Dynamic Evolution from Negative to Positive Photocharging in

Apr 3, 2017 - ... Charles J. Barrows , Arthur Graf , Alexander Schmitz , Christian S. Erickson , Daniel R. Gamelin , and Gerd Bacher. Nano Letters 201...
0 downloads 0 Views 1MB Size
Letter pubs.acs.org/NanoLett

Dynamic Evolution from Negative to Positive Photocharging in Colloidal CdS Quantum Dots Donghai Feng,*,†,‡ Dmitri R. Yakovlev,*,†,§ Victor V. Pavlov,§ Anna V. Rodina,§ Elena V. Shornikova,† Johannes Mund,† and Manfred Bayer†,§ †

Experimentelle Physik 2, Technische Universität Dortmund, 44221 Dortmund, Germany State Key Laboratory of Precision Spectroscopy, East China Normal University, Shanghai 200062, China § Ioffe Institute, Russian Academy of Sciences, 194021 Saint Petersburg, Russia ‡

S Supporting Information *

ABSTRACT: The optical properties of colloidal semiconductor nanocrystals are largely influenced by the trapping of charge carriers on the nanocrystal surface. Different concentrations of electron and hole traps and different rates of their capture to the traps provide dynamical charging of otherwise neutral nanocrystals. We study the photocharging formation and evolution dynamics in CdS colloidal quantum dots with native oleic acid surface ligands. A time-resolved technique with three laser pulses (pump, orientation, and probe) is developed to monitor the photocharging dynamics with picosecond resolution on wide time scales ranging from picoseconds to milliseconds. The detection is based on measuring the coherent spin dynamics of electrons, allowing us to distinguish the type of carrier in the QD core (electron or hole). We find that although initially negative photocharging happens because of fast hole trapping, it eventually evolves to positive photocharging due to electron trapping and hole detrapping. The positive photocharging lasts up to hundreds of microseconds at room temperature. These findings give insight into the photocharging process and provide valuable information for understanding the mechanisms responsible for the emission blinking in colloidal nanostructures. KEYWORDS: Photocharging, carrier trapping, electron spin, ultrafast transient spectroscopy, colloidal quantum dots

P

and the charging dynamics is important for understanding the mechanisms of PL blinking in colloidal QDs, which were commonly suggested to involve Auger recombination of excitons in charged QDs or trap-assisted nonradiative recombination of the charge separated state.12−16 For example, it was suggested that the gray and dark states of blinking originate from different types of charging.17 The formation and release rate as well as lifetime of the charge separation are crucial parameters for different blinking models.12−16,18,19 However, a comprehensive knowledge on the formation and evolution dynamics of the explicit charging states is commonly lacking. Charge assignment has been reported by electrostatic force microscopy20−22 and by analyzing the polarization,23,24 Zeeman splitting structure,25 and recombination dynamics of the charged exciton (trion) emission.17 In this letter, we develop a time-resolved technique to study the photocharging dynamics with a picosecond resolution on wide time scales spanning from ps to ms, which allows us to distinguish whether the dots are negatively or positively photocharged. We unambiguously

hotocharging of colloidal semiconductor quantum dots (QDs) results from the charge separation of the electron− hole pairs generated by light absorption in the dot, in which the core keeps a net negative or positive charge (negative or positive photocharging, respectively), while the opposite charge is trapped at the dot surface or ejected from the dot into the surrounding matrix.1,2 Because of the large surface-to-volume ratio in colloidal QDs, charge separation is a common phenomenon with significant consequences for the optical and electrical properties. In particular, it is detrimental for optical applications, such as QD-based light-emitting devices or fluorescent biological labeling, as the charge separation reduces the photoluminescence (PL) quantum yield. The charge separation in QDs may be of use for spin-based quantum information processing as the carrier spin lifetime is not limited by electron−hole recombination and exchange interactions.3,4 It may also promote photocatalytic activity in which selective charge transfer and suppression of charge recombination are required.5 PL blinking is a universal and intriguing phenomenon in colloidal QDs. The QD emission is randomly switched between bright (emission on) and dark (emission off) states, and in some cases, additional gray states (with emissivity between the bright and dark ones).6−11 Information on the type of charge © 2017 American Chemical Society

Received: December 21, 2016 Revised: March 16, 2017 Published: April 3, 2017 2844

DOI: 10.1021/acs.nanolett.6b05305 Nano Lett. 2017, 17, 2844−2851

Letter

Nano Letters

Figure 1. Formation and decay dynamics of negative photocharging. Laser repetition rate: 1 kHz, pump fluence: Fp = 50 μJ/cm2. (a) Time-resolved absorption (TRA) measured with 455 nm pump and 472 nm probe. Inset shows the steady-state absorption and TRA spectra for 455 nm pump with a fixed pump−probe delay of 1 ns. (b) TRA for 455 nm degenerate pump−probe. Inset shows TRA for two pump fluences, indicating that the fast dynamics with 140 ps are due to biexciton Auger recombination. (c) Electron spin signal as a function of pump-orientation delay in pumporientation-probe configuration with 455 nm pump and 472 nm orientation probe. The orientation-probe delay Δt is fixed at the time point corresponding to the first oscillation peak in the lower inset (Δt = 86.7 ps), which shows the spin dynamics for three different pump-orientation time delays. Upper inset shows their normalized FFT spectra. (d) Spin amplitude as a function of long pump-orientation time delays. Inset: corresponding FFT spectra.

Commercially available CdS colloidal QDs (see the Supporting Information) in a toluene solution (Hangzhou Najing Technology Co., Ltd.) are used for the measurements without further sample processing. The QDs are surface stabilized with oleic acid ligands, and have a QD diameter of about 5.5 nm (estimated from the wavelength of the first excitonic absorption peak).29 Spin measurements are performed using a three beam, pump-orientation-probe (POP) transient spectroscopic technique.30 The pump pulses are linearly polarized and generate electron−hole pairs in the QDs. The orientation and probe pulses are used to identify presence and type of carriers in the QDs provided by the photocharging. Varying their time delay relative to the pump pulse gives access to the photocharging dynamics. The circularly polarized orientation pulses generate spin polarization, while the subsequent dynamics of this spin polarization is monitored by the change of ellipticity of the linearly polarized probe pulses by tuning their time delay from the orientation pulses. The ellipticity signals are recorded using an optical polarization

identify negative and positive photocharging in CdS colloidal QDs and observe an unexpected evolution from initially created negative to long-lasting positive photocharging states. Different from the general expectation,26 this finding reveals that positive photocharging can be formed in colloidal QDs with efficient surface traps for holes. The experiments are based on monitoring the spin transients of negatively and positively charged QDs. Thereby, we can distinguish whether the resident carrier in the core is an electron or a hole by measuring the spin beats in a transverse magnetic field. When the electron spatially overlaps with the hole, the electron spin relaxes fast due to the strong electron− hole exchange interaction.27,28 When the electron is spatially separated from the hole, the exchange interaction vanishes, prolonging the electron spin lifetime both for negatively and positively charged dots, which, in turn, can be distinguished via their different Larmor precession frequencies, i.e., different g factors. 2845

DOI: 10.1021/acs.nanolett.6b05305 Nano Lett. 2017, 17, 2844−2851

Letter

Nano Letters

(corresponding to the exciton buildup time), which can be distinguished from the charge separation dynamics. Figure 1b with 455 nm pump−probe shows the photobleaching dynamics, which is exclusively sensitive to the electron population in the dot core because of the large number of nearly degenerate hole energy levels at the band-edge.34,37 The fast component with 140 ps decay time, observed for Fp = 50 μJ/cm2, is absent at lower pump fluences of ∼Fp/10 (see inset of Figure 1b), which allows us to attribute it to biexciton Auger recombination process. The slow component of 7.1 ns results from the depopulation process of electrons in the dot core. Figure 1c shows POP results. The oscillating ellipticity signal shown in the lower inset is due to Larmor spin precession of photogenerated electrons in the external magnetic field of 0.43 T. The electron g factor can be evaluated from the oscillation frequency: g = hvL/(μBB), where h, vL, and μB are the Planck constant, Larmor precession frequency, and Bohr magneton, respectively. The obtained g-factor value of 1.955 is in agreement with previous measurements of the electron g factor in CdS QDs.38 When the pump and orientation pulses coincide in time (ΔT = 0), the spin precession amplitude is small. Increase of the pump-orientation delay ΔT from 0 to 1 ns results in a drastic enhancement of the spin precession amplitude (lower insert of Figure 1c), but the g-factor values remain constant (upper insert of Figure 1c). A pump-injected exciton, in which the wave functions of electron and hole overlap spatially, can be excluded as origin of the spin enhancement due to the fact that the exciton has maximum population at ΔT = 0, while we detect the weakest spin signal at this moment. In fact, the exciton spin decay is fast at room temperature because of the strong electron−hole exchange interaction27,28 and cannot be resolved in our measurements. Therefore, the increase of spin signal has to be attributed to charge separation processes. We fix the orientation-probe delay to the first ellipticity oscillation peak (Δt = 86.7 ps) and scan the pump-orientation delay. The corresponding dynamics of charge separation in Figure 1c exhibits three characteristic times of ∼7, 55, and 560 ps. The monotonous increase and the constant g factor at changed ΔT suggest that these three times are corresponding to the same fundamental mechanism. In our case, it is the negative photocharging; that is, the electron is kept in the dot core while the hole becomes trapped at the surface. The fast dynamics of 7−60 ps, seen in Figure 1a,c, do not show up in the electron depopulation dynamics in Figure 1b, confirming that only holes can be trapped during this time. Even though panels a and c of Figure 1 both reveal the negative photocharging process, a small disparity between them exists and is discussed in the Supporting Information. About 2 ns after the pump pulse, the electron spin amplitude starts to decrease slowly as shown in Figure 1c, which reflects the decay of the negative photocharging. The electron in the dot core is depopulated due to electron−hole recombination and electron trapping (the existence of electron trapping will be discussed below in the part on positive photocharging). Electron depopulation also reduces the bleaching signal in the TRA measurements in Figure 1b, indicating a decay with time constant ∼7 ns and a longer decay, which continues beyond measurability. Due to the finite length of the mechanical delay line, it is not possible to continuously increase the pump-orientation delay beyond 3 ns. Therefore, we change it in discrete steps to 18 and 53 ns. Figure 1d shows that the electron spin amplitude decreases during 53 ns to ∼40% of its maximum at 1 ns delay. Note that in the whole

bridge combined with lock-in detection, similar to the commonly used Faraday rotation detection but providing stronger spin-dependent signals for laser wavelengths resonant to the charged exciton transitions.31 The laser excitation is based on a regenerative laser amplifier system combined with a broadband fs-OPA (optical parametric amplifier) and a narrowband ps-OPA. The pump pulses with a duration around 200 fs and wavelength fixed at 455 nm (line width of ∼2.5 nm) are taken from the fs-OPA. The degenerate orientation and probe pulses are emitted by the ps-OPA at a wavelength tunable around the QD bandgap (pulse duration of ∼2.5 ps and line width of ∼0.23 nm). Both OPA repetition rates are 30 kHz but can be divided by integers via a pulse picker inside the amplifier without changing the pulse energy. In the POP measurements, the orientation beam is modulated between σ+ and σ− circular polarization by an electro-optical modulator to eliminate background32 and nuclear spin effects. The time delays of pump-orientation ΔT and orientation-probe Δt are adjusted by two independent mechanical delay lines. All timeresolved measurements are performed at room temperature applying a transverse external magnetic field B = 0.43 T supplied by a permanent magnet pair. The measurements are typically done using a pump fluence F p = 50 μJ/cm 2 (corresponding to ∼0.4 eletron−hole pairs per dot on average; see the Supporting Information) and an orientation fluence F0 = 40 μJ/cm2 unless otherwise stated. When blocking the orientation beam, the POP setup can be easily transferred to a conventional two beam pump−probe configuration. More experimental details are provided in the Supporting Information. Negative Photocharging. To isolate the negative photocharging process, we fix the laser repetition rate at 1 kHz. In this case any pile-up effect from previous pulses can be neglected. CdS colloidal QDs have been demonstrated to show fast hole surface trapping processes,30 making them negatively photocharged. Charged QDs give a much stronger signal in the pump−probe ellipticity compared with neutral QDs. Therefore, through monitoring the dynamics of the spin-dependent signal by varying the time delay between the pump and orientation pulses, the hole trapping dynamics, i.e., the formation of negative photocharging, can be measured by the POP technique. In the current study, the pump wavelength is set to the first exciton absorption peak, and the orientation and probe wavelength is set to the first absorption peak of the negative photocharging states. These absorption features of the neutral and photocharged QDs are obtained from pump−probe time-resolved absorption (TRA) measurements after blocking the orientation beam. The inset of Figure 1a shows a TRA spectrum 1 ns after excitation of the first absorption peak, in comparison with a steady-state absorption measurement. Positive values of − ΔA correspond to an absorption decrease, i.e., to a “bleaching” effect. The bleaching peak at 455 nm corresponds to the first exciton transition, in agreement with the steady-state absorption data. The negative value of − ΔA around 472 nm is dominated by the absorption of photocharged QDs with a red shift due to the Stark effect induced by the trapped carriers.33,34 In the main panel of Figure 1a, the kinetics of the negative feature probed at 472 nm reveals the charge separation dynamics. Note that the exciton-to-biexciton transition may also induce absorption, but in the case of ground-state-exciton excitation, it is easily compensated by the inhomogeneous broadening of the state-filling induced bleaching peak,35,36 and its buildup time is instantaneous 2846

DOI: 10.1021/acs.nanolett.6b05305 Nano Lett. 2017, 17, 2844−2851

Letter

Nano Letters

Figure 2. Emergence of positive photocharging at Fp = 50 μJ/cm2. (a) Electron spin dynamics at laser repetition rate 30 kHz for pump-orientation delays ΔT = 53 and −5 ns (equivalent to 33.328 μs). (b) FFT spectra of the data in panel a reveal two precession frequencies: 11.77 GHz due to negative photocharging and 11.17 GHz from positive photocharging. (c) FFT amplitude of g2 component as a function of pump-orientation delay measured by changing the laser repetition rate using a fixed pump-orientation delay of ΔT = −5 ns. Blue line is single exponential fit.

Figure 3. (a) Time-resolved absorption measurements with laser repetition rate of 30 kHz for pump−probe delays of 1 and −5 ns (33.328 μs), respectively; Fp = 50 μJ/cm2. The absence of a bleaching signal at −5 ns indicates that the carrier remaining in the dot core is a hole. For comparison, pump-orientation-probe results for pump-orientation delays ΔT = 1 and −5 ns are shown in the inset, where the g2 amplitude at ΔT = −5 ns is around 1/5 of the g1 amplitude at ΔT = 1 ns. (b) Laser repetition rate dependence of FFT amplitude of the negative and positive charging components; ΔT = 1 ns and Fp = 50 μJ/cm2. Increasing the laser repetition rate decreases the amplitude of negative photocharging and simultaneously increases that of positive photocharging.

indicating that the g2 component has a long lifetime. Roughly, this lifetime should exceed 33 μs (the period between pulses at 30 kHz) but be shorter than 1 ms (the period at 1 kHz), which suggests a remarkable pile-up effects from action of previous pulses for 30 kHz that is negligible for 1 kHz rate. For the pump-orientation delay ΔT = 33.328 μs (equal to ΔT = −5 ns for 30 kHz rate), the g2 component is relatively strong (red curves in Figure 2a,b) and significantly surpasses the negative photocharging g1 component in the FFT spectrum. We note that the weak g1 signal for ΔT = −5 ns in Figure 2b is not induced by the pump pulses whose effect has totally decayed for the g1 component but arises from the orientation pulses. Blocking the pump beam for the orientation-probe configuration, the signal shows a comparable g1 contribution even for the 1 kHz rate (see Figure S8). The g2 component for ΔT = −5 ns is even stronger than the one for 53 ns delay (Figure 2b), implying that it likely originates at delays longer than 53 ns after the pump pulse. Fixing the pump-orientation time delay to ΔT = −5 ns and changing the laser repetition rate gives us information about the lifetime of the g2 component. Decreasing the laser repetition rate increases the effective pump-orientation time delay and thus decreases the amplitude of the g2 component as shown in Figure 2c. The amplitude of the g2 component as a function of

range from 1 to 53 ns, the Larmor precession frequency does not change (inset of Figure 1d), and no additional frequency components appear. As discussed above, upon photoexcitation negative photocharging takes place due to fast hole trapping processes. Next, the negative charge in the dot core decays due to electron−hole recombination as well as electron trapping, so that the dot core becomes neutral again. Nevertheless, the evolution does not stop here, but even more interestingly, the dot core gradually becomes positively charged subsequently. In the following, we will discuss the formation and evolution of the positive photocharging. Positive Photocharging. We use the same laser parameters as in Figure 1 but change the laser repetition rate from 1 to 30 kHz. For a pump-orientation delay ΔT = 53 ns, a remarkable modulation of the precession spin dynamics appears (black curve in Figure 2a), which evidence the contribution of a second Larmor precession frequency to the signal as clearly seen also in the FFT spectrum in Figure 2b. Apart from the frequency of 11.77 GHz (corresponding to g1 = 1.955) related to negative charging, a new frequency of 11.17 GHz (g2 = 1.856) appears. We will show below that the g2 component originates from positively charged dots. A mere change of the laser repetition rate from 1 to 30 kHz leads to its appearance, 2847

DOI: 10.1021/acs.nanolett.6b05305 Nano Lett. 2017, 17, 2844−2851

Letter

Nano Letters pump-orientation delay yields a lifetime of 175 μs. It does not reach zero, even at a 1000 μs delay, indicating that the signal contains an even longer lasting component with small amplitude. Note that the exponential fit only takes into account the effect from the next previous pulse, which gives a lower limit for the decay time, because higher repetition rates could lead to a stronger pile-up effect from previous pulses. At present we cannot say exactly how large the accumulated signal from preceding pulses is. (Supposing every re-excitation by the next pump pulse only gives a simple time-dependent accumulation to the generation from previous pulses,

( ) + exp(− + ··· ≈ 1/(exp( ) − 1)

( t)

exp − τ + exp −

t + (t + 0.005 μs) τ

t + 2(t + 0.005 μs) τ

One possibility to explain the coexistence of signals from negatively and positively charged QDs is to suggest that the QD ensemble comprises two fractions of dots: a small fraction with an electron trap surface for positive charging and a large fraction with a hole trap surface for negative charging. In both cases, the charging is initiated directly after laser irradiation of empty charge-neutral dots. The signal from the minority dots may be too weak to be detected at low laser repetition rates but could be enhanced by pile-up effects for high laser repetition rates. However, in this case an uncorrelated behavior of the g1 and g2 components is expected, which contradicts the experimental data from Figure 3b. With increasing laser repetition rate the positive photocharging signal increases, but the negative one decreases. This means that the positive and negative photocharging originate from the same type of QDs dominated by a hole trap surface. Therefore, the positive photocharging cannot be initiated by direct irradiation of empty charge-neutral dots but must come from the evolution of negative photocharging. More specifically, at 1 ns pumporientation delay in Figure 3b, the negative charging is initiated from empty neutral dots, while the positive charging evolves from negatively photocharged dots due to the previous pulse irradiation. When the number of positively charged dots increases, the number of empty neutral dots continuously drops, leading to a decreasing negative charging. In the following, we suggest a model that can qualitatively describe the main experimental features of the photocharging evolution. While hole trapping results in negative charging, hole detrapping and electron trapping are two necessary processes for the transformation from negative to positive charging. The evolution of photocharging can be simulated by solving a system of rate equations (see the Supporting Information for more general rate equations). As already shown in Figure 2c, the positive photocharging is long-lived; in other words, the trapped electron has a long lifetime (probably due to the fact that the electron is trapped into capping organic ligands). Thus, electron detrapping and recombination of the trapped electron with a core hole or a trapped hole can be neglected in the modeling of the formation of positive photocharging. We also assume that the trapping and detrapping time constants do not change during the whole evolution. Next, we have the following rate equations after the above approximation:

)t h e

t τ

curve in Figure 2c cannot be fitted, indicating a strong deviation from the above supposition.) As will be discussed below, the g2 component is not directly generated by the pulsed excitation. Therefore, its accumulation from previous pulses can hardly be quantified and is beyond our present investigation. We note that both frequencies in Figure 2b have constant values at fixed magnetic field and do not depend on delay time and laser repetition rate in the Figure 2a,c measurements and on laser power as well (Figure S9). As already discussed in the negative photocharging part, spatially overlapping exciton states (with maximum population at zero pump-orientation delay) will not lead to a measurable spin coherence. The g2 component should also come from charge separated states, similar to the g1 component. There are three cases of charge separation: (I) negative photocharging, (II) both electron and hole are trapped at the surface while the dot core is empty, and (III) positive photocharging. The TRA measurement at ΔT = −5 ns (Figure 3a) shows that there is no electron population in the dot core anymore, from which case I can be excluded. Otherwise, the state-filling-induced bleaching signal due to the electron population should be seen simultaneously. Comparing the amplitude of the g1 component for ΔT = 1 ns and of the g2 component for ΔT = −5 ns, a difference of about 5-fold in spin signals (inset of Figure 3a) should give comparable bleaching signals if the g2 component still comes from negative photocharging. As pointed out above, note that the small negative photocharging signal at ΔT = −5 ns is not due to the pump laser, so that it will show no response in the TRA measurements. The case II can also be excluded as neither the trapped electron nor the trapped hole can be spinpolarized by the orientation pulse (more discussion can be found in the Supporting Information). Our POP technique only monitors the charge states in the QD core, which is different from optically detected magnetic resonance by which trapped carrier can be detected.39,40 Consequently, the g2 component can only be attributed to positive photocharging. A σ+ (σ−) polarized orientation pulse generates a spin-down (spin-up) electron, independent of positive or negative charging of the QD (see the Supporting Information). The difference is that the spin coherent signal comes from a pure electron not bound in a negative trion in negatively charged QDs,41,42 while in positively charged QDs, it arises from an electron within a positive trion. Compared with negative charging, positive charging will modify the single-particle electron wave function owing to the presence of the two additional holes and will lead to a slightly smaller electron g factor.43

dPex P P P Pn = − ex − ex − ex + dt τr τtrap,h τtrap,e τdetrap,h

(1)

P dPn Pn P P = ex − − nn − n dt τtrap,h τdetrap,h τr τtrap,e

(2)

dPp dt

=

dPtrap dt

Pex τtrap,e =

+

Pn τtrap,e

Ptrap τdetrap,h −



Ptrap τdetrap,h

Pp τtrap,h +

(3)

Pp τtrap,h

dPem P P = ex + nn dt τr τr

(4)

(5)

and the unity condition Pex + Pn + Pp + Ptrap + Pem = 1 2848

(6) DOI: 10.1021/acs.nanolett.6b05305 Nano Lett. 2017, 17, 2844−2851

Letter

Nano Letters

Figure 4. (a) Simulation of dynamical evolution from negative to positive photocharging. Parameters include 1/α = 30 ns, τtrap,e = 60 ns, τtrap,h = 0.2 ns, and τdetrap,h = 1 ns. (b) Scheme of the photocharging evolution. Dashed lines show surface trap states.

where Pex is the probability to find states of both electron and hole in the dot core, Pn is the probability for one electron in the core and one hole trapped, Pp is the probability for one hole in the core and one electron trapped, Ptrap is the probability for both electron and hole being trapped, and Pem is the probability for empty dot. At t = 0, Pex(0) = P0, Pem(0) = 1 − P0, and Pn(0) = Pp(0) = Ptrap(0) = 0. τr and τnr are the recombination times of a core electron−hole pair and a core electron with a trapped hole, respectively. τtrap,e is the electron trapping time. τtrap,h and τdetrap,h are the hole trapping and detrapping times, respectively. τtrap,h is much shorter than the electron−hole recombination time including both τr and τnr , as confirmed by the experimental data in Figure 1. τdetrap,h should be comparable to τtrap,h; otherwise, the positive photocharging will be too small. The solutions for Pex(t), Pn(t), and Pp(t), as they are concerned in the following discussion, are Pex(t ) =

Pn(t ) =

Pp(t ) =

S10). Knowles et al.44 studied time-resolved photoluminescence together with transient absorption in the visible and nearinfrared spectral regions for CdSe colloidal QDs with native ligands. They found that each component in the transient spectra is given by a competition between radiative and nonradiative (trapping) pathways, and the electron trapping time is comparable to the radiative time in a range from ∼1 ns to tens of nanoseconds. These findings are in line with our parameter setting of 1/α and τtrap,e. 1/kh = τtrap,hτdetrap,h/(τtrap,h + τdetrap,h) = 167 ps is roughly the average of the times for the second and third processes shown in Figure 1c. Our modeling parameters suggest that the hole is in a shallow trap with a low activation energy, on the order of kBT; thus, detrapping from hole traps is easily thermally activated,45−47 setting up an equilibrium between surface and core states, while the electron is probably in a deep trap with a high activation energy. Recently, temperature-dependent PL measurements revealed a small activation energy of ∼35 meV in CdS nanoparticles, although the trapped charge type was not unambiguously identified.48 In our work, the surface ligands of carboxylic acids (oleic acid) mainly passivate the Cd2+ dangling bonds to prevent electron trapping and leave unpassivated anion dangling bonds as hole traps,49,50 supporting fast trapping rates for the hole and slow ones for the electron. Fast hole trapping was typically found in many other CdS and CdSe nanoscale systems.5,44,51,52 It was also revealed in CdSe colloidal QDs with native ligands that the hole became trapped much faster than the electron.44 Figure 4a shows the simulated photocharging evolution of (Pn(t) − Pp(t))/P0. After pump excitation, the QD becomes negatively charged due to hole trapping. The negative charging signal reaches maximum around 1 ns after excitation, then decays to 0 by up to 70 ns. This is in line with the experimental results shown in Figure 1c,d. At longer times, the QD becomes positively charged. The maximum ratio of positive charging to negative charging is around 1/10. In comparison, it is ∼1/5 in the estimate from experimental data in the inset of Figure 3a. The simulation takes into account only a single pulse contribution, but in experiments, cumulative effect from the pulse train can be expected for long positive charging lifetimes. We also note that, the above simulation only utilizes fixed time constants. The hole trapping rate could be different when the electron is in the dot core (i.e., when the QD is negatively photocharged) or at the surface (i.e., when the QD is positively photocharged). For instance, the hole-trapping rate 1/τtrap,h is

⎡ τ P0 ⎢exp( −k ht ) + trap,h τdetrap,h 1 + τtrap,h /τdetrap,h ⎢⎣ ⎤ exp( −αt )⎥ ⎥⎦

(7)

P0 [exp(−αt ) − exp(−k ht )] 1 + τtrap,h /τdetrap,h

(8)

P0

1 [1 − exp(−αt )] 1 + τdetrap,h /τtrap,h ατtrap,e

(9)

where kh = 1/τtrap,h + 1/τdetrap,h, α = 1/2 [kex + kn − (kex − k n)2 + 4kdetrap,h × k trap,h ], kex = 1/τr + 1/τtrap,h + 1/ τtrap,e, kn = 1/τnr + 1/τdetrap,h + 1/τtrap,e, and kdetrap,h = 1/τdetrap,h, ktrap,h = 1/τtrap,h. The hole trapping and detrapping, electron trapping, and electron−hole recombination could be characterized by several different time constants due to the inhomogeneity of the dot ensemble. We only take into account medium to large time constants because these times control the turning point of the photocharging evolution in our time-resolved experiments on an ensemble of QDs. Thus, we use for simulation 1/α = 30 ns, τtrap,e = 60 ns, τtrap,h = 0.2 ns, and τdetrap,h = 1 ns. α can be obtained from the neutral exciton recombination dynamics, as expressed in eq 7. 1/α = 30 ns, corresponding to the third time constant obtained from transient PL measurements (Figure 2849

DOI: 10.1021/acs.nanolett.6b05305 Nano Lett. 2017, 17, 2844−2851

Letter

Nano Letters ORCID

fast when both electron and hole stay in the dot core because of the assistance by transferring energy to electron excitation,13 but in positively charged dots, 1/τptrap,h slows when the electron becomes trapped. The lengthening of the hole-trapping time τptrap,h helps to generate stronger positive photocharging (for the limit case 1/τptrap,h = 0, the maximum ratio of positive charging to negative charging becomes ∼3/5; see the Supporting Information). From the above simulation, a scheme of photocharging processes after single pulse photoexcitation is suggested in Figure 4b. Upon laser irradiation electron−hole pairs are generated in the dot core. The hole can become easily trapped by the dot surface, making the dot negatively charged. Meanwhile, due to thermally activated hole detrapping, an equilibrium is built up between core and surface hole states. The core electron depopulates owing to electron−hole recombination and electron trapping at the dot surface. Finally, the QD becomes positively charged when electrons are trapped at the surface, while a part of the holes remain resident in the dot core and another part of holes is trapped at the surface. In conclusion, we have performed an ultrafast dynamics monitoring of the photocharging processes in CdS colloidal QDs covering a large time range from picoseconds to milliseconds by detecting the coherent spin signal of electrons. We demonstrate an evolution process from negative to positive photocharging. Initially, QDs become negatively photocharged due to fast hole surface trapping processes within ∼7−500 ps. During ∼100 ns, hole detrapping and electron trapping converts the QDs to positive photocharging with a long lifetime of hundreds of microseconds. Our findings help us understand the photophysical processes in colloidal QDs. The developed technique is powerful for investigating photocharging processes and suitable for application to a broad variety of colloidal nanosystems: dots, rods, platelets, and core−shell structures. Preliminary experiments on CdSe QDs with native ligands evidence a similar photocharging evolution from negative to positive. While in CdSe/ZnS core/shell QDs, the spin signal is strongly reduced due to much-less-efficient charge separation.



Donghai Feng: 0000-0003-3272-6482 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was partially supported by Deutsche Forschungsgemeinschaft in the frame of TRR 142 and ICRC TRR 160 (projects A1 and B1), and by National Natural Science Foundation of China (grant no. 11374099). V.V.P thanks RFBR 16-02-00377, and A.V.R thanks RFBR 15-52-12015. The authors are thankful to Alexander L. Efros for valuable discussions and Evgeny A. Zhukov for help in using the Cary spectrophotometer for steady-state absorption measurements.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.nanolett.6b05305. Additional information on the sample; optical and spectroscopic measurements; the calculation of excited electron−hole pair number; discussion on the slight disparity between TRA and POP measurements; background subtraction in FFT spectra; more discussion on the origin of g2 component; spin excitation scheme; and general rate equations and approximate solutions in different cases. Figures showing steady-state absorption, PL, and laser spectra; energy-level diagrams and overall scheme of spin excitation; simulation of photocharging evolution; a comparison of two beam orientation−probe and POP measurements; power dependence; and timeresolved photoluminescence. (PDF)



REFERENCES

(1) Padilha, L. A.; Robel, I.; Lee, D. C.; Nagpal, P.; Pietryga, J. M.; Klimov, V. I. ACS Nano 2011, 5, 5045−5055. (2) McGuire, J. A.; Sykora, M.; Robel, I.; Padilha, L. A.; Joo, J.; Pietryga, J. M.; Klimov, V. I. ACS Nano 2010, 4, 6087−6097. (3) He, J.; Lo, S. S.; Kim, J.; Scholes, G. D. Nano Lett. 2008, 8, 4007− 4013. (4) He, J.; Zhong, H.; Scholes, G. D. Phys. Rev. Lett. 2010, 105, 046601. (5) Wu, K.; Zhu, H.; Liu, Z.; Rodríguez-Córdoba, W.; Lian, T. J. Am. Chem. Soc. 2012, 134, 10337−10340. (6) Nirmal, M.; Dabbousi, B. O.; Bawendi, M. G.; Macklin, J. J.; Trautman, J. K.; Harris, T. D.; Brus, L. E. Nature 1996, 383, 802−804. (7) Efros, Al. L.; Nesbitt, D. J. Nat. Nanotechnol. 2016, 11, 661−671. (8) Frantsuzov, P.; Kuno, M.; Jankó, B.; Marcus, R. A. Nat. Phys. 2008, 4, 519−522. (9) Spinicelli, P.; Buil, S.; Quélin, X.; Mahler, B.; Dubertret, B.; Hermier, J.-P. Phys. Rev. Lett. 2009, 102, 136801. (10) Gómez, D. E.; van Embden, J.; Mulvaney, P.; Fernée, M. J.; Rubinsztein-Dunlop, H. ACS Nano 2009, 3, 2281−2287. (11) Kuno, M.; Fromm, D. P.; Johnson, S. T.; Gallagher, A.; Nesbitt, D. J. Phys. Rev. B: Condens. Matter Mater. Phys. 2003, 67, 125304. (12) Efros, Al. L.; Rosen, M. Phys. Rev. Lett. 1997, 78, 1110−1113. (13) Frantsuzov, P. A.; Marcus, R. A. Phys. Rev. B: Condens. Matter Mater. Phys. 2005, 72, 155321. (14) Zhao, J.; Nair, G.; Fisher, B. R.; Bawendi, M. G. Phys. Rev. Lett. 2010, 104, 157403. (15) Rosen, S.; Schwartz, O.; Oron, D. Phys. Rev. Lett. 2010, 104, 157404. (16) Galland, C.; Ghosh, Y.; Steinbrü c k, A.; Sykora, M.; Hollingsworth, J. A.; Klimov, V. I.; Htoon, H. Nature 2011, 479, 203−207. (17) Park, Y.-S.; Bae, W. K.; Pietryga, J. M.; Klimov, V. I. ACS Nano 2014, 8, 7288−7296. (18) Cordones, A. A.; Leone, S. R. Chem. Soc. Rev. 2013, 42, 3209− 3221. (19) Rabouw, F. T.; Kamp, M.; Van Dijk-Moes, R. J. A.; Gamelin, D. R.; Koenderink, A. F.; Meijerink, A.; Vanmaekelbergh, D. Nano Lett. 2015, 15, 7718−7725. (20) Krauss, T. D.; Brus, L. E. Phys. Rev. Lett. 1999, 83, 4840−4843. (21) Krauss, T. D.; O’Brien, S.; Brus, L. E. J. Phys. Chem. B 2001, 105, 1725−1733. (22) Li, S.; Steigerwald, M. L.; Brus, L. E. ACS Nano 2009, 3, 1267− 1273. (23) Javaux, C.; Mahler, B.; Dubertret, B.; Shabaev, A.; Rodina, A. V.; Efros, Al. L.; Yakovlev, D. R.; Liu, F.; Bayer, M.; Camps, G.; Biadala, L.; Buil, S.; Quelin, X.; Hermier, J.-P. Nat. Nanotechnol. 2013, 8, 206− 212. (24) Liu, F.; Biadala, L.; Rodina, A. V.; Yakovlev, D. R.; Dunker, D.; Javaux, C.; Hermier, J.-P.; Efros, Al. L.; Dubertret, B.; Bayer, M. Phys. Rev. B: Condens. Matter Mater. Phys. 2013, 88, 035302.

AUTHOR INFORMATION

Corresponding Authors

*E-mail: [email protected]. *E-mail: [email protected]. 2850

DOI: 10.1021/acs.nanolett.6b05305 Nano Lett. 2017, 17, 2844−2851

Letter

Nano Letters (25) Fernée, M. J.; Sinito, C.; Louyer, Y.; Potzner, C.; Nguyen, T.-L.; Mulvaney, P.; Tamarat, P.; Lounis, B. Nat. Commun. 2012, 3, 1287. (26) Qin, W.; Guyot-Sionnest, P. ACS Nano 2012, 6, 9125−9132. (27) Huxter, V. M.; Kovalevskij, V.; Scholes, G. D. J. Phys. Chem. B 2005, 109, 20060−20063. (28) Scholes, G. D.; Kim, J.; Wong, C. Y. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 73, 195325. (29) Yu, W. W.; Qu, L.; Guo, W.; Peng, X. Chem. Mater. 2003, 15, 2854−2860. (30) Li, X.; Feng, D. H.; Tong, H. F.; Jia, T. Q.; Deng, L.; Sun, Z. R.; Xu, Z. Z. J. Phys. Chem. Lett. 2014, 5, 4310−4316. (31) Glazov, M. M.; Yugova, I. A.; Spatzek, S.; Schwan, A.; Varwig, S.; Yakovlev, D. R.; Reuter, D.; Wieck, A. D.; Bayer, M. Phys. Rev. B: Condens. Matter Mater. Phys. 2010, 82, 155325. (32) Feng, D. H.; Shan, L. F.; Jia, T. Q.; Pan, X. Q.; Tong, H. F.; Deng, L.; Sun, Z. R.; Xu, Z. Z. Appl. Phys. Lett. 2013, 102, 062408. (33) Norris, D. J.; Sacra, A.; Murray, C. B.; Bawendi, M. G. Phys. Rev. Lett. 1994, 72, 2612−2615. (34) Klimov, V. I.; McBranch, D. W.; Leatherdale, C. A.; Bawendi, M. G. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 60, 13740−13749. (35) Klimov, V.; Hunsche, S.; Kurz, H. Phys. Rev. B: Condens. Matter Mater. Phys. 1994, 50, 8110−8113. (36) Sewall, S. L.; Cooney, R. R.; Anderson, K. E. H.; Dias, E. A.; Kambhampati, P. Phys. Rev. B: Condens. Matter Mater. Phys. 2006, 74, 235328. (37) Klimov, V. I.; Schwarz, Ch. J.; McBranch, D. W.; Leatherdale, C. A.; Bawendi, M. G. Phys. Rev. B: Condens. Matter Mater. Phys. 1999, 60, R2177−R2180. (38) Feng, D. H.; Li, X.; Jia, T. Q.; Pan, X. Q.; Sun, Z. R.; Xu, Z. Z. Appl. Phys. Lett. 2012, 100, 122406. (39) van Schooten, K. J.; Huang, J.; Baker, W. J.; Talapin, D. V.; Boehme, C.; Lupton, J. M. Nano Lett. 2013, 13, 65−71. (40) Lifshitz, E.; Glozman, A.; Litvin, I. D.; Porteanu, H. J. Phys. Chem. B 2000, 104, 10449−10461. (41) Zhukov, E. A.; Yakovlev, D. R.; Bayer, M.; Glazov, M. M.; Ivchenko, E. L.; Karczewski, G.; Kossut, J.; Wojtowicz, T. Phys. Rev. B: Condens. Matter Mater. Phys. 2007, 76, 205310. (42) Yakovlev, D. R.; Bayer, M. Coherent spin dynamics of carriers. In Spin Physics in Semiconductors; Dyakonov, M. I., Eds.; Springer: Berlin, Germany, 2008; pp 135−177. (43) Oberli, D. Y.; Byszewski, M.; Chalupar, B.; Pelucchi, E.; Rudra, A.; Kapon, E. Phys. Rev. B: Condens. Matter Mater. Phys. 2009, 80, 165312. (44) Knowles, K. E.; McArthur, E. A.; Weiss, E. A. ACS Nano 2011, 5, 2026−2035. (45) Bawendi, M. G.; Carroll, P. J.; Wilson, W. L.; Brus, L. E. J. Chem. Phys. 1992, 96, 946−954. (46) O’Neil, M.; Marohn, J.; McLendon, G. J. Phys. Chem. 1990, 94, 4356−4363. (47) Banin, U.; Bruchez, M.; Alivisatos, A. P.; Ha, T.; Weiss, S.; Chemla, D. S. J. Chem. Phys. 1999, 110, 1195−1201. (48) Mohamed, N. B. H.; Haouari, M.; Zaaboub, Z.; Nafoutti, M.; Hassen, F.; Maaref, H.; Ouada, H. B. J. Nanopart. Res. 2014, 16, 2242. (49) Lifshitz, E. J. Phys. Chem. Lett. 2015, 6, 4336−4347. (50) Peterson, M. D.; Cass, L. C.; Harris, R. D.; Edme, K.; Sung, K.; Weiss, E. A. Annu. Rev. Phys. Chem. 2014, 65, 317−339. (51) Klimov, V.; Haring Bolivar, P.; Kurz, H. Phys. Rev. B: Condens. Matter Mater. Phys. 1996, 53, 1463−1467. (52) Logunov, S.; Green, T.; Marguet, S.; El-Sayed, M. A. J. Phys. Chem. A 1998, 102, 5652−5658.

2851

DOI: 10.1021/acs.nanolett.6b05305 Nano Lett. 2017, 17, 2844−2851