First author: Ming-Hua Li, Shuo Wang
Corresponding author: Hu Jinsong
Correspondence unit: University of Chinese Academy of Sciences
Research highlights:
1. A dimethylammonium extraction strategy utilizing hydrogen bonding promotion was developed.
2. The preparation humidity and temperature range windows are significantly expanded.
3. Efficiency of undoped poly(3-hexylthiophene) (P3HT)-based CsPbI3 perovskite solar cells
reaches 20.25% and exhibits excellent humidity and operational stability.
1. Problems and challenges of inorganic perovskites
Although organic-inorganic hybrid perovskite solar cells (PSCs) have achieved certified efficiencies of up to 26.1%, the organic components are unstable under moisture, light and high temperature conditions. By replacing organic cations with inorganic cesium ions, the intrinsic stability is fundamentally improved, providing great hope for the preparation of stable PSCs. However, the commonly used DMAPbI3 (dimethylammonium [DMA]) or “HPbI3” assisted crystallization method to prepare CsPbI3 films often results in DMAPbI3 residue, thereby reducing photovoltaic performance and stability.
2. Introduction to results
In view of this, Professor Hu Jinsong’s team at the University of Chinese Academy of Sciences developed a universal hydrogen bond-promoted DMA extraction strategy to prepare high-quality γ-CsPbI3 without film residues of DMAPbI3. This environmentally friendly crystallization process significantly extends the manufacturing humidity and temperature window, improving device stability. PSCs of dopant-free poly(3-hexylthiophene) (P3HT) achieve a high efficiency of 20.25% and exhibit excellent storage and illumination stability.
In this study, EQE was measured using Enlitech-QE-R3011 product.
3. Results and Discussion
Point 1: DMAPbI3-assisted CsPbI3 crystallization kinetic engineering under the action of PAA
First, the author dissolved a certain amount of PAA in a DMF solution and added it to the CsPbI3 precursor to prepare a PAA-treated film. Pure CsPbI3 and PAA-treated CsPbI3 were labeled as control and PAA samples, respectively. As shown in Figure 1A, the PAA sample exhibited a faster color change from light yellow to deep black compared to the control sample. The crystal structure and phase transition process of the resulting perovskite film (Figure 1B) were studied using X-ray diffraction (XRD) technology and matched the standard γ-CsPbI3 phase. The disappearance of the (002) peak of the PAA film (Figure S2) indicates that the PAA treatment helps to form a structure that is more inclined to the (110) orientation. It is worth noting that there is still a small peak at about 11.8° in the control sample, which corresponds to DMAPbI3. In addition, both perovskite films show weak Cs4PbI6 diffraction peaks, which help stabilize the lattice structure. The effect of PAA treatment on the DMAPbI3-assisted CsPbI3 crystallization process was further studied by time-dependent XRD.
Figure 1 Crystallization kinetics study of CsPbI3
Furthermore, the effect of PAA treatment on the crystallization process of DMAPbI3 assisted CsPbI3 was studied through time-dependent XRD (Figures 1C and 1D). As annealing proceeded, the peaks of the two samples showed significant changes at a high angle of approximately 14.4 °, indicating that the lattice shrinkage caused by DMA evaporation led to the complete disappearance of the decomposition of the intermediate phase (CsxDMA1 − x) PbI3. At the initial stage of 0 minutes, the two cast perovskite films showed similar diffraction peaks, which are called DMAPbI3 phases. As annealing progresses, the DMAPbI3 phase begins to decompose, γ- The CsPbI3 phase gradually forms. Upon closer observation of the diffraction intensity evolution of the DMAPbI3 peak located at approximately 11.8 ° (Figure 1E), the diffraction intensity improved with increasing annealing time in a short period of time, but sharply decreased over a longer annealing time, as was the case for both the control and PAA samples. For the (110) peak of CsPbI3 located at approximately 14.4 °, the diffraction intensity of the two samples gradually increased, indicating the formation of CsPbI3. This phenomenon indicates that after PAA treatment, both the control and PAA samples experienced three stages, namely (1) crystallization of DMAPbI3, (2) decomposition of DMAPbI3, and (3) γ- The formation of CsPbI3. Interestingly, after PAA treatment, the decomposition and high crystallinity of DMAPbI3 γ- The phase transition process of CsPbI3 accelerates.
In situ oblique incidence wide angle X-ray diffraction (GIWAXS) measurements were conducted, and the intensity evolution of DMAPbI3 (100) and CsPbI3 (110) peaks was extracted and plotted as shown in Figure 1F. For the control sample, after annealing for 390 seconds, DMAPbI3 began to decompose, showing a decrease in the intensity of the DMAPbI3 (100) peak. Through the introduction of PAA, the decomposition of DMAPbI3 began at 310 seconds, indicating that PAA effectively advanced the decomposition of DMAPbI3. From the perspective of the evolution of CsPbI3, the formation of CsPbI3 began at 310 seconds for the control and PAA samples. Without the introduction of PAA, the intensity of the CsPbI3 (110) peak continued to increase until 1240 seconds, resulting in a relatively long crystallization period of 840 seconds. On the contrary, the PAA sample fully crystallized within 540 seconds of a significant decrease. The steep slope of the PAA sample during CsPbI3 crystallization indicates that PAA accelerates the formation kinetics of CsPbI3 (Figure 1F). The crystallization process of DMAPbI3 assisted CsPbI3 is determined by the decomposition kinetics of DMAPbI3 (i.e. DMAI sublimation) and the formation kinetics of CsPbI3 (i.e. Cs ion insertion into the Pb-I framework). The mismatch between the decomposition of DMAPbI3 and the formation of CsPbI3 may lead to the appearance of small pores (Figure 1G, Figure S8, and Figure S9) in the CsPbI3 film observed in the control sample, which typically serve as carrier recombination sites and hinder the improvement of PCE.
Key point 2: Hydrogen bonding promotes DMA extraction
In order to gain a deeper understanding of the influence of PAA on the crystallization of CsPbI3 from a thermodynamic perspective, the effect of PAA additives on the escape properties of CH3NH2CH3+(DMA+) from DMAPbI3 was studied using density functional theory (DFT). The (100), (010), and (001) surfaces of the DMAPbI3 model were established (Figure 2A and Figure S10) to calculate the escape potential barrier (Eb) with or without PAA treatment. Obviously, the Eb values of all three surfaces treated with PAA were significantly lower than those of the control DMAPbI3 (Figure 2B; Table S1), attributed to the strong hydrogen bonding between PAA and DMAPbI3 (N-H ··· O, indicated in orange). Further thermal gravimetric analysis (TG) measurements were conducted on samples treated with or without PAA to quantitatively evaluate the effect of hydrogen bonding on the thermal decomposition of DMAPbI3. The temperature at the 1% mass loss location is defined as the decomposition temperature. Compared with the control samples at different heating rates, PAA samples significantly reduced the decomposition temperature at various heating rates. The activation energy (Ea) for thermal decomposition of two samples was calculated using the Ozawa Flynn Wall method. The Ea of the PAA sample is significantly lower than the 141.6 kJ/mol of the control sample, which is 94.3 kJ/mol (Figure 2C), confirming the theoretical results.
Figure 2 Hydrogen bond-promoted extraction mechanism of dimethylammonium
Perform X-ray photoelectron spectroscopy (XPS) spectroscopy measurements to detect changes in the composition of the formed perovskite surface. The control sample shows a peak located at approximately 402 eV (Figure 2D), which is N 1s of residual DMA cations. For PAA samples, the N 1s peak disappears, indicating that there is no residual DMA on the surface of the perovskite film. Then, the prepared CsPbI3 film was dissolved in DMSO-d6 solution and residual DMA in the entire film was detected using nuclear magnetic resonance (NMR) measurement. The peak located at 8.17 ppm is a DMA molecule. For the control sample, the DMA signal still exists (Figure 2E). For PAA samples, the DMA signal disappears. Transmission electron microscopy (TEM) confirmed the presence of PAA in CsPbI3 thin films. Compared with the control sample (Figure 2F), the CsPbI3 sample treated with PAA is covered with an amorphous layer of several nanometers, because large-sized PAA cannot enter the lattice and aggregate between grain boundaries. The formation of amorphous layers may also help improve device stability. Based on these analyses, a three-stage crystallization process is described in Figure 2G: (1) The wet film is initially composed of DMAPbI3, PAA, and CsI. Then, during the annealing process, the hydrogen bonds caused by PAA facilitate the extraction of DMA and accelerate the insertion of CsI into the PbI6 skeleton, resulting in a rapid transition to CsPbI3. (2) Extended annealing leads to complete removal of DMA, while PAA is located at the grain boundary, forming γ- CsPbI3. (3) The concept of hydrogen bonding promoted DMA extraction was further validated by using two similar molecules, polyacrylonitrile (PAN) and poly (4-vinylpyridine) (PVP), which can form N-H ··· N hydrogen bonding interactions with DMAPbI3. As expected, both PAN and PVP can accelerate the crystallization of CsPbI3 and inhibit the residue of DMAPbI3, demonstrating the universality of this method.
Key point 3: Characterization of thin film optics and devices
The Pb 4f peak of the PAA sample shows a significant shift towards lower binding energy compared to the control sample (Figure 3A). The shift of binding energy indicates a favorable passivation effect, which may be caused by carbonyl groups in PAA. Steady state and time resolved photoluminescence (PL) measurements were performed on control and PAA samples. The PL emission intensity of CsPbI3 film treated with PAA is three times higher than that of the control film (Figure 3B). Average Life of PAA-CsPbI3 Thin Films by Fitting PL Attenuation Curve（ τ- The average is 29.12 ns, which is significantly longer than the 13.29 ns of the control sample, indicating inhibition of non radiative recombination losses caused by defects. For space charge limited current (SCLC) measurement, after PAA treatment, the defect density of CsPbI3 sample was significantly reduced (4.02 × 1014 cm-3), relative to the control sample (1.17 × 1015 cm-3) (Figure 3D). The CsPbI3 PSC treated with PAA had a higher built-in potential (Vbi, 0.91 V), while the control sample had a value of 0.69 V (Figure 3E), which is attributed to the suppressed defect density shown in the PL and SCLC results. The increased Vbi helps to form longer depletion regions and stronger driving forces for carrier transport and separation. Further J-V measurements of light intensity dependence were conducted to investigate the relationship between Voc and light intensity. At 300 K, the slope of the PAA-CsPbI3 device decreased to 1.28 kT/q, while the control sample was 1.73 kT/q, due to the improvement of membrane quality and the decrease in defect density (Figure 3F).
Figure 3 Optical and device electrical characteristics
Point 4: PV performance and stability of PSCs
The authors prepared inorganic PSCs to evaluate the impact of PAA treatment on device photovoltaic performance.To avoid device degradation caused by aqueous additives, the impurity-free polymer P3HT (Figure 4A) was chosen as the hole transport layer. Interestingly, PAA-treated CsPbI3 PSCs achieved a PCE of 20.25%, Voc of 1.16 V, Jsc of 20.99 mA cm-2, and FF of 0.831 (Fig. 4B), which is reported for n-i-p structured inorganic CsPbI3PSCs without doping HTL The highest PCE in (Fig. 4C; Table S3). The average efficiency also significantly improved from 16.42% to 19.55% (Figures 4D and 4E; Table S4), mainly due to the increase in Voc (from 1.08 to 1.16 V) and FF (from 0.73 to 0.81), which was due to the suppression of Defect complex caused by DMAPbI3 residues. The current density integrated from the external quantum efficiency (EQE) spectrum (Figure 4F) is 20.7 mA cm-2, which closely matches the Jsc value obtained from the J-V curve (deviation does not exceed 2%). The champion device provided a steady-state PCE of 20.08% (Figure 4G) and small hysteresis. With the current hydrogen bond-promoted DMA extraction method, high-performance CsPbI3PSCs (PCE exceeding 20%) can be prepared under surprisingly broad processing windows, such as a broad annealing temperature range from 180°C to 210°C and a wide humidity range from 20% to 60% relative humidity (Figures 4H and 4I).
Figure 4 Photovoltaic performance of CsPbI3 PSCs without doped HTLs
To explore the universality and compatibility of current hydrogen bond-promoted DMA extraction methods, two other commonly used HTLs and various device structures were studied, namely spiro-OMeTAD for n-i-p structures and poly[3-(4 -Carboxybutyl)thiophene-2,5-diyl] (P3CT) was used for the p-i-n structure. For n-i-p structured PSCs, using spiro-OMeTAD HTL, the PAA-treated device achieved a high PCE of 20.60%, Voc of 1.18 V, Jsc of 20.95 mA cm-2, and FF of 0.834 (Figure 5A). Similar to P3HT-based PSCs, PAA-treated PSCs exhibited a significant increase in average PCE (Figure 5B) from 17.77% to 19.85%, which was attributed to the increase in Voc from 1.11 to 1.17 V (Figure 5C). Furthermore, the PAA-treated devices exhibited small hysteresis (Figure S23). For p-i-n structured PSCs using P3CT HTL, PAA-treated PSCs also showed improved device performance, reaching a high PCE of 19.65% (Figure 5D) with small hysteresis (Figure S24). Inverted PSCs showed a significant increase in average PCE (from 16.23% to 18.83%; Figure 5E) and Voc (from 1.08 to 1.13 V; Figure 5F).
Figure 5 Performance and stability of CsPbI3 PSCs with different device structures and HTLs
After being stored in a humidity environment (<15% relative humidity) for 1000 hours, the PCE of unpackaged PSC doped with spiro OMeTAD HTL treated with PAA sharply decreased to 40% of the initial PCE (Figure 5G and Figure S25). However, the control PSC of the untreated spiro OMeTAD HTL doped control decreased to 81.7% of the initial PCE after 1000 hours. In contrast, the undoped P3HT HTL PSC treated with PAA maintained an initial efficiency of about 94% after aging for 10224 hours, demonstrating excellent stability. From the perspective of operational stability, after continuous illumination for 161 hours, the PCE of PSC doped with spiro OMeTAD HTL treated with PAA rapidly decreased to about 20% of the initial PCE (Figure 5H and Figure S26). The control device with undoped P3HT HTL decreased to approximately 48% of the initial PCE after 214 hours of illumination. In contrast, the undoped P3HT HTL PSC treated with PAA maintained an initial efficiency of 93% after 576 hours of continuous illumination. The improvement of stability may be related to the inhibition of iodine vacancies by forming hydrogen bonds (O-H ··· I) between PAA and CsPbI3. These results indicate that the combination of hydrogen bonding promoted DMA extraction method with undoped P3HT HTL significantly improves the stability of PSCs, as it improves the quality of CsPbI3 thin films and eliminates hygroscopic dopants.
4、 Summary
The author has developed a hydrogen bonding promoted DMA extraction method for preparing high-quality γ- CsPbI3 thin film. The hydrogen bond (N-H ··· O) between PAA and DMAPbI3 reduces the escape energy of DMA, accelerates the crystallization kinetics of CsPbI3, and leads to the complete elimination of DMAPbI3 residues and the absence of pinholes γ- Formation of CsPbI3 thin film. This concept was validated by using similar PAN and PVP molecules, which can also form N-H ···· N hydrogen bonds with DMAPbI3. In addition, the hydrogen bond (O-H ··· I) between PAA and CsPbI3 suppresses iodine vacancies, making the crystallization process more environmentally friendly and significantly improving the stability of the device. Under conditions of wide window humidity (20% to 60% RH) and annealing temperature (180 ° C to 210 ° C), PSCs with undoped P3HT HTLs can obtain over 20% PCE (champion PCE is 20.25%). After aging under low humidity conditions (<15% RH) for 10224 hours, PSCs maintained an initial PCE of 94%, while still maintaining over 93% PCE after 570 hours of continuous illumination. This strategy opens up new avenues for the preparation of efficient and stable inorganic PSCs through hydrogen bonding engineering.
5. References
Ming-Hua Li, Shuo Wang et al. Hydrogen-bonding-facilitated dimethylammonium extraction for stable and efficient CsPbI3 solar cells with environmentally benign processing, Joule
Doi: 10.1016/j.joule.2023.09.009 (2023).
E-mail: info@chemborun.com
Tel: +86-574-87178138 
No. 1558, Jiangnan Road,, Ningbo, Zhejiang, China (Mainland)/31
