Signi�cantly suppressing ion migration in metal halide perovskites via trace of multivalent interstitial doping


 Cations with suitable sizes to occupy an interstitial site of perovskite crystals have been widely used to inhibit ion migration and promote performance and stability of perovskite optoelectronics. However, the interstitial doping accompanies inevitable lattice strain to impair long-range ordering and stability of the crystals to cause a sacrificial trade-off. Here, we unravel the evident influence of the valence states of the interstitial cations on their efficacy to suppress the ion migration. Incorporation of a trivalent neodymium cation (Nd3+) effectively mitigates the ion migration in the perovskite lattice with significantly reduced dosage (0.08%) compared to a widely used monovalent cation dopant (Na+, 0.45%). Less but better, the prototypical perovskite solar cells incorporated with Nd3+ exhibits significantly enhanced photovoltaic performance and operational stability.

2 Cations with suitable sizes to occupy an interstitial site of perovskite crystals have been widely used to inhibit ion migration and promote performance and stability of perovskite optoelectronics. However, the interstitial doping accompanies inevitable lattice strain to impair long-range ordering and stability of the crystals to cause a sacrificial trade-off. Here, we unravel the evident influence of the valence states of the interstitial cations on their efficacy to suppress the ion migration. Incorporation of a trivalent neodymium cation (Nd 3+ ) effectively mitigates the ion migration in the perovskite lattice with significantly reduced dosage (0.08%) compared to a widely used monovalent cation dopant (Na + , 0.45%). Less but better, the prototypical perovskite solar cells incorporated with Nd 3+ exhibits significantly enhanced photovoltaic performance and operational stability.
Organometal halide perovskite solar cells have experienced eye-catching improvements in its power conversion efficiency (PCE) 1-3 . To implement its final step to the commercialization, functional longevity becomes the dernier continent to conquer 4-8 . The displacement of the ions and/or charged defects in the perovskite layer has been recognized as the major origin of device performance degradation under its working environment with electric field, heat, and excess charge carriers 9-14 . Therefore, the suppression of the ion migration should be the key to promote the operational stability of the perovskite solar cells.
Interstitial doping of Alkali metal cations such as lithium (Li + ), sodium (Na + ), potassium (K + ) or rubidium (Rb + ) have been widely used to inhibit the migration of halide ions to relieve current-voltage hysteresis and improve operational stability [15][16][17] . However, the interstitial doping possibly distorts perovskite lattice and induces strain that undermines long-range ordering and stability of the desired phase 1, 18,19 . Perhaps for this reason, state-of-the-art record efficiency devices did not incorporate this approach. Hence, there has been a strong need to develop a new approach for preventing these side effects while maximizing the benign ion migration inhibition effect to promote both performance and the operational stability of the perovskite solar cells. Here, we demonstrate an efficient interstitial doping strategy based on trivalent neodymium cation (Nd 3+ ) to prominently prevent the halide migration in the perovskite lattice with a minimal dose addition.
Compared with cations with similar ionic radii but the lower valences, i.e. calcium (Ca 2+ ) and sodium cations (Na + ), the higher valence state of Nd 3+ provides better capability to obstruct the halide migration and is proven to superiorly passivate the negatively charged defects in the perovskite lattice with substantially less dopant concentration.
Compared to iodide anion (I -), Nd 3+ , Ca 2+ , and Na + have cation to anion size ratios of 0.447, 0.455, Submitted Manuscript: Confidential 3 and 0.464, respectively, and thus octahedral (Oh) geometry with 6 coordination is expected as 'A' site cation in ABX3 perovskite 20 . However, their radii are much smaller than those of CH3NH3 + (MA + , 217 pm) or HC(NH2)2 + (FA + , 253pm) 21 , and thus expected to occupy Oh interstitial sites rather than substitution of existing MA + or FA + . We simulated the effect of the different interstitial cations on iodide ion migration energetics to the nearby iodine vacancy (VI) in perovskite crystal cells (Fig. 1A, Supplementary Note 1). Each migration route was divided into eighteen steps and the relative energy differences compared with the initial stage are recorded in Fig. 1B. Without the interstitial cation, the energy barrier for the VI migration was only about 0.37 eV. The presence of Nd 3+ enhances the energy barrier significantly (2.80 eV), and Ca 2+ (0.81 eV) and Na + (0.43 eV) are followed. Considering the similar sizes of the cations, the distinction of the ion migration impedance capability might be majorly related to the valence states of the cations. Nd 3+ with three positive charges are more likely to restrict the motion of negative iodide ions due to stronger electrostatic attraction force. This impeded halide migration by the cation doping is favorable for the stability enhancement and hysteresis elimination of the perovskite solar cell devices.
Regardless of the beneficial ion migration mitigation effect, the interstitial cation doping can introduce tensile strain to the perovskite lattice that can potentially destabilize the lattice 1, 18,19 .
In X-ray diffraction (XRD) measurement in Fig. 1C, the progressive shift of the (100) orientation peak of -FAPbI3 to lower two theta angle with increasing cation dopant concentration supports interstitial incorporation of the cation and thus induced lattice volume expansion due to the tensile strain. According to Willamson-Hall analysis ( Fig. S1-S4, Table S1-S3 and Supplementary Note 2), the lattice strain rapidly rises even with relatively low dopant concentration (<1%, by molar ratio), and then decreases with further addition probably due to relaxation by generation of dislocations, grain boundaries or secondary phases. From XRD patterns in Fig. 1D, as dopant concentration reaches 1%, it was noticed that the intensity of (001) -FAPbI3 peak is reduced while that of (001) PbI2 peak intensifies. Further increase in dopant concentration to 5% induces the appearance of (010) non-perovskite δ-FAPbI3 peak, indicating destabilization of the -FAPbI3   2C and Table S7). External quantum efficiency (EQE) spectra of the devices were compared in Fig. 2D. An integrated JSC of 25.12 mAcm −2 from the Nd 3+ doped device matched well with the value measured from the J-V scan (<5% discrepancy), whereas that of the control device (24.06 mA cm −2 ) showed relatively large discrepancy of 6.7%, probably due to pronounced J-V hysteresis. Fig. 2E compares the surface morphology of the perovskite films incorporated with different cation dopants. While overall grain sizes of the films are comparable, closer inspection of the images revealed that small particulates are segregated on the surface of the films with 0.25% Ca 2+ and 0.45% Na + . We speculate relatively higher strain (>0.19%, Table S2 and S3) induced by 0.25% Ca 2+ and 0.45% Na + might cause clustering and segregation of the dopant on the crystal surface and/or grain boundary, which might limit performance enhancement by sacrificial trade-off 22 . On the other hand, the film with 0.08% Nd 3+ showed neat surface comparable to the reference film, and thus the side effects are probably minimized.
The interstitial cations are likely to interact with negatively charged defects. We investigate the interaction energies of the different cations with the defects including iodine-formamidinium antisites (IFA), lead vacancies (VPb), formamidinium vacancies (VFA) and iodine-lead antisites (IPb) based on the density functional theory (DFT) calculations (models are shown in Fig. S9). As Submitted Manuscript: Confidential 5 summarized in Fig. 3A, the interaction energies between the defects and Nd 3+ was the highest and those for Ca 2+ and Na + are followed. Considering the similar cation radii, the higher valence charge probably contributes to relatively stronger interaction energies of the Nd 3+ with the negatively charged intrinsic defects. The stronger interaction energy of the Nd 3+ might enable more effective defect mitigation with the minimal dosage for the less side effects (e.g. strain, defects, dopant segregation) and thus higher PCE of the device. The time-resolved photoluminescence (TRPL) decay curves of the films with the optimized dopant concentration are compared with that of the reference film in Fig 3B. The curves were fitted to a single exponential decay model to quantify the PL lifetime (Table S7). Overall, we observed an elongation of the PL lifetime for the films with the three cations compared with the reference perovskite film; the PL lifetimes of the Nd 3+ (7.75 μs), Ca 2+ (5.47 μs), and Na + (4.78 μs) films were significantly longer than that of the reference sample (1.68 μs). Regardless of the least addition amount, the most largely elongated PL lifetime for the film with Nd 3+ supports the effectiveness of the Nd 3+ dopant for mitigation of the defects. The trend in PL lifetimes correlated with the enhanced steady-state PL intensity (inset of Fig. 3B). On the other hand, the cation-containing films showed no noticeable difference in their PL peak positions and absorption spectra (Fig. S10). We further investigate charge recombination kinetics in the corresponding devices by the transient photovoltage (TPV) decay measurement shown in Fig. 3C. The fitted time constants for reference, Nd 3+ -, Ca 2+ -and Na + -incorporated devices were 0.50, 0.86, 0.79, and 0.65 ms, respectively, which in line with the trend observed from the measured PL lifetimes.
Positron annihilation spectroscopy (PAS) was employed to compare the density of negatively or neutral charged defects in the films. The positively charged positrons are implanted from the perovskite surface, which is annihilated upon interaction with electrons from a free lattice site or after trapping at negatively charged or neutral (but not positive) vacancies and/or interstitial defects to emit two gamma photons 23 . The shape parameter is extracted from the doppler broadening of the gamma-ray spectrum, and it increases with the increasing density of negatively charged or neutral defects. By changing the kinetic energy of the incident positron, we were able to investigate the depth-dependent defect density of the films incorporated with different cations (Fig. 3D). The results show that the shape parameters of the films with the cations doping in the bulk (mean depth between 10 nm and 500 nm, shaded region in Fig. 3D) is lower than that of the reference film, indicating a decreased density of negatively charged or neutral defects in the bulk Submitted Manuscript: Confidential 6 region by the cation doping. The average shape parameters of different films in the bulk region are compared in Fig. 3E. The lowest shape parameter of the perovskite film with Nd 3+ closely correlates with the TRPL and TPV measurements. We further cross-checked the trap density by measuring the total density of states (tDOS) for the as-fabricated devices incorporated with different cations using angular frequency-dependent capacitance measurement. As shown in Fig.   3F, the density of in-gap states for the devices with Nd 3+ , Ca 2+ and Na + dopants are decreases compared with the reference device where the tendency in the measured trap density coincides with the PL and PAS measurements.
To experimentally verify the effect of the different cation dopants on the ion migration energetics, we performed a temperature-dependent conductivity measurement on lateral perovskite devices with structure of Au/perovskite (100 μm)/Au. The activation energy ( ) for ion migration can be extracted by linear fitting of the data based on the Nernst-Einstein equation given by is the conductivity as a function of temperature T, 0 is a constant, is Boltzmann constant 12 . The Ea was extracted from the slope of the fitted lines at relatively higher temperature (Fig. 4A). The calculated for the film incorporated with Nd 3+ , Ca 2+ , and Na + were 0.812 eV, 0.703 eV, and 0.626 eV, respectively, which are substantially higher than that of the reference film (0.425 eV). The trend of the measured s also agrees well with the simulation results. The in-situ PL measurement was performed using the lateral device under 440 nm illumination and applied an electric field of 150 mV/μm to visualize the ion migration ( Fig.   4B and C, Supplementary Movie 1). As time passes, an obvious PL quenching from the reference film is observed and worsens rapidly. The PL quenching is attributed to the destroyed stoichiometry and/or structure of the crystals due to pronounced migration of charged defects in the reference film 24 . By contrast, the bright PL signal of the film incorporated with Nd 3+ remains unchanged even after applying the bias voltage for the 10 minutes, indicating the migration of the charged defects is significantly suppressed by the addition of 0.08 % Nd 3+ .
To investigate the operational stability of the solar cell devices, we exposed the encapsulated devices to continuous illumination under ambient atmosphere and open-circuit condition (Fig. 4D). Versus maximum power point tracking, the performance degradation is expected to be more accelerated under the open-circuit condition 25 . Notably, the encapsulated device incorporated with Nd 3+ maintained 84.3% of its initial PCE after continuous illumination Incorporation of Ca 2+ and Na + also improved the operational stability to retain 68.0% and 48.6% of their initial PCEs, respectively. To further explore the thermal stability of the devices, we kept the unencapsulated devices in a nitrogen-filled glove box at 85 ℃. After 1008 hours, the reference device completely degraded while the Nd 3+ , Ca 2+ , and Na + incorporated devices retained 86.4%, 72.7% and 60.3% of their initial PCEs, respectively. The observed improvement in both photo and thermal stability of the devices indicates that incorporation of an interstitial cation with the stronger interaction with the charged defects is beneficial for the mitigation of defect while minimizing the side effects. Thus, regardless of the defect (impurity) tolerant nature of the perovskite, this study highlights the importance of minimizing the dosage of the interstitial dopant to maximize both photovoltaic performance and operational stability of perovskite optoelectronics.