were fabricated from the nanopowders synthesized by laser pyrolysis. A suspension of TiO2 or TiO2/graphene powder was prepared in ethanol at a weight ratio of 1:30 with regard to the solvent. Then, α-terpineol and ethyl-cellulose (previously dissolved in absolute ethanol) were added, following conventional recipes already published.[19,26] The obtained paste is sonicated for 1 h and stirred overnight before being spin-coated in air at 4000 rpm for 60 s and various substrates. The deposited films are then progressively annealed up to 430 °C to remove all of the organic components, leaving a high porosity and a mesoporous structure. Finally, the porous films are treated with TiCl4 that is known to improve the surface states of nanostructured titania while improving the electronic percolation of adjacent particles, influencing consequently charge recombination and transport in the electrode (see the Supporting Information for experimental details).[27−30]

The main issue regarding the TiO2/graphene composite is associated with the narrow thermal window that can be exploited for the post-sintering of the TiO2 layers, required to remove the organic additives. The SEM cross section of the mesoporous TiO2/graphene layer obtained after sintering at 430 °C is presented in Figure a.

Figure 3

SEM cross section of the TiO2/graphene mesoporous layer deposited on glass (a) before and (b) after perovskite (CH3NH3PbI3–Cl) infiltration.

The SEM image clearly demonstrates the achievement of a high-porosity film, similar to those generally reported using such recipes.[31,32] The absence of large aggregates is associated with a high transparency of the layers in the visible region (>90%) with almost no contribution from scattering, as observed by eye or through UV–visible transmission spectroscopy (Figure a). This observation is crucial to ensure that most of the incoming light reaches the perovskite material (the solar absorber) that will be infiltrated in the mesoporous electrode, as described in the next step.

Figure 4

(a) UV–vis spectroscopy (transmission) of the mesoporous layer based on TiO2 and 0.04G/TiO2 powders and top-coated by perovskite. (b) Tauc’s plot of TiO2/perovskite and 0.04G/TiO2/perovskite.

We also note that the presence of graphene does not change the optical properties of our mesoporous layers, considering the low amount used here.

The CH3NH3PbI3–Cl (MAPI-Cl) perovskite is deposited on top of this layer in a one-step process, described in Section and Supporting Information (experimental details). Figure b shows that the pores of our TiO2-based layers are well filled by the perovskite, ensuring a good interface between the two materials so that efficient electron injection can take place. UV–visible spectra of infiltrated TiO2 and 0.04G/TiO2-based mesoporous layers are reported in Figure a and shows the expected contributions of the metal oxide electrode and the perovskite layer (absorption edge close to 790 nm in our case). From this UV–visible spectra and using Tauc’s equation (Figure b), the perovskite band gap is evaluated at 1.56 eV in both cases. We therefore assume that the presence of graphene does not significantly affect the perovskite crystallization.

Steady-state photoluminescence (PL) spectroscopy was conducted to better investigate the photoinduced charge transfer mechanisms at the TiO2 layer/perovskite interface, as a function of the presence of graphene. A comparison was made with the pristine perovskite layer deposited on a mesoporous Al2O3 thin film, which acts as an insulating scaffold preventing any charge transfer process. A quenching of the perovskite photoluminescence emission (centered at 792 nm) is evidenced in the presence of a TiO2-based electron transport layer, indicating that both TiO2 and TiO2/graphene materials can efficiently collect electrons photogenerated in the perovskite (Figure ). Moreover, this quenching is significantly enhanced in the presence of graphene, which indicates a better electron transfer efficiency compared to our pure TiO2 reference. This observation is further confirmed through transient photoluminescence measurements presented in Figure S4 in the Supporting Information, which shows a significantly faster PL decay kinetics in the presence of graphene compared to pure TiO2. Although there is still a debate in the literature about the relevance of mesoporous Al2O3 films as noninjecting reference substrates, the PL analysis clearly confirms that a better electronic interaction occurs between the perovskite and the graphene-loaded TiO2 electrode compared to pure TiO2.

Figure 5

Steady-state photoluminescence spectroscopy: comparison of an electron-blocking layer (Al2O3) and an electron transport layer (TiO2 and 0.04G/TiO2).

From Mesoporous Electrodes to Photovoltaic Devices and Performance Measurements

Finally, the mesoporous TiO2 layers, with and without graphene, were used as electron transport layers in perovskite solar cells using a glass-fluorine doped tin oxide (FTO)/dense TiO2/mesoporous TiO2 (without or with graphene)/MAPI-Cl perovskite/Spiro-OMeTAD/Au n–i–p architecture (see Figure S5b in the Supporting Information). A preliminary analysis on the influence of the graphene content in TiO2 was conducted by adding cells containing an intermediate amount of graphene (0.02 wt %) for the comparison (the sample is labeled as 0.02G/TiO2). Further details of the experimental procedure used for device preparation are reported in Section and Supporting Information. Figure S5a presents the SEM cross section of a full device, showing that the expected sandwich structure is obtained.

Figure a–c represents typical current density–voltage (J–V) curves obtained from backward current–voltage characteristics measured under simulated solar emission (standard conditions, see experimental details in the Supporting Information) of perovskite solar cells based on TiO2, 0.02G/TiO2, and 0.04G/TiO2 porous electrode. Table presents the corresponding photovoltaic parameters (power conversion efficiency (PCE), short-circuit current density (Jsc), open-circuit voltage (Voc), and fill factor (FF)), including those extracted from the forward J–V sweep. We emphasize that these parameters are average values obtained from a limited number of devices that were processed in similar conditions for this study. It is important to note that the cells show significant hysteresis effect between forward and backward scans. This effect is mainly attributed to ionic migrations (mainly halides) in the perovskite layer, which can be directly affected by the charge extractions layers (either for electrons and holes).[33] Furthermore, it is also generally associated with interfacial defect states, as suggested by recent numerical simulation of perovskite solar cells.[34] Our mesoscopic device architecture, based on MAPI-Cl, is particularly sensitive to hysteresis as generally reported in the literature. Therefore, an adimensional hysteric index (HI), established by Sanchez et al., is also presented in Table .[35] HI is comprised between 0 (no hysteresis) and 1 (strong hysteresis).

Figure 6

J–V curves of solar cells with TiO2 (a), 0.02G/TiO2 (b), and 0.04G/TiO2 (c) based electron transport layers.

Table 2

Photovoltaic Parameters Extracted from Forward and Backward Measurements and the Hysteresis Index of the Solar Cells

		PCE (%)	Jsc (mA/cm2)	Voc (V)	FF (%)	HI
FW	TiO2	11.4 ± 1.2	20.6 ± 2.3	0.91 ± 0.3	61 ± 11
	0.02G/TiO2	10.6 ± 1.7	20 ± 0.2	0.89 ± 0	60 ± 10
	0.04G/TiO2	12.3 ± 0	22.3 ± 0.1	0.91 ± 0	61 ± 1
BW	TiO2	13.8 ± 1.2	21.7 ± 1.3	0.91 ± 0	70 ± 2	0.16 ± 0.06
	0.02G/TiO2	13.2 ± 1.3	20.6 ± 0.2	0.92 ± 0.07	70 ± 0	0.09 ± 0.02
	0.04G/TiO2	15.3 ± 0.1	22.3 ± 0.1	0.93 ± 0.01	74 ± 0	0.07 ± 0.02

In general, we clearly observe the improvement of solar cell performance with the increase of the graphene content. A high FF up to 74% (compared to 70% for the reference) and a decreased series resistance, from 26 to 22 Ω, are observed for cells containing graphene, evidencing a better charge extraction ability. In addition, the Voc increases from 0.91 to 0.93 V and the short-circuit current density measured under standard illumination increases from 21.7 to 22.3 mA/cm2 in the presence of graphene. This behavior is compatible with a better charge collection efficiency in the case of graphene-containing cells. This has a direct influence on power conversion efficiencies that reach an average of 15.3% efficiency compared to the 13.8% obtained for the reference. This trend is also confirmed through incident photon to charge carrier efficiency (IPCE) spectra measured on the devices, which are presented in the Supporting Information (Figure S6). The clear improvement of current generation over the whole spectral range is consistent with better charge extraction efficiency in the presence of the highest content of graphene (0.04 wt %).

Our results are consistent with the literature, for which an optimal graphene content exists.[15,36,37] Indeed, with a quite similar architecture of cell, Han et al. incorporated rGO into the mesoporous TiO2 layer with three different concentrations (0.2, 0.4, and 1.0 vol %). The best content appeared to be the intermediate one (0.4 vol %) for which an efficiency of 13.5% was obtained, compared to the 11.5% of their reference cell (pure TiO2).

Finally, the reduction of hysteresis in the case of TiO2/graphene-based porous electrode is evidenced by the reduction of the HI by a factor 2. Zhang et al. pointed out the role of interfacial oxygen vacancies in TiO2 as a potential cause of hysteresis in perovskite solar cells.[38] Our observations are therefore consistent with a reduction of oxygen vacancies on the TiO2 particle surface by the introduction of graphene. The intimate contact induced by the laser pyrolysis process between TiO2 and graphene is crucial in this context. Another positive influence of graphene incorporation is the fact that the standard deviation of all photovoltaic parameters is significantly reduced for solar cells containing the 0.04G/TiO2 composite material. Further in-depth analyses are currently undertaken to better interpret these effects; however, all of these indicators confirm the positive influence of graphene on perovskite solar cell operation.

Our first results are therefore very promising and clearly demonstrate the relevance of the laser pyrolysis process for the production of functional composites for energy conversion. Moreover, the optimal graphene content remains to be determined, suggesting that further performance improvement can be expected.

Conclusions

We show that laser pyrolysis enables to continuously produce TiO2/graphene nanocomposites with controlled properties (such as TiO2 crystallinity). More importantly, the technique provides an improved TiO2–graphene interface and a coverage of the graphene sheets by the TiO2 nanoparticles that induces a better electronic contact between them. Moreover, the technique can be expanded to produce one-step nanocomposites with different graphene contents or different graphene properties. The high quality of the obtained nanocomposites was applied to perovskite solar cells that exhibit a significant increase of their performance in average in the presence of graphene compared to pure TiO2, as well as lower hysteresis. Moreover, these nanocomposites might be of interest for the study of aging or degradation of perovskite solar cells in which mesoporous TiO2 is known to play an important role.[39,40] More generally, these results demonstrate the versatility of the laser pyrolysis process for the production of high-quality graphene nanomaterials and composites for solar energy conversion.

Experimental Section

To synthesize the nanocomposites, we choose a graphene material with suitable features for our application: a high conductivity and a few layers that offer a high specific area. Therefore, the “industrial G-200” material produced by the SIMBATT Company was used. Indeed, this industrial graphene powder is a reduced graphene oxide (rGO) with an oxygen content of less than 8 at.%, providing thus a high conductivity. Moreover, this graphene material is of a few layers (<10 layers), leading to a high specific area (>600 m2/g). For convenience, we speak about graphene to mention this reduced graphene oxide (rGO). The titanium precursor is liquid and is the titanium (IV) isopropoxide 87560-500 ML (TTIP) purchased from Sigma-Aldrich (≥97% purity).

The synthesis of the TiO2/graphene nanocomposite has been performed in a one-step method by laser pyrolysis. Its principle is based on the resonant interaction between a high-power infrared laser (CO2) and a precursor mixture that can be either gas or liquid-nebulized microdroplets, carried into the reactor zone thanks to an inert gas (argon).[20] The TiO2/graphene powders were obtained from TTIP as a liquid precursor for TiO2 formation (a sensitizer gas, C2H4, is added to the carrier gas in this case, as TTIP does not absorb well the laser radiation at 10.6 μm) and our commercial few-layered reduced graphene oxide powder (SIMBATT, Shanghai, China). The CO2 laser power was set to 520 W, and the pressure in the reactor was maintained at atmospheric pressure (105 Pa). Pure TiO2 nanoparticles were synthesized by laser pyrolysis as a reference material from pure TTIP. To produce TiO2/graphene nanocomposites, 0.04 wt % of reduced graphene oxide (rGO) was dispersed into TTIP using ultrasound. The obtained nanocomposite-based powder is then labeled as 0.04G/TiO2.

The morphology of the powders were then evaluated by a Carl Zeiss ULTRA55 scanning electron microscope (SEM) and by a JEOL 2010 high-resolution transmission electron microscope (HRTEM) operated at 200 kV. For SEM analysis, the powder was directly observed on the carbon tape. For HRTEM measurements, the powder was dispersed in ethanol and nanoparticles were separated with intensive ultrasound irradiation using a Hielscher Ultrasound Technology VialTweeter UIS250V. Then, the dispersion was dropped on a grid made of a Lacey Carbon Film (300 mesh Copper, S166-3H).

The materials have been integrated into perovskite solar cells and tested under standard conditions (AM1.5G, 100 mW/cm2) in ambiant atmosphere and without encapsulation to illustrate their quality and interest. Perovskite solar cells used in this work are composed of a stack of layers deposited onto glass substrates (see the structure in Figure S6). They comprise an FTO transparent electrode, followed by a compact TiO2 acting as hole blocking layer, an electron transport mesoporous TiO2 layer infiltrated by the perovskite, a hole transport Spiro-OMeTAD layer (doped with lithium salt (Li-TFSI) and tert-butylpyridine),[41] and finally a gold electrode. In this case, methylammonium lead iodide perovskite containing a small fraction of chlorine (CH3NH3PbI3–Cl, or MAPI-Cl) was deposited using a single-step procedure following a reported procedure.[42] This reference perovskite active layer was indeed found to allow a simple fabrication process in ambient conditions of devices with reasonable efficiencies due to the beneficial influence of chlorine on both charge diffusion length[43] and layer morphology and structure.[44] Further experimental features are detailed in the Supporting Information.