As shown in Scheme S1, the compound FDY was synthesized from commercially available compounds 1,3,5,7-tetrakis(4-iodophenyl) adamantane and (N, N-bis(4-methoxyphenyl)-4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)) via Suzuki cross-coupling with good yield. The structure was characterized by nuclear magnetic resonance (NMR) spectroscopy (Supplementary Figures S1–2, Supporting Information). The thermal properties were determined by thermogravimetric analysis and differential scanning calorimetry, respectively. The compound FDY exhibits an appropriate thermal stability with a phase transition temperature at 91°C and a decomposed temperature at 400°C (Supplementary Figures S4A,B, Supporting Information). Due to the bulky three-dimensional structure of adamantane, the intermolecular interaction is weak and the glass transition temperature is low. This maybe a benefit to form a better stacking mode of side arms when the film of FDY is treated with thermal annealing and enhance the hole extraction and the transport efficiency of FDY films.

As shown in Supplementary Figure S3A, the UV–vis absorption of FDY film is mainly at 300–400 nm, and the absorption positions of the solution and the film are similar, indicating that the intermolecular π–π stacking of FDY in the film is relatively weak. According to the onsite of the UV–vis absorption spectrum, we calculated the optical band gap of FDY is 3.15 eV, and the HOMO/ LUMO energy levels of FDY were determined by ultraviolet photoelectron spectroscopy (UPS) measurement, as shown in Supplementary Figure S3. The valence regions (E _onset) and secondary electron cutoff edges (E _cutoff) are determined to be −8.94 eV (Supplementary Figure S3D) and 7.04 eV (Supplementary Figure S3E), respectively. The HOMO/LUMO energy levels of FDY are calculated to be −5.24 eV and −2.09 eV, respectively. Compared with PEDOT:PSS (−5.0 eV) and PTAA (−5.22 eV), the HOMO energy level of FDY matches much better with the valence band maximum of MAPbI₃; meanwhile, the lower LUMO energy level of FDY can effectively block the migration of electrons and avoid a nonradiative recombination at the interface between perovskite and HTM, which usually may lead to a higher Voc.

The mobility of FDY was determined by the space charge limited current (SCLC) method with a hole-only device based on a structure of ITO/PEDOT: PSS/FDY/Au, and it is determined to be 7.62 × 10^–5 cm² V⁻¹s⁻¹ (Supplementary Figure S5, Supporting Information) (Snaith and Grätzel, 2006). All optical and electrochemical data of FDY are summarized in Table 1.

To evaluate the performance of FDY as dopant-free HTM in inverted perovskite solar cells, we fabricated the corresponding devices with a structure of ITO/ FDY/ MAPbI₃/ PC₆₁BM/ BCP/ Ag and the control devices with PEDOT:PSS as HTM for comparison (Figure 1A), and the energy levels of respective layers are shown in Figure 1B. For all devices, the light-absorbing layer was prepared with MAPbI₃ by the one-step antisolvent method. The concentrations of FDY were optimized to improve the performance of PSCs (Supplementary Table S1), and the best PCE of 18.69% with a J_SC of 22.42 mA cm⁻², V_OC of 1.05 V, and FF of 79.31% was achieved with a concentration of 2.0 mg ml⁻¹ in toluene and the hysteresis was negligible, which is much better than that of the control device based on PEDOT:PSS (PCE = 15.41% with a J_SC of 19.38 mA cm⁻², V_OC of 1.02 V, and FF of 78.08%). Figure 1C displays the current density–voltage (J–V) curves for the optimized devices under the standard AM 1.5G illumination, and the results are summarized in Table 2. The EQE spectrum was executed to confirm J_SC values obtained from the J–V curve tested under the solar simulation, as shown in Figure 1D. The integrated J_SC from EQE is calculated to be 21.52 mA cm⁻² for FDY and 18.71 mA cm⁻² for PEDOT:PSS-based devices, respectively. Both match very well with the current from the J–V curve. Furthermore, we measured the stabilized power outputs for 300 s at the maximum power point. The PCEs obtained from the stabilized power outputs are 18.31 and 14.61% for the devices based on FDY and PEDOT:PSS (Figure 1E), respectively, indicating a good reliability of the devices. As shown in Figure 1F, the dark current of the devices with FDY and PEDOT:PSS was measured, and the device with FDY in a thickness of 10 nm (2 mg ml⁻¹) showed a lower leakage current, implying there was less recombination in the device.

To understand the role of FDY on PSCs further, steady-state photoluminescence (PL) and time-resolved photoluminescence (TRPL) were measured. From Figure 2A, it is obviously found that the peak intensity of PL of perovskite on FDY (with a quench efficiency of 95.5%) was weaker than that on PEDOT:PSS (85.4%), indicating that the FDY as HTM can extract holes from the perovskite layer much easier than PEDOT:PSS. As literature reported, HTM as the substrate in inverted planar devices usually affects the quality of the perovskite film (Lai et al., 2019b). From the TRPL tests (Figure 2B), we obtained the carrier lifetimes of perovskite films based on FDY and PEDOT:PSS, and the details are summarized in Supplementary Table S2. The photocarrier lifetimes of perovskite films depositing on FDY are τ₁ = 2.86 ns and τ₂ = 25.28 ns, and those of PEDOT:PSS are τ ₁ = 8.11 ns and τ₂ = 14.48 ns. Shorter τ₁ lifetime means that the FDY film has stronger hole extraction ability and the hole can be transferred from perovskite to HTL more quickly. The results of τ₂ indicate that the perovskite films prepared on FDY films have fewer defects and excitons can exist for a longer time in perovskite films before recombination.

Transient photovoltage (TPV) and transient photocurrent (TPC) tests were also carried out, which are often used to characterize carrier lifetimes and carrier extraction efficiency in organic solar cells (Clarke et al., 2014). The longer lifetime of TPV indicates the longer lifetime of generated carriers in an open-circuit state, demonstrating the lesser nonradiative recombination in the device. The TPV measurement in Figure 2C shows that the carrier lifetime of the device on FDY is significantly longer than that on the PEDOT: PSS. It means that the nonradiative recombination at the interface of FDY and perovskite is less than that of PEDOT:PSS, which is consistent with the higher V_OC of devices based on FDY HTL. Meanwhile, as shown in Figure 2D in the TPC test, the lifetime represents the stay time of the photogenerated carriers in the short-circuit state of the device, and the device on FDY has a TPC decay faster than that on PEDOT:PSS, indicating that FDY has a stronger hole extraction ability and faster hole transfer ability from the perovskite layer to the electrode. The longer decay TPV and shorter decay TPC of the device on FDY indicate that the FDY HTL improves the extraction efficiency of holes while reducing the nonradiative recombination in the interface, which may lead to a higher Voc than PEDOT:PSS.

To further investigate the charge recombination of the devices based on FDY, the light intensity dependence of J–V characteristics was conducted. The slope of V_OC versus natural logarithm of illumination intensity gives kT/q in a trap-free system, where k, T, and q are the Boltzmann constant, temperature in Kelvin, and the elementary charge, respectively (Cao et al., 2019). As shown in Supplementary Figure S6, the fitting results for n of PEDOT:PSS and FDY are 1.51 and 1.36, respectively. A higher value of n indicates that PEDOT: PSS causes a serious monomolecular recombination in the device when used as HTL; in contrast, FDY can reduce the trap density of the device and inhibit charge recombination when used as HTL, which corresponds to the PL and TRPL measurements.

In order to compare the different effects of FDY and PEDOT:PSS on the perovskite layer, AFM was used to characterize the roughness of the HTM layer. As shown in Supplementary Figure S7 (Supporting Information), the root mean square roughness of PEDOT:PSS is 2.70 nm, while that of FDY is 2.88 nm, and both films are very smooth with almost the same roughness. However, the tests of contact angle demonstrate that the surface property of two films are very different, which affects the morphology of the perovskite layer noticeably. As shown in Figures 3A,B, the contact angle of PEDOT:PSS is only 9° with water, due to its high hydrophilicity, while that of FDY is 74°, which is much more hydrophobic and can effectively prevent the water from eroding the perovskite film and can also keep a suitable wettability with the perovskite precursor solution. All these will be conducive to the preparation of a homogeneous and dense perovskite film and enhancing the stability of devices (as discussed in the following).

The morphology of perovskite films on FDY and PEDOT:PSS was tested by Scanning Electron Microscopy (SEM) (Figures 3C,D). Compared with the film on PEDOT:PSS, the MAPbI₃ perovskite film on FDY has much larger grain sizes and less grain boundaries, which leads to higher crystal quality of the perovskite layer and improves the charge extraction from perovskite to HTM. The results agree very well with that of PL, TRPL, TPV, and TPC.

In addition, X-ray diffraction (XRD) characterization also shows that the perovskite films deposited on FDY have better crystallinity. As shown in Figure 4, the intensities of peaks of (110) and (220) of the perovskite film on FDY film are much higher than those on PEDOT:PSS comparing with the peak of (312). It shows that the hydrophobicity of the FDY film can not only increase the grain size of the perovskite film but also improve the crystallinity of the perovskite film.

As discussed above, the suitable hydrophobicity of FDY should be beneficial to the stability of PSCs not only by preventing the perovskite film from the moisture but also by improving the crystal quality of the perovskite film. The stabilities of PSCs based on FDY and PEDOT:PSS were measured as shown in Figure 5. Interestingly, the PCE of the device based on FDY was retained more than 90% of its initial one after storing for 60 days in a nitrogen glove box without any encapsulation while that of device based on PEDOT:PSS in the same condition dropped down to 50% in 50 days (Figure 5A), which may be caused by the corrosion of acidic PEDOT:PSS on the perovskite film. Moreover, when the devices based on FDY were stored for 188 h in an ambient air with 50–60% relative humidity (Figure 5B), the PCE still remained at 81% of its initial one, while that of the PEDOT:PSS-based device dropped down to 43% of its initial one. The excellent long-term stability of FDY-based devices may benefit from the high hydrophobicity of the FDY film and high quality of the perovskite film.

Poly(3,4-ethylenedioxythiophene)/poly(styrene sulfonic acid) (PEDOT:PSS), PbI₂ (99.99%, TCI), MAI (Aus), PC₆₁BM (Xi’an P-OLED), 2,9-dimethyl-4,7-diphenyl-1,10-phenanthroline (BCP), (98%+, Aldrich), anhydrous N, N, dimethylformamide (DMF), anhydrous dimethyl sulfoxide (DMSO), toluene (99.8%), and anhydrous chlorobenzene (CB) were purchased from Sigma-Aldrich. All reagents were used directly without any further purification.

650 mg PbI₂ (99.99%, TCI) and 224.2 mg MAI (Aus) were dissolved in 1.0 ml mixed solvent of DMF:DMSO = 4:1. Then, the solution was stirred overnight at room temperature and then filtered with 0.22 µm syringe PTFE filters.

PSCs with the structure of ITO/ PEDOT:PSS or FDY /MAPbI₃/PC₆₁BM/BCP/Ag were fabricated, wherein the ITO was the bottom layer. ITO glass substrates were sequentially washed with soap water, acetone, and isopropanol for 30 min and dried with N₂. Then, the glasses were treated with a UV cleaner for 20 min. The PEDOT:PSS film was fabricated by spin-coating the PEDOT:PSS aqueous dispersion at 5,000 rpm for 45 s after filtering. Different concentrations of FDY were dissolved in toluene, and the films were spin-coated onto the ITO substrate at 5,000 rpm for 45 s and then were annealed at 150°C for 10 min. After being transferred into a glovebox, the perovskite films were spin-coated on substrates with a solution of 50 μl perovskite precursor at 4,000 rpm for 20 s and the antisolvent of 120 μl chlorobenzene was dropped on the perovskite layer during the 6th second. The films were annealed at 80°C for 10 min, immediately. An electron transport layer of PC₆₁BM (99.5%, Xi’an P-OLED) was deposited from a chlorobenzene solution (20 mg ml⁻¹) at 1,000 rpm for 40 s, and then, a bathocuproine (98% +, Aldrich) solution (0.5 mg ml⁻¹ in ethanol) was spin-coated at 4,000 rpm for 40 s. Finally, device fabrication was completed by thermal evaporation of a 80 nm thick film with Ag as the cathode.

The 1H NMR spectra were measured on an MECUYR-VX300 spectrometer. The UV–vis absorption spectra were recorded on a Shimadzu UV-2500 recording spectrophotometer. Ultraviolet and UPS measurements were performed in an ultrahigh vacuum surface analysis system equipped with a SCIENTA R3000 spectrometer with a base pressure of 10⁻¹⁰ mbar. UPS employed the He I 21.22 eV as the excitation source with an energy resolution of 50 meV. XRD patterns were recorded using a PANalytical X-ray diffractometer with Cu Ka radiation for crystallinity comparison. The scanning step size was 0.01°.

The current–voltage (J-V) curves of the solar cells were characterized after fabrication without any preconditioning via a Keithley 2,400 system under an Enli Solar Simulator (AM 1.5 G,100 mW cm⁻²). The certified device active area of 0.11 cm² was defined by the designed mask to avoid overestimation of the photocurrent density. NREL certified Si cells (KG-5) were used to calibrate the light intensity. The J–V curves were measured from 1.2 to 0 V (reverse scan) and 0–1.2 V (forward scan) with a step of 0.02 V in a glove box at RT. The EQE spectra were obtained using a QE-R system (Enli Tech.) under the AC mode. All of the PSCs had no encapsulation. Transient photovoltage decay was recorded by using a digital oscilloscope (DOS-X 3104A) at an open-circuit condition.