LETTER pubs.acs.org/NanoLett
Spectral Properties of Multiply Charged Semiconductor Quantum Dots Sibel Ebru Yalcin,†,‡ Joelle A. Labastide,† Danielle L. Sowle,† and Michael D. Barnes*,†,‡ †
Department of Chemistry and ‡Department of Physics, University of Massachusetts, Amherst, Massachusetts 01003, United States
bS Supporting Information ABSTRACT: Spectrally resolved fluorescence imaging of single CdSe/ZnS quantum dots (QDs), charged by electrospray deposition under negative bias has revealed a surprising net blue shift (∼60 meV peak-to-peak) in the distribution of center frequencies in QD band-edge luminescence. Electrostatic force microscopy (EFM) on the electrospray QD samples showed a subpopulation of charged QDs with 4.7 ( 0.7 excess electrons, as well as a significant fraction of uncharged QDs as evidenced by the distinct cantilever response under bias. We show that the blue-shifted peak recombination energy can be understood as a first-order electronic perturbation that affects the band-edge electron- and hole-states differently. These studies provide new insight into the role of electronic perturbations of QD luminescence by excess charges. KEYWORDS: Charged quantum dots, single molecule, electrospray, Stark effect, blue shift, electrostatic force microscopy (EFM)
C
olloidal semiconductor quantum dots (QDs) continue to attract enormous interest for their applications in optoelectronics,2,3 light emitting,49 and light harvesting1012 devices. In addition, there has been much recent interest in temporal and spectral properties of QDs coupled with other semiconductor systems such as conjugated organic ligands,1315 or conducting substrates,16 where charge or energy exchange processes lead to qualitatively different luminescence dynamics. It is well-known that the Stark interaction induced by an asymmetric electric field (either by DC field or single point charge) affects the electron and hole energies in second order by mixing states of different symmetry, thus reducing the recombination energy by a factor proportional to the magnitude of the perturbing field,1720 as well as inducing a linear polarization.19 The interaction of electron/hole wave functions with external fields has also been studied on different sized and shaped nanostructures such as quantum wells,21 quantum wires,22 quantum dots,17 and quantum rods.23 The idea of manipulating the luminescence properties of single semiconductor quantum dots in a DC electric field has been proposed by Seufert and his co-workers using a capacitor-like geometry,24,25 and Stark-modified emission from colloidal QDs in a planar electrode geometry has been recently studied by Bawendi and Bulovic.26,27 The distinct response of colloidal QDs to an applied external field not only makes them attractive candidates for light-emitting diodes (QD-LEDs),26 but also presents a promising prospect for biological imaging since QDs can be used as nonblinking systems. 16,28,29 In atomic systems, second-order Stark corrections to the electronic energies scale like the Bohr radius and thus are generally quite small even for extremely large electric fields. In quantum dot systems, the spatial extent of the electron/hole r 2011 American Chemical Society
envelope wave functions are hundreds of times larger than those of atomic systems, giving rise to proportionally larger Stark interactions for a given electric field strength. However, the characteristic signature of the DC Stark effect observed in all previous studies is a red shift in the recombination energy associated with a second order mixing of electron and hole states of different symmetry. Previously observed blue-shifted band-edge luminescence has generally been attributed to either QD oxidation,30,31 or a size decrease in QD core (changes in quantum confinement).32 In this Letter, we report on wavelength-resolved imaging of single CdSe/ZnS QDs charged via electrospray under negative bias. We observe a surprising spectral blue shift (∼60 meV peak-to-peak) in the distribution of QD emission frequency maxima, which can be understood as a first-order correction to the electron and hole band-edge states that is proportional to Ær2æ, the expectation value of the square of the radius (see Supporting Information eqs S1 and S2). Electrostatic force microscopy (EFM) studies on electrospray QD samples showed both charged and uncharged dots in a population ratio of ∼1:3, as evidenced by the distinct cantilever response under tip bias. For the charged QD subpopulation, quantitative charge state measurements showed that 4.7 ( 0.7 excess electrons were present. These results provide insight into understanding the role of electronic perturbations of single QD luminescence by excess charges, which can be used to develop future local charge reporters. Recent work in our laboratory examined the correlation of QD band-edge emission maxima with excitation polarization from Received: July 29, 2011 Revised: September 7, 2011 Published: September 09, 2011 4425
dx.doi.org/10.1021/nl2026103 | Nano Lett. 2011, 11, 4425–4430
Nano Letters
Figure 1. (a) Schematic design of experimental configuration for electrospray used to charge the quantum dots. (b) Emission spectra of electrosprayed charged QDs (blue) and drop-cast QDs (red). The observed spectral shift is about 60 meV peak-to-peak.
single CdSe-oligo phenylene vinylene nanostructures (CdSeOPV),19 where induced linear polarization and red-shifted emission were hypothesized to originate from an electric field resulting from a single electron (donated from photoexcited OPV) trapped near the QD surface. Because of the short (≈ 1.5 nm) distance of the charge from the QD surface, the Stark-perturbed electron and hole energies give rise to a redshifted band-edge luminescence (ΔE ≈ 70 meV). In the work reported here, we intentionally charged QDs using an electrospray deposition technique33 to correlate QD charge state and spectral luminescence properties. We find a surprising spectral blue shift in the band-edge luminescence of the charged QDs that can be explained as a result of a first order correction to the band-edge electron/hole states due to multiple charges over the surface. We used single CdSe/ZnS QDs (Evident Technology, EviDots 605) with 6 nm “nominal diameter” (∼4.5 nm core/ ∼1.5 nm shell) that can support 46 excess carriers.33 To charge the QDs, a dilute (∼107 M) quantum dot/toluene solution was sonicated for ∼60 min and loaded into a Hamilton syringe. The syringe was connected to a fine gauge inner tube that was biased at 400 V to 1 kV, through which the solution was injected (see Figure 1a). The inner tube was enclosed by a larger bore outer needle (33 gauge needle = 0.2 mm outer diameter), ensuring that the dry nitrogen nebulizer gas flowed evenly. The end of the inner tube was connected to the sprayer needle (0.1 mm inner diameter) where strong electric field builds up due to the applied high voltage.34,35 Particles were collected on plasma-cleaned glass coverslips, mounted about one inch away from the sprayer needle. This distance was chosen so that the nitrogen gas would dry the toluene in the aerosolized solution as soon as the nanoparticles reached the glass surface.33 An inverted microscope coupled with CCD and spectrometer was used for single QD spectral measurements (see Supporting Information Figure S1a). A 440 nm continuous wave diode laser
LETTER
Figure 2. (a) Emission energies for single drop-cast dot (DC-QD10) and electrospray charged dot (ES-QD5) (see Supporting Information Table S1 and Table S2 for QD number labeling). (b) Emission intensities for the drop-cast dot (DC-QD10) and electrospray charged dot (ES-QD5). Drop-cast QDs are represented by red, while blue color represents the electrospray charged dots. All the intensity values are normalized to the maximum value of the emission intensity during the measurements between 10 to 40 s. Trajectories of these QDs are shown in Supporting Information Figure S3.
(Crystalaser Model BCL-005-440) focused on the back aperture of a 1.4 NA, 100 microscope objective, was used as an excitation source. The fluorescence studies were done in wide field imaging mode with ∼10 μm spot size. Luminescence from individual CdSe/ZnS QDs was collected through the same objective and imaged through a dichroic mirror and 530 nm long pass emission filter to filter out the laser light in the detected signal. The QD luminescence was routed to either a high-speed imaging charge-coupled device (CCD) or an imaging spectrograph for wavelength-resolved measurements. The distribution of center frequencies of electrosprayed dots compared with those of drop-cast dots is presented in Figure 1b as a histogram for ∼500 dots. The observed spectral blue shift from peak to peak is ∼60 meV. Spectra for single charged QDs with an exposure time of 10 s is shown in Supporting Information Figure S1b. To confirm that the observed blue shift is indeed due to the excess electrons on the QDs, we have used electrospray source under different bias voltages but for the same spraying conditions and with the same sample concentration to deposit the QDs on glass substrate. Comparison of the histograms of peak emission energy from a sample of ∼150 single QDs for each (needle) bias voltage showed that blue shift is observable only for the QDs deposited through negative bias. QDs deposited using positive needle bias did not show any blue shift in their peak fluorescence energies confirming that the observed blue shift is due to the excess electrons on the QD surface (see Figures S2a and S2b in Supporting Information). We investigated the spectral and intensity trajectories of drop-cast and electrospray charged QDs in order to gain information about the spectral and intensity fluctuations for 4426
dx.doi.org/10.1021/nl2026103 |Nano Lett. 2011, 11, 4425–4430
Nano Letters the different QD samples. Ten drop-cast and 15 electrosprayed QDs were observed for 0.2 s exposure time per frame for a total 200 frames. The average spectral diffusion for drop-cast QDs was ∼7 meV (see Supporting Information Table S1). Measurements of spectral diffusion for electrosprayed QDs showed bimodal behavior with histograms of mean square deviation of spectral fluctuation showing peaks at ∼12 meV and ∼5 meV. Some of the QDs from the electrospray charged sample exhibited spectral diffusion with a rms spectral deviation nearly twice that of the drop-cast QDs with emission maxima less than 590 nm (see Supporting Information Table S2). We observed that the average intensity (Iavg) reduced by 23% for the electrospray charged QDs with an average spectral diffusion of ∼12 meV (Iavg = 0.41 fwhm = 0.23) compared to drop-cast QDs (Iavg = 0.53 fwhm = 0.20) and reduced 18% for the electrospray charged QDs with an average spectral diffusion of ∼5 meV (Iavg = 0.43 fwhm = 0.24). In previous reports, GuyotSionnest and co-workers have observed quenching in the photoluminescence of semiconductor nanocrystals due to injection of charge (electron) to either localized surface states or delocalized quantum-confined states.36,37 Lian and co-workers measured time-resolved luminescence from charged QDs on indium tin oxide (ITO) surfaces, showing suppressed blinking, shortened fluorescence lifetimes and quenched emission intensities for similar CdSe/ZnS QDs. They argued that the trapped electrons on the QD surface can provide fast nonradiative decay pathways, quenching the QD emission.16 Our time dependent emission intensity characterizations on single electrospray charged QDs are consistent with this picture. Figure 2 shows a comparison of spectral diffusion and intensity fluctuations of single QDs from drop-cast and electrospray charged samples. The average intensity fluctuations for electrospray charged dots increased by ∼15% with respect to drop-cast dots (see Supporting Information). Supporting Information Tables S1 and S2 provide details on the number labeling of QDs and further details are given in Supporting Information Figures S4 and S5. Figure 3 shows the effect of QD charging on photoluminescence (PL) lifetime. The fluorescence lifetime histograms, each consisting of ~50 single QDs, were constructed for both dropcast and electrosprayed QDs. Time-correlated single photon counting (TCSPC) histograms were satisfactorily fit with biexponential fits. From these fits, we extracted the four parameters used in Figure 3 to characterize the PL decay dynamics, which are the exponential decay constants and their associated amplitudes. We observed lifetime shortening in both decay components in the electrospray QD PL with respect to the drop-cast dots, consistent with the recent reports by Lian and his co-workers.16 In addition, the presence of external charges on the electrospray charged QDs notably alters the relative contributions of the decay components to the total decay. Statistical analysis of the histograms of long decay components revealed the mean and standard deviation (first and second moments, respectively) of the distributions. Drop cast QDs showed decay constant distributions with an average of 26 ns and fwhm of ∼5.3 ns. In contrast, electrosprayed charged QDs showed decay dynamics with an average of 18 ns and fwhm of ∼3 ns. The long decay components of charged QDs are shortened by ∼31% and form a narrower distribution when compared with drop-cast QDs. In addition, the process that produces the fast component of the QD photoluminescence decay produces a converged distribution of decay constants for the electrosprayed dots, whereas
LETTER
Figure 3. Histogram of decay constants obtained from four-parameter (exponential decay constants and their associated amplitudes for long and short components), biexponential fits to photoluminescence decay curves of ∼50 single (a) electrospray charged vs (b) drop-cast QDs. The areas under the histograms have been scaled to reflect their relative contributions to the total PL decay, thus, each pair of histograms approximates a lifetime probability distribution.
the distribution of fast decay components of the drop-cast dots appear to span a broad range of values. As can be seen by comparing the relative areas under the histograms in Figure 3, the short decay component in the drop-cast sample is responsible for approximately 33% of the total decay, while in the electrosprayed dot sample, this component contributes the majority, approximately 65%, of the total decay. We observed some overlap of the lifetime distributions of the drop cast and electrosprayed dots that presumably derives from a combination of the finite charging efficiency of the electrospraying process in combination with the distribution of charge states that we expect as a result of charging process (Figures 1b and 3). Electrostatic force microscopy (EFM) was used to extract quantitative charge information on the electrosprayed QD sample.3840 In general, EFM measures the long-range electrostatic forces by combining atomic force microscopy (AFM, Asylum Research, MFP-3D) with conductive biased probes (SCM-PIT from Bruker Nano) and interleaved scanning techniques.41 Conductive EFM tips (oscillation frequency f0 ≈ 60 kHz and spring constant k ≈ 15, nominally 2.8 N/m) were coated with 20 nm Pt/Ir to provide an electrical connection from the cantilever to the tip apex. Unfortunately, the mechanical stability requirements for precise EFM measurements of this type preclude direct correlation of emission spectra with charge state for a particular quantum dot. Therefore, we have maintained similar conditions such as ambient environment and room temperature for EFM studies as well as spectral measurements. In interleaved EFM measurements, the first pass is the height mode, where the conductive AFM tip is maintained in the van der Waals force regime and the height information is obtained via intermittent contact imaging. In the second pass of EFM, the oscillation amplitude of the tip was reduced significantly compared to the first pass, and the lift height is optimized so that the tip is maintained in the electrostatic force regime, then the tip is biased by applying a dc voltage (Figure 4a). The optimized lift height between the tip and the surface for our experiment 4427
dx.doi.org/10.1021/nl2026103 |Nano Lett. 2011, 11, 4425–4430
Nano Letters
LETTER
Figure 4. (a) Schematic of EFM setup. EFM phase images of electrospray charged CdSe/ZnS QDs (b) for 2.5 V bias applied to the tip, VEFM = 2.5 V and (c) for +2.5 V bias applied to the tip, VEFM = +2.5 V. (d) Phase shift vs VEFM measured for drop-cast (red triangles) and electrospray charged (blue circles) QDs. Note that the data point for VEFM = 0 V for drop-cast and electrospray QDs both have zero phase shift therefore they overlap. The red and blue curves are the best fit for the polynomial law1 of ΔΦ = AVEFM + BVEFM2, where A = 0.032 /V and B = 0.034 /V2 for the drop-cast dots and A = 0.111 /V and B = 0.057 /V2 for electrospray charged QDs. Inset: scheme showing the effect of QD excess charge distribution on the EFM tip. Surface charge was measured using EFM bias ranging from 2.5 e VEFM e + 2.5 V. EFM images of charged dots at different bias applied to the tip are shown in Supporting Information Figure S8. Note that the phase-force response of the Asylum Research MFP-3D is opposite to commonly reported literature.
was ∼25 nm. During the EFM scan, the phase shift of the cantilever changes relative to the tip driving oscillation due to electrostatic forces acting between the tip and the surface. These forces include the capacitive (always attractive) and the Coulombic (either attractive or repulsive) interactions.40 For our EFM studies, we used a noncontact mode imaging that allows extraction of surface electronic properties of the sample by varying the dc bias applied to the tip (VEFM).1,41 EFM data is collected for different voltages applied to the tip that allow extraction of the Coulombic forces. We measure the phase shift due to electrostatic forces at different voltages was measured from the cross section of the EFM data. The tip bias response to the individual QDs provides a signature of excess charges. In the electrospray sample, approximately 1 out of 3 QDs showed a clear sign change in the measured electrostatic force under reversing the bias from 2.5 V to +2.5 V (see Figure 4b,c). This is completely consistent with the emission energy histogram (Figure 1b) that shows a strongly perturbed subpopulation (with a narrow charge state distribution) and an unperturbed subpopulation. We have also included EFM images of drop-cast QDs in Supporting Information Figures S9 and S10. The measured phase shift vs EFM tip bias (VEFM) is shown in Figure 4d. The red triangles correspond to the phase shift of the drop-cast QDs, while the blue circles represent electrospray charged QDs. The data was fit to a polynomial of the form of ΔΦ = AVEFM + BVEFM2 where ΔΦ is the phase shift of the resonant peak (degree).1 A and B are the constants that
give the charge and polarizability information respectively. The coefficient A and B are defined as A = [Λ/(kz2)]q and B = (Λ/k)[(3α)/z4], respectively.42 Here Λ is the quality factor, k is the spring constant, z is the lift height, and α is the electric polarizability. For the conductive tip SCM-PIT silicon cantilevers with a nm layer of Pt/Ir coating, Λ = 186, k = 2.8 N/m and z = 25 nm. In our case A = 0.032 /V and B = 0.034 /V2 for drop-cast dots. For electrospray charged dots, A = 0.111 /V and B = 0.057 /V2. The relative errors were