Incorporation of carbon nanotubes in hole transport materials for perovskite solar cells

ABSTRACT Many loss processes in perovskite solar cells (PSCs) take place at interfaces. One effective mitigation strategy is introducing carbon derivatives at the interface of the charge transport layers with the halide perovskite absorber and/or with the contacts. In this work, single-walled carbon nanotubes are incorporated into the hole transporting material (HTM), either as composites or as a separate layer on top of the HTM. Both organic and inorganic hole transport materials are studied to validate this approach for different types of HTMs. It is shown that the CNTs facilitate charge transport from perovskite to the back contact, regardless of the HTM used. The use of CNTs results in an increase in a significant increase in short circuit current and power conversion efficiency. This study illustrates how SWCNTs can be utilised to improve the properties of a wide range of HTMs, which will pave the way for more efficient and PSCs.


Introduction
The increasing global electricity demand and the negative impact of using fossil fuels on the environment motivate the development of clean and renewable energy technologies.Solar energy is a promising alternative to fossil fuels due to its sheer abundance (Cui et al. 2015;Ibn-Mohammed et al. 2017;Zhou et al. 2018).The past decade has seen the emergence of lead halide perovskite solar cells (PSCs).The ease of fabrication, abundance and low cost of materials, and high lab-scale power conversion efficiencies (PCEs) of up to 25.7% (Green et al. 2021) make PSCs a promising photovoltaic technology.The high PCE achieved in PSCs can be ascribed to the unique set of optoelectronic properties of lead halide perovskites, including; a tunable bandgap (E g = 1.5-2.3eV for pure leadbased perovskites) (Park 2015;Tu et al. 2017;Zhu et al. 2016;Mosconi et al. 2013), long charge carrier diffusion lengths, high charge carrier mobilities (∼35 cm 2 V −1 s −1 ) and high absorption coefficient (Herz 2017;Noh et al. 2013;Yin, Shi, and Yan 2014).PSCs are constructed by sandwiching the perovskite absorber between a hole and an electron transporting material (HTM and ETM).In the n-i-p structure, the ETM is deposited first, followed by perovskite and HTM, and vice versa for the p-i-n structure (Zhou et al. 2018;Green et al. 2021;Park 2015;Zhu et al. 2016;Herz 2017;Tong et al. 2016).PSCs using the n-i-p architecture generally show higher PCEs than their p-i-n counterparts (Best Research-Cell Efficiency Chart 2022).In n-i-p devices, the HTM plays a crucial role in charge extraction and stability, but not many materials possess the ability to execute both roles sufficiently (Zhou et al. 2018;Green et al. 2021;Park 2015;Tong et al. 2016;Cronin et al. 2017;Leijtens et al. 2015;Rajeswari et al. 2017).An ideal HTM needs energy levels matched to the perovskite, high hole mobility, thermal and chemical stability, and hydrophobicity to provide environmental stability (Zhou et al. 2018;Green et al. 2021;Park 2015;Tong et al. 2016;Cronin et al. 2017;Leijtens et al. 2015;Rajeswari et al. 2017).Two classes of HTMs have been used extensively in n-i-p mesoporous structures: (i) organic p-type semiconductors such as P3HT (Habisreutinger et al. 2014), Poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine (PTAA) (Khorasani et al. 2019), and 2,2',7,7'-Tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9'-spirobifluorene (spiro-OMeTAD) (Ko et al. 2018), and (ii) inorganic p-type semiconductors such as CuO, Cu 2 O, NiO x , CuI, CuSCN and CuIn (1-x) Ga x S 2 (Zhou et al. 2018;Green et al. 2021;Tong et al. 2016;Rajeswari et al. 2017;Juarez-Perez et al. 2016).Organic HTMs generally yield higher PCEs, but their high cost (due to complicated synthetic processes and high purity requirement), the need to use additives to optimise their conductivity, and their poor thermal, chemical, and environmental stability are significant drawbacks (Tong et al. 2016;Rajeswari et al. 2017).Inorganic HTMs on the other hand benefit from inexpensive fabrication using cost-effective and abundant materials and exhibit excellent stability.But a significant drawback is that inorganic HTMs need high processing temperatures to achieve high charge carrier mobility and low defect densities, which is incompatible with processing on top of lead halide perovskite (Tong et al. 2016;Rajeswari et al. 2017;Roose, Wang, and Abate 2019).There is thus much need to further improve the properties of HTMs for n-i-p PSCs.
In recent years, carbon derivatives have been used as a means to improve the performance of PSCs (Gong et al. 2021;Ferguson, Silva, and Zhang 2019;Hadadian, Smått, and Correa-Baena 2020).This includes graphdiyne (GDY) (Zhang et al. 2018), multi-walled carbon nanotubes (MWCNT) (Lee et al. 2015), single-walled carbon nanotubes (SWCNT, from here on CNTs) (Habisreutinger et al. 2014;Gatti et al. 2016;Shadrokh et al. 2020), reduced graphene oxide (RGO) (Gatti et al. 2016), and graphene oxide (GO) (Oo and Debnath 2017).These materials have been employed in several different ways: i) in the photoactive layer combined with perovskite, ii) as electrodes to replace costly Au and Ag electrodes or to improve and enable the charge transfer process by the electrode, iii) as an interlayer between perovskite and HTMs, and iv) as a composite with the HTM (Habisreutinger et al. 2014;Zhang et al. 2018;Lee et al. 2015;Gatti et al. 2016;Shadrokh et al. 2020;Oo and Debnath 2017;Nouri, Mohammadi, and Lianos 2018;Yang et al. 2018;Li et al. 2014;Wei et al. 2015).For example, Kartikay et al. (2021) reported the application of functionalised NiO x (f-NiOx) and MWCNT composite as hole transport material in n-i-p low-temperature processed carbon counter electrode-based PSCs, which resulted in the 11.36% PCE (see Table S1).Since the MWCNTs are multi-layer of graphene with larger diameters and have lower electronic properties than SWCNTs, in which by increasing the concentration of MWCNTs, the series resistance of the layer is increased which results in the suppression of the photovoltaic parameters e.g.fill factor (Charles et al. 1981;Lee et al. 2014).However, the SWCNTs are a single layer of graphene with ultra-high surface and distinct electronic properties, where the charge carriers can be transported through SWCNTs without scattering.Moreover, SWCNTs have beneficial mechanical and thermodynamical properties (Heise et al. 2009;Bati et al. 2019;Asgari and Lohrasbi 2012).
In this work, the integration of SWCNTs in NiO x and P3HT (additive-free) HTMs was studied, both as a composite (HTM&CNT) and a separate layer between the HTM and the back contact (HTM/CNT), with the aim to better understand the influence of SWCNT incorporation on device efficiency and stability.NiO x was chosen, as its properties are representative of inorganic HTMs, including a wide bandgap and a simple and cost-effective synthesis method (Oo and Debnath 2017;Zhang et al. 2016;Zhang et al. 2016;Yang et al. 2017;Abdollahi Nejand, Ahmadi, and Shahverdi 2015;Wang et al. 2016;Cao et al. 2017).Among the organic HTMs, P3HT was selected as it does not need any additives to boost conductivity (Habisreutinger et al. 2014), which would complicate studying the impact of CNTs.SWCNTs were chosen, as these can be easily processed into a web-like surface morphology, providing a facile and low-cost method to create an interconnected network of carbon derivatives (Lee et al. 2015;Kartikay et al. 2021).SWCNTs are either incorporated in the HTM, or used as a separate layer between the HTM and back electrode.Both strategies lead to a marked increase in efficiency, mainly by increasing J sc .SEM images show a uniform distribution of SWCNTs, both when they are incorporated into the HTMs, and when they are deposited as a separate layer on top, indicating that they can act as a charge carrier bridge between perovskite, HTM and back contact.PL results show increased quenching when SWCNTs are used for both HTMs, indicating charge extraction is enhanced.The P3HT&CNT composite shows the strongest quenching.In addition, electrochemical impedance spectroscopy (EIS) shows the lowest charge transfer resistance for P3HT&CNT.As a result, all HTMs show improved efficiency in combination with SWCNTs.NiO and SWCNTs perform equally well as a composite and as separate layers, but for P3HT the composite shows superior performance, which is ascribed to the strong interaction of P3HT with CNTs, leading to the highest efficiency.

Synthesis of nio x
For the preparation of the NiO x solution, 240 mg of Ni (NO 3 ) 2 •6H 2 O was dissolved in 20 ml of pure isopropanol (IPA) and stirred continuously overnight.The green NiO x solution was filtered by a 0.45 μm PTFE filter.

Preparation of nio x &CNT ink
To prepare NiO x &CNT, at first, 5 mg of CNT was added into 1 ml IPA under sonication for 3 h, 150 μl of the resulting suspension was added to 1 ml NiO x solution under stirring for 3 h, followed by 60 min sonication.

Preparation of P3HT&CNT ink
To prepare P3HT&CNT composite solution, first, 15 mg of P3HT was dissolved in 1 ml chlorobenzene at 70°C for 3 h under continuous stirring to obtain a homogeneous solution, then 150 μl of 5 mg ml −1 of CNT in chlorobenzene was added to P3HT solution under stirring for 3 h, followed by 60 min sonication.

Solar cell fabrication
Fluorine-doped tin oxide (FTO) glasses (10 Ω □ −1 ) were etched with zinc powder and HCl (4 M) to form the electrode pattern.The FTO glasses were cleaned separately in boiling DI water and 4 steps of an ultrasonic bath for 15 min, including deionised water with detergent, acetone, ethanol, 2-propanol, and finally dried by clean and hot air.The glass was exposed to a vacuum oven at 200 °C for 2 h before use.A 50-70 nm thin layer of TiO 2 was spin-coated from 250 µl TTIP and HCl (2 M) solution in ethanol at 2500 rpm for 30 s and then annealed at 500 °C for 30 min.A mesoporous TiO 2 layer was deposited by spin-coating a diluted TiO 2 paste in ethanol (ratio 1:6) at 2000rpm for 30 s.After the spin coating procedure, the substrates were heated at 120 °C for 10 min and then gradually annealed at 500 °C for 30 min.
The PbI 2 films were made from dissolving PbI 2 powder in DMF (1 M) filtered by a 450 nm PTFE filter and then spin-coated at 5000 rpm for 30 s, the films were annealed at 100 °C for 15 min.After cooling down, the PbI 2 films were placed in a petri dish at a distance of about 1 cm above the CH 3- NH 3 I (MAI) powder at 150°C for 3 h in an oven to convert the PbI 2 film to MAPbI 3 .After cooling to room temperature, the MAPbI 3 films were washed with isopropanol to remove excess MAI and were dried at 100°C for 10 min.For control devices, NiO x and P3HT were deposited by spin-coating at 3000 rpm for 30 s.For NiO x /CNT or P3HT/CNT devices, NiO x or P3HT was spin-coated at 3000 rpm for 30 s followed by deposition of CNTs.NiO x &CNT and P3HT&CNT composites were spin-coated at 3000 rpm for 30 s.Then, the samples were kept at 70°C for 10 min.Finally, an 80 nm layer of gold was deposited on the top of HTM to form a back contact.The active area of the electrodes was fixed with a 0.09 cm 2 shadow mask.

Characterisation
The thin films of CH 3 NH 3 PbI 3 perovskite were characterised by field-emission scanning electron microscopy (FESEM, TESCAN, Mira 3-XMU), X-ray diffraction (XRD, D8-Advance Bruker), The UV-Vis absorbance properties was determined by an Avantes spectrometer (AVASPEC-ULS2048L) in a range from 200 nm to 900 nm.Photoluminescence (PL) measurements were performed by using a nitrogen laser (NL 100) with an excitation wavelength of 337 nm.
Current-voltage (J-V ) curves of the CH 3 NH 3 PbI 3 PSCs were measured by a homemade parameter analyzer sweeping from -0.1 to 1.2 V a t a scan rate of 50 mV s −1 , illuminated under a solar simulator (AM 1.5G, 100 mW cm −2 , SharifSolar).The active area of the devices was defined by a 0.09 cm 2 aperture mask.Impedance spectroscopy measurements were carried out under illumination with an Autolab 302 N, with different DC bias potentials from 100 to 1000 mV and from 1 Hz to 1 MHz.The spectra were fitted using Z-View software.

Device architecture
Figure 1 shows the device architecture of HTM&CNT (Figure 1(a)) and HTM/CNT (Figure 1(b)) devices and the energy band alignment of PSCs employing P3HT, NiO x, and CNTs (Figure 1(c)).As can be seen from Figure 1(c), the energy levels of perovskite and the HTMs need to be well aligned to facilitate efficient charge transport from perovskite to the Au electrode.In addition, CNTs create a separate pathway for the extraction of holes toward the Au electrode.

Scanning electron microscopy
Figure 2 shows FESEM images of the pristine perovskite layer (verified by XRD pattern, see Figure S1a) and in combination with the different HTM layers (verified by XRD pattern, see Figure S1b for NiO x HTM) deposited on top.As shown in Figure 2(a), the perovskite layer is uniform and pinhole-free, and consists of grains with a diameter of several hundred nanometer.Figure 2(b) shows a NiO x layer on top of perovskite, consisting of small particles forming a uniform and dense layer.The morphology of the underlying perovskite layer cannot be seen, further suggesting the NiO x particles form a thick conformal layer.Figures 2(c,d) show a conformal covering of the perovskite layer by P3HT over multiple length scales.In contrast to NiO x , the underlying perovskite morphology is still visible, suggesting the P3HT layer is relatively thin.For NiO x &CNT (Figure 2(e)) CNTs are randomly distributed throughout the NiO x layer.Some aggregation of CNTs is taking place, which may be due to the low solubility of CNTs in isopropanol alcohol (IPA), and the magnetic properties of NiO x that lead to the aggregation of CNTs. Figure 2(f) illustrates the uniformly dispersed layer of CNTs on top of NiO x and perovskite.In this case, CNTs are much less aggregated than for the composite.Figure 2(g) shows the P3HT&CNT layer on perovskite, showing excellent complete coverage of the CNTs by P3HT to form a uniform layer of the P3HT&CNT composite.P3HT/CNT is shown in Figure 2(h), forming a uniform and dense layer, similar to the layer of CNTs formed on top of NiO x in Figure 2(f).Figure 2(i,j), demonstrate the cross-section of prepared devices based on NiO x /CNT and P3HT&CNT, respectively.The FTO, TiO 2 , perovskite, HTMs and gold layers can be clearly distinguished.Thus, as can be seen in FESEM images, by applying the CNTs into the HTM layers and/or depositing them on HTMs separately, CNTs form a carbon layer as an interface.The CNTs can act as a bridge that connects the perovskite layer and HTM, or HTM and back contact, to transfer holes in the perovskite layer to the HTM and back contact, aiding charge extraction.

Optical characterization
To investigate the influence of the HTMS on the optical properties of perovskite, we studied their UV-Vis absorption and photoluminescence (PL) spectra.Figure 3(a,b) demonstrate UV-Vis absorption spectra of the samples consisting of NiO x -based HTMs and P3HT-based HTMs on perovskite, respectively.As can be seen, all the samples exhibit a higher absorption in the visible region (400-700 nm) as compared with the pristine perovskite.This increase is mainly caused by scattering by the HTM and CNTs, as is evidenced by the below bandgap signal.In addition, the annealing step (70°C) after HTM deposition may enhance the quality of the perovskite crystals and improve absorption.Figure 3(c,d) illustrates the PL spectra of the samples in a Glass/Perovskite/HTM structure, excited at a wavelength of 337 nm.The high-intensity PL peak of the perovskite lies at 765 nm, and it tends to be quenched by the presence of different HTMs.The quenching of the PL peak is due to the charge transfer between perovskite and HTM.Thus, a more pronounced quenching of PL is indicative of better charge extraction from the perovskite to the HTM.As can be seen, both HTM&CNT and HTM/CNT strategies lead to increased quenching (and thus improved charge extraction).It is interesting to note that HTM/CNT shows similar quenching for both HTMs, but that for NiO x &CNT quenching decreases, whereas for P3HT&CNT quenching increases.This could be due to how P3HT effectively encapsulates the CNTs, preventing direct contact between perovskite and the CNTs.Because CNTs are not charge-selective, direct contact between perovskite and CNTs can lead to relatively high resistance and compromised hole transport (Habisreutinger, Nicholas, and Snaith 2017).

Electrochemical impedance spectroscopy
To gain a better understanding of the effect of CNTs on the recombination rate and charge transfer resistance at the perovskite/HTM interface, the EIS spectra of the samples were measured under  illumination.Figure 4(a,b) show EIS data for P3HT&CNT and NiO x /CNT samples by applying different voltages in the frequency range of 10 Hz to 1 MHz.The devices have only one semicircle in their Nyquist plots, which can be attributed to the charge transfer resistance (R t ) at the interface between the perovskite and HTM (Thakur et al. 2017;Wang et al. 2019;Zheng et al. 2017), and displaying no obvious semicircle for charge recombination (Thakur et al. 2017;Wang et al. 2019;Zheng et al. 2017).As the voltage increases towards the open-circuit voltage (V oc ), the radius of the semicircle decreases.A smaller radius indicates a reduction of R t and faster charge transfer at the interface between the HTM or ETM and the perovskite (Thakur et al. 2017;Wang et al. 2019;Ahn et al. 2018).The series resistance (R s ), which includes the resistances of FTO, TiO 2 , perovskite, and Au electrodes, is 12 Ω for the P3HT&CNT device and slightly above 13 Ω for the NiO x / CNT device and remains unchanged at different voltages due to the similar device structure.In Figure 4(c) reference devices employing NiO x and P3HT as HTM are compared to those using NiO x /CNT and P3HT&CNT, which showed the most PL quenching.Both HTMs show a significant reduction in R t upon the introduction of CNTs.According to previous reports (Zheng et al. 2017), a smaller R t is indicative of more efficient charge injection and.This translates into an enhancement of J sc , as can be seen in the J-V plot (see Figure 5, and Table 1), where J sc follows the same trend as R t , i.e. the reference HTMs have higher R t and lower J sc than the HTMs with CNTs, and NiO x HTMs have higher R t and lower J sc than the P3HT HTMs.

Solar cell performance
Figure 5 (a, b) shows the photocurrent density-voltage (J-V ) characterisation of PSCs fabricated by using the different HTMs.Table 1 presents the results of the photovoltaic parameters for all samples.In PSCs employing NiO x , the record efficiency is achieved for NiO x /CNT devices (Figure 5(a)), but average efficiencies are similar for NiO x /CNT and NiO x &CNT, which may indicate that the primary role of the CNTs is to improve charge transfer between the HTM and the back contact.In contrast, in PSCs employing P3HT, the highest record and average device efficiency is achieved for P3HT&CNT devices (Figure 5(b)), in line with the trend in charge transport that was found from PL.
The presence of CNTs results in a significant increment of J sc and PCE (see Figure 5, and Table 1) due to the improvement of fast hole extraction.Moreover, as can be seen in the band energy diagrams (see Figure 1(c)), the energy alignment of P3HT and Au (5.0 eV) can be another reason for the higher photovoltaic performance of P3HT&CNT than P3HT/CNT, since the energy level of P3HT and Au are the same, making injection harder, the CNT provides a more aligned pathway for the charges to reach Au.P3HT&CNT shows the best performance, which may be due to the ability of CNTs to be functionalised with P3HT (Yang et al. 2017) and result in a uniform layer (Figure 2(g)).The J-V characteristics of the P3HT&CNT and NiO x /CNT cells were monitored over 90 days in 35-40% humidity at room temperature to investigate their stability.Figure 5(c) shows the PCE of P3HT, NiO x , P3HT&CNT and NiO x /CNT devices for 90 days.It can be seen that after eight days the CNT devices stabilised and reached the highest efficiencies, which is consistent with previous reports (Gatti et al. 2016;Nouri, Mohammadi, and Lianos 2018).Moreover, P3HT&CNT devices showed a slower decrease in PCE compared to NiO x /CNT, where the devices retained 80% and 65% of their initial PCE after 90 days, respectively.In contrast, devices using NiO x or P3HT as HTM only retain 10% and 38%.This difference in stability is ascribed to the hydrophobic nature of the CNTs, which protects the perovskite against moisture.The presence of P3HT/CNT leads to delay of degradation because of two reasons, (i) CNTs play as a shielding layer against the penetration of moisture due to their hydrophobic property, and (ii) polymeric properties of P3HT and the formation of supramolecular CNT-polymer hybrids with high hydrophobicity form a protecting layer on top of the perovskite.Overall results demonstrate the promising application of CNTs as a composite with different HTMs and/or as an interlayer between HTM and electrode.

Conclusions
HTMs used in PSCs need both excellent charges for transporting properties and long-term stability.
In n-i-p PSCs, many HTMs only offer excellent charge transport (organic HTMs) or long-term stability (inorganic HTMs).In this study, we have shown that CNTs can be used to enhance the properties of organic (P3HT) and inorganic (NiO x ) HTMs in n-i-p PSCs.The CNTs can act as a bridge between the HTM and the back contact.PL and EIS results suggest this aids charge extraction, resulting in a marked increase in J sc and PCE.The fabricated devices exhibit promising stability as a result of the formation of a more uniform HTM layer by the introduction of CNTs.We show the merit of CNTs as a low-cost material that can be used to boost the performance of a wide range of HTMs, which will be important for the commercialisation of PSCs.

Figure 2 .
Figure 2. FESEM top views of (a) perovskite film applied on a TiO 2 mesoporous layer, (b) NiO x film on perovskite (c), and (d) P3HT film on perovskite (e) NiO x &CNT layer on perovskite, (f) NiO x /CNT layer on perovskite (g) P3HT&CNT layer on perovskite, (h) P3HT/CNT layer on perovskite.The scale bar corresponds to 500 nm.Cross-sectional FESEM image of the perovskite solar devices based on (i) NiO x /CNT, and (j) P3HT&CNT (a)-(i) are secondary electron images, (j) is a backscattered electron image.The scale bar corresponds to 250 nm.

Figure 4 .
Figure 4. Nyquist plots of the devices with (a) NiO x /CNT and (b) P3HT&CNT HTMs for different applied bias, and (c) comparison between Nyquist plots of devices using NiO x , P3HT, NiO x /CNT and P3HT&CNT as HTM, under illumination at the biased voltage of 0.8 V (inset: the equivalent circuit model used for EIS analysis).

Figure 5 .
Figure 5. J-V curves of perovskite solar devices with different HTMs a) P3HT and b) NiO x , after 8 days of fabrication kept at room temperature, 35-40% RH, and dark condition.and (c) comparison of photoconversion efficiency of NiO x , P3HT, NiO x /CNT and P3HT&CNT devices during 90 days.

Table 1 .
Photovoltaic parameters of the perovskite solar devices with different HTMs, 8 devices were used for each condition.