The incorporation of potassium into perovskite solar cells (PSCs) has been empirically validated to mitigate hysteresis phenomena and boost the power conversion efficiency (PCE). However, the doping mechanism of potassium ions in the perovskite film and their effect on photocarrier recombination remains a topic of debate. Here, we grew doped MAPbI3: K single crystals by inverse temperature crystallization using KI as a dopant, and then perovskite thin films were spin-coated with dissolved MAPbI3: K crystals as a precursor. The doped MAPbI3: K perovskite films exhibit better crystal quality with large columnar grains and lower defect density. Employing Hall effect, ultraviolet photoelectron spectroscopy, and Kelvin probe force microscopy measurements, we definitively demonstrate that K-doping transforms the conductivity type of the perovskite film from a marginally N-type to a distinct P-type semiconductor. Furthermore, this doping strategy induces a concurrent downward shift in both the conduction band minimum and valence band maximum. As a result, the PCE of the PSCs increases from 15.15% to an impressive 20.66%, and the J–V curve hysteresis almost disappears. Additionally, theoretical simulations using SCAPS-1D software reveal a profound modification in the device's energy band diagram after K+-doping. Specifically, the energy level offset between the perovskite layer and the electron transport layer diminishes from 0.24 to 0.14 eV, with a result of bigger quasi-Fermi energy level splitting. This, in turn, elevates the open-circuit voltage (Voc) of the doped perovskite solar cell, underscoring the profound impact of potassium doping on enhancing PSC performance.
I. INTRODUCTION
Organic–inorganic hybrid perovskite semiconductors have attracted extensive attention in recent decades owing to their excellent properties such as high absorption coefficient, a tunable bandgap, long carrier diffusion length, and small Urbach energies.1–5 The power conversion efficiency (PCE) of perovskite solar cells (PSCs) increased from the initial 3.8% to 26.1% only in the past ten years.6,7 Although the PCE of PSCs is very close to that of silicon-based cells, abnormal current (J)–voltage (V) hysteresis exists in the voltage scanning under different directions.8 At present, there are several opinions on the origin of such hysteresis, e.g., ferroelectric polarization,9,10 capacitance effect,11,12 Shockley–Read–Hall nonradiative recombination,13–15 and ion migration.16 Among them, deep-level traps caused by specific charged defects are mainly nonradiative recombination centers, which is the most important factor limiting the solar cell PCE to the Shockley–Queisser (S–Q) theoretical limit17 as well as one of the most important reasons of hysteresis in PSCs.18
One of the most effective routes to eliminating hysteresis in PSCs is to introduce potassium into the perovskite film or the interfaces of the PSC devices.19–21 Segawa and co-workers showed that K+ ions in the perovskite film can both modify the crystal lattice and the grain boundaries. Moreover, the incorporation of K+ changed the energy level position and minimized the charge accumulation at the interface with TiO2, thus reducing the hysteresis of PSC devices.22 Based on the x-ray diffraction (XRD) and density functional theory (DFT) calculations, Park and co-workers claimed that K+ occupied the interstitial site in the perovskite lattice and prevented the Frenkel defect formation of iodide.23 Zhao and co-wokers also proved through experiments and DFT calculations that K+ is the best interstitial cation of FAPbI3, which inhibits the diffusion of I− by partially blocking the diffusion pathway, increases the vacancy formation energy of iodide, and reduces the hysteresis of PSCs.24 Stranks and co-workers demonstrated that K+ can inhibit transient light-induced ion migration and prevent halide segregation, ultimately reducing hysteresis in PSCs.25 Prezhdo and co-workers proposed a mechanism of alkali ion passivation of metal halide perovskite structural defects by means of non-adiabatic molecular dynamics (NA-MD) and time-dependent density functional theory (TD-DFT). They also suggest that the B-site occupation of the alkali ion is more favorable than the inter-gap occupation, and that the Pb-p bond state is broken by replacing the Pb atom near the iodine vacancy, thus alleviating the defect state.26 Until now, the position of K+ in the perovskite lattice and its effect on the photocarrier recombination dynamics and device performance is still a topic of debate. Recently, we found the monovalent Cu+ in the MAPbI3 crystal at the Pb site can impel the change of conductive type from weak N-type to P-type obviously,27 which encourages us to explore the effects of monovalent alkaline earth metal doping on the intrinsic property of perovskites and the device performance.
In this paper, we grew MAPbI3: K doped thin films using redissolved single crystals as a precursor28 and studied the effect of K+ doping on the electrical and optical properties of perovskite semiconductor single crystals and doped film-based solar cell devices. It was found that K+ doping elevates the work function of perovskite and transforms the conductivity type from a weak N-type to a pronounced P-type. Furthermore, the position of conduction band minimum (CBM) and valence band minimum (VBM) in the doped samples are shifted downward. Combined with the results of optical absorption, XRD, UPS, and Kelvin probe force microscopy (KPFM), we believe that K+ mainly replaces Pb2+, and thus induces P-type doping. In addition, the crystal quality of doped perovskite films was further improved with large columnar grains, and the defect density was significantly reduced as proved with the SEM and space charge limited current (SCLC) measurements. The PCE of the final champion device (FTO/NiOx/perovskite/PC61BM/BCP/Ag) increased from 15.15% to 20.66%, and the hysteresis factor decreased from 0.0884 to an impressive 0. Using SCAPS-1D simulation software, we observe that K+ doping reduces the energy level offset between the perovskite layer and PC61BM from 0.24 to 0.14 eV, enabling a more efficient quasi-Fermi level splitting (QFLS). Accordingly, the simulated Voc increases from 0.98 to 1.04 eV, which is in excellent agreement with the experimentally observed Voc rise from 0.98 to 1.06 eV. These results indicate that the addition of K+ to perovskite can both significantly reduce hysteresis and boost open-circuit voltage. Our champion device represents one of the highest efficiencies (19.8%–21.4%) for MAPbI3-based thin film solar cells with a typical planar P-i-N structure.
II. EXPERIMENTS
A. Materials
Lead iodide (99.99%), methylamine iodide (95%), PC61BM (99.8%) were purchased from Xi'an Yuri Solar Co., Ltd. γ-Butyrolactone (GBL, ≥99%), potassium iodide (98%), chlorobenzene (AR), nickel nitrate hexahydrate (99.9%), zinc nitrate hexahydrate (99.9%) were purchased from Aladdin. Methylamine aqueous solution (AR, 25.0%–30.0%), ethylene glycol (AR), ethylenediamine (AR) was purchased from Sinopharm. Dimethylformamide (99.8%) was purchased from Sigma-Aldrich. Acetonitrile (99.9%) was purchased from McLean. All drugs were used as is without further processing.
B. K-doped CH3NH3PbI3 single crystal growth
PbI2, MAI, and KI (5%) were dissolved in 10 ml of γ-butyrolactone to obtain a mixed solution of 1.23M. The solution was heated and stirred for 3 h at 60 °C on a heating table. After that, the precursor solution was removed, and the solution was filtered into a clean Petri dish using a syringe and a PTFE filter tip, and seed crystals with a diameter of about 1–2 mm were placed in the center of the above solution and covered with a lid. The solution was transferred to a heating table preheated to 100 °C, and the crystals were allowed to grow for 24 h to obtain centimeter-sized single crystals with regular shapes and smooth surfaces.
C. Device fabrication
The FTO glass was first cleaned with glass cleaning solution, acetone, isopropyl alcohol, and de-ionized water for 30 min, and then blown dry with a nitrogen gun and plasma treated. The prepared NiOx precursor solution was deposited onto the FTO glass at a speed of 5000 rpm in an air environment, and then it was moved into a fast annealing furnace for 1 h at 400 °C. The substrate was then annealed in a glovebox. Subsequently, the substrate was transferred to a glovebox, and the deposition of the perovskite layer was divided into two types: A-film was deposited by one-step spin-coating method at 5000 rpm, and chlorobenzene was added as a counter-solvent to purify the film during the spin-coating process, and the perovskite layer was formed by annealing at 100 °C after spin-coating, and the C-film was deposited by one-step spin-coating method, and when the deposition speed was 5500 rpm, the precursor solution was rapidly dripped onto the substrate. The precursor solution was rapidly dripped onto the substrate at 5500 rpm, and the perovskite layer was formed after 60 S without any treatment. PC61BM was deposited on the perovskite layer at 3000 rpm, annealed at 100 °C for 10 min, and then BCP was deposited on the electron transport layer. Finally, the substrate was removed from the glovebox and Ag electrodes were deposited in a vacuum coater to obtain a perovskite solar cell with a complete device structure.
D. Simulation of solar cells
Simulations were performed using SCAPS-1D software (http://scaps.elis.ugent.be). The current density–voltage (J–V) characteristics of the solar cells were simulated under standard AM 1.5 G spectroscopy with an intensity of 100 mW cm−2 in the voltage range of 0–1.2 V.
E. Characterization
Characterization of the XRD patterns was carried out using an Ultima IV x-ray diffractometer with Cu-Kα radiation at 10° min−1. Optical absorption spectroscopy (UV) was performed using a UV-3600 manufactured by Shimadzu, Japan. Photoluminescence (PL) was measured using a photoluminescence spectrometer (Edinburgh Instrument, FLS1000) with a xenon lamp (Edimburgh Instrument, Xe2) at an excitation wavelength of 450 nm. For time-resolved PL (TRPL) spectroscopy measurements, a 450 nm diode laser was used as the excitation source. Hall effect measurements were performed using a Hall measurement system (Lake Shore 8404) with the Van Der Pauw protocol. Ultraviolet photoelectron spectroscopy (UPS) measurements of the films were performed using an ESCALAB 250Xi from Thermo Fisher, USA. The photoelectric conversion efficiency (PCE) of the solar cells was recorded and measured by a solar cell photoelectric conversion efficiency test system, which in this thesis was provided with AM1.5 simulated sunlight using an XES-40S1 sunlight simulator manufactured by San–Ei with a light intensity magnitude of 100 mW/cm2.The PV performance was mainly tested by a Keithly 2612A digital source meter. The external quantum efficiency (EQE) was measured using a QEX10 quantum efficiency test system manufactured by PV Measurements Inc. in the US, with a wavelength range of 350–850 nm.
III. RESULTS AND DISCUSSIONS
Three groups of thin films were prepared by the traditional solution (T-film) method and the crystal redissolved method (C-film: 0% K+; C-film: 5% K+), respectively. The MAPbI3: K doped single crystals and films were analyzed by XRD as shown in Fig. 1(b). No new diffraction peaks appeared after K+ doping, indicating that no impurities were detected after K+ doping the perovskite semiconductor. The (110) diffraction peak at 2θ = 14.2° was further magnified as shown in Fig. 1(c), and it was found that the diffraction peaks of both the 5% K+ doped crystals and the thin film were shifted to smaller angles. The results indicated that K+ doping makes the crystal cell expand and the lattice constant becomes larger. X-ray photoelectron spectroscopy (XPS) tests of Figs. S1 and S2 in the supplementary material indicate the presence of K+ in doped MAPbI3 crystals and films. Meanwhile, SEM-EDX of Fig. S3 in the supplementary material demonstrates that K+ is distributed uniformly throughout the perovskite film. Since the ionic radius of K+ (0.138 nm) is larger than that of Pb2+ (0.119 nm), and in conjunction with our previous research,27 we believe that the reason for the shift of XRD diffraction peaks to a smaller angle is that K+ replaces Pb2+. If more K+ cations substitute A sites, the perovskite structure will collapse as the Goldschmidt tolerance factor will be smaller than 0.8 due to the huge difference in ionic radius between K+ (0.138 nm) and MA+ (0.217 nm). This supposition is also consistent with the findings reported by Lin et al.,29 who attributed that the XRD peak shift to a lower angle is mainly due to the substitution of Pb2+ with K+ on the B-site, which reduces the bandgap by changing the lattice field. In addition, Prezhdo and co-workers.26 found the B-site occupation of the alkali ion is more favorable than the inter-gap occupation, using a combination of nonadiabatic molecular dynamics (NA-MD) and time-domain density functional theory (TD-DFT).
To further investigate the doping mechanism and verify our conjectures, several optical spectra of the doped crystals and thin films were measured as shown in Fig. 2. The optical absorption edges of the samples exhibited a significant red shift after K+ doping [Figs. 2(a) and 2(c)], indicating that K+ doping can enter into the perovskite lattice and cause a change in bandgap. Figures 2(b) and 2(d) show the PL spectra of crystal and thin film samples. The PL emission peak also exhibited a red shift, in accordance with the optical absorption result and previous report.22 Combined with the results of XRD, we believe that the redshift of UV and PL is also due to the replacement of Pb2+ by K+, which causes the lattice expansion and thus reduces the optical bandgap.
To investigate the potassium doping effect on the solar cell device performance, we fabricated the PSCs devices with a FTO/NiOx/perovskite/PC61BM/BCP/Ag structure using three different perovskite films, as shown in Fig. 5(a). Previous reports have indicated that the performance of the device is obviously affected when the doping concentration of Rb+ reaches a certain threshold.37 To address analogous issues, we prepared devices with a wide K+-doping concentration from 0% to 10%. As illustrated in Fig. 6, the device performance is optimal when the K+ doping concentration is 5%. Thereafter, the device performance declines. Consequently, this concentration was selected as the experimental group for further analysis of the impact of K+ doping on MAPbI3 PSC performance.
Figure 5(b) shows the current density–voltage (J–V) curves of the champion devices fabricated with T-film and 0% or 5% K-doped perovskite film. The PCE increases sequentially from 15.15% to 18.61% and finally to 20.66%, and the hysteresis factor decreases sequentially from 0.0884 to 0.0736 and 0. The detailed device parameters are summarized in Table I. The primary long-term stability of unpackaged devices is presented in Fig. S6 in the supplementary material. Indicating that K-doped devices exhibit enhanced stability. Combined with the previous analysis of the film quality, we believe that the improvement of Jsc mainly depends on the highly ordered lattice structure of the perovskite film grown by the crystal redissolved method. Moreover, the films further showed perfect columnar crystallization after the addition of K+. In the external quantum efficiency (EQE) test as shown in Fig. 5(c), the improvement of EQE and integrated current is also observed due to the preferred orientation of the grains.
Film type . | . | PCE (%) . | Jsc (mA/cm2) . | Voc (V) . | FF (%) . | Hysteresis factor . |
---|---|---|---|---|---|---|
T-Film | F | 13.81 | 21.47 | 1.02 | 62.90 | 0.0884 |
R | 15.15 | 20.00 | 1.02 | 73.98 | ||
C-Film:0%K+ | F | 17.24 | 24.91 | 0.98 | 70.35 | 0.0736 |
R | 18.61 | 24.23 | 0.98 | 78.00 | ||
C-Film:5%K+ | F | 20.66 | 25.78 | 1.06 | 75.82 | 0 |
R | 20.66 | 25.63 | 1.06 | 75.96 |
Film type . | . | PCE (%) . | Jsc (mA/cm2) . | Voc (V) . | FF (%) . | Hysteresis factor . |
---|---|---|---|---|---|---|
T-Film | F | 13.81 | 21.47 | 1.02 | 62.90 | 0.0884 |
R | 15.15 | 20.00 | 1.02 | 73.98 | ||
C-Film:0%K+ | F | 17.24 | 24.91 | 0.98 | 70.35 | 0.0736 |
R | 18.61 | 24.23 | 0.98 | 78.00 | ||
C-Film:5%K+ | F | 20.66 | 25.78 | 1.06 | 75.82 | 0 |
R | 20.66 | 25.63 | 1.06 | 75.96 |
In order to assess the reproducibility of the performance of the K+ doped cell devices, 20 cells were prepared using the each film growth method. The corresponding device PV parameters and hysteresis factor statistics are shown in Fig. S5 in the supplementary material and Fig. 5(d). The average hysteresis factor of the cell device decreased in a sequential manner from 0.103 to 0.056 and 0.01 with K+ doping. Previous reports have indicated that the migration of I− and the formation of defects and vacancies caused by I− migration are one of the main reasons for the J–V hysteresis in the MAPbI3 solar cell devices.23–25 In the previous section, we have deduced that after K+ enters the MAPbI3 lattice, and K+ primarily replaces Pb2+ at the B position. The bond energy of K–I (322.5 ± 2.1 KJ/mol) is larger than that of the Pb–I bond (194 ± 38 KJ/mol),34 which indicates that I− is more stable in the perovskite lattice after K+ partially replaces Pb2+. This prevents the migration of I− ions, which inhibits the formation of halide vacancies and effectively reduces the hysteresis phenomenon of PSC. In word, our champion device represents one of the highest efficiencies (19.8%–21.4%, as shown in Table II) for MAPbI3 thin film-based solar cells with a planar P-i-N structure. More importantly, K-doping can effectively eliminate the J–V hysteresis effect in such devices.
Device configuration . | PCE (%) . | J-V hysteresis . | Reference/year . |
---|---|---|---|
ITO/NiOx/MAPbI3/PC61BM/AM-TiOx/Ag | 20.84 | Yes | Ref. 38/2020.5 |
FTO/NiOx/MAPbI3/PC61BM/BCP/Ag | 21.4 | Yes | Ref. 39/2021.4 |
FTO/NiOx/MAPbI3/PC61BM/BCP/Ag | 19.8 | Yes | Ref. 28/2022.1 |
ITO/NiOx/MAPbI3/PCBM/BCP/Ag | 19.92 | Yes | Ref. 40/2022.6 |
ITO/NiOx/MAPbI3/PCBM/BCP/Ag | 19.91 | Yes | Ref. 41/2022.7 |
ITO/NiOx/MAPbI3/PCBM/BCP/Ag | 20.8 | Yes | Ref. 42/2024.8 |
FTO/NiOx/MAPbI3/PC61BM/BCP/Ag | 20.66 | No | This work |
Device configuration . | PCE (%) . | J-V hysteresis . | Reference/year . |
---|---|---|---|
ITO/NiOx/MAPbI3/PC61BM/AM-TiOx/Ag | 20.84 | Yes | Ref. 38/2020.5 |
FTO/NiOx/MAPbI3/PC61BM/BCP/Ag | 21.4 | Yes | Ref. 39/2021.4 |
FTO/NiOx/MAPbI3/PC61BM/BCP/Ag | 19.8 | Yes | Ref. 28/2022.1 |
ITO/NiOx/MAPbI3/PCBM/BCP/Ag | 19.92 | Yes | Ref. 40/2022.6 |
ITO/NiOx/MAPbI3/PCBM/BCP/Ag | 19.91 | Yes | Ref. 41/2022.7 |
ITO/NiOx/MAPbI3/PCBM/BCP/Ag | 20.8 | Yes | Ref. 42/2024.8 |
FTO/NiOx/MAPbI3/PC61BM/BCP/Ag | 20.66 | No | This work |
As shown in Fig. S5 in the supplementary material, the main improved parameter in the doped solar cell is Voc compared with the pristine device. To explore the reason, we used solar cell capacitance simulator (SCAPS-1D)43,44 to validate our experimental results. First, we simulate the undoped perovskite cell device, and the simulated J–V curves are shown in Fig. 7(a). The photovoltaic parameters obtained were PCE of 18.39%, Voc of 0.98 V, Jsc of 24.32 mA/cm2, and FF of 77.45%. These values are comparable to the experimental results, indicating the validity of the parameters chosen (see Table S3 in the supplementary material for specific parameters). Among them, the thickness was obtained from SEM images, the bandgap from UV spectroscopy, the mobility and carrier concentration from Hall measurements (see Table S1 in the supplementary material), Nt from the SCLC test, and the rest of the parameters were referenced from the literature.45–51 Then, we similarly simulated the 5% K+ doped film-based cell device with updated parameters and the J–V curve is shown in Fig. 7(c). The obtained photovoltaic parameters are as follows: PCE of 21.98%, Voc of 1.04 V, Jsc of 24.48 mA/cm2, FF of 86.25%, which are in general agreement with the experimental results. As illustrated in the energy band diagrams presented before and after doping in Figs. 7(b) and 7(d). The energy level offset between the perovskite layer and the PC61BM decreases from 0.24 to 0.14 eV after 5% K+ doping, a superior alignment of the energy levels is achieved, along with a smoother charge transfer pathway and a reduction in electron accumulation at the interfaces. This results in a notable decline in nonradiative recombination at the interfaces and achieves a more efficient QFLS.52 In the Shockley–Queisser theory, QFLS and are two interchangeable quantities that are usually considered equal.53 The simulation demonstrates a enhancement from 0.98 to 1.04 eV, which is in alignment with the experimental enhancement of from 0.98 to 1.06 eV. Concurrently, the simulation shows that the QFLS is augmented from its original value of 0.98–1.06 eV. It can, therefore, be posited that the observed increase in Voc can be attributed to the enhanced QFLS in the doped device. In addition, K+ doping makes the energy level offset at the interface between perovskite and PC61BM lower, which is also a contributing factor to the near-elimination of J–V hysteresis.
IV. CONCLUSION
In conclusion, our study has demonstrated the viability of K+ substitutional doping in MAPbI₃ crystals, which can be employed for the fabrication of highly efficient and hysteresis-free PSCs. Besides the improved film crystal and optical quality after doping, K+ doping also resulted in a more aligned energy level between MAPbI₃ and PC₆₁BM, leading to a smoother and efficient charge transfer. This reduces nonradiative recombination at the interface and realizes a more efficient QFLS, thus improving Voc of PSCs devices. Moreover, due to the higher binding energy of KI over PbI₂, the partial substitution of Pb2+ by K+ makes I− more stable in the perovskite lattice, preventing the migration of I− ions. This doping process reduces the formation of halide vacancies, thereby significantly reducing the J-V hysteresis phenomenon in PSCs. In addition, we adopted the crystal redissolved method to grow perovskite films, which ensures the precise attainment of stoichiometric equilibrium during the doping process. Our research presents a simple and reproducible method for producing high-efficiency, hysteresis-free PSCs, opening up avenues for promising applications in the broader landscape of perovskite photovoltaic devices.
SUPPLEMENTARY MATERIALS
See the supplementary material for additional XPS, EDS, KFM, and PL characterizations on perovskite films and more detailed solar cell device performance data.
ACKNOWLEDGMENTS
This work was supported by NSFC (No. 12174211), the Key R&D Program of Shandong Province, China (No. 2024SFGC0102), Jinan Bureau of Education (JNSX2023015), and Jinan Bureau of Science and Technology (202333042).
AUTHOR DECLARATIONS
Conflict of Interest
The authors have no conflicts to disclose.
Author Contributions
Weijie Li: Data curation (equal); Writing – original draft (equal). Ting Liu: Data curation (equal). Guanwen Chen: Data curation (equal). Ning Li: Investigation (supporting). Xia Wang: Funding acquisition (equal); Supervision (equal). Zongming Liu: Funding acquisition (equal); Project administration (equal). Bingqiang Cao: Project administration (equal); Writing – review & editing (equal).
DATA AVAILABILITY
The data that support the findings of this study are available from the corresponding author upon reasonable.