Nanoscale Imaging of Exciton Transport in Organic Photovoltaic

Sep 14, 2009 - National Renewable Energy Laboratory (NREL), 1617 Cole Boulevard, Golden, Colorado 80401-3393. Nano Lett. , 2009, 9 (11), pp 3904– ...
1 downloads 0 Views 4MB Size
NANO LETTERS

Nanoscale Imaging of Exciton Transport in Organic Photovoltaic Semiconductors by Tip-Enhanced Tunneling Luminescence

2009 Vol. 9, No. 11 3904-3908

Manuel J. Romero,* Anthony J. Morfa, Thomas H. Reilly III., Jao van de Lagemaat, and Mowafak Al-Jassim National Renewable Energy Laboratory (NREL), 1617 Cole BouleVard, Golden, Colorado 80401-3393 Received July 1, 2009

ABSTRACT In organic solar cells, the efficiency of the exciton transport and dissociation across donor-acceptor (D/A) interfaces is controlled by the nanoscale distribution of the donor and acceptor phases. The observation of photoluminescence quenching is often used as confirmation for efficient exciton dissociation but provides no information on the nanoscopic nature of the exciton transport. Here we demonstrate nanoscale imaging of the exciton transport in films consisting of the conjugated polymer poly(3-hexylthiophene) (P3HT, electron donor) blended with the C60 derivative 1-(3-methoxycarbonyl)-propyl-1-phenyl-(6,6)C61 (PCBM, electron acceptor) by a tunneling luminescence spectroscopy based on atomic force microscopy. The excitonic luminescence is significantly enhanced when the conjugated polymer is coupled to the plasmon excitation at the tip (tip-enhanced luminescence). This effect allows one to dramatically improve the detection efficiency of the excitonic luminescence and, consequently, resolve individual domains of the conjugated polymer in which the exciton will recombine before dissociation at the D/A interface. Under thermal annealing conditions promoting the segregation of the donor and acceptor phases, a clear increase of the luminescence is seen from polymer-rich regions, consistent with domains of dimensions much larger than the exciton diffusion length. The described scanning luminescence microscopy can thus be applied to the optimization of the blends used in solar cells.

Photon absorption by a molecule or conjugated polymer in an organic solar cell creates a singlet exciton.1 This exciton diffuses and dissociates at the interface with another organic semiconductor (the donor-acceptor (D/A) interface),2,3 and the resulting electron (in the electron-acceptor phase) and hole (in the donor phase) are collected at their respective electrodes. This is in simple terms the principle of operation behind most organic solar cells,4 in which semiconducting fullerene derivatives are commonly used as the acceptor phase, while the donor properties are supplied by conjugated polymers. The limited exciton diffusion length in conjugated polymers (nanometers to tens of nanometers at best),5 in combination with the requirement of absorbing a significant fraction of the solar spectrum, imposes a number of criteria to the optimum morphology of the organic blends: first, the dimensions of the conjugated polymer domains must be of the order of the exciton diffusion length (tens of nanometers) to avoid the current loss due to exciton recombination; second, the D/A interfacial area must be maximized to increase the probability of exciton dissociation; and third, * Corresponding author: phone, (303) 384-6653; fax, (303) 384-6604; e-mail, [email protected]. 10.1021/nl902105f CCC: $40.75 Published on Web 09/14/2009

 2009 American Chemical Society

the domains of the donor and acceptor phases must be continuously connected to the appropriate electrode to ensure efficient electron (hole) transport.6,7 Following these criteria, record photoconversion efficiencies have been reported for biphase (D/A) organic semiconductors in which chemically homogeneous donor and acceptor regions form a bicontinuous and interpenetrating network with dimensions in the nanoscale.8 The extinction of the donor’s photoluminescence signature (commonly referred to as photoluminescence quenching)9,10 is often used as confirmation for efficient exciton dissociation but provides no information on the nanoscopic nature of the exciton transport. Also, it is very difficult to assess the efficiency of the exciton dissociation from the degree of quenching. It is thus desirable to probe the exciton transport with nanometer resolution. In this contribution we demonstrate nanoscale imaging of the excitonic luminescence in films consisting of the conjugated polymer poly(3-hexylthiophene) (P3HT, electron donor) blended with the C60 derivative 1-(3-methoxycarbonyl)-propyl-1-phenyl-(6,6)C61 (PCBM, electron acceptor). The radiative and nonradiative properties of a molecule (or conjugated polymer) can be controlled by the precise

Figure 1. (a) Schematics of the AFM developed for tunneling luminescence measurements. The oscillation amplitude of the tuning fork (TF) sensor near its resonance frequency (monitored by the current Itf(ωr)) is fed to the Z control of the AFM. A voltage is applied to the metallic tip to create pulses of tunneling current in sync with the TF oscillation. The optics for the detection of the tunneling luminescence consists of a parabolic mirror, focusing optics, spectrograph, and a cryogenic multichannel detector (CCD). (b) Molecular luminescence from the P3HT blended with PCBM is excited by the localized mode of the plasmon excitation (LSP), which is basically an oscillating dipole aligned with and confined to the tunneling gap. A propagating mode (PSP) also exists.

excitation of surface plasmons (SPs) in the local environment of the molecule. This is a result of the more general principle that the fluorescence emission rate of a molecule depends on the distribution of the electromagnetic field in its surroundings (Purcell effect).11 Plasmon-enhanced luminescence is exploited in the detection of single quantum dots12 and other nanoscopic objects such as fluorophores,13 molecules,14 and carbon nanotubes,15 opening the possibility of performing luminescence spectroscopy with unprecedented resolution when a nanoantenna is used for the plasmon excitation. This is the case when a laser illuminated metallic tip (with the local amplification of the electromagnetic field) is used to locally excite luminescence from the specimen in a near-field scanning microscope. Here, we demonstrate nanometer resolution of the excitonic luminescence in P3HT:PCBM blends by a tunneling luminescence spectroscopy based on atomic force microscope (AFM). Force feedback is used instead of current feedback in these experiments because P3HT and PCBM present large dissimilarities in conductivity, causing unstable operation of the STM under constant tunneling current conditions. STM is therefore impracticable for most organic blends used in photovoltaics. We have solved this problem with tunneling luminescence measurements based on AFM instead. The schematic for these measurements is shown in Figure 1a. The critical component is a force sensor consisting of an ultrasharp metallic tip attached to a self-sensing and -actuating piezo tuning fork (TF).16 The TF is driven by an oscillating voltage source near its resonance frequency Vtf(ωr). The interaction between the tip and the specimen causes a reduction of the oscillation amplitude (monitored by the current Itf(ωr)) and a shift in the resonance frequency and phase of the oscillation. We maintain constant the distance Z by controlling the amplitude of the TF oscillation during the scanning of the tip while applying a voltage to the tip to create pulses of tunneling current in synchronization with the TF oscillation. Tunneling current pulses as high as ∼100 nA have been observed without significant degradation of the AFM operation. Nano Lett., Vol. 9, No. 11, 2009

When the supporting substrate of the P3HT:PCBM film is metallic (gold or silver in our case), hot electrons tunneling across the metallic gap can excite plasmon modes. As shown in this Letter, luminescence is mainly observed from the conjugated polymer (P3HT) by coupling with the plasmon mode confined to the tip; basically an oscillating dipole aligned with the tunneling gap (see Figure 1b). The tip therefore assembles a tunable plasmonic resonator in which a cross section of the P3HT:PCBM under the tip is fully immersed. This plasmon-enhanced luminescence allows one to dramatically improve the detection efficiency of the excitonic luminescence and, consequently, resolve regions in which the exciton will recombine before dissociation at the D/A interface. The role of the plasmon excitation in the luminescence is elucidated by comparing substrates with different plasmonic properties. The influence of the annealing temperature on the nanoscopic nature of the exciton transport in P3HT:PCBM will be discussed. We have chosen this polymer-fullerene system because it is one of the more exhaustively investigated for organic photovoltaics. Regioregular P3HT (>93%, EE grade) is synthesized by Rieke Metals Inc. and used as received. Blend solutions of P3HT:PCBM (1:1 w/w) are prepared using o-dichlorobenzene (50 mg/mL). P3HT-PCBM films are then prepared by drop casting this solution on three different substrates: gold, silver, and indium tin oxide (ITO). The substrates are left tilted to dry overnight in order to achieve a gradient in the film thickness from top to bottom. Finally, these films are thermally annealed for 15 min at 90, 120, and 150 °C. For comparison purposes, one of the films is not annealed at all (pristine). The films are preserved in a nitrogen atmosphere prior to the AFM measurements, which are performed under high vacuum (10-6 Torr). The AFM, specifically designed for these measurements, is based on a X-Y-Z nanopositioning platform and is compatible with the TF sensors described above (Figure 1a). For the detection of the tunneling luminescence, the tip is positioned in the focal point of a parabolic mirror providing optimum collection efficiency. The light is guided through 3905

Figure 2. Tunneling luminescence spectroscopy of P3HT-PCBM films on gold substrates. V ) +3 V, It ∼ 25 nA. These tunneling luminescence spectra are selected from spectrum imaging measurements, as described in the text. The luminescence associated to the plasmon excitation (SP) is shown in (a). Molecular luminescence from the conjugated polymer, which is accompanied by the extinction of the SP luminescence, is shown in (b-d). Spectra are shifted for clarity.

a spectrograph equipped with a cryogenic multichannel detector (Roper Scientific Silicon EEV 1340 × 400 CCD). Spectrum imaging combines spectroscopy and imaging in one single measurement by acquiring the emission spectrum (the output of the CCD) in synchronization with the scanning of the tip. After the acquisition is complete, the resulting spectrum series can be processed to reconstruct maps of the photon intensity (resolved in energy), photon energy, or extract the spectrum for a selected area on the AFM image. More details can be found elsewhere.17 First, we need to demonstrate that the plasmon excitation at the tip can stimulate excitonic luminescence from organic semiconductors. Figure 2 shows a series of tunneling luminescence spectra extracted from spectrum imaging (spectrumper-pixel) measurements on P3HT-PCBM films on gold substrates. An external bias of V ) +3 V is applied to the metallic tip, and the set point of the Z feedback (the distance from the tip to the specimen) is adjusted until tunneling current (It) pulses of ∼25 nA are obtained. Tunneling electrons can excite a SP localized at the tip, which can be visualized as an oscillating dipole aligned with the tunneling gap. Such an oscillating dipole can radiate at optical frequencies, in which case, luminescence associated with the plasmon excitation is observed. This is shown in the emission spectrum of Figure 2a. The spectrum reflects the energy distribution of the radiating plasmon (which is basically equivalent to the energy distribution of the tunneling electrons undergoing inelastic scattering) with a maximum energy corresponding to the most energetic tunneling electron transferring all its energy to the excitation of the plasmon mode. The excitation of tunneling luminescence from the organic semiconductors is evidenced by the examination of the 3906

Figure 3. Tunneling luminescence spectra for P3HT-PCBM films deposited on different substrates: gold (a), silver (b), and indium tin oxide (c). The films were annealed at 90 °C for 15 min prior to the measurements. V ) +3 V, It ∼ 25 nA. Spectra are shifted for clarity.

spectrum series, as illustrated by spectra b-d of Figure 2. In this case, not only does the wavelength of the luminescence agree with that obtained from photoluminescence measurements (∼630-640 nm) but this wavelength does not shift at all during the measurement. This suggests that the observed luminescence is molecular in origin and not due to a higher energy mode of the SP excitation; in this case, spectral shifts will be observed due to local fluctuations of the tunneling gap. In addition to this, the excitation of molecular luminescence is accompanied by the extinction of the SP luminescence seen in Figure 2a, which suggests that excitonic luminescence from the organic semiconductors is resulting from the direct coupling with the plasmon excitation. The SP coupling seems to be more effective with the conjugated polymer than with the fullerene (the photon intensity from P3HT is higher than that of PCBM by a factor of 20 or so) as confirmed by tunneling luminescence measurements done on pristine P3HT and PCBM films. P3HT is a much better insulator than PCBM, and that might help in the local confinement of the SP excitation at the tip. In order to substantiate our hypothesis that the molecular luminescence is significantly enhanced when the conjugated polymer is coupled to the plasmon excitation at the tip and, consequently, that the plasmon excitation is indeed responsible for the excitation of the molecular luminescence, we compare different substrates with different plasmonic properties. Figure 3 shows the tunneling luminescence spectra (acquired under similar conditions (V ) +3 V, It ∼ 25 nA)) for P3HT-PCBM films (annealed at 90 °C for 15 min) cast on gold, silver, and ITO substrates. Luminescence from the polymer is observed when using gold substrates (Figure 2b-d) although the overlap with the SP spectrum (Figure 2a) is limited: only the high energy end of the SP spectrum lies near the fundamental transition of the conjugated polymer (see also Figure 3a). Under similar V and It settings in the STM, SPs in silver are more energetic than those of gold meaning that the coupling between the SP and the conjugated Nano Lett., Vol. 9, No. 11, 2009

polymer should be largely improved. This seems to be the case, and we observe a significant enhancement (a factor of 5 or so) of the molecular luminescence when the supporting substrate is silver (Figure 3b). The SP coupling is therefore absolutely critical to the excitation of the excitonic luminescence. The anode of most practical solar cells is made of ITO, and the possibility of using this substrate for our measurements is obviously of great interest. Furthermore, ITO presents a high electron density and is, therefore, plasmonically active. We have succeeded in exciting SP luminescence from ITO, and the overlap with the fundamental transition of the P3HT seems to be adequate but the plasmon excitation is much weaker than that seen in noble metals. In the case of ITO as supporting substrate, we have no evidence for SP-enhanced molecular luminescence (see Figure 3c). Because of the electron tunneling excitation of the SP, these measurements are restricted to “plasmonic-active” supporting substrates and ultrathin organic films (20 nA). On the other hand, nearly the entire cross section of the organic film under the tip is excited, providing information about the exciton transport and recombination from top to bottom, not limited to the surface of the film like in other AFM-based measurements. This tip-enhanced luminescence allows one to dramatically improve the detection efficiency of the excitonic luminescence and, consequently, resolve regions in which the exciton will recombine before dissociation at the D/A interface. Spectrum imaging measurements in which the scanning of the tip is synchronized with the spectrum acquisition can be run at 100-200 ms per spectrum, opening the possibility of performing nanoscale imaging of the exciton transport in organic semiconductors. The total acquisition time of the spectrum series over the AFM image is about 1-2 h. We have investigated the influence of the annealing temperature on the nanoscopic nature of the exciton transport in P3HT-PCBM blends using the tip-enhanced luminescence effect. Figure 4 shows a series of AFM and corresponding photon-intensity images of the molecular luminescence for pristine P3HT-PCBM films and films annealed during 15 min at 90 and 120 °C, respectively, deposited on gold. The horizontal lines along the images are a consequence of the cross-talk between the tunneling current flowing through the tip and the electrodes of the tuning fork. This artifact can be removed from the images by standard probe microscopy image processing, but we chose to not do this for reason of completeness. The observed morphologies for the pristine and T ) 90 °C films are consistent with previous results indicating phase separation on the order of 50-100 nm. Annealing at T ) 120 °C results in a completely different morphology, with a considerable segregation and crystallization of PCBM leading to P3HT-rich regions among the PCBM single crystals, which has been reported previously as well.18 The photon intensity images of the exciton recombination for films other than T ) 120 °C contain features of the order of 20-30 nm. This is well below the phase separation of the order of 50-100 nm that the AFM seems to suggest (see the Nano Lett., Vol. 9, No. 11, 2009

Figure 4. AFM and corresponding photon intensity images of the molecular luminescence for P3HT-PCBM films deposited on gold and annealed at different temperatures. V ) +3 V, It ∼ 100 nA. The observed morphologies are consistent with previous results indicating phase separation of the order of 50-100 nm. The photon intensity images can resolve features of the order of 20-30 nm.

scale bars on Figure 4). Because the diffusion length is in the tens of nanometers at best and these domains present dimensions much larger that the projected exciton diffusion length, exciton recombination and subsequent photon emission (SP-enhanced) are expected when the exciton is created further than one diffusion length from the interface, and this seems to be the case. The photon intensity image thus reproduces the distribution of the polymer domains within the film and can be used to optimize the morphology. Domains of the order of the diffusion length with efficient dissociation at their interfaces will not luminesce. Inefficient exciton dissociation will cause an increase in the exciton recombination rate and possibly an increase in the intensity of the luminescence (depending on the contribution of the nonradiative recombination at the interface). Furthermore, the local exciton diffusion length can be estimated when comparing the dimensions of one particular feature on the tunneling luminescence image with the dimensions of the associated P3HT domain obtained from the AFM phase image.19 Annealing at T ) 90 °C increases the overall photon intensity (increased recombination) and promotes the segregation of regions with different exciton recombination in 3907

correspondence with the increase in the fraction of the polymer, as seen in the previous case. In summary, we have observed excitonic luminescence from P3HT blended with PCBM in tunneling luminescence measurements based on AFM. The SP excitation confined to the metallic tunneling gap at the tip is responsible for the large enhancements in the intensity of the molecular luminescence. This effect allows for a dramatic improvement in the detection sensitivity and, in combination with AFM, the acquisition of photon intensity images of the excitonic luminescence with nanometer resolution. For a series of P3HT-PCBM annealed at different temperatures, we have found an excellent correlation between the nanoscale morphology and the predicted exciton transport properties reflected in the photon intensity images of the molecular luminescence. This scanning luminescence microscopy can thus be applied to the investigation of the exciton transport and dissociation across donor-acceptor interfaces, all with nanometer resolution.

Figure 5. Similar measurements to those shown in Figure 4 but for P3HT-PCBM films deposited on silver substrates. V ) +3 V, It ∼ 25 nA.

Acknowledgment. This work was funded by the Photochemistry and Radiation Research Program of the U.S. Department of Energy, Office of Science, Basic Energy Sciences, Division of Chemical Sciences, Geosciences and Biosciences, under Contracts DE-AC36-99GO10337 and DE-AC36-08-GO28308. References

the 200-500 nm range (much larger than individual domains) (see A and B regions on the corresponding image of Figure 4: the overall recombination in A is higher than in B). This must be correlated with changes in the nanoscale morphology of the P3HT:PCBM blend, and the beginning of the segregation of P3HT and PCBM in larger domains. When increasing the annealing temperature to T ) 120 °C, the P3HT-rich region outlined on the AFM image does luminescence very efficiently (see Figure 4). This is consistent with the loss of D/A interface and the obvious increase in the fraction of the polymer. We see a good agreement between the exciton transport properties predicted from the nanoscale morphology and the photon intensity images of the molecular luminescence. Results for P3HT-PCBM films undergoing similar processing but deposited on silver are shown in Figure 5. In this case, as we described above, we benefit from larger SP enhancements and the improvement in the signal-to-noise ratio of the tunneling luminescence measurements. The morphology of these films is very similar to those deposited on gold substrates with the distinction of the film annealed at T ) 120 °C, in which the PCBM single crystals are smaller than and not as elongated as those seen on gold. No molecular luminescence was detected for pristine blends, an indication of the high degree of dispersion of the P3HT and PCBM achieved during the preparation of the films. Annealing at T ) 90 °C results in the detection of luminescence from the P3HT domains, most probably because of the coarsening of the nanoscale morphology, although their distribution seems to be more uniform that in the previous case of gold. Finally, for the PH3T-rich regions in the film annealed at T ) 120 °C, the distribution of the excitonic luminescence is very much spread out over the image, in 3908

(1) Arkhipov, V. I.; Ba¨ssler, H. Phys. Status Solidi A 2004, 201, 1152. (2) Rand, B. P.; Burk, D. P.; Forrest, S. R. Phys. ReV. B 2007, 75, 115327. (3) Scharber, M.; Mu¨hlbacher, D.; Koppe, M.; Denk, P.; Waldauf, C.; Heeger, A.; Brabec, C. AdV. Mater. 2006, 18, 789. (4) Coakley, K. M.; McGehee, M. D. Chem. Mater. 2004, 16, 4533–4542. (5) Scully, S. R.; McGehee, M. D. J. Appl. Phys. 2006, 100, 034907. (6) Yang, F.; Shtein, M.; Forrest, S. R. Nat. Mater. 2005, 4, 37–41. (7) Ma, W.; Yang, C.; Gong, X.; Lee, K.; Heeger, A. J. AdV. Funct. Mater. 2005, 15, 1617–1622. (8) Shaheen, S. E.; Brabec, C. J.; Sariciftci, N. S.; Padinger, F.; Fromherz, T.; Hummelen, J. C. Appl. Phys. Lett. 2001, 78, 841. (9) Haugeneder, A.; Neges, M.; Kallinger, C.; Spirkl, W.; Lemmer, U.; Feldmann, J.; Scherf, U.; Harth, E.; Gu¨gel, A.; Mu¨llen, K. Phys. ReV. B 1999, 59, 15346. (10) Peumans, P.; Yakimov, A.; Forrest., S. R. J. Appl. Phys. 2003, 93, 3693. (11) Purcell, E. M. Phys. ReV. 1946, 69, 681. (12) Farahani, J. N.; Pohl, D. W.; Eisler, J.-J.; Hecht, B. Phys. ReV. Lett. 2005, 95, 017402. (13) Anger, P.; Bharadwaj, P.; Novotny, L. Phys. ReV. Lett. 2006, 113002 (2006), 96. (14) Rossel, F.; Pivetta, M.; Patthey, F.; Schneider, W.-D. Opt. Express 2009, 17, 2714. (15) Hartschuh, A.; Qian, H.; Meixner, A. J.; Anderson, N.; Novotny, L. Nano Lett. 2005, 5, 2310. (16) A wire of 25 mm in diameter (either of gold or PtIr) is attached to the tuning fork. A sharp tip is first obtained electrochemically and is then ultrasharpened by focused ion beam (FIB) until the desired geometry is obtained. The main criterion to be satisfied is that the tip does not eclipse the luminescence from the collection optics. (17) Romero, M. J.; van de Lagemaat, J.; Mora-Sero, I.; Rumbles, G.; AlJassim, M. M. Nano Lett. 2006, 6, 2833. (18) Savenije, T. J.; Kroeze, J. E.; Yang, X.; Loos, J. AdV. Funct. Mater. 2005, 15, 1260. (19) In the ultrathin films used here (∼20 nm in thickness and with domains in the 50-100 nm range) the observed AFM morphology can be extrapolated to the entire cross section of the P3HT:PCBM blend. If the diameter of one particular domain is estimated to be 100 nm from the AFM phase image and the corresponding feature on the photon intensity image is 50 nm, then the exciton diffusion length Lx is Lx ) (100 nm50 nm)/2 ) 25 nm. Therefore, quantification of the diffusion length is possible using this scanning tunneling luminescence microscopy.

NL902105F Nano Lett., Vol. 9, No. 11, 2009