LETTER pubs.acs.org/NanoLett
Spontaneous Formation of Bulk Heterojunction Nanostructures: Multiple Routes to Equivalent Morphologies Ji Sun Moon, Christopher J. Takacs, Yanming Sun,* and Alan J. Heeger* Center for Polymers and Organic Solids, University of California at Santa Barbara, Santa Barbara, California 93106, United States
bS Supporting Information ABSTRACT: Bulk heterojunction (BHJ) layers based on poly(3-hexylthiophene-2,5-diyl) (P3HT) and [6,6]-phenyl C61 butyric acid methyl ester (PCBM) were fabricated by two methods: codeposition of P3HT/PCBM from a common solvent (conventional BHJ) and by sequential, layer-by-layer deposition of P3HT/PCBM from separate solvents (layer-evolved BHJ). Thermally annealed layer-evolved BHJ solar cells show power conversion efficiencies and electron/hole mobilities comparable to conventional BHJ solar cells. The nanomorphology of both active layers is compared in situ by transmission electron microscopy (TEM) using a multilayer cross-sectional sample architecture. No significant difference is observed between the nanomorphology of the conventional BHJ and layer-evolved BHJ material implying that the bulk heterojunction forms spontaneously and that it is the lowest energy state of the two component system. KEYWORDS: Organic solar cell, polymer, bulk heterojunction, cross-section transmission electron microscopy, nanomorphology
A
s a promising alternative to inorganic solar cells, organic photovoltaic (OPV) devices have developed rapidly during recent years.1-8 OPV devices have been fabricated either in the form of a donor, acceptor bilayer heterojunction or a phase separated bulk heterojunction (BHJ) of the two components. The first bilayer heterojunction solar cell was reported by Tang in 1986.2 In a bilayer heterojunction structure, the pure donor thin film is deposited on top of the pure acceptor thin film (or vice versa). The planar bilayer heterojunction does not provide a pathway to high power conversion efficiency (PCE) however because photons absorbed far from the planar heterojunction interface do not yield charge separated carriers. Tang’s bilayer heterojunction exhibited a PCE of about 1%.2 Bulk heterojunction solar cells were developed to increase the interfacial area between the donor and acceptor and to reduce the distance required for photoexcitations to reach a charge separating interface. In typical BHJ devices, a solution of conjugated polymer mixed with a fullerene derivative is spin-cast and, as the solvent evaporates, nanoscale-percolated networks form spontaneously throughout the film.5 Hence, charge separation can take place throughout the active layer and the photocurrent is correspondingly increased. PCEs in the range of 6-8% have been reported for polymer-based BHJ solar cells.9-15 Recently, Ayzner et al. reported the fabrication of solution processed P3HT/PCBM bilayer heterojunction solar cells.16 The bilayer was created by first spin-casting the P3HT underlayer from o-dichlorobenzene (ODCB). The PCBM overlayer was spin-cast from dichloromethane (DCM), a poor solvent for P3HT. After optimizing the annealing conditions and the thickness of r 2011 American Chemical Society
the P3HT and PCBM layers, PCE values up to 3.5% were obtained, similar to the PCE obtained from bulk heterojunction solar cells fabricated from the same materials. Several groups17-21 have recently investigated the interdiffusion of P3HT and PCBM of annealed systems. Treat et al.18 through a combination of dynamic secondary ion mass spectrometry (DSIMS) and grazing incidence wide-angle X-ray scattering (GIWAXS) characterized the vertical composition and crystalline regions of annealed bilayers over macroscopic areas. They found significant interdiffusion of PCBM into amorphous regions of P3HT without disruption of the crystalline polymer even at modest temperatures.18 Similar results were presented by Chen et al.19 Exploiting interdiffusion as a method to create devices provides new opportunities for optimizing the nanomorphology beyond what is currently possible within conventional BHJ devices. It is unclear, however, if the formation of nanoscale morphology differs from the conventional BHJ approach. For this, imaging techniques such as transmission electron microscopy (TEM) are required. In this report, we present a direct comparison of the conventional P3HT/PCBM BHJ (C-BHJ) and an active layer created by solution processed, layer-by-layer, sequential deposition (layerevolved BHJ, LE-BHJ). A multilayer cross-sectional TEM sample incorporating both a conventional BHJ layer and a layer-evolved BHJ along with pristine polymer and fullerene layers was used for in situ Received: November 3, 2010 Published: January 26, 2011 1036
dx.doi.org/10.1021/nl200056p | Nano Lett. 2011, 11, 1036–1039
Nano Letters
LETTER
Figure 2. (a) Output characteristics of OTFTs based on P3HT/PCBM LE-BHJ layer. Inset: the device structure of OTFTs. (b) Transfer characteristics of OTFTs based on P3HT/PCBM LE-BHJ layer, pristine P3HT layer and pristine PCBM layer, respectively.
Figure 1. (a) J-V characteristics of P3HT/PCBM LE-BHJ solar cell. Inset: the device structure of P3HT/PCBM LE-BHJ solar cell. (b) IPCE spectrum of P3HT/PCBM LE-BHJ solar cell.
and simultaneous comparison of the nanomorphology. The data indicate that the nanomorphologies of the conventional BHJ and the layer-evolved BHJ are indistinguishable despite the radically different methods of formation. This implies that the structure we call the bulk heterojunction forms spontaneously and that it is the lowest energy and stable state of this two component system. Solar cells were fabricated with a procedure similar to Ayzner et al. Details are presented in the Supporting Information; the following is a synopsis. A PEDOT:PSS layer was cast onto an ITO/glass substrate. Then, the P3HT solution in ODCB was spin-cast onto a PEDOT:PSS layer and dried in the glovebox for 20 min. After that, the PCBM layer was spin-cast from DCM on the top of P3HT layer followed immediately by annealing at 100 °C for 10 min. A thin layer of TiOx was spin-cast onto the active layer in air, and finally ∼100 nm aluminum was thermally evaporated to complete the solar cell device fabrication. The current density, voltage (J-V) characteristics under AM 1.5 G irradiation (100 mW/cm2), along with the incident photon conversion efficiency (IPCE) spectra of the P3HT/PCBM layerevolved BHJ solar cells are shown in Figure 1. The device shows a PCE = 3.6%, an open-circuit voltage, Voc = 0.60 V, a short-circuit current, Jsc = 9.00 mA/cm2, and a fill factor, FF = 67%. Conventional P3HT/PCBM BHJ solar cells were fabricated for comparison. The conventional BHJ solar cell data closely matches the
data from the layer-evolved BHJ solar cell with Voc = 0.60 V, Jsc = 9.21 mA/cm2, FF = 66%, and PCE = 3.65%. Organic thin-film transistors (OTFTs) were fabricated on Si/SiO2 substrates with P3HT/PCBM LE-BHJ, the pristine P3HT and PCBM layers, respectively (see Supporting Information for details). The output and transfer characteristics are shown in Figure 2. For devices based on P3HT/PCBM LE-BHJ active layer, bipolar transfer characteristics were observed indicative of percolated structures. The hole mobility (μh) and electron mobility (μe) were nearly balanced, μh = 1.6 10-3 cm2 V-1 s-1 and μe = 1.0 10-3 cm2 V-1 s-1. For devices based on the pristine layers, only unipolar transport characteristics were observed. The hole mobility and the electron mobility were 1.0 10-3 and 6.0 10-3 cm2 V-1 s-1 for pristine P3HT and PCBM, respectively. The hole and electron mobilities from P3HT/PCBM LE-BHJ active layer agreed well with the values reported by Cho et al.22 for OTFTs based on P3HT/PCBM C-BHJ active layer. Cross-section TEM samples were prepared using a focused ion beam (FIB) for direct view of the vertical morphology. Application of FIB to organic materials was first demonstrated in block copolymers23 and has subsequently been applied to OPVs. Studies of bilayer heterojunction24 and bulk heterojunction25-27 active layers by FIB cross-section have elucidated the phaseseparation between the donor and acceptor. The formation of cross sections by FIB preparation suffers, however, from low throughput, high cost-per-sample, and most importantly milling artifacts. To overcome these difficulties and to eliminate ambiguities from milling artifacts, a multilayer stacked architecture was developed that allowed multiple experiments within a single FIB cross-section. To separate each component of the stack, to prevent solvent cross-contamination, and to maintain the appropriate interface chemistry that is known to benefit the performance of 1037
dx.doi.org/10.1021/nl200056p |Nano Lett. 2011, 11, 1036–1039
Nano Letters
LETTER
Detailed fabrication procedures of the multilayer stack appear in the Supporting Information along with cross-section TEM results obtained from a single LE-BHJ device. The FIB milling produced some artifacts such as undercutting near interfaces and minor vertical striations, both of which were easily identified in this image. The in-focus contrast is generated by the exclusion of scattered electrons by an aperture in the objective plane.32 Within the active layers, the measured intensity at each pixel is proportional to an (approximately) exponentially decreasing function of the mass, thickness, that is, the density integrated in the beam direction. Fullerene rich areas will therefore appear darker. For the small mass, thickness fluctuations characteristic of semiconducting polymers and fullerenes,26,27 contrast within a range of spatial frequencies determined by the contrast transfer function (CTF) can be significantly increased by defocusing.33-35 As the defocus is increased, longer wavelength fluctuations become visible. The CTF is given by the following relation CTFðuBÞ ¼ AðuBÞEðuBÞ2 sinðπΔZλu2 Þ
Figure 3. (a) Layer schematic of the multilayer cross-section sample. (b) In-focus, (c) -10 μm defocus, and (d) -30 μm defocus TEM image of the multilayer stack sample. The layers of the multilayer stack sample from top-to-bottom: aluminum, pristine P3HT, thick PEDOT, TiOx, pristine PCBM, thick PEDOT, TiOx, C-BHJ, thick PEDOT, TiOx, LEBHJ layer, normal PEDOT, silicon dioxide, and silicon (not shown).
polymer solar cells, the stack uses a sequentially deposited thin TiOx layer and two PEDOT:PSS layers (similar to a tandem cell6) to act as blocking layers. The FIB sample will have a nearly uniform cross-sectional thickness and milling effects can be detected and examined by including reference layers (e.g., pristine polymer/fullerene layers). Thus, with only a modest increase in processing time, the results of experiments that investigate the nanomorphology of the individual layers can be compared in situ and simultaneously. The FIB sample preparation procedure26,27 was modified for use with the FEI Helios 600 Dual Beam FIB and integrated Omniprobe Autoprobe 200 nanomanipulator. Low accelerating voltages (down to 5 kV) were found to reduce gallium implantation and damage of the FIB surfaces.28-30 The possibility of platinum redeposition was eliminated by using a ∼700 nm thick layer of aluminum (seven individual evaporations) to protect the sample surface during FIB processing. A silicon substrate with a 200 nm thermal oxide was used instead of ITO/glass in conjunction with the aluminum overlayer to reduce temperature rise during FIB milling.31 A thick cross-section (12 μm wide by 3 μm tall by 500 nm thick) sample was cut, attached/moved with the nanomanipulator, and welded to an Omniprobe copper lift-out grid for subsequent thinning. Both sides of the sample were thinned at 16 kV with a low beam current. One side of the sample was cleaned at 5 kV, but due to erosion of the protective aluminum layer we were unable to clean the other side. The final thickness of the multilayer cross-section was
