An experimental study on the blinking suppression mechanism of organic-inorganic formamidinium lead halide perovskite quantum dots on N-Type semiconductors

Colloidal perovskite quantum dots (PQDs) exhibit appealing optical properties of high photoluminescence (PL) quantum yield (QY), broadly tunable emission wavelength that includes the visible spectral range, and a narrow emission line width, which enables their wide applications in optoelectronic devices, such as solar cells, light-emitting diodes, lasing, and single photon sources.^1–5 To optimize their utilization in optoelectronic devices, it is necessary to investigate the fundamental photophysical properties, such as the fluorescence intermittency (or blinking) and interfacial charge transfer. It has been known that interfacial engineering plays an important role in improving the efficiency of semiconductor optoelectronic devices. In particular, the outstanding improvements in perovskite-based optoelectronic devices can be primarily ascribed to a careful choice of constructing the interfacial layer materials, such as TiO₂ and ZnO as the electron transport layers and NiO and P3HT as the hole transport layers.^6–8 However, random switching between the high (ON)- and low (OFF)-emission states (blinking behaviors) in single-dot emission is a significant drawback in some PQD-based devices.^5,9,10 In general, the OFF periods in single-dot emission reduce the PL QY of the PQD ensemble from their ideal values.¹¹ In addition, for a high quality single photon generation, a stable photon stream from a single PQD is required without any interruption.¹² 

The blinking behaviors of PQDs have been extensively studied in recent times.^9,13–21 Although the blinking mechanism of PQDs seems to be diverse depending on their compositions, they demonstrate many similarities with that of conventional colloidal semiconductor QDs (e.g., CdSe-based). Therefore, the prior methodologies having been successful in controlling/suppressing conventional QD blinking can also be applied in the case of PQDs.⁹ The modifications of PQD blinking have been reported by several groups so far.^{13,17,22–26} Although surface treatment is a widely applied method for controlling the blinking of PQDs, it eliminates only the flickering and is ineffective in suppressing blinking.^13,17 To date, the effective suppression of PQD blinking has been demonstrated at low temperatures^22,26 or by controlling the synthetic parameters of PQD.²³ 

In this work, we studied the blinking behaviors and photon emission properties of a single formamidinium lead bromide PQD (FAPbBr₃ PQD) placed on the TiO₂ n-type semiconductor to deduce the influence of surface (environment) conditions on the blinking behaviors of the PQDs. We observed the efficient blinking suppression of PQDs on the TiO₂ substrate at room temperature. Particularly, the long OFF time duration is dramatically suppressed. To explain our observation, we proposed two possible mechanisms. First, the electrons on TiO₂ filled the trap states of PQDs during Fermi-level equilibrium. Second, the extra non-radiative (NR) electron transfer (El_T) pathways between PQDs and TiO₂, including charge separation and charge recombination, could eliminate the charged states (OFF state) of the PQDs.²⁷ In addition, such extra NR El_T pathways could influence the photon emission statistics of the single FAPbBr₃ PQDs. We consider that this study provides some significant insights into the charge carrier transport between PQDs and TiO₂, which can be a key ingredient in the practical applications of PQD-based optoelectronic devices.

The atomic layer deposition (ALD) method was used to coat a 10 nm-thick-TiO₂ film on a cover glass substrate. The operating temperature was 100 °C. Titanium tetrachloride (TiCl₄, Aldrich, ≥ 99.995%) and water (H₂O) were used as two precursors. Each ALD cycle followed TiCl₄ dosing (2 s), Ar purging (20 s), H₂O dosing (2 s), and Ar purging (40 s). The TiO₂ film thickness was measured by ellipsometry, and the growth per cycle (GPC) was 0.5 Å/cycle.

FAPbBr₃ PQD was synthesized by adopting the method from Ref. 28. Briefly, 0.1 mmol (0.0445 g) of PbBr₂-DMSO, 0.1 mmol (0.0125 g) of FAPbBr₃, and 20 µl of OLA in 7 ml anhydrous N,N-dimethylformamide (DMF) were combined to form the precursor solution. The precursor solution was, then, added dropwise into a large amount of toluene (175 ml)/OLAc (787 µl) mixture with vigorous stirring. Upon mixing with toluene, the solution immediately turned pale green. Finally, the FAPbBr₃ nanocrystal solution was centrifuged at a relative centrifugal force of 5300 RCF for 10 min to remove some large particles. The average size of the finally obtained PQDs was found to be ∼11 nm from transmission electron microscope images (not shown, please see Ref. 18).

A diluted solution of 3 wt. % PMMA (Mw = 350 000 g/mol) in toluene with sufficient FAPbBr₃ PQD particles was spin-coated onto the bare glass substrate or the TiO₂ substrate at the speed of 2000 rpm for 45 s. The PMMA film thickness of ∼150 nm was measured by using an atomic force microscope (not shown).

Single PQDs were excited using a pulse laser (380 nm wavelength, 76 ps pulse width, and 20 MHz repetition rate) focused by an oil objective lens (Nikon Plan Apo VC, 100×, NA = 1.40) aligned in a customized confocal microscope system. The emitted photons from PQDs were collected by the same objective lens and detected using avalanche photodiodes (APDs, SPCM-AQ4C, PerkinElmer). The data were recorded using the time-tagged time-resolved (TTTR) T3 mode single photon counting of PicoHarp300 (PicoQuant). All the collected data were reconstructed and analyzed using algorithms, coded using the MATLAB software, and fitted by using the ORIGIN 2016 software. The temporal resolution of the photon detection system used in this study was estimated to be approximately 0.65 ns by temporally measuring the short laser pulses (∼3 ps) and de-convoluting the measured curves. More experimental details can be found in Refs. 18 and 21.

Figures 1(a) and 1(b) show the PL intensity time trajectories of single FAPbBr₃ PQDs located on a bare glass substrate and on a TiO₂ substrate, respectively. The average exciton number per pulse was estimated to be ⟨N⟩ ∼ 0.1 from the PL intensity saturation curve as a function of the excitation pump energy.⁹ The bin time was 10 ms in both cases. The gray lines near the bottom indicate the background level, including the dark-count of the APDs. The PL intensity trace of a single PQD on the bare glass substrate (PQD1_glass) shows a strong fluctuation between two states (blinking), i.e., high (ON) and low (OFF) emission levels. The counts of the OFF periods are almost close to the background level. In contrast, the PL intensity trace of the single PQD on the TiO₂ substrate (PQD1_TiO₂) is substantially far above the background level, indicating a significant suppression of blinking, especially the long lasting OFF durations. These features can be clearly observed in the histograms in Fig. 1(c) (PQD1_glass) and Fig. 1(d) (PQD1_TiO₂).

The black and red lines in Fig. 2(a), respectively, represent the PL decay lifetime of the selected ON states of PQD1_glass and PQD1_ TiO₂ [corresponding to the green and red areas in Figs. 1(a) and 1(b), respectively]. The black line can be fitted using a single exponential decay function, yielding the decay lifetime of 20.8 ns. This value can be attributed to the single exciton lifetime (τ_X) of PQD1 on the bare cover glass substrate.^9,13,21 The red line for PQD1_TiO₂, in comparison, follows a bi-exponential decay function, which yields a long and a short lifetime of 15.3 ns and 3.6 ns, respectively. The long lifetime component can be referred to as the single exciton lifetime of PQD1 on the TiO₂ substrate.^13,20 The short lifetime component could have resulted from the negative trion (⁠$τX*−$⁠), which will be discussed later in this work. Figure 2(b) demonstrates the single exciton lifetime distribution of 34 PQDs on the bare glass substrate and 38 PQDs on the TiO₂ substrate. The blue curves are Gaussian fitting curves with the peak values of 20 ns for the bare glass substrate and 12 ns for the TiO₂ substrate. The shortened single exciton lifetime in the case of the TiO₂ substrate is possibly attributed to the presence of NR El_T from PQD to TiO₂, as illustrated in Fig. 2(c) (process P1). The conduction band (CB) and valence band (VB) energy values of FAPbBr₃ PQD are estimated at −3.4 eV and −5.6 eV, respectively.^4,29 The band-edge energy of TiO₂ is −4 eV (CB) and −7.2 eV (VB),^2,29 suggesting that the El_T from PQD to TiO₂ is energetically enabled, as indicated in Fig. 2(c) (P1, red arrow). Following this scenario, the modified single exciton lifetime of PQD on TiO₂ can be expressed by the following function:

where $τXT$ and $τXg$ are the single exciton lifetimes of PQD on the TiO₂ and bare glass substrates, respectively. τ_{El_T} is the NR El_T lifetime of the transfer from PQD to the TiO₂ substrate. From the average value of $τXT$ = 12 ns and $τXg$ = 20 ns obtained from the Gaussian fittings shown in Fig. 2(b), we estimated the average El_T rate k_{El_T} (=$1τEl_T$⁠) to be 3.3 × 10⁷ Hz. This El_T rate is very close to the average El_T rate from CdSe/ZnS QDs to TiO₂ particles (3.2 × 10⁷ Hz).³⁰ In addition, the CB energy values of CdSe/ZnS and TiO₂ reported in Ref. 30 were −3.7 eV and −4.2 eV, respectively, which are similar to our PQD/TiO₂ energy levels. Therefore, these results further support our explanation on the shorter lifetimes of PQD on the TiO₂ substrate.

However, the blinking behaviors of single CdSe/ZnS QD on the TiO₂ nanocrystalline thin film reported in previous studies are different from those of PQDs on TiO₂ observed in this study. In the case of CdSe/ZnS QDs, blinking dominated by the OFF state was observed to be more active on TiO₂ than on the bare glass substrate.³⁰ This clearly contradicted the results with respect to PQDs on TiO₂. Figure 2(d) represents the statistic of ON-time fraction $=ONdurationsONdurations+OFFdurations$ of 34 PQDs on the bare glass substrate (black bar) and 38 PQDs on the TiO₂ substrate (red bar).^25,31 The threshold intensity (red dashed line in Fig. 1) is defined as I_th = I_b + 6σ_b to separate the ON and OFF states. I_b and σ_b denote the average and standard deviation values of the background, respectively. While the ON-time fraction of PQDs on the bare glass substrate spans a wide range (20%–95%), majority of PQDs on TiO₂ (71%) demonstrate an ON-time fraction of over 95%. This clearly demonstrates that the TiO₂ thin film can efficiently suppress the blinking of FAPbBr₃ PQDs. To explain this experimental finding that contradicts with what was reported in Ref. 30, in addition to NR El_T from PQD to TiO₂, additional carrier (electron) dynamics that affects the PL emission of PQDs on TiO₂ is required. Our explanation for this phenomenon is as follows: first, the free electrons in TiO₂ can transfer into the trap states of PQD, as depicted by process P2 (green arrow) shown in Fig. 2(c). This process can partially/totally fill up the trap states depending on the relevant energy level of TiO₂ and trap states. Therefore, hot electrons (excited to an energy value higher than that of CB) or electrons in the CB of PQD demonstrate a lower possibility to be captured by the trap state, thus suppressing blinking. Second, when El_T occurs from PQD to TiO₂ leaving a hole in PQD (resulting in a positively charged QD), the extra electron from TiO₂ or shallow trap states can easily neutralize the PQD by the NR recombining with the hole [process P3 in Fig. 2(c) shown by blue arrows], thereby eliminating the long OFF periods and suppressing blinking. Third, if process P3 occurs before process P1, a negatively charged QD is formed, and this state continues until El_T happens (process P1). The negative trion triggers the Auger recombination pathway, but enables the partial photon emission process, resulting in a short lifetime component (⁠$τX*−=3.6$ ns), as introduced in Fig. 2(a).²⁴ 

The El_T model was proposed to play a crucial role in suppressing the blinking of the CdSe-based QD on an indium-tin-oxide (ITO) n-type semiconductor substrate in the previous study.^32,33 In contrast, several groups pointed out that a different mechanism of energy transfer (ET) from the excited QD to the ITO surface could be a possible reason for the changes in single QD emission behaviors.^34,35 To elucidate that El_T was the dominant mechanism for our samples (FAPbBr₃ PQDs on the TiO₂ substrate), we analyzed the PL intensity time traces corresponding to their lifetime dynamics of another PQD particle on the TiO₂ substrate (PQD2_TiO₂), and the results are shown in Fig. 3. Figure 3(a) shows the PL intensity time trace of PQD2_TiO₂ together with the histogram in Fig. 3(b). Although the ON state dominates in the complete range of time trace, a few OFF or intermediate states are also observed. The lifetimes of the OFF or intermediate states are derived by selecting only the OFF or intermediate regions [pink dashed area denoted as @1 and orange area denoted as @2, respectively, in Fig. 3(a)]. The decay curves of @1 and @2 can be fitted by a single exponential decay function, as illustrated in Figs. 3(c) and 3(d), respectively. The OFF state lifetime of 0.7 ns is assigned to the positive trion lifetime (⁠$τX*+$⁠), while the intermediate state lifetime of 2.9 ns is attributed to the negative trion lifetime (⁠$τX*−$⁠). From these trion lifetimes, we can evaluate the biexciton lifetime $1τXX=21τX*−+1τX*+$ to be 280 ps, which is consistent with the biexciton lifetime of FAPbBr₃ PQD calculated by different methods.^18,36 The different emission levels of positive and negative trion states were also observed in CdSe-based QD and FAPbBr₃ PQD, thereby supporting our suggestion with respect to the partially radiative negative trion states.^24,37 In addition, the ON state lifetime of PQD2_TiO₂ [green area denoted as @3 in Fig. 3(a)] exhibits bi-exponential components of 2.8 ns and 10.7 ns [Fig. 3(e)], which is similar to the case of PQD1_TiO₂. Therefore, the ON state of PQD on the TiO₂ substrate can be regarded as the combination of a single exciton and negative trion states. The photons selected from the uppermost emission level [violet region (@4) in Fig. 3(a)] yield a single exponential decay with the decay lifetime of 10.7 ns [Fig. 3(f)], corresponding to the single exciton lifetime. Furthermore, the infrequently appearing long lasting OFF state (positive and negative trion states) is well understandable, if the role of El_T processes is included, as mentioned earlier [P1 and P3 in Fig. 2(c)]. Contrary to El_T, it is well accepted that ET would annihilate the exciton energy by transferring it to the exterior materials via dipole–dipole energy transfer or dipole–surface energy transfer, thus keeping QDs neutral.^34,35 Therefore, El_T is considered the most dominant process that manages the suppressed blinking of FAPbBr₃ PQD on the TiO₂ substrate.

To further analyze the blinking behaviors of FAPbBr₃ PQD on the bare glass and TiO₂ substrates, we performed the fluorescence lifetime intensity distribution (FLID) measurements on PQD1_glass, PQD1_TiO₂, and PQD2_TiO₂, and the results are shown in Figs. 4(a)–4(c), respectively. Measuring the FLID is an effective approach to identify the different types of blinking mechanisms of QDs, which are commonly divided into type-A and type-B blinking.^11,38–40 In type-A blinking, when the QD is charged due to the charge carrier (e.g., electron) transfer to trap states via the ionization process, the NR Auger recombination of trion (e.g., positive trion) quenches the photon emission, leading to an OFF state. The ON state is recovered, when the QD is re-neutralized.^38–40 Type-A blinking requires long-lived trap states (e.g., deep trap states).⁴¹ In type-B blinking, the time dependent fluctuations in the carrier trapping rate (electron or hole) lead to PL intensity fluctuations.^38–40 The fluctuation of the trapping rate was initially attributed to the slow energetic diffusion⁴² and later to the activation/deactivation of the multicarrier recombination center (MRC).⁴³ When the trapping rate is much higher than the exciton recombination rate, the charge carrier (e.g., electron) is captured into the trap states and non-radiatively recombines with the hole to render the PQD neutral. These models require short-lived traps (e.g., shallow traps) that enable the trapped electron to non-radiatively recombine with the hole before forming the trion.^40,42,43 In addition, type-B blinking is also sub-categorized into type-B-BC or type-B-HC blinking, in which the charge carrier transfers from the band-edge or hot carrier band to trap states.^38,39 Type-A blinking can be identified by the curved FLID trajectory (green line) shown in Fig. 4(a), while type-B-HC blinking leads to quasi-constant lifetimes in the ON and OFF states.^21,38 Although PQD1_TiO₂ and PQD2_TiO₂ remain mainly in the ON state, their lifetime distributions in the ON state are observed to linearly depend on their PL intensity [white lines in Figs. 4(b) and 4(c)]. If we assume that the signal is composed of only the contributions of the negative trion and single exciton transition, then the FLID histogram is expected to follow a curvature line similar to that in Fig. 4(a) (green line).^21,38,39 Therefore, the linear dependence between the intensity and average lifetime suggests that more transition mechanisms are involved in the ON states of PQDs placed onto the TiO₂ substrate. This linear correlation is predicted by Frantsuzov et al.⁴³ and assigned to type-B-BC blinking.³⁹ In this type of blinking, intensity I is proportional to PL QY and related to average lifetime τ and radiative lifetime τ_r. This relationship is defined as follows:³⁹ 

where k_r is the radiative relaxation rate and k_t(t) is the NR trapping rate expressed as follows:⁴² 

where k_i is the trapping rate in the ith trap, k₀ is the background NR relaxation rate, σ_i(t) is assigned to 1(0) for the activated (inactivated) ith trap, and M is the total number of traps (or recombination centers: RC) at time t.

From our previous study,²¹ it was established that PQDs on the bare glass substrate demonstrated only type-A and/or type-B-HC blinking and not type-B-BC. For type-B-HC, the trapping rate k_t,HC of the hot carriers is higher than the hot electron cooling rate k_cooling (k_cooling < k_t,HC).³⁸ However, for type-B-BC, k_t,HC is expected to be lower than k_cooling to enable the carriers to attain the band-edge. The radiative decay, then, competes with NR recombination [process P4 in Fig. 2(c)] to form a linearly elongated trajectory in the FLID.³⁹ If the cooling rates (k_cooling) are not much different among both types of substrates, the hot carrier trapping rate k_t,HC for PQDs on TiO₂ must be smaller than that in the case of the bare glass substrate. This can be explained by the reduction in the number of active RCs M for the TiO₂ substrate as the RCs are occupied by the electrons transferred from TiO₂ [process P2 in Fig. 2(c)]. Therefore, the FLID results also support our suggestion on the carrier transition mechanism shown in Fig. 2(c). In addition, we verified the carrier transfer from TiO₂ to PQDs in a separate experiment (not shown). When the PQDs on the TiO₂ substrate was excited by a pump energy (e.g., 405 nm) lower than the band-edge energy of TiO₂, their blinking behavior showed more increased OFF durations than on the glass substrate due to the reduced electron sources from TiO₂.

Blinking suppression could be beneficial for achieving a successful application of PQDs in optoelectronics devices. In addition, the elimination of capturing hot electrons to trap states may increase the efficiency of carrier multiplication during the hot electron cooling process.⁴⁴ However, to utilize PQDs as a single photon source, they must guarantee a good antibunching nature, i.e., only single photons are emitted at a specific point of time, leading to the second order photon correlation function g⁽²⁾(τ) ≈ 0 at delay τ = 0. Figure 5(a) shows g⁽²⁾(τ) of PQDs on the bare glass substrate with g⁽²⁾(0) = 0.1, and the ratio between the central peak area (τ ∼ 0) and side peak area (τ ∼ 50 ns) R = $centralareasidearea=QXXQX$ = 0.01.⁴⁵ In this case, Q_XX and Q_X are biexciton and single exciton PL QY, respectively. This strong photon antibunching for PQD on the bare glass substrate results from a fast NR Auger decay, which quenches the photon emission from multiexciton states.^9,18 In contrast, the purity of single photon emission degrades in the case of PQDs on TiO₂ [g⁽²⁾(0) = 0.5 and R = 0.45], as shown in Fig. 5(b). This implies that the modified photon emission property results possibly from the introduction of extra NR El_T processes. Photon antibunching degradation was also observed in the case of conventional QDs (e.g., CdSe QD) on ITO or metallic substrates, which was attributed to the extra NR processes.^34,46 Further systematic studies on g⁽²⁾(τ) of PQD on different substrates are currently being conducted.

In summary, we experimentally observed a highly suppressed blinking from FAPbBr₃ PQDs placed on a TiO₂ n-type semiconductor substrate. Particularly, long lasting OFF states are almost eliminated. In addition, the results of FLID measurements showed that type-A blinking and type-B-HC blinking were dominant when the PQDs were on a glass substrate, while type-B-BC blinking was observed in the case of the TiO₂ substrate. To explain our experimental findings, we proposed a simple and intuitive model that explained the charge carrier dynamics of PQD on the TiO₂ substrate. According to our model, the trap states responsible for type-A and type-B-HC blinking could be preferentially occupied by electrons transferred from the TiO₂ substrate. This reduced the number of trap states, thus reducing the probabilities of capturing carriers from PQD to trap states. The NR El_T processes, such as charge recombination and charge separation, between PQD and TiO₂ led to a rapid competition between the charged and neutral states of the PQD. Therefore, the long lasting OFF (charged) state can be efficiently eliminated. The practical applications of PQD-based optoelectronic devices depend on the successful suppression of blinking. Therefore, it is important to conduct a quantitative experimental study on the suppressed blinking of PQDs on n-type semiconductor substrates.

The authors thank Professor M. Sung for his experimental support. This study was supported by the Basic Science Research Program through the National Research Foundation (NRF) of Korea and funded by the Ministry of Science and ICT (Grant Nos. 2019R1F1A1060582, 2018R1A2B6001449, 2018K1A3A1A32053991, and 2017R1A2B2007618).