Letter pubs.acs.org/JPCL
Probing Exciton Diffusion and Dissociation in Single-Walled Carbon Nanotube−C60 Heterojunctions Anne-Marie Dowgiallo,† Kevin S. Mistry,†,‡ Justin C. Johnson,† Obadiah G. Reid,†,§ and Jeffrey L. Blackburn*,† †
National Renewable Energy Laboratory, Golden, Colorado 80401, United States Physics Department and §Renewable and Sustainable Energy Institute, University of Colorado, Boulder, Colorado 80309, United States
‡
S Supporting Information *
ABSTRACT: The efficiency of thin-film organic photovoltaic (OPV) devices relies heavily upon the transport of excitons to type-II heterojunction interfaces, where there is sufficient driving force for exciton dissociation and ultimately the formation of charge carriers. Semiconducting single-walled carbon nanotubes (SWCNTs) are strong near-infrared absorbers that form type-II heterojunctions with fullerenes such as C60. Although the efficiencies of SWCNT−fullerene OPV devices have climbed over the past few years, questions remain regarding the fundamental factors that currently limit their performance. In this study, we determine the exciton diffusion length in the C60 layer of SWCNT−C60 bilayer active layers using femtosecond transient absorption measurements. We demonstrate that hole transfer from photoexcited C60 molecules to SWCNTs can be tracked by the growth of narrow spectroscopic signatures of holes in the SWCNT “reporter layer”. In bilayers with thick C60 layers, the SWCNT charge-related signatures display a slow rise over hundreds of picoseconds, reflecting exciton diffusion through the C60 layer to the interface. A model based on exciton diffusion with a Beer−Lambert excitation profile, as well as Monte Carlo simulations, gives the best fit to the data as a function of C60 layer thickness using an exciton diffusion length of approximately 5 nm. rganic thin film photovoltaic devices comprised of singlewalled carbon nanotubes (SWCNTs) and fullerenes show promise for efficient solar capture over the visible and near-infrared (NIR).1−3 Devices consisting of a thin film of semiconducting SWCNTs with an overlying C60 layer have demonstrated AM1.5 power conversion efficiencies of ∼1%.4,5 Because free charges are not directly created in these “excitonic” solar cells, the diffusion of excitons in each phase to the interface where they are dissociated is an important process.5−8 Thus, recent studies have shown that bulk heterojunction geometries can increase power conversion efficiencies to the range of 2−3% by reducing the exciton transit distance.9,10 Even in bulk heterojunction devices, where excitons should travel shorter distances to reach a dissociation interface, knowledge of the exciton diffusion lengths provides crucial insight into appropriate morphologies for device optimization. At a more fundamental level, the exciton diffusion length is a critical property of any excitonic semiconductor that determines the kinetic constraints inherent in converting photons to useful work within macroscopic films.11,12 Exciton diffusion lengths in both SWCNTs and fullerene films have been previously investigated.13−15 The singlet exciton diffusion length in thin evaporated C60 films was determined to be 7 nm for a Zn-phthalocyanine/C60 bilayer film using time-resolved microwave conductivity (TRMC).13 The authors attribute this short diffusion length to poor crystallinity of the C60 film. In addition, C60 exciton diffusion
O
© XXXX American Chemical Society
lengths of 19 nm14 and 7.7 nm15 were reported for PEOPT/ C60 and CuPc/C60 bilayer films, respectively. Discrepancies in these reported values might be due to several other important factors including material purity, film crystallinity, and the efficiency of exciton dissociation at the donor−acceptor interface.13 Hence, it is imperative to characterize the exciton diffusion length for each specific donor−acceptor pair being considered. In contrast with C60 films, the exciton diffusion length along the longitudinal axis of isolated SWCNTs can be hundreds of nanometers.16,17 Nevertheless, in thin-films exciton diffusion is predominantly in the transverse direction between neighboring tubes, and LD is small, typically < 10 nm for films with residual polyfluorene.6,7,18 Importantly, all of these previous studies used techniques that cannot follow the exciton diffusion process in real time, but instead rely upon a reporter (photocurrent, microwave absorption) that reflects the charge carrier density resulting from exciton diffusion and dissociation. In the current study, we utilize femtosecond transient absorption (TA) spectroscopy to track exciton diffusion and dissociation in real time in a model SWCNT−C60 donor− acceptor bilayer system. To avoid any spectral congestion caused by polychiral SWCNT samples, we focused on monochiral (6,5) and (7,5) SWCNT dispersions to determine Received: March 15, 2016 Accepted: April 25, 2016
1794
DOI: 10.1021/acs.jpclett.6b00604 J. Phys. Chem. Lett. 2016, 7, 1794−1799
Letter
The Journal of Physical Chemistry Letters
Figure 1. (a) Linear absorption spectra for the neat (6,5) SWCNT film, and (6,5) SWCNT−C60 bilayer films with C60 thicknesses of 6, 16, 42, 51, and 71 nm. The blue arrow indicates the 450 nm pump wavelength utilized to selectively excite C60 to probe exciton diffusion within this layer. (b) Schematic of trion formation resulting from photoinduced hole transfer (PHT), following photoexcitation of the C60 layer at 450 nm. The important processes (1−4) leading to signal in the TA measurement are detailed on the right side of the panel.
Figure 2. (a) Transient absorption spectra for neat (6,5) SWCNT and SWCNT−C60 bilayer films at long delay times (∼300 ps) after excitation at 450 nm. Photoinduced hole transfer from C60 to the SWCNT layer results in a long-lived S11 bleach and an induced absorption at 1174 nm corresponding to the trion transition, X+. (b, c) Transient absorption dynamics for neat and bilayer films probed at the position of either the S11 bleach (b) or X+ induced absorption (c). All transients in panels b and c are normalized at a pump−probe delay where the transients for the neat films have reached their maximum values. As labeled in panels b and c, the slow rise of the signal in the bilayer films (first ∼150 ps) results from exciton diffusion to the interface, followed by exciton dissociation. Then, the decay of each signal beyond ∼150 ps corresponds to charge recombination.
used to track the charges that result from exciton dissociation in real time, with a slow rise of signal intensity reflecting exciton diffusion within the C60 phase to the interface. We simulate both the magnitude and temporal dynamics of SWCNT charge generation by modeling exciton diffusion in the C60 layer of (6,5) SWCNT−C60 bilayer films, and extract an exciton diffusion length by fitting our model to the experimental data. Both strategies yield an exciton diffusion length of 5 nm in the C60 layer that is in line with several recent estimates. For this study, we prepared nearly monochiral (6,5) and (7,5) SWCNT dispersions using fluorene-based polymers, poly[(9,9-dioctylfluorenyl-2,7-diyl)-alt-co-(6,60-[2,20-bipyridine])] (PFO-BPy) and PFO, respectively. Thin films of (6,5) and (7,5) SWCNTs (∼10 nm) were prepared with and without an overlying C60 layer of varying thicknesses: approximately 6, 16, 42, 51, and 71 nm, on quartz substrates. The approximate thickness of each layer was determined using the optical density at the S11 transition for the (6,5) and (7,5) SWCNTs5 and the absorption coefficient for C60, 1.21 × 105 cm−1 (at ca. 450 nm).26 We focus on the results for the (6,5) bilayers here, but as shown in the Supporting Information, the results are similar for (7,5) bilayers (Figures S1−S3). A neat (6,5) SWCNT film exhibits strong absorption at the first excitonic transition (S11) at 1000 nm, and other notable features include the X1 phonon sideband at 800 nm27 and second excitonic transition (S22) at
the diffusion length of photogenerated C60 excitons. As we previously demonstrated for an interface between semiconducting SWCNTs and a perovskite absorber layer,19 the sharp optical transitions of monochiral s-SWCNT layers can serve as sensitive reporters for exciton diffusion in this model system. The SWCNT−C60 bilayer configuration forms a type-II heterojunction where there is sufficient driving force for electron transfer from the donor species, SWCNTs, to the acceptor molecules, C60.20−22 Excitons that are generated upon photoexcitation of the first or second SWCNT exciton transitions (S 11 and S 22 , respectively) diffuse to the SWCNT−C60 interface where exciton dissociation occurs5,23 because of ultrafast electron transfer.20 On the other hand, selective excitation of the C60 layer using pump pulses that are resonant with C60 excitonic transitions can also lead to exciton dissociation due to photoinduced hole transfer (PHT).5 In PHT, excitons are first generated in the C60 layer and then diffuse to the interface, where the energetic driving force established by the type-II heterojunction drives hole transfer from C60 to the SWCNT. Our recent study20 demonstrated that SWCNT holes resulting from the ultrafast exciton dissociation process can bind to excitons from the probe pulse to form a trion24,25 with an associated induced absorbance (IA) peak that is red-shifted from the S11 by ∼170 meV. In the current study, both the S11 exciton bleach and the trion IA are 1795
DOI: 10.1021/acs.jpclett.6b00604 J. Phys. Chem. Lett. 2016, 7, 1794−1799
Letter
The Journal of Physical Chemistry Letters
this initial rapid decay, the S11 bleach rises very slowly (over ca. 150 ps) for the bilayer, deviating dramatically from the transient observed for the neat (6,5) film. This deviation reflects the contribution of a new species to the SWCNT−C60 bleach dynamics, namely, holes created by diffusion-limited interfacial exciton dissociation where the slow rise reflects exciton diffusion through the relatively thick C60 layer. The temporal evolution of the trion IA at 1174 nm (Figure 2c) provides similar information, with a slow rise observed over the course of ca. 150 ps for the bilayer sample. The slow rise of both signals demonstrates that the sharp TA signals of the s-SWCNT layer can be used as a reporter to follow exciton diffusion in the C60 layer in real time. To model exciton diffusion within the C60 layer, we can use two strategies that consider either the thickness-dependent magnitude or transient dynamics of charge-related spectral features. The former strategy is similar to that employed in recent microwave conductivity and device studies.12−15 The latter strategy capitalizes upon the time resolution of our TA measurements that, unlike the TRMC and device studies, allow us to follow the exciton diffusion process in real time. Considering the thickness-dependent magnitude of charges resulting from exciton diffusion, we first consider the method described by Kroeze et al.30 In this case, the fraction of photons that enter the film and create excitons that can reach the interface by diffusion, SFS (denoted as such to represent frontside (FS) illumination), can be modeled by the following equation based on a Beer−Lambert excitation profile:
574 nm (Figure 1a). Upon deposition of the C60 layer, a red shift of the S11 occurs and a peak at 442 nm appears, corresponding to lowest-energy allowed exciton transition of the C60 layer (Figure 1a). The red-shifted S11 can be attributed to an increase in the local dielectric environment due to the presence of C60.5,20 Figure 1b displays a schematic representation of the pump− probe measurement employed here to probe the exciton diffusion dynamics within the C60 layer. The bilayer films, having a constant (6,5) SWCNT thickness of ∼10 nm and C60 thickness ranging from 6 to 71 nm, were excited at 450 nm to primarily excite C60 molecules. The exciton population generated in the C60 layer, with spatial distribution determined by the Beer−Lambert law, then diffuses to the interface where exciton dissociation occurs via hole transfer to the SWCNT layer. We use two spectroscopic signatures to temporally track the arrival of these holes in the SWCNT layer, either the bleach of the S11 transition or an induced absorption generated by the probe-induced formation of trions. In the main text of the Letter, we discuss experiments in which the 450 nm pump pulse excites the bilayers through the C60 layer first (termed “frontside illumination” here). Experiments performed with “backside illumination” produced similar results, as shown in Figure S3. Figure 2 compares TA spectra and dynamics for (6,5) films with and without a ca. 71 nm overlayer of C60, following selective excitation of C60 at 450 nm. In Figure 2a, the TA spectrum at a pump−probe delay of 300 ps for the (6,5) SWCNT−C60 bilayer displays a prominent S11 bleach (ca. 1000 nm) and an IA at 1180 nm corresponding to the trion transition, X+. In contrast, the S11 bleach for the neat film has almost completely decayed by 300 ps, and no trion IA can be observed. These results are similar to what we observed recently (see also Figure S1) for selective excitation of the sSWCNT layer, where charges are generated by PET when the C60 acceptor layer is present.20 In both cases (PET and PHT), the charge-separated state involves a hole on the SWCNT phase and an electron in the C60 phase, and the spatial separation of the two charge carriers typically makes this charge-separated state long-lived (see also Figure S2).20−22 The small S11 bleach observed in the neat film in Figure 2a arises from an unavoidable low level of “off-resonance” excitation of the (6,5) s-SWCNTs at 450 nm. While the oscillator strength for the dominant exciton transitions (polarized parallel to the SWCNT axis, and either direct or phonon-assisted) is low at 450 nm, there is nonzero oscillator strength for continuum (nonexcitonic)28 and cross-polarized (perpendicular to the SWCNT axis) exciton transitions.29 Importantly, however, the short-lived S11 bleach and the lack of a trion IA indicate that charge generation is negligible in the neat film. For the bilayer with the 71 nm C60 layer, the S11 bleach dynamics (green trace, Figure 2b) are characterized by an initial fast exciton decay in the first few picoseconds that is then followed by a slow rise up to ca. 150 ps and a slow decay beyond ca. 150 ps. The initial fast exciton decay is similar to the transient obtained when pumping the neat (6,5) film at 450 nm (black trace, Figure 2b), where directly generated SWCNT excitons should dominate the signal. In contrast to the trion peak, which only reflects the presence of charges in the SWCNT layer, the S11 exciton bleach reflects occupation of either exciton or hole levels. Thus, the fast decay in Figure 2b results from the non-negligible contribution from directly generated SWCNT excitons in all of the bilayer films. Following
SFS
d ⎛ ⎛ tanh L ⎞⎞⎟ ⎛ α 2L 2 ⎞⎜ ⎛ d ⎞ −αL ⎜ D D ⎟ ⎟⎜sech⎜ ⎟ − e ⎜1 + =⎜ 2 2 α L D ⎟⎟⎟ ⎝ LD ⎠ ⎝ α L D − 1 ⎠⎜ ⎝ ⎠⎠ ⎝
(1)
In eq 1, α is the linear absorption coefficient of C60, LD the exciton diffusion length, and d the layer thickness.30 Figure 3a plots the peak intensities of both the S11 bleach and trion IA following FS excitation at 450 nm. As discussed above, pumping at 450 nm unavoidably excites a small excited-state population in SWCNTs as well as C60; hence, there is a nonzero contribution from SWCNT excitons in all of the bilayer films. To correct the bleach data for this contribution, we took the raw kinetics traces for the (6,5) SWCNT−C60 bilayers (probed at the bleach at 1000 nm, signal from both excitons and charges) and subtracted the contribution from the neat (6,5) SWCNT film (signal from only excitons) that was pumped at 450 nm using the same fluence of 3 × 1012 photons pulse−1 cm−2. This method resulted in a corrected S11 intensity, where the corrected data now represent only the kinetics of the charges created by interfacial exciton dissociation (see Figure S4).31 The best fit to our data was achieved using an exciton diffusion length of 5 nm (black trace in Figure 3a). We note that the trion IA magnitude for the 5 nm C60 thickness is likely artificially inflated because of the nonzero SWCNT contribution discussed above. We adopted a second strategy, based on a Monte Carlo (MC) simulation of both the magnitude and transient dynamics of the TA data, to model exciton diffusion in the C60 layer. The C60 exciton lifetime (τ) is needed as an input parameter for the MC simulations, because the diffusion constant (D) is tied to both LD and the exciton lifetime by the equation D = LD2/τ. In the Supporting Information, we fit the kinetics of the main peak observed in the TA spectrum (545 1796
DOI: 10.1021/acs.jpclett.6b00604 J. Phys. Chem. Lett. 2016, 7, 1794−1799
Letter
The Journal of Physical Chemistry Letters
trion IA decay. There is a spike at the beginning of the corrected transient because the decay of the S11 at early times is faster in the bilayer film compared to the neat film because of photoinduced electron transfer from the “off-resonance” excited SWCNTs (Figure S7). The gray trace in Figure 3b, corresponding to the MC simulation, does not fit this positive ΔA spike in the corrected data but shows a good fit to the remainder of the data at longer times (see also Figure S8). As demonstrated above, the ability to track the diffusion process in real time enables two strategies for approximating the exciton diffusion length in the C60 layer, using either the magnitude or the temporal dynamics of the SWCNT reporter peaks. Both methods arrive at a similar C60 exciton diffusion length of approximately 5 nm, providing an internal consistency that goes beyond experiments that are limited (due to time resolution) to measuring only the magnitude of charges following photoexcitation.11,13−15 The large oscillator strengths of the sharp SWCNT optical transitions32 enable relatively high signal-to-noise ratios for the SWCNT radical cations (in this experiment) or anions. While spectro-electrochemical measurements have demonstrated that C60 has a photoinduced absorption in the range of 1000 nm for the radical anion,33 this transition has a very low oscillator strength. In fact, we do not observe any signal that could be attributed to the formation of C60−, making the strong and narrow SWCNT optical transitions imperative for this study. The magnitude of the C60 exciton diffusion length found here, ca. 5 nm, is on the low end of estimates determined in a number of recent studies (7−19 nm),5,13−15 although we note that two of these studies find LD values in the range of 7 nm,13,15 commensurate with our finding of 5 nm. Our own recent study successfully modeled the quantum efficiency of similarly prepared SWCNT−C60 devices (based on (7,5) sSWCNTs) by utilizing an upper limit of ca. 15 nm for the C60 exciton diffusion length.5 These relatively low values have a number of implications for bilayer and bulk heterojunction organic photovoltaic devices utilizing C60. It is first important to note that the relatively short exciton diffusion lengths measured here, and the exciton diffusion times of a few hundred picoseconds for all of the studied bilayers, are consistent with singlet excitons being the main contributor to charge carrier formation following C60 excitation.13 It is possible that the short exciton diffusion lengths measured here may reflect poor crystallinity of the evaporated C60 layer. Thus, it may be important for future studies (and bilayer PV devices) to attempt to improve the crystallinity within this phase, e.g. by sublimation purification of the C60 source material. However, it is also important to consider that C60 has both a low luminescence quantum yield and very low absorption coefficient at the emission energy (ca. 725 nm).34 As such, the contribution of Förster resonant energy transfer in films of C60 should be negligible, and the primary exciton transport mechanism is likely Dexter transfer. Because collisional Dexter transfer is inherently short-range, it is reasonable to expect short exciton diffusion lengths in C60, even in well-ordered films. Finally, it is also possible that the higher fluences required for pump−probe measurements such as those reported here and elsewhere13 result in an appreciable reduction in the exciton lifetime within the C60 layer due to nonlinear loss processes such as exciton−exciton annihilation. Other studies have reported lifetimes in the range of 650 ps for C60 in solution.35 In this case, the exciton diffusion length reported here would be a lower limit, and slightly longer LD may be
Figure 3. (a) Dependence of maximum differential absorption at S11 exciton (diamonds) and X+ trion (circles) on C60 thickness for FS illumination at 450 nm. The calculated dependence based on eq 1 is shown as a black line, and Monte Carlo simulations are shown as a gray line, assuming an exciton diffusion length (LD) of 5 nm. (b) Corrected transient absorption dynamics for the (6,5) SWCNT-70 nm C60 bilayer film (green points) with a Monte Carlo simulation (gray trace), assuming an exciton diffusion length of 5 nm. The orange and blue shaded regions denote the regions of the dynamics dominated by exciton diffusion and dissociation (orange) or charge recombination (blue).
nm) of a neat C60 layer to estimate the C60 exciton lifetime (Figure S5). This analysis required a triexponential decay with an average lifetime of ca. 40 ps. We applied our MC simulations to the data in Figure 3a to determine the magnitude of the exciton population reaching the SWCNT−C60 interface after FS illumination (gray trace). The MC simulation shown utilizes an exciton diffusion length of 5 nm and overlaps well with the simulation based on eq 1. Longer exciton diffusion lengths were also tested, but they did not fit the data as well (Figure S6). The good agreement between both the MC simulations and eq 1 suggests that exciton diffusion through these evaporated C60 layer is well described by a relatively short exciton diffusion length of 5 nm, in line with previous estimates of ∼7 nm for similar evaporated thin films.13,15 Next, we extracted the time evolution of the exciton population at the interface. The TA signal at 1000 nm for the (6,5) SWCNT−C60 bilayer with ca. 71 nm C60 is shown as an example (green points, Figure 3b), where the data have been corrected to account for the contribution from SWCNT excitation (see above). The MC simulation shown (gray trace) uses an exciton diffusion length of 5 nm and an average exciton lifetime of 40 ps in C60 (Figure S5) to simulate exciton arrival at the interface, and these dynamics are convoluted with a charge decay rate of 10 ns as determined by an independent fit of the 1797
DOI: 10.1021/acs.jpclett.6b00604 J. Phys. Chem. Lett. 2016, 7, 1794−1799
Letter
The Journal of Physical Chemistry Letters
and a deposition rate of 1 Å/s. A thin film of 72 nm was deposited on the (7,5) SWCNT film. To ensure the SWCNT and SWCNT−C60 films were in an inert atmosphere during the transient absorption experiments, the quartz slide containing the samples were introduced into a helium glovebox, and another quartz slide was sealed on top of the sample using a polymer film (Surlyn, Solaronix) heated to 90 °C. The polymer was cut into a hollow frame so that it only sealed the outer edges of the slide and did not interfere with the sample. Steady-state absorbance measurements were carried out on a Varian Cary 500 spectrophotometer. Femtosecond pump− probe TA experiments were performed on a 1 kHz regeneratively amplified Ti:sapphire laser system that produces 4 mJ laser pulses at 800 nm. The Ti:sapphire laser pumps an optical parametric amplifier (OPA) to generate 450, 1000, and 1050 nm light, which was chopped at a rate of 500 Hz and used as the excitation pump pulse. The fluence for each pump wavelength was 3 × 1012 photons pulse−1 cm−2. Near-infrared (800 nm < λprobe < 1700 nm) and visible (400 nm < λprobe < 800 nm) continuum probe pulses were generated by passing a portion of the amplified 800 nm light through a sapphire plate. The probe pulse was delayed in time with respect to the pump pulse using a motorized translation stage mounted with a retroreflecting mirror. The pump and probe pulse were spatially overlapped at the quartz slide, and the sample was excited either through the nanotube or C60 layer first. Typical averaging times for data collection were 5−8 s, at each pump−probe delay, to achieve high signal-to-noise ratios. The instrument response function was 117 ± 5 fs. The measurements were carried out from 5 ps before time zero to 5.6 ns after time zero.
realized at solar fluences.5,14,15 The low values found for LD, in both SWCNT and C60 thin films, suggest that SWCNT/ fullerene device efficiencies will likely benefit from bulk heterojunction geometries, as demonstrated in recent studies. 9,10 However, the low fill factor of these initial demonstrations of bulk heterojunctions9,10 suggests that the source of significant recombination losses should be explored and mitigated.22 In this study, we utilized femtosecond transient absorption spectroscopy and kinetic modeling to determine a singlet exciton diffusion length of approximately 5 nm in thin films of C60. Using sharp spectroscopic signatures of a monochiral sSWCNT reporter layer, namely, the S11 exciton bleach and trion induced absorbance, we were able to track exciton diffusion in real time by observing the slow increase in signal intensity over the first few hundred picoseconds following C60 photoexcitation. The trion spectral signature is an especially powerful spectroscopic tool because it unambiguously indicates the presence of charge carriers in the SWCNT. The ability to track the diffusion process in real time enables two strategies for approximating the exciton diffusion length in the C60 layer, using either the magnitude or the temporal dynamics of the SWCNT reporter peaks. Both methods arrive at a similar C60 exciton diffusion length, providing an internal consistency that is not possible for experiments that are limited (due to time resolution) to measuring only the magnitude of charges following photoexcitation. For photovoltaic applications, higher power conversion efficiencies will rely upon increasing the exciton diffusion length through these layers and/or optimization of the film thicknesses and device geometries based on both the exciton diffusion length and optical field distribution within a particular device.
■
■
ASSOCIATED CONTENT
* Supporting Information S
EXPERIMENTAL SECTION CoMoCat SG65i SWCNTs were purchased from Southwest Nanotechnologies, Inc., and PFO-BPy and PFO were purchased from American Dye Source. SWCNTs were dispersed in ∼2.5 mg/mL polyfluene polymer (PFO-BPy to isolate the (6,5) chirality, PFO for the (7,5)) in toluene through tip sonication for 30 min (Cole-Palmer CPX 750) in a bath of cool (18 °C) flowing water, followed by ultracentrifugation for 5 min using an SW32Ti motor (Beckman) at 13 200 rpm and 20 °C. The supernatant was retained and contained predominantly (6,5) or (7,5) SWCNTs. To remove excess polymer, the supernatant containing the chirality-sorted SWCNTs was centrifuged for 20 h at 24,100 rpm and 0 °C. In this case, the resulting supernatant was discarded and the pellet containing (6,5) or (7,5) SWCNT material was redispersed in toluene. This polymer removal process was repeated until a 1:1 SWCNT:PFO-BPy mass ratio was obtained. Next, quartz substrates (22 × 10 × 1 mm3) were cleaned by bath sonication in separate solutions of toluene and acetone and then exposed to oxygen plasma for 5 min. (6,5) and (7,5) SWCNT thin films ∼10 nm thick were prepared through ultrasonic spray deposition onto the quartz substrates using a dispersion flow rate of 0.25 mL/min and gas flow rate of 7.0 std L/min. The nozzle power was fixed at 0.8 W, and the substrate was heated to 130 °C to allow for evaporation of the solvent. After spraying the films, they were soaked in a hot toluene bath (80 °C) to remove some residual polymer and better couple SWCNTs within the film. A thin layer of C60 having thicknesses of 6, 16, 42, 51, and 71 nm were deposited on (6,5) SWCNT films via thermal evaporation at a pressure of
