Photoinduced Dark Fraction Due to Blinking and Photodarkening

Article pubs.acs.org/JPCC

Photoinduced Dark Fraction Due to Blinking and Photodarkening Probability in Aqueous CdTe Quantum Dots A. V. R. Murthy, Padmashri Patil, Shouvik Datta, and Shivprasad Patil* h-cross building, Department of Physics, Indian Institute of Science Education and Research, Pashan, Pune 411008, India ABSTRACT: We provide a novel approach to measure the photoinduced dark fraction due to blinking in semiconductor quantum dots in dilute aqueous solution using fluorescence correlation spectroscopy. We find that below certain threshold excitation intensity, addition of β-mercaptoethanol (BME) does not have a measurable effect on the average number of luminescent quantum dots in a diffraction-limited detection volume. However, above this threshold, it was found to increase and saturate with increasing BME concentration. From these measurements we calculated the photoinduced dark fraction due to blinking and the probability of quantum dot entering into a dark state upon photon absorption (∼10−6).

■

as the electron transfer rate that causes the “on−off” states of luminescence.32 We demonstrate that measurement of this rate can become a method to reveal the electron transfer mechanism in blinking QDs.

INTRODUCTION The emission wavelength of colloidal quantum dots (QDs) depends on its band gap energy and can be tuned by varying its size. Due to excellent wavelength tunability and high photostability compared to conventional organic fluorescent dyes, QDs are being explored as efficient markers for biological imaging applications.1−9 Despite having tremendous applications in bioimaging and devices such as tunable lasers, solar cells, and LEDs,10,11 luminescence intermittency known as “blinking”12−14 restricts advances in QD technology where a single QD is used.15 The random “on” and “off” states in luminescence blinking occur at all time scales.12−14 The “off” state also affects ensemble quantum yield,16 and it is important to understand the photophysics of luminescence blinking from a single QD in aqueous solution.3 Although there is no clear physical picture of blinking mechanism yet,12,17−20 researchers have found that blinking can be suppressed by the addition of thiols to a colloidal solution of QDs.21 It has also been demonstrated that thiolates, formed due to dissociation of thiols in aqueous medium, suppresses blinking.16 Fluorescence correlation spectroscopy (FCS) is emerging as a useful single molecule tool for investigating photophysical properties of QDs.3,22−29 Blinking affects FCS measurements, and many researchers have tried to model this effect using Monte Carlo simulations,3 stretched exponents,24,25 and multiple exponents.29 In general, FCS can measure average number of luminescent particles in its detection volume30−33 and it can be used to measure number of luminescent QDs before and after thiol addition such as β-mercaptoethanol (BME). Since addition of thiols to solution of colloidal QDs is known to suppress blinking,21,16 this provides a new approach toward measuring photoinduced dark fraction in blinking QDs. In this paper we describe the methodology of such a measurement. Further, from intensity dependence of this fraction, we compute photodarkening probability at band-edge photoexcitation (∼ 10−6). We interpret the reaction rate used in conventional FCS analysis © XXXX American Chemical Society

■

MATERIALS AND METHODS

Fluorescence Correlation Spectroscopy. A home-built FCS setup is used in all the measurements reported in this paper. We used a diode laser of 532 nm, TEM00 mode (Dream lasers, China) as an excitation source. Excitation intensity is varied using a combination of neutral density filters. The initial laser beam was expanded to a collimated beam using two lenses having focal lengths of 2.54 and 20 cm (THORLABS, USA) to overfill the back aperture of the objective. The collimated laser beam is deflected by a couple of mirrors (THORLABS, USA) onto a dichroic mirror (Omega Filters, USA), which reflects incident light and transmits the emitted fluorescent light. A laser beam passed through a high NA objective (60X, 1.2NA, Olympus, Japan) is used to excite the sample. The emitted fluorescent light was collected back from the same path, and it passed through the dichroic mirror. This is further filtered by a band-pass filter around emission maximum (550−650 nm for red emission (Omega filters, USA) and 480 to 550 nm for green emission (Thorlabs, USA) and focused onto a pinhole using an achromatic lens (THORLABS, USA). An optical fiber (QMMJ-3S3SUVVIS-25/125-3-1, OZ optics, USA) collects the focused fluorescent light and carries it to the detector (SPCM, PerkinElmer, USA). A correlator card (Correlator.Com, USA) was used to compute autocorrelation of the recorded fluorescent intensities. We used LABVIEW for recording correlated data on a computer. The FCS setup was calibrated by a standard dye Received: March 21, 2013 Revised: June 3, 2013

A

dx.doi.org/10.1021/jp402792f | J. Phys. Chem. C XXXX, XXX, XXX−XXX

The Journal of Physical Chemistry C

Article

Figure 1. A schematic for measurement of dark fraction f due to blinking using FCS. Since some of the QDs are dark at any given time due to blinking, the average number of QDs measured using FCS is less than the actual value. For instance in panel a, there are six QDs in the detection volume, but FCS measures only four. Two QDs are dark due to blinking. The addition of thiol molecules to QD solution suppresses blinking, and one can measure the actual number of QDs. The difference in N divided by total number is dark fraction f.

and QD565. Figure 2 shows the absorption and emission spectra of four different QD sets labeled with the wavelength of the

Rhodamine 6G (Rh6G) at all intensities prior to actual experiments. Along with diffusion measurements, FCS can determine the average number of fluorophores in the detection volume.30−33 Rochira et al. have used FCS to characterize photophysical properties of QDs.29 In the case of blinking QDs, this average number is less than the actual number. The addition of thiols suppresses blinking,16,21,22 and we can recover dark QDs in the detection volume due to blinking. Such a measurement can thus provide dark fraction f due to blinking in aqueous solution of QDs. Figure 1 shows a schematic of the measurement of dark fraction f using FCS. Synthesis and Characterization of CdTe Quantum Dots. All chemicals were used as purchased without further purification. Cadmium chloride, sodium borohydride, mercaptosuccinic acid, and boric acid were purchased from Merck, sodium telluride was purchased from LobaChemie, and trisodium citrate was purchased from Thomas Baker. Deionized water (18 MΩ/cm) was used as solvent. Mercaptosuccinic acid (MSA)-capped CdTe QDs were synthesized by the modifying procedure developed by Ying, et al.34 All reactions proceeded in buffer solutions of boric acid and trisodium citrate. Fifteen millimolars of boric acid and 15 mM of trisodium citrate were used in 50 mL Millipore water to make buffer of pH 8. The precursor solution for CdTe QDs was made by adding cadmium chloride (1 mM), sodium telluride Na2TeO3 (0.25 mM), and MSA (3 mM) in 50 mL of buffer solution at pH 8 at 25 °C. After vigorous stirring of precursors for 5 min, 20 mg of sodium borohydride (NaBH4) was added in to the precursor solution. After the reaction proceeded for 10 min at room temperature, the flask was attached to a condenser and refluxed at 80 °C under open air conditions. Aliquots were collected at different time intervals after starting the reaction. All aliquots are first characterized by an ultraviolet/visible/ infrared (UV/vis/NIR) spectrometer (Lambda 950 PerkinElmer) and then a photoluminescence spectrometer (Fluorolog3, HORIBA). QDs having four different diameters corresponding to absorption peaks at 456, 535, 546, and 565 nm were synthesized. They are referred to as QD456, QD535, QD546,

Figure 2. (a) UV/vis absorption spectra for CdTe samples. The peak absorption wavelengths are 456, 535, 546, and 565 nm corresponding to four different QD diameters. (b) Photoluminescence spectra for QD456, QD535, QD546, and QD565.

absorption peak. QD546 is used to calculate the absorption cross section. A series of absorption spectra were recorded at various dilutions. A molar extinction coefficient of 5.6 × 104 mol−1 cm−1 was measured from the concentration versus peak absorbance data. The absorption cross section was determined following ref 35. The absorption cross section of QD546 is 2 × 10−16 cm2. After complete characterization, QD samples were further diluted for FCS study.

■

RESULTS AND DISCUSSION Dependence of Autocorrelation Curves on Excitation Intensity. We measured autocorrelation curves for QD546 in B



Article

and luminescent states, is higher than this, and we refer to it as Nactual. The measurements in Figure 3a provide us Napp. Figure 3c shows the values of τD obtained by fitting eq 1 for both QD and Rh6G. In contrast to Rh6G, τD for QDs reduces with increasing excitation intensity. We attribute this reduction in τD to an apparently reduced detection volume due to a blinking QD. The cartoon in Figure 4 describes this concept. When a QD

extremely dilute aqueous solution at different excitation intensities. We fit an analytical expression describing diffusion for all intensities. This expression for autocorrelation function is given by30−33 1

G (τ ) =

(

N 1+

τ τD

)(

τ

1 + (r /l)2 τ

D

1/2

)

(1)

The inverse of autocorrelation at zero time lag gives the average number of fluorescent particles in the detection volume, N = 1/G(0). Here τD is the diffusion time or average residence time spent by luminescent QDs in the detection volume. r and l are radial and axial dimensions of the detection volume respectively. Figure 3a shows measured autocorrelations and fits to eq 1 at different excitation intensities. The residual plot for a representative fit at 145 kW/cm2 shows the quality of the fit.

Figure 4. The schematic describing the “apparently reduced detection volume” for a blinking QD. The continuous line represents the trajectory of the QD in its luminescent state, whereas the dotted line represents the the trajectory of the QD in the dark state. The apparent detection volume in FCS measurement is the volume traversed by QDs while they are luminescent. The excluded volume is the volume traversed by QDs while they are dark. The actual detection volume is measured by nonblinking organic fluorphore, Rh6G. Although eq 1 fits to intensity autocorrelations of blinking QDs, due to “apparently reduced detection volume”, τD is less than the actual average residence time.

enters into the detection volume it starts to blink. While calculating autocorrelation, a QD becoming dark is not different from a QD leaving the detection volume. Similarly, a QD becoming luminescent again is identical to a QD entering the detection volume. Therefore, the effective detection volume seen by a blinking QD is less than the actual detection volume. This actual detection volume is usually measured using a standard fluorophore (Rh6G) with a known diffusion coefficient in a given medium. Due to the apparently reduced detection volume, the measured average residence time of blinking QD is less than its actual value. A close inspection of Figure 3b reveals that at low excitation intensity, the τD does not change much with intensity and remains constant around 200 μs. From this value of τD and using the Stokes−Einstein relationship we determined the hydrodynamic radius to be 3 nm. This matches well with the estimated radius of these QDs (2.2 nm) using the effective mass approximation to its excitonic absorption spectrum.36 Note that the effective mass approximation gives the QD core size, while the hydrodynamic radius using FCS also includes the thickness of the capping ligands. The above result suggests that blinking does not affect measurement of the hydrodynamic radius of these CdTe QDs using FCS at lower intensities. Effect of BME Addition on Autocorrelation Curves. The “apparently reduced detection volume” for blinking QDs should not appear in FCS measurements on nonblinking QDs. Moreover, the diffusion time τD should provide the correct hydrodynamic radius. Since thiol addition is known to suppress blinking,16,21,22 we added 1 mL of 100 μM BME to an aqueous solution of QDs. Since blinking affects ensemble quantum yield,

Figure 3. (a) Autocorrelation curves fitted to eq 1 to determine N and τD at different excitation intensities. The continuous lines are best fits, and dotted lines are data. The lower inset shows residuals of a representative fit at 145 kW/cm2 to show the quality of the fit. (b) The dependence of parameter N, the average number of QDs in the detection volume on excitation intensity. We attribute the reduction in N for higher intensities to blinking. (c) The dependence of average residence time τD on excitation intensity. For larger intensities, τD for QDs is smaller than Rh6G. We explain this using the concept of “apparently reduced detection volume” described in Figure 4.

An important parameter for subsequent analysis and conclusion in this work is N, the average number of QDs in the detection volume. Figure 3b shows N at various intensities. The values of N are recorded by fitting eq 1 to measured autocorrelation curves. N initially decreases with intensity and further remains constant. We argue that blinking QDs under FCS measurement will show effectively less average number of QDs in the detection volume than actual number (see schematic in Figure 1). We refer to this number as the apparent average number of QDs Napp. The actual number of QDs, both in dark C



Article

the per particle brightness (PPB) is expected to increase after the addition of BME. We observed that the count rate as well as the PPB increased from 19 × 103 s−1 to 36 × 104 s−1 and from 6 × 103 s‑1 to 18 × 103 s−1, respectively. Figure 5a shows

Table 1. Fit Parameters Are Described before and after BME Addition to QDs Solution fit parameters at 145 kW/cm2 τD (μs)b

τr (μs)c

Fd

2.96 ± 0.01

40.96 ± 0.67

-

-

4.74 ± 0.02

160.96 ± 11

28.3 ± 1.67

0.426 ± 0.01

sample

N

QDs without BME QDs with BME above 70 μM a

a

Average number of particles (N). bDiffusion time (τD). ckr =1/ τr. Equilibrium dark fraction (F).

d

of BME is recovered and now determines the hydrodynamic radius reasonably well. This supports the concept of “apparently reduced detection volume” presented in the previous section. Additionally, the fitting of eq 2 means that luminescence after the addition of BME fluctuates with a single rate kr = 1/τr.32 τr = 28.3 ± 1.67 μs in the present experiment. The fit value of 0.426 ± 0.01 for F also suggests that nearly 40% of the QDs are dark in equilibrium dark−bright conversion, yet all of them contribute to N.30−32 A nonzero value for τr indicates that the addition of BME does not suppress luminescence fluctuations at all time scales. This does not contradict the previous evidence of blinking suppression using BME since those measurements employ millisecond binning times.21 The average number of QDs in the detection volume increases from 2.96 ± 0.01 to 4.74 ± 0.02 after addition of BME (Table 1). Yao et al. have shown that in collection of QDs, there are blinking, nonblinking, and nonradiant QDs.3 Nonradiant QDs remain dark at all times. In our measurement, we are concerned only with blinking QDs and nonblinking QDs. Assuming that all blinking QDs are recovered from their dark state with thiol addition, the actual number of QDs in the detection volume Nactual = 4.74 ± 0.02. So far, there is no evidence that BME addition recovers the luminescence of long-lived nonradiant QDs and therefore Nactual does not include them. ΔN = Nactual − Napp = 4.74 − 2.96 = 1.78. The dark fraction due to blinking is f = ΔN/Nactual= 0.37. It should be noted here that F, the equilibrium dark fraction, and f, the photodarkened fraction due to blinking, are different. At 3.3 kW/cm2, the addition of BME does not have any effect on FCS measurements. Figure 5b shows autocorrelations for both before and after addition of BME at 3.3 kW/cm2. The two curves overlap with each other, and the analytical expression of eq 1 describing diffusion alone fits to measured autocorrelations before and after BME addition. Note that we are measuring τD and N by fitting the equations to data starting from 10 μs. This range is typically avoided in FCS measurements owing to photobleaching in the case of fluorophores32 and rotational dynamics in the case of nanostructures.26 The laser intensity used in our experiments is in the range of a few microwatts to 500 μW. We monitor the fluorescence intensity trace at the time of experiment. In the case of photobleaching, average fluorescence intensity and per particle brightness should decrease within the time (100 s) needed to record autocorrelation curves. We did not observe these effects related to photobleaching. Second, in the case of photobleaching, the equation describing only diffusion does not fit autocorrelation properly. However, it fits very well in our experiments. Our experimental setup is not equipped to capture

Figure 5. (a) Autocorrelations without addition of BME and with addition of BME at 145 kW/cm2. The continuous lines are fits with eqs 1 and 2 for without and with BME addition, respectively. (b) Autocorrelations without and with BME addition for excitation intensity of 3.3 kW/cm2. Both curves overlap each other, and eq 1 fits to both curves yielding τD of 200 ± 5 μs. The hydrodynamic radius corresponding to this τD is 3 nm. This matches reasonably well with radius determined from effective mass approximation.

autocorrelation curves before and after addition of BME at145 kW/cm2. Interestingly, eq 1 (simple 3D diffusion fit) does not fit to the autocorrelation curve after the addition of BME. Instead, an analytical expression of eq 2 describing a single “on−off” rate including diffusion kinetics fits to autocorrelation data. This equation is typically used in FCS analysis where a chemical reaction involving the luminescence “on−off” state along with diffusion contributes to the equilibrium intensity fluctuations.30−33 ⎛ ⎞ F G (τ ) = ⎜ 1 + e−τ / τr⎟ ⎝ ⎠ 1−F 1 × τ τ N 1+ τ 1 + (r /l)2 τ

(

D

)(

D

1/2

)

(2)

Here luminescence fluctuations are caused due to (a) fluorophores moving in and out of the detection volume because of diffusion, and (b) switching between “on” and “off” luminescence states of fluorophores with a single rate to attain equilibrium. This rate is reaction rate kr = 1/τr, where τr is a characteristic time needed to reach an equilibrium number of dark and bright populations. The kr is the sum of forward and backward rates in reaction causing luminescence.32 F is the equilibrium dark fraction in dark−bright conversion.32 Table 1 shows various parameters obtained by fitting eqs 1 and 2 to measured autocorrelations in Figure 5a. As mentioned above, eq 1 fits to autocorrelations without BME addition, whereas eq 2 fits to autocorrelations with BME addition. Both of these measurements are performed at 145 kW/cm2 excitation intensity. From fit parameters, it is clear that the diffusion time after the addition D



Article

4. At excitation intensity I = 145 kW/cm2 in Figure 5a, the average number of photons absorbed by a QD while in transit in the detection volume is 1.6 × 105. Since the probability of photodarkening is around 10−5 to 10−6, QDs are likely to enter into the photoinduced dark state while in transit. This explains the result in Figure 5 that the blinking starts to affect FCS measurements only at higher intensities and correctly determines the hydrodynamic radius of QDs at lower intensities. The Number of BME Molecules Required Per QD to Suppress Blinking. In order to determine the minimum number of BME molecules needed per QD to suppress blinking, we add a controlled amount of BME to the dilute solution of QDs. We use an excitation intensity of 145 kW/cm2 and QD546. BME is added to the 1 mL aqueous solution of QDs in steps, and autocorrelations are measured after each such addition. Figure 7

polarization of emission light. This excludes effects of rotational dynamics on autocorrelation curves. Dependence of Dark Fraction on Excitation Intensity and Photodarkening Probability. We measured dark fraction ( f) due to blinking at different photoexcitation intensities. The data is plotted in Figure 6. The excitation

Figure 6. The photodarkened fraction versus excitation intensity. The slope of straight line fitted to data is Pρt/hν. It gives the probability of the QD becoming dark after absorption of a photon (10−6).

intensity is varied using neutral density filters. A straight line fit to the data in Figure 6 yields a slope of 1.2 × 10−6 cm2/W. The dependence of dark fraction ( f) on excitation intensity suggests that the dark state in blinking is photoinduced, and the linear dependence on intensity suggests that it is a one-photon process. From the data in Figure 6, we can compute the probability of a QD becoming dark after absorption of a photon. We refer to this as photodarkening probability. The photoinduced dark fraction measured in our experiments f = ΔN/Nactual must be the product of photodarkening probability P and the number of photons absorbed n, while the QD is illuminated with excitation light: f = P × n. The number of photons absorbed n can be computed if we know the absorption cross section ρ, the excitation intensity I in kW/cm2, the frequency of excitation ν, and the illumination time t: n = Iρt/hν.18 Tthe he plot in Figure 6 is photoinduced dark fraction f versus excitation intensity. According to the aforementioned discussion, f = P × Iρt/hν. Then theoretically, the slope of the straight line in Figure 6 is Pρt/hν. From Figure 6, the slope is 1.2 × 10−6 cm2/W. Pρt /hν = 1.2 × 10−6 cm 2/W

Figure 7. The effect of the gradual addition of BME on diffusion time τD and the average number of QDs N in the detection volume. (a) τD versus BME concentration. As blinking slowly disappears with addition of BME, τD starts to recover. (b) The average number of QDs in the detection volume versus the BME concentration. All blinking QDs that were not contributing to the total average number of luminescent QDs are recovered with gradual BME addition. At 70 μM BME concentration, N starts to saturate, indicating a total recovery of QDs. At 70 μM, there are 4 × 104 BME molecules in our detection volume of 0.95 fL. This implies that roughly 104 BME molecules per QD are required to suppress blinking.

(3) −6

From the slope, we compute P to be 9 × 10 . We used hν = photon energy for 532 nm wavelength, which is the excitation wavelength of our laser, and measured absorption cross section ρ = 2 × 10−16 cm2 (see section II). We have taken illumination time t to be the average diffusion time τD measured at intensity 3.3 kW/cm2. It is 210 μs. The photodarkening probability of 9 × 10−6 implies that, typically a QD needs to absorb ∼105 to 106 photons before it enters into a photodarkened “off” state. We now explain the result in Figure 5 in the light of the value obtained for photodarkening probability. We note that in Figure 5b, there is no discernible effect of BME addition on autocorrelation curves at I = 3.3 kW/cm2. The autocorrelations overlap with each other, and eq 1 fits to both correlation curves. For excitation intensity of 3.3 kW/cm2, the average number of photons absorbed by a QD in its transit through the detection volume is IρτD/hν ∼ 1.0 × 103. The QDs rarely photodarken since the probability of photodarkening is 9 × 10−6. There are no photodarkened QDs to recover after addition of BME. Since QDs do not enter into a dark state while in transit, there is no “apparently reduced detection volume” effect discussed in Figure

shows diffusion time τD and average number of QDs N versus BME concentration. Figure 7a shows that, after enough BME addition, τD nearly saturates. Figure 7b shows N versus BME concentration. The number of fluorescent QDs at fixed laser intensity gradually increases with BME addition. The growth is initially slow, then changes rapidly, and finally saturates. The growth in N with respect to BME concentration is modeled with a sigmoid growth given by the Hill equation, θ = θmax[xn/(kn + xn)].37 Here θ is the number of QDs recovered from the photodarkened state by BME addition, and x is BME concentration. The parameter k gives the concentration of BME molecules needed to recover half of the QDs from the photodarkened state. The value of k for experiments with different excitation intensities is in the range of 10 to 20 μM. As shown in Figure 7b, we need a BME concentration of around 70 μM. The detection volume in our measurement is 0.95 fL. This translates into 4 × 104 BME molecules in our detection volume. E



Article

As seen in Figure 7b, there are four QDs in the detection volume. This implies that we need 104 BME molecules per QD to suppress blinking under dilute conditions. Photoionization of QDs and Physical Meaning of kr. Do charged QDs have no photoluminescence at all? Recently, the charging model for loss of luminescence is challenged by Zhao et al.20 The present FCS measurement of photoinduced dark fraction and photodarkening probability does not answer this question in a direct way. However, it is remarkable that the measurement of photoionization (charging) rate using electrostatic force microscopy (EFM) suggests a one-photon process of photoionization.8 The linear dependence of photoinduced dark fraction on intensity in our FCS measurements suggests that photodarkening is also a one photon process (Figure 6). Moreover, the ionization probability measured using EFM for band-edge excitation is of the same order as photodarkening probability measured in our experiments (∼ 10−5 to 10−6).8 Therefore, it is tempting to suggest that the photodarkening is caused by photoionization. However, it is difficult to gather direct evidence of charging on an individual QD in the solution. In this section, we attempt to understand the relation between photoionization and photodarkening purely through FCS analysis. In conventional FCS measurement of chemical kinetics, the kr in eq 2 is associated with reaction rate causing “on” and “off” luminescence states; kr = 1/τr. We have seen that after addition of enough BME to suppress blinking, eq 2 fits well to autocorrelations. What does kr represent in present experiments? Two mechanisms are possible. The first possibility is that kr represents the association−dissociation of BME molecules with QDs. In this picture, a BME molecule attaches itself to a QD, and if it is in the dark state it will be turned “on”. Here kr should depend on the concentration of BME. We found that once the BME concentration to recover all photodarkened QDs is reached, then kr does not depend on BME concentration. If kr is the association−dissociation rate between BME and QD, then it is expected to depend on temperature. However, we found no dependence on temperature. We measured kr in the temperature range of 25−65 °C. Figure 8a shows kr versus temperature. The temperature range is somewhat limited because the experiment is performed in aqueous phase with water immersion objectives. The second possibility for the interpretation of kr is as follows. In the charging model of blinking, a charged QD core loses luminescence and recovers it after charge neutralization. The addition of BME is thought to neutralize, or render the electronaccepting traps inactive, by its electron-donating ability. It is wellknown that BME addition prevents the long-lived charged state and enhances neutralization rates. Such suppression of blinking is experimentally measured with millisecond binning times, and the luminescence from QDs appears continuous.21 The ionization− neutralization can happen at a much faster rate than can be captured in millisecond binning times. We interpret kr as the ionization−neutralization (electron transfer) rate. The kr describes the rate of electron transfer between the core and surface trap states. The above-mentioned interpretation of kr suggests that it should depend on the QD core diameter and also on the excitation rate, the number of photons absorbed per second. Figure 8b shows dependence of kr on QD core diameter. We measured autocorrelations for four different QDs; QD456, QD535, QD546, and QD565. The data point indicated with an arrow is for QD456 having an absorption peak at 456 nm. It is excited with a blue laser of 446 nm. The calibration for blue

Figure 8. (a) Dependence of kr on temperature. For the modest range of temperatures allowed in this experiment, kr does not depend on temperature. This excludes the possibility of interpreting kr as the association−dissociation rate of BME molecules and QDs. (b) Dependence of kr on QD diameter estimated using effective mass approximation. The arrow indicates a data point measured using blue laser. The dependence strengthens the interpretation of kr as the electron transfer rate between the core and the surface trap states. The continuous line is a guide to the eye. (c) Dependence of kr on excitation rate. This suggests a one-photon process for electron transfer. The slope of this line is the probability of electron transfer after a photon is absorbed in the presence of BME molecules. It is roughly 10−4.

excitation is performed using the standard dye Coumarin. The other three points represent QDs excited with a green (532 nm) laser. The QD core diameters are estimated using the effective mass approximation as described in the Materials and Methods section. The kr depends on QD core diameter and decreases for larger QD sizes. Figure 8c shows the dependence of kr on excitation rate. The excitation rate is computed as discussed in ref 18. The linear dependence on excitation rate suggests that electron transfer from core to surface states is a one-photon process. A similar conclusion about charging is reached using EFM measurements in ref 18. In general, the FCS measurements support the charging model of luminescence blinking proposed by Nirmal et al. and others.12,13,17 Does Photoionization Occur via Tunneling? The ejection of an electron from the QD core is a complex process, and many mechanisms have been proposed to explain this phenomenon.12−14,20 Tunneling is thought to be a dominant process of ionization.12−14 If kr represents the electron exchange rate between core and surface trap states, then its measurement becomes a useful experimental tool to reveal the process of photoionization. In the case where the photogenerated electron tunnels through to surface states, the well width affects attempt frequency. The tunneling rate then has inverse square dependence on well width.38,39 The effective well width for electrons in the QD core is the diameter of the core itself. The linear dependence of kr on excitation rate, its inverse dependence on QD core diameter, and its insensitivity toward temperature in a modest range in our experiments suggests that tunneling is the F



Article

luminescence. The insensitivity of this rate toward temperature and its strong dependence on QD size supports the tunneling theory of electron exchange between core and surface trap states.

likely mechanism of electron exchange between core and surface trap states. Figure 9 shows a schematic describing the photoionization process. It is based on results using FCS to explore photophysical

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Phone: +91 20-25908034. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The research is funded by a Department of Science and Technology (DST) nanoscience unit grant (SR/NM/NS-42/ 2009) and a Department of Information Technology (DIT) grant (No. 12(4)/ 2007-PDD).

■

REFERENCES

(1) Bruchez, M., Jr.; Moronne, M.; Gin, P.; Weiss, S.; Alivisatos, A. P. Semiconductor Nanocrystals as Fluorescent Biological Labels. Science 1998, 281, 2013−2015. (2) Larson, D. R.; Zipfel, W. R.; Williams, R. M.; Clark, S. W.; Bruchez, M. P.; Wise, F. W.; Webb, W. W. Water-Soluble Quantum Dots for Multiphoton Fluorescence Imaging in Vivo. Science 2003, 300, 1434− 1436. (3) Yao, J.; Larson, D. R.; Vishwasrao, H. D.; Zipfel, W. R.; Webb, W. W. Blinking and Nonradiant Dark Fraction of Water-Soluble Quantum Dots in Aqueous Solution. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 14284−14289. (4) Medintz, I. L.; Uyeda, H. T.; Goldman, E. R.; Mattoussi, H. Quantum Dot Bioconjugates for Imaging, Labelling and Sensing. Nat. Mater. 2005, 4, 435−446. (5) Han, M.; Gao, X.; Su, J. Z.; Nie, S. Bar-Coding Biomolecules with Fluorescent Nanocrystals. Nat. Biotechnol. 2001, 19, 621−622. (6) Anas, A.; Akita, H.; Harashima, H.; Itoh, T.; Ishikawa, M.; Biju, V. Photosensitized Breakage and Damage of DNA by CdSe−ZnS Quantum Dots. J. Phys. Chem. B 2008, 112, 10005−10011. (7) Chan, W. C. W.; Nie, S. M. Quantum Dot Bioconjugates for Ultrasensitive Nonisotopic Detection. Science 1998, 281, 2016−2018. (8) Michalet, M.; Pinaud, F. F.; Bentolila, L. A.; Tsay, J. M.; Doose, S.; Li, J. J.; Sundaresan, G.; Wu, A. M.; Gambhir, S. S.; Weiss, S. Quantum Dots for Live Cells, in Vivo Imaging, and Diagnostics. Science 2005, 307, 538−544. (9) Wu, X. Y.; Liu, H. J.; Liu, J. Q.; Haley, K. N.; Treadway, J. A.; Larson, J. P.; Ge, N. F.; Peale, F.; Bruchez, M. P. Immunofluorescent Labeling of Cancer Marker Her2 and other Cellular Targets with Semiconductor Quantum Dots. Nat. Biotechnol. 2003, 21, 41−46. (10) Klimov, V. I.; Mikhaelovsky, A. A.; Xu, S.; Mlko, A.; Hollingsworth, J. A. Optical Gain and Stimulated Emission in Nanocrystal Quantum Dots. Science 2000, 290, 314−317. (11) Colvin, V. L.; Schlamp, M. C.; Alivisatos, A. P. Light-emitting Diodes Made from Cadmium Selenide Nanocrystals and a Semiconducting Polymer. Nature 1994, 370, 354−357. (12) Nirmal, M.; Dabbousi, B. O.; Bawendi, M. G.; Macklin, J. J.; Trautman, J. K.; Harris, T. D.; Brus, L. E. Fluorescence Intermittency in Single Cadmium Selenide Nano crystals. Nature 1996, 383, 802−804. (13) Kuno, M.; Fromm, D. P.; Hamann, H. F.; Gallagher, A.; Nesbitt, D. J. “On”/“Off” Fluorescence Intermittency of Single Semiconductor Quantum Dots. J. Chem. Phys. 2001, 115, 1028−1040. (14) Shimizu, K. T.; Neuhauser, R. G.; Leatherdale, C. A.; Empedocles, S. A.; Woo, W. K.; Bawendi, M. G. Blinking Statistics in Single Semiconductor Nanocrystal Quantum Dots. Phys. Rev. B 2001, 63 (205316), 1−5. (15) Dahan, M.; Lévi, S.; Luccardini, C.; Rostaing, P.; Riveau, B.; Triller, A. Diffusion Dynamics of Glycine Receptors Revealed by Single−Quantum Dot Tracking. Science 2003, 302, 442−445. (16) Jeong, S.; Achermann, M.; Nanda, J.; Ivanov, S.; Klimov, V. I.; Hollingsworth, J. A. Effect of the Thiol-Thiolate Equilibrium on the

Figure 9. A schematic describing the findings in this paper regarding the photophysics of QDs using FCS. It describes electron transfer process in photoexcited QD in aqueous solution. (a) A photoexcited electron− hole pair and its recombination gives photoluminescence. This radiative recombination rate is represented with thick red arrow. Occasionally, an electron is transferred to surface trap states with a probability of ≈10−6 to 10−5. The neutralization rate kn (thin black arrow) is negligibly small compared to the ionization rate ki (thick black arrow). (b) After addition of BME , both the ionization and neutralization rates (thick black arrows) increase, and the probability of electron transfer becomes ≈10−4. The QD luminescence now fluctuates due to this single and fast ionization−neutralization rate in microsecond time-scale.

properties of QDs reported in this paper. In the absence of BME, the ionization rate (core to surface states) is more compared to the neutralization rate (surface states to core). The probability of electron transfer after absorption of photon in the presence of BME, the slope of the line in Figure 8c, is found to be ∼10−4. This is roughly 10 times more than the photodarkening (∼ photoionization) probability of 10−5 calculated from Figure 6. Therefore we infer that the addition of BME enhances the neutralization rate, and this in turn increases the ionization rate as well. The kr is the addition of both ionization and neutralization rate, kr = ki + kn.32 However, it should be noted that the equilibrium dark fraction, the fraction of QDs remaining dark at any given time, is still around 40% even after BME addition. This indicates that the appearance of continuous luminescence after BME addition on millisecond time-scales is not continuous on microsecond time scales.

■

CONCLUSION To summarize, we found that blinking of colloidal CdTe QDs does not play a significant role in FCS measurements below the threshold intensity for band-edge photoexcitation. This threshold is actually related to the number of photons absorbed by a QD under illumination. Above this threshold excitation intensity, we measured the photoinduced dark fraction of QDs by recovering their luminescence with BME addition. The linear dependence of this fraction on intensity of band-edge photoexcitation suggests a one-photon photodarkening process. This fraction is also used to calculate the probability of photodarkening. The chemical kinetics rate in the reaction-diffusion model used in conventional FCS analysis is interpreted as the electron exchange rate that causes the “on−off” states of G



Article

Photophysical Properties of Aqueous CdSe/ZnS Nanocrystal Quantum Dots. J. Am. Chem. Soc. 2005, 127, 10126−10127. (17) Gomez, D. E.; Califano, M.; Mulvaney, P. Optical Properties of Single Semiconductor Nanocrystals. Phys. Chem. Chem. Phys. 2006, 8, 4989−5011. (18) Krauss, T. D.; O’Brien, S.; Brus, L. E. Charge and Photoionization Properties of Single Semiconductor Nanocrystals. J. Phys. Chem. B. 2001, 105, 1725−1733. (19) Li, S.; Steigerwald, M.; Brus, L. E. Surface States in the Photoionization of High-Quality CdSe Core/Shell Nanocrystals. ACS Nano 2009, 3, 1267−1273. (20) Zhao, J.; Nair, G.; Fisher, B. R.; Bawendi, M. G. Challenge to the Charging Model of Semiconductor-Nanocrystal Fluorescence Intermittency from Off-State Quantum Yields and Multiexciton Blinking. Phys. Rev. Lett. 2010, 104 (157403), 1−4. (21) Hohang, S.; Ha, T. Near-Complete Suppression of Quantum Dot Blinking in Ambient Conditions. J. Am. Chem. Soc. 2004, 126, 1324− 1325. (22) Jay, L. N.; Carlini, L.; Suffern, D.; Ivanova, O.; Bradforth, S. E. Effects of β-Mercaptoethanol on Quantum Dot Emission Evaluated from Photoluminescence Decays. J. Phys. Chem. C 2012, 116 (4), 2728− 2739. (23) Doose, S.; Tsay, J. M.; Pinaud, F.; Weiss, S. Comparison of Photophysical and Colloidal Properties of Biocompatible Semiconductor Nanocrystals Using Fluorescence Correlation Spectroscopy. Anal. Chem. 2005, 77, 2235−2242. (24) Ito, S.; Toitani, N.; Pan, L.; Tamai, N.; Miyasaka, H. Fluorescence Correlation Spectroscopic Study on Water-soluble Cadmium Telluride Nanocrystals: Fast Blinking Dynamics in the μs−ms Region. J. Phys.: Condens. Matter 2007, 19, 486208−10pp. (25) Dong, C.; Ren, J. Measurements for Molar Extinction Coefficients of Aqueous Quantum Dots. Analyst 2010, 135, 1395−1399. (26) Dong, C.; Qian, H.; Fang, N.; Ren, J. Study of Fluorescence Quenching and Dialysis Process of CdTe Quantum Dots, Using Ensemble Techniques and Fluorescence Correlation Spectroscopy. J. Phys. Chem. B 2006, 110, 11069−11075. (27) Tsay, J. M.; Doose, S.; Weiss, S. Rotational and Translational Diffusion of Peptide-Coated CdSe/CdS/ZnS Nanorods Studied by Fluorescence Correlation Spectroscopy. J. Am. Chem. Soc. 2006, 128, 1639−1647. (28) Zhang, P.; Liang, L.; Dong, C.; Qian, H.; Ren, J. Sizes of Watersoluble Luminescent Quantum Dots Measured by Fuorescence Correlation Spectroscopy. Anal. Chim. Acta 2005, 546, 46−51. (29) Rochira, J. A.; Gudhti, M. V.; Gould, T. J.; Laughlin, R. R.; Nadeu, J. L.; Hess, S. T. Fluorescence Intermittency Limits Brightness in CdSe/ ZnS Nanoparticles Quantified by Fluorescence Correlation Spectroscopy. J. Phys. Chem. C 2007, 111, 1695−1708. (30) Magde, D.; Elson, E.; Webb, W. W. Thermodynamic Fluctuations in a Reacting System-Measurement by Fluorescence Correlation Spectroscopy. Phys. Rev. Lett. 1972, 29, 705−708. (31) Magde, D.; Elson, E.; Webb, W. W. Fluorescence Correlation Spectroscopy. I. Conceptual Basis and Theory. Biopolymers 1974, 13, 1− 27. (32) Krichevsky, O.; Bonnet, G. Fluorescence Correlation Spectroscopy: The Technique and its Applications. Rep. Prog. Phys. 2002, 65, 251−297. (33) Sengupta, P.; Balaji, J.; Maiti, S. Measuring Diffusion in Cell Membranes by Fluorescence Correlation Spectroscopy. Methods 2002, 27, 374−387. (34) Ying, E.; Li, D.; Guo, S.; Dong, S.; Wang, J. Synthesis and BioImaging Application of Highly Luminescent Mercaptosuccinic AcidCoated CdTe Nanocrystals. PLoS One 2008, 3 (e2222), 1−7. (35) Leatherdale, C. A.; Woo, W. −K.; Mikulec, F. V.; Bawendi, M. G. On the Absorption Cross Section of CdSe Nanocrystal Quantum Dots. J. Phys. Chem. B 2002, 106, 7619−7622. (36) Brus, L. E. Electron−Electron and Electron−Hole Interactions in Small Semiconductor Crystallites: The Size Dependence of the Lowest Excited Electronic State. J. Chem. Phys. 1984, 80, 4403−4409.

(37) Hills, A. V. The Possible Effects of the Aggregation of the Molecules of Hemoglobin on Its Dissociation Curves. J. Physiol. 1910, 40, iv−vii. (38) Fox, A. M.; David, A. B.; Livescu, J.; Cunningham, J. E.; Jan, W. Y. Quantum Well Carrier Sweep Out: Relation to Electroabsorption and Exciton Saturation. IEEE J. Quantum Electron. 1991, 27, 2281−2295. (39) Larsson, A.; Anderkson, P. A.; ENG, S. T.; Yariv, A. Tunable Superlattice p−i−n Photodetectors: Characteristics, Theory, and Applications. IEEE J. Quantum Electron. 1988, 24, 787−801.

H


Photoinduced Dark Fraction Due to Blinking and Photodarkening

Recommend Documents