A Low-Cost and Lithium-Free Hole Transport Layer for Efficient and Stable Normal Perovskite Solar Cells

The most widely used material as a hole-transport layer (HTL) for effective normal perovskite solar cells (PSCs) is still 2,2′,7,7′-Tetrakis[N, N-di(4-methoxyphenyl)amino]-9,9′-spirobifluorene (Spiro-OMeTAD), which requires heavy doping with the hydroscopic Lithium bis(trifluoromethanesulfonyl)imide (Li-ΤFSI). However, the long-term stability and performance of PCSs are frequently hampered by the residual insoluble dopants in the HTL, Li+ diffusion throughout the device, dopant by-products, and the hygroscopic nature of Li-TFSI. Due to the high cost of Spiro-OMeTAD, alternative efficient low-cost HTLs, such as octakis(4-methoxyphenyl)spiro[fluorene-9,9′-xanthene]-2,2′,7,7′-tetraamine) (X60), have attracted attention. However, they require doping with Li-TFSI, and the devices develop the same Li-TFSI-derived problems. Here, we propose Li-free 1-Ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide (EMIM-TFSI) as an efficient p-type dopant of X60, resulting in a high-quality HTL with enhanced conductivity and deeper energy levels The optimized X60:EMIM-TFSI-enabled devices exhibit a higher efficiency of 21.85% and improved stability, compared to the Li-TFSI-doped X60 devices. The stability of the optimized EMIM-TFSI-doped PSCs is greatly improved, and after 1200 hr of storage under ambient conditions, the resulting PSCs maintain 85% of the initial PCE. These findings offer a fresh method for doping the cost effective X60 as the HTL with a Li-free alternative dopant for efficient, cheaper, and reliable planar PSCs.


Introduction
Over the last decade, perovskite solar cells (PSCs) have monopolized the interest of the scientific community as they have reached a power conversion efficiency (PCE) of 25.7% [1]. However, despite their excellent performance and low production cost, the long-term stability and durability against environmental factors, and the presence of a number of defects, are barriers on the road to commercialization [2][3][4]. PSCs are exhibiting excellent results in terms of efficiency compared to those of Si and CIGS solar cells [5][6][7]. The most efficient PSCs employ the n-i-p device architecture and incorporate 2,2 ,7,7 -Tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9 -spirobifluorene (Spiro-OMeTAD) as the hole transport layer (HTL), which must be doped with various additives, the most well-known being Lithium bis(trifluoromethanesulfonyl)imide (Li-TFSI) and 4-tert-Butylpyridine (t-BP) [5][6][7]. Spiro-OMeTAD suffers from low conductivity and unfavorable energy levels for perovskite devices, and in order to increase the hole mobility and to tune the energy levels of Spiro-OMeTAD, various p-dopants are employed. As well as Li-TFSI, Tris(pentafluorophenyl)borane (BCF), 2,3,5,6-tetrafluoro-7,7,8,8-tetracyanoquinodimethane good performance of X60 in PSCs, we used it as the HTL, and we endeavored for high performance devices without Li-TFSI as a dopant. The aim of this work is to show a cheaper alternative to Spiro-OMeTAD, exhibiting PCE greater than 21% and improved stability, without using the problematic Li-TFSI but with an effective Li-free dopant. The EMIM-TFSI can oxidize X60 efficiently, increasing the conductivity by two orders of magnitude and lowering the energy levels of X60 by about 0.3 eV. The EMIM-TFSI-doped X60 exhibits better energy level alignment with the perovskite and the top metal electrode and thus better performance.

Device Fabrication
A sequence of washes with distilled water, acetone, and isopropanol were used to clean the ITO substrate (Naranjo). After coating the ITO substrate with SnO 2 (diluted to 2.47 wt% in deionized water), the layer was spin coated at 6000 rpm for 45 s and subsequently annealed in air at 150 • C for 30 min to form the electron transport layer. The substrate was brought down to room temperature, treated with ultraviolet ozone for 10 min, and then coated with a PbI 2 solution. The PbI 2 powder was dissolved in DMF/DMSO (v/v, 94/6) at a concentration of 1.5 M and stirred at 70 • C for 15 h. FAPbI 3 perovskite was prepared by a two-step process. First, the PbI 2 solution was deposited by spin coating at 1900 rpm for 30 s, dried at 70 • C for 1 min, and then cooled to room temperature. On top of the PbI 2 layer, a solution of FAI/MAI/MACl/Ph-FACl (76.5:13.5:18:3.5 mg) in 1 mL of isopropanol was spin-coated at a rotation speed of 2500 rpm for 40 s [48,49]. This was followed by thermal annealing at 160 • C for 10 min in ambient conditions (approximately 40% RH). The HTL solution was then spin-coated at 3000 rpm for 30 s. The HTL for the control devices was prepared by dissolving 80 mg of X60, 30 µL of 4-tert-butylpyridine (tBP), and 35 µL of 260 mg mL −1 Li-TFSI in acetonitrile in 1 mL chlorobenzene. For the EMIM-TFSI-doped X60 devices, various percentages (wt%) of EMIM-TFSI (4%, 8%, 12%, 16%) were added to the X60 solution instead of Li-TFSI, with the molar ratio of tBP remaining constant. The as-deposited incomplete devices were aged for ∼12 h under dry air conditions (30% RH). Finally, a 15 nm MoO x layer and a 80 nm Ag layer were deposited by thermal evaporation under a pressure of 1.0 × 10 −4 mbar. The effective area was 0.04 cm 2 , as defined by an aperture mask.

Device Characterization
The FS5 spectrofluorometer was used to measure steady-state photoluminescence (PL) (Edinburgh Instruments, Livingston, UK). The work function (WF) and the higher occupied molecular orbital (HOMO) level were estimated by ambient photoemission spectroscopy (APS) using an APS04 N2-RH setup (KP Technology, Wick, UK). Specifically, a vibrating gold alloy probe was used to assess the contact potential difference (CPD) (2 mm in diameter). The absolute WF of a polished silver reference was measured, and its WF was determined using APS to calibrate the tip WF, which was calculated to be 4.54 eV. By extrapolating the cube root of the photoemission signal to zero and using a UV light excitation source (D2) with an excitation energy in the region of 3.8-6 eV, HOMO was obtained. The bandgap of Perovskite was calculated by the 1st derivative of the EQE spectrum. The LUMO level of X60 was calculated by adding the HOMO values as estimated from APS to the optical bandgap provided from the supplier (Dyenamo) using the following formula: where E g is the bandgap.

Results
The chemical structures of the Li-TFSI and EMIM-TFSI dopants are depicted in Figure 1a. The EMIM-TFSI contains an imidazolium cation instead of the Li + of Li-TFSI. Li + is known to diffuse through the PSC layers and to cause stability and performance issues [50]. Therefore, an efficient dopant of the HTL without the Li + cation is highly desirable. To assess the direct impact of EMIM-TFSI on the photovoltaic performance, different percentages of EMIM-TFSI were added to the X60 solution, and PSCs were fabricated with EMIM-TFSIdoped X60 as the HTL. The molecular structure of X60 and the planar device structure used in this work are depicted in Figure 1b,c respectively. Statistical findings from the same batch of ten devices are shown in Figure 1c. The control devices contain X60 doped with Li-TFSI. Compared to the control devices, all photovoltaic parameters, i.e., open circuit voltage (V oc ), short circuit current (J sc ), fill factor (FF), and PCE improve with increasing concentration of EMIM-TFSI, reaching an optimal performance at 12 wt% and an increased average PCE by +9.1%. In particular, the V oc significantly increases, which is most likely a result of the dopant's higher WF and surface passivation effect. The PSCs based on X60:EMIM-TFSI with an optimal doping percentage of 12 wt% exhibited the best PCE performance of 21.85%, with a high FF of 0.807, Voc of 1.132 V, and Jsc of 23.90 mA/cm 2 . This performance is comparable to that of devices that incorporate Spiro-OMeTAD as HTL doped with Li-TFSI and Co-TFSI [7,51,52]. The statistical distribution of the best performing devices is depicted in Figure 1d, which attests to the excellent repeatability of the devices. Moreover, in Table 1 the mean and champion photovoltaic parameters are listed. The photovoltaic parameters were extracted from the corresponding J-V curves. Figure 2a shows the reverse J-V scans of the control and best performing EMIM-TFSI devices. The hysteresis index can be calculated using the formula HI = (PCE (RS) − PCE (FS) )/PCE (RS) , where PCE (RS) and PCE (FS) represent the reverse and forward scans, respectively (HI). According to Table 2, both PSCs with X60:Li-TFSI and X60:EMIM-TFSI exhibit low HI values of 0.051 and 0.027, respectively. As we will show later, the decreased defect density at the perovskite/HTL interface and the absence of the Li + cation in the device may be the cause of the hysteresis reduction observed in the EMIM-TFSI-containing devices. The EQE spectra for the control and best performing PSC are shown in Figure 2b. The integrated current densities obtained in the range of 300-850 nm were calculated to be 22.85 and 23.97 mA.cm −2 for the control and optimal-doped devices with EMIM-TFSI, respectively (Figure 2b).   (b) EQE spectra for the best performing devices. The integrated current density depicted on the right axis was calculated from the device's spectral response and the AM1.5G photon flux spectrum. Integrated current density was estimated at 22.85 and 23.97 and mA cm −2 for the control and EMIM-TFSI (12 wt%), respectively. A crucial electrical property that affects the performance of HTLs is conductivity (σ), which can be estimated through direct current-voltage (I-V) measurements. We adopted the ITO/X60/Ag design for the I-V measurements. The conductivity can be determined from the following equation.
where A is the area of the electrodes (0.04 cm 2 ), d is the thickness of the X60 layer (200 nm), V is the bias that was applied, and I is the real-time current. The acquired I-V curves are shown in Figure 3a, and the slope of the linear curve gives the conductivity. Therefore, σ was estimated at 7.3 × 10 −6 S/cm, 3.35 × 10 −4 S/cm, and 6.83 × 10 −4 S/cm for X60: dopant-free, X60:Li-TFSI, and X60:EMIM-TFSI, respectively. The undoped X60 shows low conductivity but increases by two orders of magnitude upon doping with Li-TFSI and EMIM-TFSI, with the EMIM-TFSI-doped X60 exhibiting the highest electrical conductivity. The high electric conductivity of EMIM-TFSI-doped X60 indicates that the former readily oxidizes X60, p-type doping it [53,54]. To determine the energy levels of X60, APS and dark work function measurements were conducted. The valence band maximum (VBM) of the perovskite and the HOMO of X60 were estimated from the spectra demonstrated in Figure 3b,c. The undoped X60 shows a HOMO at −5.12 eV, which decreases to −5.25 eV upon Li-TFSI doping. However, when X60 is doped with EMIM-TFSI, the HOMO of X60 decreases to −5.44 eV, getting closer to the VBM of perovskite at −5.91 eV. This energy level adjustment with a reduced energy level difference between the HOMO of HTL and VBM of perovskite enhances the hole transport at the perovskite/HTL interface [55]. Figure 3e shows the correlation of the HOMO energy level position of X60 with PCE. With increasing EMIM-TFSI concentration, the HOMO has lower values, and the PCE increases, reaching a maximum at 12%. A higher doping concentration slightly increases the HOMO, and the PCE declines. Additionally, the X60:EMIM-TFSI has a higher WF value (5.32 eV) than X60:Li-TFSI (5.09 eV), as depicted in Figure 3d, which aligns better with the WF of MoO x and facilitates hole transport. Figure 4 shows the corresponding schematic energy level diagram. As is well known, the difference in Fermi Level (E F ) between the electron transport materials (ETMs) and the hole transport materials (HTMs) in PSCs determines the device's V oc . Therefore, it is assumed that X60:EMIM-TFSi-based devices with deeper WF will display higher V oc in PSCs. Moreover, the V oc is affected by the not-radiative recombination in the bulk and at the surfaces [56][57][58]. The enhanced performance and higher V oc of the X60:EMIM-TFSI devices can be related to reduced non-radiative recombination. As will be shown later in the manuscript with additional measurements, the X60:EMIM-TFSI devices exhibit reduced trap density and enhanced built-in potential, indicating reduced nonradiative recombination in the devices. The previous results show that EMIM-TFSI has the ability to act as an efficient dopant to control interfacial energy levels and X60 oxidation. The PL spectra in Figure 5a support the superior hole extraction capabilities of the EMIM-TFSI-doped X60. Specifically, the quenching of the PL intensity of the ITO/perovskite/X60:EMIM-TFSI sample indicates faster hole extraction compared to the control. The quenching of the PL spectra follows the concentration of EMIM-TFSI, that is, higher dopant concentration comes with reduced PL intensity. The PL results correlate with the improved PCE upon doping with EMIM-TFSI, and a concentration of 12% is needed for optimal performance. Although the doping with 16% induces further PL quenching, the PCE does not increase further due to the none-optimized energy level alignment at the perovskite/HTL interface for high doping concentration, as we have shown previously. Figure 5b depicts the dark J-V scans of holeonly devices exhibiting the trap-filled limit voltage (VTFL) characteristic. The following equation, which considers the elementary charge (1.6 × 10 −19 C), the trap-state density (N t ), the thickness of the perovskite film (L), estimated to be 450 nm, the relative dielectric constant (ε) (30 for perovskite), and the vacuum permittivity (ε o ), was used to estimate the trap density [59].   The trap density was calculated at 1.20 × 10 16 cm 3 and 5.90 × 10 15 cm 3 for the Li-TFSI and EMIM-TFSI-doped X60 devices, respectively. Interestingly, the EMIM-TFSI-doped devices exhibited lower trap density compared to the control, and consequently the trap assisted nonradiative recombination must have been reduced. Indeed, the optimized EMIM-TFSI-doped devices deliver higher V oc , which is the result of reduced non-radiative recombination in the device. In addition, the degeneration of the transient photovoltage (TPV) was investigated, as shown in Figure 5c. All photogenerated charge carriers will recombine when the device is maintained in the open-circuit condition; as a result, the recombination process will reflect the device's characteristics. It is abundantly clear that the charge recombination lifetime of the device based on X60:EMIM-TFSI is longer than that of the device based on X60:Li-TFSI, which is in agreement with the increased Voc. These results indicate that the non-radiative pathway of the EMIM-TFSI-doped HTL was reduced, and the charge extraction was improved, which is compatible with the PL results. Additionally, the non-radiative pathway of the EMIM-TFSI-doped HTL was reduced. The built-in potential (V bi ) of the two devices was estimated following Mott-Schottky analysis, as illustrated in Figure 5d. The following equation describes the relationship between capacitance and DC voltage bias [60]: where V represents the applied DC bias, N denotes the impurity doping density, A denotes the active area of the device, and ε o and ε r represent vacuum and relative permittivity, respectively. The Vbi can be determined using the X-intercept of the linear regime of the M-S plot. The device with EMIM-TFSI-doped X60 exhibited a higher V bi of 0.96 V than the control device's 0.88 V, which is consistent with the higher V oc of the EMIM-TFSI-doped devices. The enhanced V bi can be attributed to the reduction of nonradiative recombination due to the efficient passivation of charge defects at the perovskite surface and to the deeper HOMO of EMIM-TFSI-doped X60 [52]. The transient photocurrent measurements (TPC) involve the photo-excitation of the complete device with a short light pulse, slightly perturbing the otherwise constant bias illumination and creating charges out of the equilibrium. Then, after turning off the light, the perturbation charges will recombine, and after some time the current will reach equilibrium. The speed of the process is determined by charge carrier mobility, trap states, and doping levels [47]. The extracted charge from the TPC measurements on the full devices was plotted as a function of current density, as seen in Figure 5e. The EMIM-TFSI facilitates charge collection in comparison with Li-TFSI, which is in a good agreement with the non-radiative recombination lifetime extracted from the TPV measurements [61][62][63]. The analysis of the steady-state and transient measurements indicate the high capability of EMIM-TFSI to p-type-doped X60 and to improve the charge collection in the device.

Stability
The stability of PSCs is an important performance metric for the commercial potential of PSCs. The stabilized photocurrent density at maximum power point (MPP) as a function of time is depicted in Figure 6b. Devices based on HTL doped with 12% EMI-TFSI and Li-TFSI exhibit a steady current density of 22.02 mAcm −2 and 21.01 mAcm −2 ,

Stability
The stability of PSCs is an important performance metric for the commercial potential of PSCs. The stabilized photocurrent density at maximum power point (MPP) as a function of time is depicted in Figure 6b. Devices based on HTL doped with 12% EMI-TFSI and Li-TFSI exhibit a steady current density of 22.02 mAcm −2 and 21.01 mAcm −2 , respectively, keeping in tune with the PCE values tested by J-V measurements. PSCs degrade when exposed to humidity, thermal stress, and UV light, limiting their lifetime. Numerous researchers have participated actively in efforts to improve the stability of perovskite solar cells. [56][57][58][59][60][61][62][63]. The long-term operational stability of the devices was investigated under ambient conditions, according to the ISOS D-1 protocol [60]. Unencapsulated devices were stored in the dark under ambient conditions and approximately 50-60% RH and evaluated at regular intervals. As depicted in Figure 6c, the PCE degradation curve for the X60 demonstrates that the EMIM-TFSI-based device retains approximately 88% of its initial efficiency after more than 1100 h of storage. After 950 h of operation under the same aging conditions, the PCE of devices with X60:Li-TFSI declines to 45% of its initial value. To further evaluate the stability of EMIM-TFSI-doped devices, we measured the water contact angle. Figure 6a demonstrates that the X60:EMIM-TFSI-based film has a larger contact angle (78.1 • ) than the Li-TFSI-doped films (61.9 • ). Li-TFSI is known for its high hydrophilicity and has a detrimental effect on the stability [64][65][66][67][68]. Thus, the higher hydrophobicity of X60:EMIM-TFSI shields the HTL from water penetration and hence protects the underlying perovskite from the ambient humidity. In Table 3 are listed recent articles regarding PSCs that incorporate Spiro-OMeTAD as the HTL, doped with Li-free dopants, along with devices that used X60-and SFX-based HTLs. Compared to the Spiro-OMeTAD devices, our devices exhibit similar PCEs, but better stability. Moreover, our approach delivers devices with the highest PCE among the X60-based PSCs.

Conclusions
In conclusion, we have shown that X60 is an efficient and cheaper alternative to Spiro-OMeTAD for high-performance and stable PSCs. We propose the EMIM-TFSI as an efficient dopant for the X60 HTL that provides high PCE up to 21.85%, V oc of 1.132 V, increased J sc of 23.90 mAcm −2 , and an FF of 80.77, instead of the hygroscopic Li-TFSI. According to our findings, EMIM-TFSI has the potential to provide an effective route for X60 oxidation without using Li + cations, which are detrimental to the performance and stability of the device. The conductivity of X60 increases by two orders of magnitude upon doping with EMIM-TFSI. Moreover, the transient and steady-state characterizations of X60:EMIM:TFI PSCs indicated reduction in the non-radiative recombination and more efficient charge extraction. The superior performance of EMIM-TFSI-doped devices is attributable to the high work function and deeper HOMO of the X60:EMIM-TFSI layer, which provide a well-matched interface energy level alignment with the perovskite valence band at the perovskite/X60 interface and WF of MoOx, hence facilitating the extraction and transport of carriers in the device. To this end, we developed an efficient oxidation method to effectively promote X60 oxidation and p-type doping using the Li-free ionic liquid 1-Ethyl-3-methylimidazolium bis(trifluoromethanesulfonyl)imide (EMIM-TFSI). The optimized devices with 12% EMIM-TFSI (mass ratio relative to X60), yield a significantly improved PCE of 21.85%. Furthermore, the unencapsulated X60:EMIM-TFSI device exhibits enhanced stability and retains 85% of its initial PCE after 1200 h of storage under ambient conditions without encapsulation due to the absence of the hygroscopic Li-TFSI in the HTL.
Funding: This work was funded by the Greece-China joint R&D project Calypso (T7∆KI-00039), co-financed by Greece, the EU Regional Development Fund, and the Chinese Ministry of Science and Technology.
Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.

Conflicts of Interest:
The authors declare no conflict of interest.