Contrasting Electron and Hole Transfer Dynamics from CH(NH2)2PbI3 Perovskite Quantum Dots to Charge Transport Layers

In this work, the ultrafast transient absorption spectroscopy (TAs) was utilized to first investigate the charge transfer from the emerging FAPbI3 (FA = CH(NH2)2) perovskite quantum dots (PQDs) to charge transport layers. Specifically, we compared the TAs in pure FAPbI3 PQDs, PQDs grown with both electron and hole transfer layers (ETL and HTL), and PQDs with only ETL or HTL. The TA signals induced by photoexcited electrons decay much faster in PQDs samples with the ETL (~20 ps) compared to the pure FAPbI3 PQDs (>1 ns). These results reveal that electrons can effectively transport between coupled PQDs and transfer to the ETL (TiO2) at a time scale of 20 ps, much faster than the bimolecular charge recombination inside the PQDs (>1 ns), and the electron transfer efficiency is estimated to be close to 100%. In contrast, the temporal evolution of the TA signals in the PQDs with and without HTL exhibit negligible change, and no substantive hole transfer to the HTL (poly[bis(4-phenyl)(2,4,6-trimethylphenyl)amine], PTAA) occurs within 1 ns. The much slower hole transfer implies the further potential of increasing the overall photo-carrier conversion efficiency through enhancing the hole diffusion length and fine-tuning the coupling between the HTL and PQDs.


Materials and Methods
The FAPbI 3 PQDs were synthesized according to the recent report [25]. For fabrication of the PTAA/FAPbI 3 /TiO 2 sample, a 50 nm compact hydrothermal TiO 2 layer was deposited above the quartz substrate. The TiO 2 film was exposed to UV-ozone for 10 min before FAPbI 3 PQDs deposition. The FAPbI 3 PQDs (45 mg/mL in chloroform) were then spin-cast on top of as-prepared TiO 2 film at 2000 rpm for 40 s. Then the PQDs film was swiftly treated by 200 µL MeOAc for 5 s to remove the surface ligands and spun at 2000 rpm for 20 s. The PQDs film deposition process, performed in the dry air-filled glovebox with relative humidity below 10%, could build up a~160 nm thick film. Before the fabrication of HTL, the PQDs film was annealed at 70 • C for 7 min in a nitrogen-filled glovebox, which can further remove the solvent. By spin-coating the tris(pentafluorophenyl)borane-doped PTAA toluene solution (15 mg/mL, dopant/PTAA = 5% in weight) at 3000 rpm for 40 s, the 40 nm HTL could be deposited above the FAPbI 3 PQDs layer in the N 2 glovebox.
We used a Perkin Elmer model Lambda 950 spectrophotometer to characterize the ultraviolet-visible (UV-vis) absorption spectra of the FAPbI 3 PQDs. Their photoluminescence (PL) spectra were measured with a FluoroMax-4 spectrofluorometer (HORIBA Scientific). The transmission electron microscopy (TEM) images were obtained by using Tecnai G2 F20 S-Twin transmission electron microscope.
The TAs were measured by using a pump−probe experimental system. The pump and probe pulses were delivered by a Ti:sapphire amplifier laser with the pulse width of 100 fs and repetition rate of 1 kHz. We used a mechanical chopper to modulate the pump beam at~300 Hz and focused the laser on the sample with a spot diameter of~3 mm. The pump beam, converted from the fundamental laser beam by using of a β-barium borate (BBO) crystal, has the central wavelength of 400 nm. Supercontinuum white light was generated by a sapphire plate irradiated by 800 nm pulses and it then passed though different color filters with 10 nm bandwidth to serve as the probe pulses with the wavelength range from 500 to 1000 nm. A silicon photodiode was used to collect the probe light transmitted through the sample. A lock-in amplifier was connected to the photodiode to extract the transmission change (∆T) between the states with and without the pump excitation. Figure 1a shows the structure of PTAA/FAPbI 3 /TiO 2 sample, where the thickness of the ETL TiO 2 , the light harvester layer FAPbI 3 PQDs, and the HTL PTAA doped by tris(pentafluorophenyl)borane is~50 nm,~160 nm, and~40 nm, respectively. As shown in Figure 1b, FAPbI 3 PQDs exhibit deep valence band (VB) energy level of −5.46 eV, whereas the PTAA exhibits shallower highest occupied molecular orbital (HOMO) energy level of −5.20 eV [24,25], and the conduction band (CB) energy level values of FAPbI 3 PQDs and TiO 2 are −3.94 eV and −4.15 eV [25,33,34], respectively, which are in favor of the charge transfer. From the TEM image of the pure FAPbI 3 PQDs (Figure 1c), we found the PQDs have an average size of about 15 nm. The UV−vis absorption spectra show a broad absorption with the first absorption peak at~770 nm as a result of optical transition from the ground state to the first excitonic state (Figure 1d). A narrow PL peak centered at~770nm can also be identified in the PL spectra. Based on this structure, the PCE of our FAPbI 3 PQDs solar cells has reached 11.6% [25]. Using PTAA as the hole transport material (HTM) instead of the conventional 2,2,7,7-tetrakis-(N,N-di-p-methoxyphenylamine)-9,9-bifluorene (Spiro-OMeTAD) can reduce manufacturing costs and avoid device instability caused by complex doping and oxidation processes [15,35,36].

Results and Discussion
We then performed the ultrafast TA measurements in the pure FAPbI 3 PQDs and PTAA/FAPbI 3 /TiO 2 . Figure 2a,b show their transmission modulation (∆T/T) spectra under excitation of 400-nm pump laser pulses. The spectra of pure FAPbI 3 display positive photoinduced bleaching (PB) signals in the range of 650-1000 nm. These PB signals are pronouncedly enhanced around 750 nm, which coincides with the wavelength of the first absorption peak, shown in Figure 1d, resulting from the transition from the ground state to the first excitonic state. We can thus ascribe the strong PB signals to the ground state bleaching (GSB). Actually, previous works have demonstrated that in the FAPbI 3 PQDs the strong PB signals was mainly contributed by the holes in the valence band after laser excitation [37][38][39]. We note a slight red shift of the PB peak with increasing time delay. This shift may be attributed to the bandgap renormalization (BGR) as a result of the Coulomb interaction of photoexcited free carriers near the band edge [40,41].
In the range of 500-650 nm, the ∆T/T spectra show negative photoinduced absorption (PA) signals. The time dependences of the PA signals at different wavelengths are nearly identical, as shown in Figure 2c. This indicates that the PA signals are mainly originated from one component of the photoexcited species, since different components typically have distinct dynamics and spectra weight. In principle, both transient absorption of conduction band electron and BGR may contribute to the PA signals [42,43]. However, both contributions in the pure FAPbI 3 PQDs are proportional to the We then performed the ultrafast TA measurements in the pure FAPbI3 PQDs and PTAA/FAPbI3/TiO2. Figure 2a,b show their transmission modulation (ΔT/T) spectra under excitation of 400-nm pump laser pulses. The spectra of pure FAPbI3 display positive photoinduced bleaching (PB) signals in the range of 650-1000 nm. These PB signals are pronouncedly enhanced around 750 nm, which coincides with the wavelength of the first absorption peak, shown in Figure 1d, resulting from the transition from the ground state to the first excitonic state. We can thus ascribe the strong PB signals to the ground state bleaching (GSB). Actually, previous works have demonstrated that in the FAPbI3 PQDs the strong PB signals was mainly contributed by the holes in the valence band after laser excitation [37][38][39]. We note a slight red shift of the PB peak with increasing time delay. This shift may be attributed to the bandgap renormalization (BGR) as a result of the Coulomb interaction of photoexcited free carriers near the band edge [40,41]. Compared to the pure FAPbI 3 PQDs, the ∆T/T spectra in PTAA/FAPbI 3 /TiO 2 also display the apparent PB and PA features. However, the PA wavelength range in the latter is broadened to above 700 nm and the PA peak is shifted to the longer wavelength of~650 nm. Consequently the strong PB signals only appear in a much narrower wavelength range, and the PB peaks show negligible red shift over a long time delay. We may postulate at this stage that a significant number of photoexcited carriers in the PQDs are quickly transferred to the charge transport layer, which leads to a strong internal electric field, resulting in a large Stark effect to modify the bandgap of the PQDs and therefore the structure of transient absorption spectra [40,41]. Another possibility is that the strong electronic coupling between the PbI 2 and the TiO 2 , favoring a fast electron transfer process, leads to a broadened exciton band in the TA spectra [34]. The quick electron transfer may result in an imbalance of the electron and hole numbers in the PQDs. This imbalance may affect the time dependences of the PA signals probed at various wavelengths in a different way. Actually, we can see from Figure 2d some small differences of the normalized time dependences of the PA signals in 500-700 nm. We also note significant PB and PA signals remained at long delay time of 800 ps in Figure 2b, pointing to the existence of long-lived holes residing in the PQDs as a result of the quick electron transfer. Assured evidences of quick electron transfer will be given later. In the range of 500-650 nm, the ΔT/T spectra show negative photoinduced absorption (PA) signals. The time dependences of the PA signals at different wavelengths are nearly identical, as shown in Figure 2c. This indicates that the PA signals are mainly originated from one component of the photoexcited species, since different components typically have distinct dynamics and spectra weight. In principle, both transient absorption of conduction band electron and BGR may contribute to the PA signals [42,43]. However, both contributions in the pure FAPbI3 PQDs are proportional to the number of electrons or holes which exist as pairs, therefore, their dynamics are expected to behave in the same way.
Compared to the pure FAPbI3 PQDs, the ΔT/T spectra in PTAA/FAPbI3/TiO2 also display the apparent PB and PA features. However, the PA wavelength range in the latter is broadened to above 700 nm and the PA peak is shifted to the longer wavelength of ~650 nm. Consequently the strong PB signals only appear in a much narrower wavelength range, and the PB peaks show negligible red shift over a long time delay. We may postulate at this stage that a significant number of photoexcited carriers in the PQDs are quickly transferred to the charge transport layer, which leads to a strong internal electric field, resulting in a large Stark effect to modify the bandgap of the PQDs and therefore the structure of transient absorption spectra [40,41]. Another possibility is that the strong electronic coupling between the PbI2 and the TiO2, favoring a fast electron transfer process, leads to a broadened exciton band in the TA spectra [34]. The quick electron transfer may result in an imbalance of the Because the average size of FAPbI 3 PQDs is about 15 nm, much larger than its exciton Bohr radius (~8 nm), the photoexcited excitons in the FAPbI 3 PQDs quickly dissociate to become free carriers [44,45]. The free carriers then decay via bimolecular recombination [37,46]. However, the Auger recombination may emerge for large carrier density when the PQDs are excited with high pump fluences [41,46]. The Auger recombination is normally much faster than the bimolecular recombination, and thus the decay of ∆T/T signals may display two stages with different time scales [41,46]. With increasing pump fluences, an even faster decaying process at the time scale of several picoseconds owing to the quick biexciton recombination may occur as previously discovered in some PQDs [41,[46][47][48][49][50]. This was also observed in our FAPbI 3 PQDs, as shown in the supplemental Figures S1 and S2 and Table S1 (Supplementary Materials). To avoid the very quick biexciton recombination and reduce the effect of the fast Auger recombination on the charge transfer processes, we have measured the temporal evolution of the ∆T/T signals for both pure FAPbI 3 PQDs and PTAA/FAPbI 3 /TiO 2 at a low pump fluence of 3.5 µJ/cm 2 . The measured time dependences of the PA signals at 570 nm and the PB signals at 770 nm are shown in Figure 3a,b. We then used a formula with the sum of two exponential decaying terms, ∆T/T = A 1 e −t/τ 1 + A 2 e −t/τ 2 , to fit the temporal evolution of the PA and PB decay signals (see solid curves as fitting results in the Figures). Both PA and PB in pure FAPbI 3 PQDs show a fast component with τ 1 ≈ 120 ps and a slow component with τ 2 ≈ 1600 ps (see Table 1), attributed to the Auger and bimolecular recombination, respectively. The slow component has much larger amplitude, indicating that the bimolecular recombination is the dominant decay mechanism. PB decay signals (see solid curves as fitting results in the Figures). Both PA and PB in pure FAPbI3 PQDs show a fast component with ≈ 120 ps and a slow component with ≈ 1600 ps (see Table   1), attributed to the Auger and bimolecular recombination, respectively. The slow component has much larger amplitude, indicating that the bimolecular recombination is the dominant decay mechanism.    We then turn to the time dependence of the PA signal at 570 nm for PTAA/FAPbI 3 /TiO 2 shown in Figure 3a. The PA in PTAA/FAPbI 3 /TiO 2 displays a much faster decaying process compared to that of the pure FAPbI 3 PQDs. The fitting using the above formula reveals τ 1 ≈ 20 ps, and this fast decay component contributes more than 53% of the total PA amplitude. In contrast, the other exponential decay component becomes much slower at the time scale of τ 2 ≈ 2600 ps, compared to that in the pure FAPbI 3 PQDs. The emergence of fast decay component gives important evidence that free charge carriers in the PQDs are quickly transferred to the charge transport layer. To distinguish which charge carriers, electrons or holes, are effectively transferred within 20 ps, we may turn to the time dependences of PB signals, which are mainly contributed from the holes. The time dependences of the PB signals in Figure 3b clearly show a much slower decay process in PTAA/FAPbI 3 /TiO 2 than in the pure FAPbI 3 . The fitting reveals the absence of fast component but only one slow component with τ 2 ≈ 2800 ps. These results strongly support that electrons are quickly transferred to the TiO 2 layer. Such transfer reduces the conduction electron absorption in the PQDs, and consequently attenuates the PA signals at 570 nm within τ 1 ≈ 20 ps. The quick electron transfer leaves the holes in the valence band of the PQDs, thus markedly slowing down the PB decaying process. Actually, we also note a small rising signal in the PB of 770 nm at a time scale of 20 ps, as can be better seen in Figure 4b with expanded time scale. This is likely due to the diminishment of conduction electron absorption at this probe wavelength in PQDs as a result of the fast electron transfer to TiO 2 . Since the PB signals and the slow decay component of the PA signals have an approximately equal decaying time scale (~2600-2800 ps), we believe that in addition to the origin of conduction electron absorption, the PA signals at 570 nm also have a significant contribution from the BGR caused by the holes. In other words, the slow decay component of the PA at 570 nm in PTAA/FAPbI 3 /TiO 2 is mainly caused by the holes residing in the PQDs. This implies that the electrons have transfer efficiency much larger than 53% at the time scale of 20 ps. Moreover, in PTAA/FAPbI 3 /TiO 2 , we did not observe the decay component at~1600 ps caused by the bimolecular recombination inside the PQDs. Hence we believe nearly all electrons have been transferred to TiO 2 , which means the transfer efficiency is close to 100%. On the contrary, in the time window of our experiments (~1 ns), we have not observed substantive hole transfer from the FAPbI 3 PQDs to PTAA.
To confirm the above discussions that electrons have quick and efficient transfer from FAPbI3 PQDs to TiO2, whereas the hole transfer from FAPbI3 PQDs to PTAA is not significant within 1 ns, we also have measured the time dependences of PA and PB signals for FAPbI3 PQDs covered with only TiO2, allowing only electron transfer, or just PTAA, allowing only hole transfer. The ΔT/T spectra of FAPbI3/TiO2 and PTAA/FAPbI3 allow us to separately investigate electron transfer and hole transfer respectively. The measured results are shown in Figure 4a,b. We can see that the PA decaying processes in FAPbI3/TiO2 and PTAA/FAPbI3/TiO2 resemble each other very closely, and the PA decaying processes in PTAA/FAPbI3 and pure FAPbI3 PQDs are quite similar. Consistently, the variations of PB decaying processes in these samples behave in the same manner as those of PA decaying processes. Detailed fitting parameters shown in Table 1 further confirm such behaviors. These comparisons therefore confirm the above discussions of effective electron transfer at the time scale of 20 ps and insignificant hole transfer within 1 ns.
The most reasonable explanations for the different electron and hole transfer processes are described below. The strong electronic interactions between the FAPbI3 and the TiO2 is beneficial to the electron transfer [34], and the coupling between PQDs also favors the electron diffusion. Owing to the suitable energy levels and ideal interface between the FAPbI3 PQDs and TiO2 layers, the electron transfer is sufficiently faster than the bimolecular charge recombination. In contrast, the coupling between the PTAA and FAPbI3 is likely not strong enough to yield quick hole injection, though their HOMO energy levels are matched. One possibility is related to the defective physical contact. Less physical contact, due to the roughness of the interfacial planar heterojunction between the FAPbI3 PQDs and PTAA layers, would increase the need of the hole diffusion before finding a suitable site for transfer [35]. Another possibility is that the coupling between PQDs are not strong enough to support the quick transport of holes to reach the interface. Based on the above CT results and the underlying limiting factors, we therefore propose to make proper interface modification between the FAPbI3 PQDs and PTAA to largely enhance the hole injection rate. In addition, the coupling strength affecting the hole transport between neighboring FAPbI3 PQDs also needs to be further examined and improved. If the hole transfer rate and efficiency are greatly enhanced, a large increase of PCE in FAPbI3 PQDs based solar cells can be expected.  Table 1. To confirm the above discussions that electrons have quick and efficient transfer from FAPbI 3 PQDs to TiO 2 , whereas the hole transfer from FAPbI 3 PQDs to PTAA is not significant within 1 ns, we also have measured the time dependences of PA and PB signals for FAPbI 3 PQDs covered with only TiO 2 , allowing only electron transfer, or just PTAA, allowing only hole transfer. The ∆T/T spectra of FAPbI 3 /TiO 2 and PTAA/FAPbI 3 allow us to separately investigate electron transfer and hole transfer respectively. The measured results are shown in Figure 4a,b. We can see that the PA decaying processes in FAPbI 3 /TiO 2 and PTAA/FAPbI 3 /TiO 2 resemble each other very closely, and the PA decaying processes in PTAA/FAPbI 3 and pure FAPbI 3 PQDs are quite similar. Consistently, the variations of PB decaying processes in these samples behave in the same manner as those of PA decaying processes. Detailed fitting parameters shown in Table 1 further confirm such behaviors. These comparisons therefore confirm the above discussions of effective electron transfer at the time scale of 20 ps and insignificant hole transfer within 1 ns.

Conclusions
The most reasonable explanations for the different electron and hole transfer processes are described below. The strong electronic interactions between the FAPbI 3 and the TiO 2 is beneficial to the electron transfer [34], and the coupling between PQDs also favors the electron diffusion. Owing to the suitable energy levels and ideal interface between the FAPbI 3 PQDs and TiO 2 layers, the electron transfer is sufficiently faster than the bimolecular charge recombination. In contrast, the coupling between the PTAA and FAPbI 3 is likely not strong enough to yield quick hole injection, though their HOMO energy levels are matched. One possibility is related to the defective physical contact. Less physical contact, due to the roughness of the interfacial planar heterojunction between the FAPbI 3 PQDs and PTAA layers, would increase the need of the hole diffusion before finding a suitable site for transfer [35]. Another possibility is that the coupling between PQDs are not strong enough to support the quick transport of holes to reach the interface. Based on the above CT results and the underlying limiting factors, we therefore propose to make proper interface modification between the FAPbI 3 PQDs and PTAA to largely enhance the hole injection rate. In addition, the coupling strength affecting the hole transport between neighboring FAPbI 3 PQDs also needs to be further examined and improved. If the hole transfer rate and efficiency are greatly enhanced, a large increase of PCE in FAPbI 3 PQDs based solar cells can be expected.

Conclusions
In conclusion, based on the comparative studies of the photoexcited charge carrier dynamics in FAPbI 3 , PTAA/FAPbI 3 , FAPbI 3 /TiO 2 , and PTAA/FAPbI 3 /TiO 2 planar junctions by using ultrafast TAs, we unveiled evidences of quick electron transfer process from FAPbI 3 PQDs to TiO 2 ETL at the time scale of 20 ps and the transfer efficiency is estimated to be close to 100%. In contrast, no significant hole transfer to the PTAA HTL were observed within~1 ns, which is likely due to the fact that the interface coupling is not strong enough to drive quick hole transport through PQDs and fast hole injection into PTAA. Our results suggest that improving the ultrafast hole transfer at PQDs/HTL interface is essential for the high-performance of FAPbI 3 PQDs solar cells.
Supplementary Materials: The following are available online at http://www.mdpi.com/2076-3417/10/16/5553/s1, Figure S1: Temporal evolution of ∆T/T for pure FAPbI 3 PQDs film with the probe wavelength of 770 nm under different pump fluences. Solid curves represent the fitting results using the formula of ∆T/T = A 0 e −t/τ 0 + A 1 e −t/τ 1 + A 2 e −t/τ 2 . The fitting parameters are summarized in Table S1; Figure S2: Temporal evolution of ∆T/T for pure FAPbI 3 PQDs film with the probe wavelength of 570 nm under different pump fluences. Solid curves represent the fitting results using the formula of ∆T/T = A 0 e −t/τ 0 + A 1 e −t/τ 1 + A 2 e −t/τ 2 . Relative to the case of pump fluence of ≈14 µJ/cm 2 , the quick biexciton recombination is absent under the low fluences; Table S1: Parameters used to fit the PB decay processes of the pure FAPbI 3 PQDs film probed at 770 nm and excited at 400 nm with different pump fluences, shown in Figure S1.