Organic Salts as p-Type Dopants for Efficient LiTFSI-Free Perovskite Solar Cells.

Despite the ubiquity and importance of organic hole-transport materials in photovoltaic devices, their intrinsic low conductivity remains a drawback. Thus, chemical doping is an indispensable solution to this drawback and is essentially always required. The most widely used p-type dopant, FK209, is a cobalt coordination complex. By reducing Co(III) to Co(II), Spiro-OMeTAD becomes partially oxidized, and the film conductivity is initially increased. In order to further increase the conductivity, the hygroscopic co-dopant LiTFSI is typically needed. However, lithium salts are normally quite hygroscopic, and thus, water absorption has been suggested as a significant reason for perovskite degradation and therefore limited device stability. In this work, we report a LiTFSI-free doping process by applying organic salts in relatively high amounts. The film conductivity and morphology have been studied at different doping amounts. The resulting solar cell devices show comparable power conversion efficiencies to those based on conventional LiTFSI-doped Spiro-OMeTAD but show considerably better long-term device stability in an ambient atmosphere.

Introduction

Organic hole-transport materials (HTMs) are widely used in solid-state, dye-sensitized solar cells (ssDSSCs)[1,2] and perovskite solar cells (PSCs).[3,4] However, due to low intrinsic conductivities, the addition of dopants is typically needed. Recently, novel doping strategies have become an increasingly studied topic, especially for perovskite solar cells.[5] Suitable dopants are considered as key factors for solar cell performance and device stability.

Coordination complexes are widely studied because they are cost-effective and commercially available. One representative family of such dopants is based on cobalt complexes. These were first applied as p-type dopants for 2,2′,7,7′-tetrakis[N,N-di(4-methoxyphenyl)amino]-9,9′-spirobifluorene (Spiro-OMeTAD) in 2011, when tris(2-(1H-pyrazol-1-yl)pyridine)cobalt(III) (FK102) was investigated in ssDSSCs. The devices based on FK102 showed an efficiency higher than 7.2%, with more than 80% of their initial performance remaining after 40 days of aging.[6] Further work in 2013 introduced two more cobalt complexes as dopants, tris(2-(1H-pyrazol-1-yl)-4-tert-butylpyridine)cobalt(III) tri[hexafluorophosphate] (FK209) and bis(2,6-di(1H-pyrazol-1-yl)pyridine)cobalt(III) tris(bis(trifluoromethylsulfonyl)imide) (FK269);[7] the differences regarding ligands are primarily associated with differences in redox potentials. It was emphasized that the oxidation of Spiro-OMeTAD is linked to the reduction of Co(III) to Co(II), according to the UV–vis absorption spectra shown in the report. Further work including tris[2-(1H-pyrazol-1yl)pyrimidine]cobalt(III) tri[bis(trifluoromethylsulfonyl)imide] (MY11)[8] and bipyridine cobalt complexes[9] made the subsequent selection of cobalt-based dopants even more popular. A recent study shows that the role of cobalt-based dopants is not limited to the actual p-doping, i.e., the generation of Spiro+ (first oxidation state of Spiro-OMeTAD) thus improving the conductivity. In addition, the complex may also protect the perovskite layer from the detrimental effects of the commonly used additives lithium bis(trifluoromethanesulfonyl)imide (LiTFSI) and 4-tert-butylpyridine (TBP) and thereby reduce the concentration of recombination sites.[10]

Copper salts represent another family of inexpensive and widely studied dopants. In 2016, CuI and CuSCN were reported as p-type dopants for Spiro-OMeTAD.[11] The conductivity was found to increase at least 1 order of magnitude after 33% molar ratio doping with CuSCN and 32% molar ratio doping with CuI. A new absorption peak from the HTM film was noted at around 454 nm and was attributed to oxidized Spiro-OMeTAD. The use of a copper(II)–pyridine complex was first reported in a previous work from our lab,[12] in which two copper complexes bis[di(pyridin-2-yl)methane] copper(II) bis[bis(trifluoromethyl-sulfonyl) imide][Cu(bpm)2] and bis[2,2′-(chloromethylene)-dipyridine] copper(II) bis[bis(trifluoromethylsulfonyl) imide][Cu(bpcm)2] were designed, synthesized, and applied in solar cells. The introduction of chlorides into the ligands results in the complex Cu(bpcm)2 having a low-lying redox potential. The oxidation yield of Spiro-OMeTAD by Cu(bpcm)2 was determined to be 75%. Devices based on Cu(bpcm)2 perform as well as those based on the cobalt-based dopant FK209, and a PSC device efficiency up to 18.5% was obtained. Other metal salts, including AgTFSI[13] and FeCl3,[14] have also been studied and were found to be effective alternatives to partially oxidized, doped Spiro-OMeTAD.

Apart from the coordination complexes and metal salts, small organic molecules have also been investigated as HTM dopants. Tetracyanoquinodimethane (TCNQ) derivatives, such as F4-TCNQ, were successfully applied as dopants for Spiro-OMeTAD by Huang et al.[15] Although the power conversion efficiency (PCE) obtained was a bit lower than for the control cells, the long-term stability was noted to become much better than for cells containing LiTFSI. The reason was attributed to the hygroscopic properties of the lithium salts, while the halogen atoms in the TCNQ molecules were postulated to increase the hydrophobicity of the hole-transport layer (HTL). Another investigated material is 1,1,2,2-tetrachloroethane (TeCA).[16] This compound was considered as a low-cost, chlorinated organic solvent, which can be used both as a cosolvent and an effective additive to Spiro-OMeTAD. The Spiro-OMeTAD/TeCA solution needs to be activated by UV light (λmax = 400 nm), in which Spiro+Cl– is schematically formed. Although the detailed photochemical reaction involved is unclear, the average efficiency of TeCA-doped devices was higher than those obtained with FK209, showing that small organic molecules represent promising alternatives to the standard doping substances used.

Organic salts, as another type of metal-free molecules for HTM doping, are structurally flexible, and the cations can be based on any hole-transport materials. The most commonly used counteranion tetrafluorosulfonimide (TFSI–) shows very weak coordination (Lewis-base) properties and extensive solubility in organic solvents. McGehee et al. oxidized Spiro-OMeTAD into Spiro(TFSI)2, and the latter was then used as dopant for Spiro-OMeTAD itself. PSCs fabricated with Spiro(TFSI)2 as the dopant show similar PCEs as devices based on a conventionally doped Spiro-OMeTAD HTM but showed better stability when illuminated and operated in an inert atmosphere.[17] Snaith et al. designed two new HTMs (AS44 and EH44) with long alkyl chains, and by doping with their respective oxidized HTM+TFSI– salts, LiTFSI could be omitted as an additive. This work demonstrates that it is possible to use the HTL as a highly hydrophobic, protective layer in perovskite solar cells without affecting the PCEs.[18] Further work by Sellinger et al. emphasized the importance of the stability of the organic salts, suggesting that EH44+TFSI– could also be applied as doping agents for other HTMs.[19]

Despite the extensive and recent studies on p-type dopants, doping by organic salts representing a LiTFSI-free alternative is worth further exploration. In this work, we report two new organic salts (MeO-TPD)TFSI and (TBD)TFSI (Figure ) based on the substructures of Spiro(TFSI)2, with subsequent application as dopants for Spiro-OMeTAD. We demonstrate that organic salts can be sufficiently efficient dopants for Spiro-OMeTAD. In addition, the stability of radical cations from the Spiro-OMeTAD substructures demonstrates a simple but efficient route to the design of organic salts for the purpose of LiTFSI-free and stable perovskite solar cells.

Figure 1

Molecular structures of (MeO-TPD)TFSI, (TBD)TFSI, Spiro(TFSI)2, and their reactivity with respect to Spiro-OMeTAD.

Results and Discussion

The synthetic routes of (MeO-TPD)TFSI and (TBD)TFSI are shown in Figure S1, and their reactivity with respect to Spiro-OMeTAD is shown in Figure . The previously reported Spiro(TFSI)2 was also synthesized following the method described,[17] and its doping ability was compared to that of (MeO-TPD)TFSI in terms of the photovoltaic performance in perovskite solar cells.

The structures of the synthesized organic salts are based on the molecular building blocks of existing HTMs, which means that the dopant itself can also act as the HTM. In addition, (MeO-TPD)TFSI can be considered as a substructure of Spiro(TFSI)2, where one of the fluorene derivatives has been eliminated from the structure. (TBD)TFSI was considered as a substructure of (MeO-TPD)TFSI, in which the phenyl parts have been removed and replaced with methyl groups. Such a design simplifies the structure of the organic dopants and thus lowers the synthetic efforts and, of course, costs. The structures of the materials were not attainable using proton NMR spectroscopy due to the paramagnetic properties of the radical cations and were instead verified by an elemental analysis. Moreover, 19F-NMR spectroscopy could be used to confirm the fluorine presence in the reported organic salts. As shown in Figure S2 and Figure S3, both (MeO-TPD)TFSI and (TBD)TFSI only display one 19F-NMR peak in their spectra, indicating that the organic salts have been successfully formed and that the TFSI– anions are chemically stable. The paramagnetic properties of the radical cations were further confirmed by electron paramagnetic resonance (EPR) spectroscopy, as shown in Figure a. The hyperfine coupling with three positive and three negative peaks observed in the (MeO-TPD)TFSI spectrum can be resolved by considering the coupling from two nitrogen nuclei with a coupling constant aN = 4.45G. Because of the similar coupling observed (around 22G) with respect to the full-width-half-maximum (fwhm) in the spectra of (MeO-TPD)TFSI and (TBD)TFSI, it is reasonable to assign the hyperfine lines in the (TBD)TFSI spectrum to nitrogen nuclei with a similar coupling constant. The spectra of (TBD)TFSI also show more lines in both positive and negative directions compared with (MeO-TPD)TFSI, indicating that the hyperfine coupling with protons may widen the spectral response.

Figure 2

(a) EPR spectra of (MeO-TPD)TFSI and (TBD)TFSI. UV–vis spectra from (b) (MeO-TPD)TFSI as the dopant for Spiro-OMeTAD, (c) (TBD)TFSI as the dopant for Spiro-OMeTAD, and (d) (MeO-TPD)TFSI and (TBD)TFSI. All spectra were recorded from an acetonitrile solution.

In order to be effective as p-type dopants, (MeO-TPD)TFSI and (TBD)TFSI need to oxidize Spiro-OMeTAD into Spiro+, forming neutral MeO-TPD molecules in the process. Different absorption features in the UV–vis spectra can be observed before and after doping. As shown in Figure b, pristine Spiro-OMeTAD displays a maximum absorption at 377 nm in acetonitrile. After the stepwise addition of (MeO-TPD)TFSI, new peaks at around 511 nm, 686 nm, and higher than 900 nm are observed. Although these peaks are likely to originate from oxidized Spiro-OMeTAD based on previously reported results,[17,20] the spectral features are more complicated in this work because of the very similar absorption characteristics by (MeO-TPD)TFSI. However, a careful investigation of the absorption at around 511 nm reveals that there is a small change in the position after a 200% molar ratio addition of (MeO-TPD)TFSI with a peak shift to 486 nm. This is a strong indication that Spiro-OMeTAD is oxidized after the addition of (MeO-TPD)TFSI. When the concentration is above a specific threshold, the oxidation process does not proceed further, and the absorption of nonreduced (MeO-TPD)TFSI starts to become visible in the spectra. Figure c shows the effects of the addition of (TBD)TFSI, where no new peaks at neither 511 nor 686 nm appear. The peaks at 471 and 900 nm originate from the absorption of (TBD)TFSI itself, as indicated in Figure d. Moreover, since there is no peak overlap at around 377 nm with peaks from Spiro-OMeTAD, it is clear that no oxidation takes place as no decrease was observed in the peak at 377 nm. This suggests that Spiro-OMeTAD is unaffected after the addition of (TBD)TFSI in solution.

In order to gain further insights into the reactivity of (MeO-TPD)TFSI and (TBD)TFSI with respect to Spiro-OMeTAD, cyclic voltammetry (CV) and differential pulse voltammetry (DPV) were used to study the redox potentials of the materials. The results are shown in Figure S4. As indicated from the CV results, the oxidation and reduction processes are reversible for both Spiro-OMeTAD and (MeO-TPD)TFSI. However, the reduction process of (TBD)TFSI shows features more akin to a two-electron process instead of two separate one-electron processes. This indicates that (TBD)TFSI may undergo a significant structural change upon oxidation/reduction. The DPV measurements show that Spiro-OMeTAD has a redox potential of −0.036 V versus Fc/Fc+, while (MeO-TPD)TFSI has a redox potential of 0.031 V versus Fc/Fc+, 67 mV more positive in comparison with that of Spiro-OMeTAD. In comparison, the redox potential of (TBD)TFSI is determined to be −0.053 V versus Fc/Fc+, thus slightly more negative than Spiro-OMeTAD. Therefore, the latter potential dopant is expected to be unable to oxidize Spiro-OMeTAD, in accordance with the results obtained from UV–vis spectrophotometry.

It is in this context where it is interesting to compare the results from organic dopants with our previous work,[12] in which the copper complex Cu(bpm)2 was applied as a p-type dopant for Spiro-OMeTAD. It turns out that, even with a 90 mV more positive redox potential, no reaction was observed in that case. The difference in the reorganization energy (ER) between the coordination complexes and organic salts may possibly account for the observed difference in the reactivity. As indicated by the results from density-functional theory (DFT) calculations,[21] the ER of (MeO-TPD)TFSI is 300 meV, which is significantly lower than that of Cu(bpm)2 by >1500 meV. These results indicate that organic salts may be more efficient as p-type dopants for Spiro-OMeTAD than coordination complexes and, as a consequence, may require a significantly lower driving force for oxidation.

The conductivity of Spiro-OMeTAD films was investigated and compared at different doping levels to further monitor the doping effect. As shown in Figure a, when applying an external voltage to Spiro-OMeTAD films, the obtained current was improved as a function of the concentration of (MeO-TPD)TFSI. However, the increase in the conductivity is negligible below 20% doping. Accordingly, the film conductivity improves from 7.77 × 10–7 S cm–1 for the pristine Spiro-OMeTAD film to 1.50 × 10–6 S cm–1 at 20% doping, as shown in Figure b. The conductivity increases dramatically with further doping, reaching 5.24 × 10–6 S cm–1 at 30% doping and 1.66 × 10–5 S cm–1 at 40% doping. Doping levels at 50% show a continued increase in the conductivity but with a lower increase rate.

Figure 3

(a) Current response of Spiro-OMeTAD thin films doped with (MeO-TPD)TFSI under different applied voltages. (b) Conductivity of Spiro-OMeTAD thin films doped by (MeO-TPD)TFSI at different doping levels.

A low film quality can become a detrimental issue, especially at high doping ratios. The scanning electron microscopy (SEM) images of doped Spiro-OMeTAD films at different doping levels of (MeO-TPD)TFSI are shown in Figure . The top images of the bare perovskite and reference HTL (Spiro-OMeTAD/LiTFSI/TBP/FK209) are also provided. The reference HTM layer contains a 3% molar ratio of FK209. The images are shown in Figure .

Figure 4

Top-down surface SEM images of (a) a bare perovskite surface; (b) a Spiro-OMeTAD film doped by a 3% molar ratio of FK209 (reference); a Spiro-OMeTAD film doped by (MeO-TPD)TFSI at (c) 0, (d) 10, (e) 20, (f) 30, and (g) 40% molar ratios; (h) Spiro-OMeTAD film doped with a 30% molar ratio of FK209.

As shown in Figure a, the bare perovskite film surface shows a well-crystallized material with typical crystallite sizes of around 200–500 nm.[22] After being coated with the reference HTL (Figure b), the perovskite surface is effectively coated, and the top surface is flat without visible pin holes. Replacing the dopant FK209 with (MeO-TPD)TFSI has no obvious influence on the film quality, as indicated in Figure c to 4g, where different doping levels of up to a 40% molar ratio were investigated. All films appear to be of a high quality without visible pin holes, aggregations, or cracks. Considering the high doping levels, this is much better than expected. In comparison, films doped by FK209 at a 30% molar ratio display a surface morphology with obvious cracks and a tendency toward crystal formation, as shown in Figure h. These results indicate that organic salts, such as (MeO-TPD)TFSI, can qualify as very good alternatives to FK209 as a dopant for Spiro-OMeTAD, even if requiring high doping ratios. Also, considering the high conductivity of HTM films doped by the organic alternatives, the hygroscopic additive LiTFSI is no longer required as a co-dopant.

Solar cells based on Spiro-OMeTAD/TBP/(MeO-TPD)TFSI as HTLs were fabricated and optimized. PSCs based on a reference HTL were also fabricated for the sake of comparison. All studied solar cells were based on the mixed perovskite (FA)0.85(MA)0.15Pb(I3)0.85(Br3)0.15 material. Figure S5 highlights all photovoltaic parameters of the devices doped by (MeO-TPD)TFSI without the addition of LiTFSI; the doping amounts comprise molar ratios of 10 to 40%.

As shown in Figure S5a and S5b, the initial doping (10% molar ratio) resulted in a slightly lower open-circuit voltage (Voc) as compared to the reference cells. Further doping showed an increase in the voltage with a corresponding decrease in the current density. The fill factor (FF) (Figure c) increased consistently with the doping level, indicating that the charge transport ability improves with an increasing film conductivity. Light soaking appears to not be important for the device performance increase in the tested devices at high doping amounts. The recorded efficiency reaches its maximum at a 30% doping level, which is comparable to the reference cell, as shown in Figure S5d. Details of the photovoltaic parameters are shown in Table S1.

Figure 5

(a) J–V curves, (b) Steady-state output of photovoltaic champion devices based on the dopant Spiro(TFSI)2 (blue squares), the dopant (MeO-TPD)TFSI (red circles), and the reference cells (yellow triangles). (c) Incident photon-to-electron conversion efficiency (IPCE) of champion devices doped by (MeO-TPD)TFSI.

The optimization shows that a 30% molar doping ratio offers the best PCEs. Referring back to the conductivity results, 30% (MeO-TPD)TFSI-doped films show a conductivity of around 5.24 × 10–6 S cm–1. Although such a conductivity is lower than for the LiTFSI-doped reference film (Figure S6), it appears to be good enough for rendering high device PCEs. Further optimization was performed on substrates with SnO2 as the electron transport material (ETM) and with a slightly different perovskite material (FA)0.91(MA)0.09Pb(I3)0.91(Br3)0.09 as the light-absorbing layer. It has been shown that SnO2 as the ETM gives less hysteresis and may therefore reveal more information on the HTLs.[23,24] Additionally, a 15% molar ratio of Spiro(TFSI)2 as the dopant was also included as a comparison to (MeO-TPD)TFSI, the ratio was selected based on the expected reactions between Spiro(TFSI)2 and Spiro-OMeTAD, where two molecular Spiro+ would be formed instead of one with (MeO-TPD)TFSI. As shown in Figure a, devices based on the reference HTL offered a Voc of around 1.10 V and a current density (Jsc) of around 22.8 mA cm–2; the FF was determined to be 0.76, and the PCE was around 19.1%. The hysteresis was negligible, as indicated in Figure b, which shows the steady-state output of the photocurrent density and the PCE determined at the maximum power point illuminated under 1 sun AM (air mass coefficient) 1.5G conditions. The steady-state output of the PCE was determined to be 18.8% at 0.925 V, similar to the results offered by the J–V curves. In comparison with the reference cells, Spiro-OMeTAD doped by (MeO-TPD)TFSI showed a slightly lower Voc at around 1.08 V, and the Jsc value was exactly the same, 22.8 mA cm–2, suggesting that LiTFSI can be eliminated and replaced by a large amount of (MeO-TPD)TFSI as the dopant without affecting the charge-transport abilities. The FF was found to be similar at around 0.75, and a comparable PCE at 18.6% was recorded. The steady-state output PCE of 18.1% at 0.90 V indicated a minor hysteresis effect, also comparable to that of the LiTFSI-doped reference cells. As previously emphasized, the comparable efficiency may also be derived from the neutral MeO-TPD that is formed after doping, a substance that can also work as a HTM, as confirmed in Figure S7 and Table S2. The Spiro(TFSI)2-doped device was found to give a lower PCE of around 17.8%, lower than both that of the reference cells and that of (MeO-TPD)TFSI-doped devices. The lower performance recorded for Spiro(TFSI)2-based devices can possibly be linked to the fact that the chosen doping level was arbitrary and was not the result from optimization. A comparison of the average performance showed the same results, details of which are given in Table and Figure S8. The steady-state output shown in Figure b illustrates that all devices are characterized by a stable output during the 180 s time period.

Table 1

Photovoltaic Parameters of Devices Based on (MeO-TPD)TFSI, Spiro(TFSI)2, and the Reference Cella

	Voc (V)	Jsc (mA cm–2)	FF	η (%)
Spiro+LiTFSI+TBP+FK209 (reference)	1.08 ± 0.03 (1.10)	22.6 ± 0.1 (22.8)	0.74 ± 0.02 (0.76)	18.1 ± 0.7 (19.1)
Spiro+TBP+30% (MeO-TPD)TFSI	1.07 ± 0.02 (1.08)	22.6 ± 0.2 (22.8)	0.74 ± 0.01 (0.75)	17.9 ± 0.4 (18.6)
Spiro+TBP+15% Spiro(TFSI)2	1.01 ± 0.04 (1.06)	22.4 ± 0.3 (22.7)	0.71 ± 0.03 (0.74)	16.1 ± 1.0 (17.8)

All HTLs contain TBP as an additive, and in addition, the reference cell contains LiTFSI as a co-dopant. The fabricated devices are based on SnO2 as the electron transport material and (FA)0.91(MA)0.09Pb(I3)0.91(Br3)0.09 as the active layer. The average data are obtained from 15 devices. The bracketed numbers are from champion cells.

Figure c shows the IPCE spectrum of (MeO-TPD)TFSI-doped devices. The spectrum indicates a strong absorption of the active layer up to 800 nm. The integrated current density is around 21.5 mA cm–2, similar to the values obtained from the J–V data. The incident photon-to-electron conversion efficiency (IPCE) spectra show that the dopant (MeO-TPD)TFSI is capable of oxidizing Spiro-OMeTAD and providing efficient charge transport at all wavelengths of absorption.

In addition to the comparable solar cell performance, the devices doped by (MeO-TPD)TFSI show a much improved stability toward humidity, in contrast to many other reported p-type dopants.[7,12,25] Results from contact-angle measurements and X-ray diffraction (XRD) tracking illustrate the hydrophobicity of the (MeO-TPD)TFSI HTLs quite clearly.

Figure a shows the contact angle of a Spiro-OMeTAD film doped by LiTFSI, TBP, and FK209, the reference in this study. Figure b shows the contact angle of a Spiro-OMeTAD film doped by TBP and (MeO-TPD)TFSI at a 30% molar ratio. As shown, the reference film is characterized by a contact angle of around 65.8° (mean), which is similar to previously reported results.[26] The elimination of LiTFSI and the inclusion of (MeO-TPD)TFSI increased the contact angle to 76.8° (mean). The higher contact angle in (MeO-TPD)TFSI-doped Spiro-OMeTAD suggests a more hydrophobic film, which potentially can increase the humidity-resistant properties of the resulting HTLs. The increased humidity resistance was further confirmed by XRD and PCE tracking over a longer period. Figure c shows the XRD tracking results (details are shown in Figure S9) of a perovskite film coated by the reference HTL and by a (MeO-TPD)TFSI-doped Spiro-OMeTAD film, respectively. The ratio between the (110) diffraction peak of the perovskite material at 14.34° and the (001) peak of lead iodide at 12.96° was used as an indicator to follow the perovskite decomposition process with time. The degradation of the reference perovskite film was observed already in the first 5 days, with only 37% left following normalization. In contrast, the devices covered by a film doped by (MeO-TPD)TFSI retained 87% of the perovskite material. The lower degradation rate was confirmed upon further storage in an ambient atmosphere, and the difference was less prominent after 20 days. As a result, 22% of the perovskite remained in the reference cell after storage in an ambient environment for 60 days, while the (MeO-TPD)TFSI-doped device retained 65% of the perovskite. The PCE tracking experiment showed similar results. As displayed in Figure d, 85% of the initial PCE remained after storage for 30 days in an atmosphere with a relative humidity of 40 ± 10%. In comparison, the PCE of the reference cells decreased quickly during the first 10 days, and the device stopped working after storage for 15 days.

Figure 6

(a) Contact-angle measurement of the Spiro-OMeTAD film doped by the reference HTL. (b) Contact-angle measurement of Spiro-OMeTAD films doped by TBP and (MeO-TPD)TFSI at a 30% molar ratio; (c) Normalized ratios between perovskite and lead iodide diffraction peaks from XRD tracking. (d) Stability of device performances.

Conclusion

Spiro-OMeTAD was doped using organic salts based on other HTMs. The resulting mixed HTMs without the addition of LiTFSI showed comparable PCEs (18.6%) to the reference cells (19.1%) co-doped with FK209 and LiTFSI. However, (MeO-TPD)TFSI-doped devices showed a significantly improved humidity resistance because of the elimination of the hygroscopic LiTFSI additive. The present results demonstrated the possibility to achieve highly efficient and stable PSC devices using Spiro-OMeTAD doped by organic salts other than Spiro(TFSI)2.