Quantum Light Signatures and Nanosecond Spectral Diffusion from

Mar 22, 2012 - Single-walled carbon nanotubes (SWCNTs) are considered for novel optoelectronic and quantum photonic devices, such as single photon sou...
0 downloads 5 Views 1MB Size
Download PDF
Letter pubs.acs.org/NanoLett

Quantum Light Signatures and Nanosecond Spectral Diffusion from Cavity-Embedded Carbon Nanotubes William Walden-Newman, Ibrahim Sarpkaya, and Stefan Strauf* Department of Physics and Engineering Physics, Stevens Institute of Technology, Castle Point on the Hudson, Hoboken, New Jersey 07030, United States S Supporting Information *

ABSTRACT: Single-walled carbon nanotubes (SWCNTs) are considered for novel optoelectronic and quantum photonic devices, such as single photon sources, but methods must be developed to enhance the light extraction and spectral purity, while simultaneously preventing multiphoton emission as well as spectral diffusion and blinking in dielectric environments of a cavity. Here we demonstrate that utilization of nonpolar polystyrene as a cavity dielectric completely removes spectral diffusion and blinking in individual SWCNTs on the millisecond to multisecond time scale, despite the presence of surfactants. With these cavity-embedded SWCNT samples, providing a 50-fold enhanced exciton emission into the far field, we have been able to carry out photophysical studies for the first time with nanosecond timing resolution. We uncovered that fast spectral diffusion processes (1−3 ns) remain that make significant contributions to the spectral purity, thereby limiting the use of SWCNTs in quantum optical applications requiring indistinguishable photons. Measured quantum light signatures reveal pronounced photon antibunching (g2(0) = 0.15) accompanied by side-peak bunching signatures indicative of residual blinking on the submicrosecond time scale. The demonstrated enhanced single photon emission from cavity-embedded SWCNTs is promising for applications in quantum key distribution, while the demonstrated passivation effect of polystyrene with respect to the stability of the optical emission opens a novel pathway toward optoelectronic devices with enhanced performance. KEYWORDS: Single-walled carbon nanotubes, photon antibunching, blinking, spectral diffusion, single photon source, cavity, polystyrene

S

spectral purity in order to prevent two-photon interference from diminishing. A detrimental effect on spectral purity of a quantum emitter is inhomogeneous broadening of the intrinsic photoluminescence (PL) line width brought about by spectral diffusion (SD).11 Spin and charge fluctuations in the vicinity of the photoemitter create a Stark-effect induced variability in the emission wavelength over time,12 in addition to broadening caused by spontaneous emission, pure dephasing,13 and unintentional doping.14 Previous studies at room15 and cryogenic16 temperatures have found SD at a millisecond to multisecond time scale, giving rise to pronounced spectral broadening for substrate-deposited SWCNTs. It was also shown that SD can be suppressed, at least on the millisecond time scale, by suspending SWCNTs in air between supports and in the absence of surfactant wrapping.17 However, nonencapsulated, free-standing SWCNTs18 are brittle and might be disadvantageous in light of optoelectronic device integration into cavities for enhanced light extraction. To overcome this limitation, we recently demonstrated that both

ingle-walled carbon nanotubes (SWCNTs) display outstanding optical properties, such as optical recombination, that arises from excitons with binding energies of several hundred meV1 and emission wavelengths that extend into the telecom band which are controlled by the chirality.2,3 Toward optoelectronic devices utilizing individual SWCNTs,4 electrical injection was demonstrated in a field effect transistor geometry5 as well as monolithic integration in a planar cavity giving rise to directional emission and a four-fold enhanced exciton emission rate.6 Recently it was also demonstrated that individual SWCNTs display photon antibunching at cryogenic temperatures,7 i.e., emission of nonclassical light from a quantum emitter with a one-dimensional density of states. These optical properties suggest that SWCNTs are ideal candidates for the next generation single photon sources,8 which can operate at room temperature under electrical injection in the telecom band as required for practical applications in quantum cryptography.9 For optical quantum information processing, however, the requirements are more stringent and require indistinguishable single photons, if entanglement is created by bunching at the beam splitter.10 Hence, the quantum emitter must display photon antibunching combined with a high photon collection efficiency, long dephasing times, and a high © 2012 American Chemical Society

Received: December 13, 2011 Revised: March 21, 2012 Published: March 22, 2012 1934

dx.doi.org/10.1021/nl204402v | Nano Lett. 2012, 12, 1934−1941

Nano Letters

Letter

intermittent quantum blinking19 and SD were greatly suppressed at millisecond time scales by encapsulating the SWCNT in poly(methyl) methacrylate (PMMA) and subjecting the sample to a thermal annealing procedure to remove detrimental surface bound water near the SWCNTs.20 Although important for applications in quantum information processing, it is largely unknown to what extent SD is present on faster time scales and to what extent it affects the spectral purity as well as the purity of the nonclassical light emission of individual SWCNTs. One problem that complicates measurements of SD at faster time scales is that the quantum yields of SWCNTs are low,21 in part due to high exciton mobility along the 1D axis22 combined with the presence of defect sites,23,24 resulting in susceptibility to nonradiative recombination channels along the tube,25 and to a lesser extend due to the lower lying exciton dark state which affects the PL yield below 50 K.16 The low photon count rate limits the timing resolution of typical methods to record SD, such as continuous recording of PL with charged-coupled photodetector arrays11,12 or photon correlation Fourier spectroscopy26 to the ms and μs time scale, respectively. Recently it was found that the secondorder photon correlation function (g2(τ)) in combination with spectral filtering within the spectral line width of the PL emission can extend SD studies down to the nanosecond time domain.27−29 While subnanosecond SD was observed with this technique for semiconductor QDs and attributed to hopping between charged and neutral exciton states, the technique has not yet been applied to SWCNTs. Hence, to enable quantum light applications based on SWCNTs, methods must be developed to enhance the light extraction and suppress nonradiative recombination in order to improve the single photon emission rate, while simultaneously preventing SD and blinking in dielectric environments of a cavity. Here we demonstrate that utilization of nonpolar polystyrene as a cavity dielectric completely removes SD as well as blinking on the millisecond to multisecond time scale despite the presence of surfactants. With these samples we have been able to carry out photophysical studies of SWCNTs for the first time with nanosecond timing resolution and uncover that fast SD processes (1−3 ns) remain that make significant contributions to the PL line width and thus to the spectral purity in addition to residual doping and exciton dephasing processes. We also demonstrate pronounced photon antibunching for the first time under true resonant excitation of the E22 transition despite the presence of nanosecond SD and furthermore uncover for some samples a side-peak bunching signature indicative of two-state blinking on a submicrosecond time scale, which implies that the universal power law does not apply for all blinking phenomena. Results and Discussion. Our goal is to extend spectral diffusion studies to the nanosecond time domain and study its effect on quantum light generation. Since these measurements are carried out with two-photon coincidence techniques, it is essential to collect as much exciton emission as possible in the far field. To this end, we first describe an approach based on a half-cavity design providing 50-fold enhanced PL emission, which enabled the nanosecond time domain studies. Since most SD and blinking studies are carried out with millisecond timing resolution, we continue to provide such a study for our samples and demonstrate for the first time that these effects at slow time scales can be completely eliminated by proper choice of the cavity dielectric. After that we introduce the coincidence technique to record SD with nanosecond timing resolution and discuss the influence of carrier density, polymer polarity, and

substrate proximity as well as the interplay with the observed quantum light signatures. In order to enhance the light collection efficiency from individual SWCNTs and to reduce possible nonradiative recombination into the substrate channel, we embedded SWCNTs in a thin slab half-cavity design. To this end chirality, purified (6,5) SWCNTs grown by the CoMoCat technique were dispersed in sodium dodecylsulfate (SDS) solution (see Methods Section) and integrated in the two device geometries shown in Figure 1a. In the first design, SWCNTs were directly

Figure 1. (a) Schematic diagrams of an individual SWCNT deposited directly on SiO2 beneath a polymer layer (design 1) and between two polymer layers suspended above an Au-coated substrate (design 2). (b) Plots of integrated PL intensity under cw laser excitation at 780 nm recorded with a CCD camera with 4 samples from design 1 and 5 samples from design 2. Data are shown in all cases for samples with PMMA layers as a dielectric and are recorded at 80 K.

deposited onto the SiO2 substrate and capped with a 160 nm layer of spin-coated polymer similar to our previous work.20 The second design incorporates the ideas of eliminating the detrimental influence of the SiO2 and enhancing light extraction by utilizing a planar broadband Au mirror and additionally a polymer spacer layer before deposition of SWCNTs and final capping with a second polymer layer. All experiments have been carried out with individual SWCNTs in the detection volume showing bright emission with little to no sidepeaks in the emission spectrum (see Methods Section). Figure 1c shows that design 2 provides a remarkable factor 8−10 improvement in light emission from individual SWCNTs over design 1. The combined enhancement provided by the bottom mirror, the dielectric spacer to SiO2, and the capping layer of design 2 reaches up to 50-fold when compared to bare uncapped SWCNTs located on a SiO2 substrate, as further detailed in the Supporting Information. We note that the half-cavity design was optimized to provide more exciton emission in the far field rather than a Purcell effect. We now discuss the photophysics on the millisecond time scale of blinking nanotubes that are sandwiched in a polymer (design 2) using two different polymers, the nonpolar PS, and the polar PMMA. We plot occurrence of a count value within 1 ms for the entire time-correlated photon counting (TCPC) trace and define the on/off ratio as the percentage of time spent in the on state relative to the off state, or more simply, the peak ratio between on and off event counts (see Methods Section). For samples coated with polar PMMA, we observe pump power induced blinking in TCPC traces that illustrate the transition from weakly pronounced blinking to strong blinking behavior (Figure 2a) with an on/off ratio of 100 at 15 μW excitation that 1935

dx.doi.org/10.1021/nl204402v | Nano Lett. 2012, 12, 1934−1941

Nano Letters

Letter

Figure 2. Occurrence plots of a count value within 1 ms time bins for an entire TCPC trace at 3 different pump powers for PMMA (a) and PS (c). A TCPC trace is shown at the highest pump power for each polymer (b,d).

decreases to 2 at 79 μW. One out of eight PMMA-coated SWCNTs showed an absence of blinking, indicating rare circumstances of SWCNTs in a defect-free environment. Similar results were found for encapsulation in polar polyvinyl alcohol (PVA, not shown). In contrast, we see stable emission up to mW excitation powers for all six investigated SWCNT samples coated with nonpolar PS, as shown in Figure 2c,d. It is interesting to note that stable emission has been achieved despite the fact that the SWCNTs were initially dispersed using SDS surfactant, which was previously attributed as the cause of blinking.17 It is thus likely that the nonpolar character of PS increases the ionic pairing in the SDS between the positively charged sodium ions with the negatively charged sulfate end, as compared to the more polar PMMA and PVA. A reduced amount of mobile sodium ions might in turn reduce charge fluctuations in the SWCNT’s environment. As a result, blinking is completely removed up to time scales of half an hour without a single jump of the quantum emitter into the off state, implying on/off ratios of infinity even at highest pump powers similar to results achieved for nitrogen vacancy centers.30 The complete removal of blinking due to PS capping combined with the 50-fold enhanced exciton emission provided by design 2 are excellent properties for applications of SWCNTs in optoelectronic devices, such as light emitting diodes. For optical quantum information processing, however, the requirements are more stringent, and quantum emitters must display photon antibunching combined with long dephasing times and a high spectral purity, i.e., indistinguishable single photons are required. The standard approach to investigate SD is to stream spectral trajectories into a CCD camera with a typical timing resolution down to 10 ms. As shown in Figure S2 in the Supporting Information, the spectral trajectories recorded every 200 ms for SWCNTs suspended between layers of PS or PMMA show no significant SD within the resolution of the spectral line width, while uncapped SWCNTs touching a SiO2 substrate are severely affected by SD. Thus, encapsulation of the SWCNT between polymer layers effectively removes external accumulation of local charge traps, thereby preventing dynamical Stark shifts12 on the millisecond time scale that would occur if exposed to ambient

air. However, the question remains to what extent SD is still present on faster time scales and to what extent it affects the spectral purity as well as the purity of the nonclassical light emission of SWCNTs. To extend SD studies for SWCNTs to the nanosecond time domain, we adapted a technique based on photon coincidence measurements, which was recently introduced for semiconductor QDs.28,29 The second-order photon correlation function g2(τ) was recorded with a Hanbury-Brown and Twiss setup under CW laser excitation (see Methods Section). When g2(τ) is recorded using a 10 nm bandpass filter centered over the right half (R) of the emission spectrum (Figure 3c), we observe a pronounced photon bunching signature shown in Figure 3a with a decay half-width of 1.8 ± 0.2 ns at 2.1 mW excitation. This photon bunching signature was observed in all

Figure 3. Plots of second-order correlation under CW excitation using a (a) 10 nm filter across half the emission and (b) 40 nm filter across the entire emission as illustrated in (c) with blue and red shading. (d) Additional plot of second-order correlation using a 10 nm filter for an SWCNT showing a decrease in counts at zero time delay and g2(τ) > 2. All data are taken at 9 K. 1936

dx.doi.org/10.1021/nl204402v | Nano Lett. 2012, 12, 1934−1941

Nano Letters

Letter

configurations, i.e., with PMMA or with PS capping and in design 1 as well as in design 2 and seems to be inherent to SWCNTs wrapped in SDS surfactant. There are several reasons why a quantum emitter would display photon bunching signatures, such as the intrinsic bosonic nature of thermal radiation,31 the temporal sequential emission of the biexciton− exciton cascade,32 or the transition from lasing to spontaneous emission under decreasing pumping in a nanolaser.33,34 We find, however, that the bunching signature is removed without any change to the sample by switching the 10 nm bandpass filter centered over R with a 40 nm filter that covers the entire PL emission (Figure 3b,c). Hence, the cause of the bunching must lie in the method of measurement and not thermal or coherent properties of the quantum emitter itself. We therefore attribute the bunching to nanosecond SD of the intrinsic PL line into and out of the filter range, in analogy to similar experiments carried out with QDs.28 To understand this qualitatively, we assume that the line shape in the time-integrated PL spectrum in Figure 3c has its origin in PL emission from a much narrower spectral line, which rapidly changes its emission energy due to the dynamic Stark shift involved in SD. The spectral filtering of photons from a specific subregion of the PL spectrum creates then a conditional probability in the coincidence measurements. If, for example, a photon from region R is recorded by detector one, there is a high likelihood that another photon from region R will be recorded by detector two (at least at times longer than the spontaneous emission time where photon antibunching is observable). This probability decreases since the dynamic Stark shift can cause the SWCNT to emit later in region L where no photons are detected due to the spectral filtering, resulting in a decreasing coincidence rate at later times. The time delay between two events is thus a measure of the average residence time in R, corresponding to the SD time. For a quantitative analysis we utilized a four-level rate equation model, which is further detailed in the Supporting Information. The photon correlation traces can be analyzed by extracting the spectral diffusion rate γd (time τd) from the decay of the bunching peak using the relation:

Figure 4. (a) Plot comparing the dependence of spectral diffusion rate and integrated PL on excitation power. (b) Line width and line shape taken at various pump powers for the same tube. (c) Comparison of spectral diffusion rate for samples with designs 2 and 1 and for different polymers in design 2. All data are taken at 9 K.

intensity, which is also shown in Figure 4a for comparison. The observed increase of the SD rate with increasing pump power can be attributed to an effective increase of possible charge configurations in the SWCNT vicinity, which changes the effective rate for filling or depleting the randomly populated trap states. To discuss the influence of SD on line width, we plot the full width at half-maximum of the exciton emission spectrum as a function of pump power in Figure 4b. The SWCNT spectra show typically a single spectral line and the line width increases by about 12% from 3.9 to 4.4 meV in the same pump power range. The 12% broadening in Figure 4b can be attributed to the pump power-dependent contribution to dephasing via exciton−exciton scattering processes also causing a slight redshift of the spectrum, as shown in the inset of Figure 4b. According to recent studies of ultrafast exciton dephasing in SWCNT ensembles,13 pure dephasing times of T2*∼500 fs were reported in the low pump power regime, corresponding to typical line widths of 3−4 meV, which are similar to our findings of 3.9−4.4 meV for an individual SWCNT. While we measure a SD time with the coincidence measurements, the physics behind that rate is an intrinsic narrower line width which hops in and out of the spectral filter at the crossover rate γL. Therefore, there must be a contribution from SD to the time integrated PL line width in addition to pure exciton dephasing, otherwise we would not see any bunching in these measurements. These findings highlight the importance of taking into account for possible contributions of SD to the line shape when interpreting the line width of individual SWCNTs.

⎡ ⎤ ⎛γ ⎞ g 2(τ, R) = ⎢1 + ⎜⎜ d − 1⎟⎟e(−γdτ)⎥[1 − ae(−γτ)] ⎢⎣ ⎥⎦ ⎝ γL ⎠

where R is the right half of the PL emission peak for each bandpass filter before the avalanche photodiodes (APDs), as illustrated by the blue shading in Figure 3c, γL is the crossover rate for hopping in and out of the filter range which determines the bunching peak height (See Figure S4 in the Supporting Information), γ is the spontaneous emission (SE) rate, and a takes into account for the PL background. In Figure 3a we do not observe a decrease in counts at zero delay time (i.e., photon antibunching) because the measured SE lifetime of most SWCNTs is typically less than 400 ps and below the timing resolution of the APDs.35,36 One can thus neglect the last term, exp(−γτ), in the above equation when estimating γd. We demonstrate in Figure 3d that some measurements however show an effective SE lifetime in the order of the timing resolution, resulting in a drop in photocounts at zero delay time. For those cases, as in Figure 3d, the term exp(−γτ) was explicitly taken into account, with a = 0.8 and 1/γ = 2 ns. To study the influence of carrier density, we plot in Figure 4a the extracted γd values as a function of excitation pump power. Interestingly, the SD rates follow the trend of the integrated PL 1937

dx.doi.org/10.1021/nl204402v | Nano Lett. 2012, 12, 1934−1941

Nano Letters

Letter

Additionally, we observe that the polarity of the polymer material does not seem to affect the SD rates considerably since the pump power dependence shown in Figure 4c is similar for PMMA and PS capped SWCNTs for sample design 2. In contrast, the SD rate is remarkably lower at comparable pump powers for SWCNTs in contact with the SiO2 substrate of sample design 1 with γd about 4−5 times slower. The significantly slower SD rates in design 1 correlate with the 4−5 times lower PL emission efficiency of these SWCNTs when in contact with SiO2, as discussed in context of Figure 1 above. Furthermore, for a particular PMMA covered tube in design 2, we obtained measurements of both SD rate and on/ off ratio, showing a joint correlation of a faster SD rate causing an increased amount of blinking, i.e., a lower on/off ratio (see Supporting Information). This indicates that the SD contributes via the Stark shift to the establishing of the temporary resonance condition between the radiative exciton transition and the defect energy levels required for the tunneling process involved in the blinking. These studies of the SD time dynamics indicate that three different time scales exist which can be attributed to three different extrinsic mechanisms. First, SD on the millisecond time scale is effectively eliminated due to the polymer capping and removal of surface water molecules. Second, we observe SD on the time scale of about 10 ns when SWCNTs are in contact with the SiO2 substrate. In this regime, the SWCNT-substrate morphology plays an important role, which was recently shown to cause an increased charge density at the interface, where parts of the semiconducting SWCNT can become metallic.37 Third, the SD time scale reaches down to about 1 ns for cavityembedded SWCNTs which no longer touch the substrate and display enhanced light emission. In this regime the SD is independent of the choice of the polymer and seems to be intrinsic to the remaining SDS surfactant wrapping the SWCNTs. Finally we investigate the quantum light signatures of individual SWCNTs in the presence of SD and blinking. To suppress multiphoton emission events, a pump power of 10 μW is chosen, which is well below the saturation regime of the exciton emission38 (see Figure 1b at pump powers above 1 mW). Since the spontaneous emission rate is faster than the timing resolution of the APDs, nonclassical light emission can only be observed under pulsed laser excitation. To this end, we recorded the second-order photon correlation function g2(τ) at 10 μW pump power with a 40 nm bandpass filter centered on the PL emission (see Methods Section). With the bandpass considerably larger than the emission spectrum, we avoid additional bunching effects which could be caused by the spectral jumps of the emitter related to SD. The excitation wavelength was chosen to coincide with the E22 excited state (700 nm) in order to maximize the PL emission and minimize any detrimental contribution from Raman side bands. Figure 5 shows pronounced single photon antibunching with a normalized peak area of g2(0) = 0.26 ± 0.05 at zero time delay, for a SWCNT embedded into design 2. The data have been normalized by plotting the integrated peak areas over five time bins in Figure 5b and taking the norm with respect to those peaks at larger delay times which are unaffected by the bunching effect. For other SWCNTs embedded into design 2 with comparable line width and similar SD rates we often found that g2(0) ranged from 0.15 to about 0.5, and in some cases, no antibunching was found (data not shown). This degree of antibunching is similar to the results found by Högele et al.7 for

Figure 5. (a) Plot of second-order correlation under pulsed excitation at 80 MHz and 10 μW. (b) Plot of peak area for a given peak number with on and off blinking times τon,off extracted from the curve fit. All data were obtained at 9 K.

uncapped SWCNTs in contact with the substrate and in the presence of SD on the millisecond time scale. Thus it appears that the purity of single photon antibunching is not severely affected by the presence of SD but depends rather on the degree of exciton localization along the tube axis, which is known to vary from tube to tube even at room temperature.39 In addition to photon antibunching, we also observe pronounced side-peak bunching on the g2(τ) trace in Figure 5 at much longer time scales than the corresponding SD times of a few nanoseconds, which was never shown before for SWCNTs. This side-peak bunching is however typically observed in semiconductor QDs under resonant excitation of the p-shell and is interpreted to be caused by two-state submicrosecond blinking.10 Following Santori et al.,10 the photon correlation traces can be analyzed by extracting the on and off blinking times τon,off from the decay of the bunching peak using the relation: ⎛ τ ⎞ −|m τ | ( 1 + 1 ) L τoff τon PA m = 1 + ⎜ off ⎟e ⎝ τon ⎠

where PAm is the normalized peak area of the mth correlation peak and τL = 12.5 ns is the repetition period of the laser. Figure 5b shows a plot of PAm for a given peak number with on and off blinking times τon = 55 ns and τoff = 120 ns extracted from the curve fit. The time scales in quantum emitter blinking are often reported to follow a universal power law,19 which we also observe in our uncapped or PMMA-embedded SWCNTs.20 In stark contrast, our findings indicate a breakdown of this rule, since the slow pump power induced blinking on the millisecond to multisecond time scale inherent to PMMA encapsulated samples was completely removed when instead embedded in nonpolar PS. This suggests that mobile charges in the PMMA/ surfactant (e.g., Na+ ions from the SDS or charges from trap states) as well as in the SiO2 substrate are dominant in the slow regime. For the case of PS encapsulated samples of design 2, only a fast component of two-state blinking on a submicrosecond time scale remains, which might be related to charge reconfigurations in the particular defects in the SWCNT sidewalls.24 Interestingly, since the remaining blinking reveals no contributions at slower time scales, the universal power law does not apply to the particular microscopic mechanism responsible for the remaining two-state nanosecond blinking. Thus, in addition to the very recently uncovered two types of blinking at the millisecond to multisecond time scale related to charging and discharging of a QD core as well as to charge fluctuations in the electron accepting surface sites of QDs,40 a 1938

dx.doi.org/10.1021/nl204402v | Nano Lett. 2012, 12, 1934−1941

Nano Letters

Letter

800 nm high-pass filter. In order to collect more light from individual SWCNTs, we used a nearly index-matched solid immersion lens (S-LAH79 with refractive index 1.97) on top of the wafer to enhance the light extraction from the polymer medium (refractive index of 1.5−1.6), which is limited by total internal reflection. To enhance the exciton emission, the laser polarization was rotated with a half-wave plate with respect to the axis of the SWCNTS. The samples revealed predominantly SWCNTs with (9,1) or (8,3) chirality, as identified from their optical emission at 900 and 930 nm, respectively.20 Probing Individual SWCNTs. To identify individual SWCNTs, we first scanned the sample with the piezo actuators underneath the 1.5 μm laser spot of the 780 nm cw laser diode. We typically find only one emission center within about a 20 μm scan range of the actuators. When scanning over such an emission center, we find that the intensity varies strongly and is typically lost after about 2 μm scan range, indicative of an individual SWCNT within the detection volume. In addition, we confirmed a pronounced sinusoidal dependence between the polarization axis of the excitation laser and the integrated PL intensity for each SWCNT, which is maximized when the electric field component of the laser lies along the tube axis.20 This polarization dependence and the observed two-state blinking behavior for PMMA-coated samples would be washed out in the case of more than one emitter. Direct proof comes from the observed single photon antibunching below 0.5 for some of the SWCNTs (Figure 5). We do sometimes observe CNTs emitting much weaker than others as well as CNTs with various sidebands in the spectra. In those cases, there could be either defective tubes or other carbon material nearby, which quenches the PL emission. While we cannot fully rule out that such effects occur, we focused the study only on the brightest tubes which are less likely affected by nearby PL quenching. Photon correlation measurements. Measurements of the blinking statistics (on/off ratio) were performed by sending the PL emission through a 10 nm bandpass filter and tagging timecorrelated photon counts into 1 ms time bins with a single photon counting avalanche photodiode (APD) and continuously recording with a data acquisition card. Measurements of spectral trajectory over time were obtained by continuously recording the spectra every 200 ms with the CCD camera. Second-order correlation functions g2(τ) were recorded by sending PL emission either through 10 nm (spectral diffusion) or a 40 nm (quantum light) bandpass filters of a HanburyBrown and Twiss setup consisting of a fiber-coupled 50/50 beam splitter connected to two APD’s with about 400 ps timing resolution. A linear polarizer was used in the collection path for photon antibunching measurements. Without the filter, antibunching signatures were strongly degraded since the SWCNT emission appears to be only partially polarized despite the linear polarized optical excitation along the tube axis, providing a detrimental coincidence background at zero delay time. Coincidence counts were time stamped and analyzed with a high-resolution timing module (SensL).

third type of blinking at nanosecond time scales is present in SWCNTs. In summary, we studied individual cavity-embedded SWCNTs and found that elevation from the SiO2 substrate and encapsulation into nonpolar PS completely removes SD and pump power induced blinking at the millisecond time scale despite the presence of surfactants. These samples enabled for the first time SD studies of individual SWCNTs with nanosecond timing resolution. Pronounced photon bunching signatures observed in photon coincidence measurements reveal residual fast SD processes on a time scale of 1−3 ns, which significantly contribute to the PL line width (3−4 meV), i.e., the spectral purity. Despite the presence of nanosecond SD, we observe pronounced photon antibunching (g2(τ) = 0.15) under resonant E22 excitation. Moreover, side-peak bunching effects uncover residual and fast two-state blinking events at a submicrosecond time scale, which does not follow the common power law distribution of blinking times. While it is expected that inhomogeneous broadening due to the remaining SD on the nanosecond time scale will severely diminish the twophoton interference contrast, for example, when entanglement is created by bunching at the beam splitter, the demonstrated pronounced single photon emission in combination with the 50-fold enhanced exciton emission efficiency in our sample design is promising for applications in quantum key distribution, where efficient single photon sources are required. The demonstrated passivation effect of polystyrene with respect to the stability of the PL emission is also of interest beyond quantum photonics, since it opens a pathway to CNT light emitting diodes and solar cells with enhanced performance. Methods. Sample Preparation. Samples were fabricated using an ebeam evaporator (Lesker) to coat a 100 nm layer of Au on top of a standard p++ type Si wafer with a 90 nm SiO2 layer. This was followed by a 160 nm layer of spin-coated polymer (either PMMA or PS). Commercial chirality-purified CoMoCat SWCNTs were prepared with bath sonication for 1 h in a vial containing 0.4 wt % sodium dodecylsulphate (SDS) solution. The product was poured through a 5 μm filter to form a concentration of 0.2 mg/mL, deposited directly onto the first layer of polymer, and then covered with a second 160 nm layer of polymer. A polymer layer thickness of 160 nm each was chosen such that the thin slab half-cavity supports optical modes in the 900 nm wavelength range with the field maximum near the center of the thin slab where the SWCNTs are deposited. Finally, the PMMA samples were baked at 105 °C and the PS samples at 95 °C for several hours before cryogenic measurements to remove charge trap states.20 Experimental Setup. Measurements of μ-PL were taken inside a liquid helium cryostat with a 7 K base temperature. Samples were excited either with a laser diode operating at 780 nm in cw or pulsed mode (80 MHz repetition rate and 100 ps pulse length) or by utilizing a supercontinuum light source (80 MHz repetition rate and 7 ps pulse length) combined with narrow band spectral band-pass filters for resonant excitation of the E22 transition in the 700 nm band. A laser spot size of about 1.5 μm was achieved using a microscope objective of NA 0.55. The relative position between sample and laser spot was adjusted with a piezo-electric xyz-actuator (Attocube Systems) mounted directly onto the coldfinger of the cryostat. Spectral emission from the sample was dispersed using a 0.75 m focal length spectrometer and imaged by a liquid-nitrogen-cooled CCD camera resulting in a spectral resolution of 0.05 nm. Laser stray light was rejected combining a 780 nm notch filter and

■

ASSOCIATED CONTENT

S Supporting Information *

Further details on the determination of light extraction efficiency, spectral diffusion at millisecond time scales, the rate equation model to determine spectral diffusion rates, and correlations between spectral diffusion and blinking are provided. This material is available free of charge via the Internet at http://pubs.acs.org. 1939

dx.doi.org/10.1021/nl204402v | Nano Lett. 2012, 12, 1934−1941

Nano Letters

■

Letter

Revealed Through Spectral Diffusion. Phys. Rev. B 2008, 77, 193405/ 1−193405/4. (17) Kiowski, O.; Lebedkin, S.; Hennrich, F.; Kappes, M. M. SingleWalled Carbon Nanotubes Show Stable Emission and Simple Photoluminescence Spectra with Weak Excitation Sidebands at Cryogenic Temperatures. Phys. Rev. B 2007, 76, 075422/1−075422/8. (18) Lefebvre, J.; Fraser, J. M.; Finnie, P.; Homma, Y. Photoluminescence from an Individual Single-Walled Carbon Nanotube. Phys. Rev. B 2004, 69, 075403/1−075403/5. (19) Frantsuzov, P.; Kuno, M.; Jankó, B.; Marcus, R. A. Universal Emission Intermittency in Quantum Dots, Nanorods, and Nanowires. Nature 2008, 4, 519−522. (20) Ai, N.; Walden-Newman, W.; Song, Q.; Kalliakos, S.; Strauf, S. Suppression of Blinking and Enhanced Exciton Emission from Individual Carbon Nanotubes. ACS Nano 2011, 5, 2664−2670. (21) Tsyboulski, D. A.; Rocha, J.-D. R.; Bachilo, S. M.; Cognet, L.; Weisman, R. B. Structure-Dependent Fluorescence Efficiencies of Individual Single-Walled Carbon Nanotubes. Nano Lett. 2007, 7, 3080−3085. (22) Siitonen, A. J.; Tsyboulski, D. A.; Bachilo, S. M.; Weisman, R. B. Surfactant-Dependent Exciton Mobility in Single-Walled Carbon Nanotubes Studied by Single-Molecule Reactions. Nano Lett. 2010, 10, 1595−1599. (23) Hertel, T.; Himmelein, S.; Ackermann, T.; Stich, D.; Crochet, J. Diffusion Limited Photoluminescence Quantum Yields in 1-D Semiconductors: Single-Wall Carbon Nanotubes. ACS Nano 2010, 4, 7161−7168. (24) Lee, J. A.; Wang, X.; Carlson, L. J.; Smyder, J. A.; Loesch, B.; Tu, X.; Zheng, M.; Krauss, T. D. Bright Fluorescence from Individual Single-Walled Carbon Nanotubes. Nano Lett. 2011, 11, 1636−1640. (25) Perebeinos, V.; Avouris, P. Phonon and Electronic Nonradiative Decay Mechanisms of Excitons in Carbon Nanotubes. Phys. Rev. Lett. 2008, 101, 057401/1−057401/4. (26) Coolen, L.; Brokmann, X.; Spinicelli, P.; Hermier, J.-P. Emission Characterization of a Single CdSe-ZnS Nanocrystal with High Temporal and Spectral Resolution by Photon-Correlation Fourier Spectroscopy. Phys. Rev. Lett. 2008, 100, 027403/1−027403/4. (27) Messin, G.; Hermier, J. P.; Giacobino, E.; Desbiolles, P.; Dahan, M. Bunching and Antibunching in the Fluorescence of Semiconductor Nanocrystals. Opt. Lett. 2001, 26, 1891−1893. (28) Sallen, G.; Tribu, A.; Aichele, T.; André, R.; Besombes, L.; Bougerol, C.; Richard, M.; Tatarenko, S.; Kheng, K.; Poizat, J.-Ph. Subnanosecond Spectral Diffusion Measurement Using Photon Correlation. Nat. Photonics 2010, 4, 696−699. (29) Sallen, G.; Tribu, A.; Aichele, T.; André, R.; Besombes, L.; Bougerol, C.; Richard, M.; Tatarenko, S.; Kheng, K.; Poizat, J.-Ph. Subnanosecond Spectral Diffusion of a Single Quantum Dot in a Nanowire. Phys. Rev. B 2011, 84, 041405/1−041405/4. (30) Bradac, C.; Gaebel, T.; Naidoo, N.; Sellars, M. J.; Twamley, J.; Brown, L. J.; Barnard, A. S.; Plakhotnik, T.; Zvyagin, A. V.; Rabeau, J. R. Observation and Control of Blinking Nitrogen-Vacancy Centres in Discrete Nanodiamonds. Nat. Nanotechnol. 2010, 5, 345−349. (31) Morgan, B. L.; Mandel, L. Measurement of Photon Bunching in a Thermal Light Beam. Phys. Rev. Lett. 1966, 16, 1012−1015. (32) Ulrich, S. M.; Strauf, S.; Michler, P.; Bacher, G.; Forchel, A. Triggered polarization-correlated photon pairs from a single CdSe quantum dot. Appl. Phys. Lett. 2003, No. 83, 1848−11850. (33) Strauf, S.; Jahnke, F. Single Quantum Dot Nanolaser. Laser Photonics Rev. 2011, 5, 607−633. (34) Choi, Y.-S.; Rakher, M. T.; Hennessy, K.; Strauf, S.; Badolato, A.; Petroff, P. M.; Bouwmeester, D.; Hu, E. L. Evolution of the Onset of Coherence in a Family of Photonic Crystal Nanolasers. Appl. Phys. Lett. 2007, 91, 031108/1−031108/3. (35) Wang, F.; Dukovic, G.; Brus, L. E.; Heinz, T. F. Time-Resolved Fluorescence of Carbon Nanotubes and Its Implication for Radiative Lifetimes. Phys. Rev. Lett. 2004, 92, 177401. (36) Hagen, A.; Steiner, M.; Raschke, M. B.; Lienau, C.; Hertel, T.; Meixner, A. J.; Hartschuh, A. Exponential Decay Lifetimes of Excitons

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS We thank Svetlana Sukhishvili and Nan Ai for fruitful discussions, David Thielke for helping develop software, and Yashwaant Verma for assistance with sample preparation. Partial financial support was provided by the National Science Foundation CAREER award ECCS/1053537. Metal deposition and spin coating was carried out at the Center for Functional Nanomaterials, Brookhaven National Laboratory, which is supported by the U.S. Department of Energy, Office of Basic Energy Sciences, under contract no. DE-AC02-98CH10886.

■

REFERENCES

(1) Wang, F.; Dukovic, G.; Brus, L. E.; Heinz, T. F. The Optical Resonances in Carbon Nanotubes Arise from Excitons. Science 2005, 308, 838−841. (2) O’ Connell, M. J.; Bachilo, S. M.; Huffman, C. B.; Moore, V. C.; Strano, M. S.; Haroz, E. H.; Rialon, K. L.; Boul, P. J.; Noon, W. H.; Kittrell, C.; et al. Band Gap Fluorescence from Individual SingleWalled Carbon Nanotubes. Science 2002, 297, 593−596. (3) Bachilo, S. M.; Strano, M. S.; Kittrell, C.; Hauge, R. H.; Smalley, R. E.; Weisman, R. B. Structure-Assigned Optical Spectra of SingleWalled Carbon Nanotubes. Science 2002, 298, 2361−2366. (4) Avouris, P.; Freitag, M.; Perebeinos, V. Carbon-Nanotube Photonics and Optoelectronics. Nat. Photonics 2008, 2, 341−350. (5) Misewich, J. A.; Martel, R.; Avouris, Ph.; Tsang, J. C.; Heinze, S.; Tersoff, J. Electrically Induced Optical Emission from a Carbon Nanotube FET. Science 2003, 300, 783−786. (6) Xia, F.; Steiner, M.; Lin, Y; Avouris, P. A Microcavity-Controlled, Current-Driven, On-Chip Nanotube Emitter at Infrared Wavelengths. Nat. Nanotechnol. 2008, 3, 609−613. (7) Högele, A.; Galland, C.; Winger, M.; Imamoğlu, A. Photon Antibunching in the Photoluminescence Spectra of a Single Carbon Nanotube. Phys. Rev. Lett. 2008, 100, 217401/1−217401/4. (8) Strauf, S.; Stolz, N. G.; Rakher, M. T.; Coldren, L. A.; Petroff, P. M.; Bouwmeester, D. High-Frequency Single-Photon Source with Polarization Control. Nat. Photonics 2007, 1, 704−708. (9) Bouwmeester, D.; Ekert, A.; Zeilinger, A. The Physics of Quantum Information; Springer-Verlag: Berlin, Heidelberg, New York, 2000. (10) Santori, C.; Fattal, D.; Vučković, J.; Solomon, G. S.; Yamamoto, Y. Indistinguishable Photons from a Single-Photon Device. Nature 2002, 419, 594−597. (11) Klauder, J. R.; Anderson, P. W. Spectral Diffusion Decay in Spin Resonance Experiments. Phys. Rev. 1962, 125, 912−932. (12) Empedocles, S. A.; Bawendi, M. G. Quantum-Confined Stark Effect in Single CdSe Nanocrystallite Quantum Dots. Science 1997, 278, 2114−2117. (13) Graham, M. W.; Ma, Y.-Z.; Green, A. A.; Hersam, M. C.; Fleming, G. R. Pure Optical Dephasing Dynamics in Semiconducting Single-Walled Carbon Nanotubes. J. Chem. Phys. 2011, 134, 034504/ 1−034504/13. (14) Htoon, H.; O’Connell, M. J.; Cox, P. J.; Doorn, S. K.; Klimov, V. I. Low Temperature Emission Spectra of Individual Single-Walled Carbon Nanotubes: Multiplicity of Subspecies within Single-Species Nanotube Ensembles. Phys. Rev. Lett. 2004, 93, 027401/1−027401/4. (15) Matsuda, K.; Kanemitsu, Y.; Irie, K.; Saiki, T.; Someya, T.; Miyauchi, Y.; Maruyama, S. Photoluminescence Intermittency in an Individual Single-Walled Carbon Nanotube at Room Temperature. Appl. Phys. Lett. 2005, 86, 123116/1−123116/3. (16) Matsuda, K.; Inoue, T.; Murakami, Y.; Maruyama, S.; Kanemitsu, Y. Exciton Fine Structure in a Single Carbon Nanotube 1940

dx.doi.org/10.1021/nl204402v | Nano Lett. 2012, 12, 1934−1941

Nano Letters

Letter

in Individual Single-Walled Carbon Nanotubes. Phys. Rev. Lett. 2005, 95, 197401. (37) Soares, J. S.; Barboza, A. P. M.; Araujo, P. T.; Neto, N. M. B.; Nakabayashi, D.; Shadmi, N.; Yarden, T. S.; Ismach, A.; Geblinger, N.; Joselevich, E.; et al. Modulating the Electronic Properties Along Carbon Nanotubes via Tube-Substrate Interaction. Nano Lett. 2010, 10, 5043−5048. (38) Ma, Y.-Z.; Valkunas, L.; Dexheimer, S. L.; Bachilo, S. M.; Fleming, G. R. Femtosecond Spectroscopy of Optical Excitations in Single-Walled Carbon Nanotubes: Evidence for Exciton-Exciton Annihilation. Phys. Rev. Lett. 2005, 94, 157402. (39) Georgi, C.; Green, A. A.; Hersam, M. C.; Hartschuh, A. Probing Exciton Localization in Single-Walled Carbon Nanotubes Using HighResolution Near-Field Microscopy. ACS Nano 2010, 4, 5914−5920. (40) Galland, C; Gosh, Y.; Steinbrueck, A.; Sykora, M.; Hollingsworth, J. A.; Klimov, V. I.; Htoon, H. Two types of luminescence blinking revealed by spectroelectrochemistry of single quantum dots. Nature 2011, 479, 203−207.

1941

dx.doi.org/10.1021/nl204402v | Nano Lett. 2012, 12, 1934−1941