IOP PUBLISHING NANOTECHNOLOGY Nanotechnology 19 (2008) 424016 (6pp) doi:10.1088/0957-4484/19/42/424016 Nanoimprinted semiconducting polymer films with 50 nm features and their application to organic heterojunction solar cells D Cheyns1,2, K Vasseur1,3, C Rolin1, J Genoe1, J Poortmans1,2 and P Heremans1,2 1 IMEC vzw, Kapeldreef 75, 3001 Leuven, Belgium 2 ESAT, Katholieke Universiteit Leuven, Kasteelpark Arenberg 10, B-3001 Leuven, Belgium 3 MTM, Katholieke Universiteit Leuven, Kasteelpark Arenberg 44, B-3001 Leuven, Belgium E-mail: david.cheyns@imec.be Received 11 July 2008, in final form 11 August 2008 Published 25 September 2008 Online at stacks.iop.org/Nano/19/424016 Abstract Nanoimprint lithography is used to directly pattern the conjugated polymer semiconductor poly(3-hexylthiophene) (P3HT). We obtain trenches with aspect ratios up to 2 and feature sizes as small as 50 nm in this polymer. The application to organic solar cells is shown by creating an interpenetrated donor���acceptor interface, based on P3HT and N , N -ditridecyl-3,4,9,10- perylenetetracarboxylic diimide (PTCDI-C13), deposited from the vapor phase to reduce shadow effects. A planarizing layer of spin-coated zinc oxide (ZnO) nanoparticles is used to reduce the roughness of the layer stack. The response of the photovoltaic devices follows the increased interface area, up to a 2.5-fold enhancement. 1. Introduction An important breakthrough for organic solar cells was the first bilayer device [1] in 1986. Earlier organic photovoltaic devices relied upon a single layer of conjugated semiconductor, in which high electrical fields are needed to dissociate excitons. In contrast, the use of a heterojunction of two materials with different energy levels can induce efficient exciton dissociation at the interface between them. The light-generated excitons need to diffuse to the donor���acceptor interface to reach the dissociation center. The exciton diffusion length (LD) limits the width of the effective photo-active region (see figure 1(a)). Nevertheless, by clever layer optimization to increase the optical field inside the device, relatively efficient devices have been demonstrated [2���4]. The second breakthrough was the introduction of the bulk heterojunction concept [5, 6]. The photo-active layer consists of a mixture of donor and acceptor materials, as such creating dissociation centers everywhere within the active layer (figure 1(b)). Exciton diffusion will not limit the device performance, but carrier extraction is an issue. Inside the layer, two major problems arise: (1) shunt paths of a single material (being donor or acceptor) connecting both contacts, and (2) isolated islands of material. The shunt paths will reduce the parallel resistance and this has a direct effect on the cell���s performance [7���9]. The isolated islands can trap carriers, which will not contribute to the photo response. The choice of solvent and post-annealing steps influence the carrier collection efficiency, but the exact effect is difficult to predict beforehand [10, 11]. Here, we present a hybrid between a planar and bulk heterojunction solar cell, where a controlled morphology combines the advantages of both. A separate deposition of donor and acceptor material makes it possible to optimize each, and to reduce shunts or isolated islands, while the interpenetrating network provides dissociation centers throughout the whole layer (figure 1(c)). Ideally, the width of the trench should be in the range of the exciton diffusion length [12], or wa ��� LD,A and wD ��� LD,D . In this case, every created exciton is within an exciton diffusion length of a dissociation center. The depth of the trenches, di , should be as large as possible, with the absorption length of the organic semiconductor as a good guideline. Examples in the literature use vapor phase deposition [13] or nanoimprint lithography [14] to make the interpenetrating network. 0957-4484/08/424016+06$30.00 �� 2008 IOP Publishing Ltd Printed in the UK 1

Nanotechnology 19 (2008) 424016 D Cheyns et al (a) (b) (c) Figure 1. Different architectures of organic solar cells: (a) a planar heterojunction solar cell, (b) a bulk heterojunction and (c) a solar cell with an interpenetrating network. The dashed areas are the photo-active areas, determined by the exciton diffusion lengths of the donor material (L D,D ) and acceptor material (L D,A ). Two possible losses of performance are shown for (b): (A) are shunting paths and (B) are isolated islands. The important dimensions in (c) are the depth of the trench (di ) and the width of the trench in the donor (wD) and acceptor (wA). In this paper we use nanoimprinting to mold the well established donor material poly(3-hexylthiophene) (P3HT). Nanoimprint lithography uses a mold with fine structures to transfer these nanostructures by imprinting them into a polymer [15, 16]. Although the fabrication of a mold with fine structures is time-consuming and expensive (mostly done via e-beam lithography), every mold can be used many times to reproduce the nanostructures, even allowing the mold to be cloned in a fast and inexpensive process [17]. Nanoimprint lithography is not restricted to small areas, and roll-to-roll processing [18, 19] and large area imprinting [20, 17] are possible. Structures below 10 nm are demonstrated in the literature [21, 22], making nanoimprint lithography a candidate to challenge regular photo-lithography. Moreover, these small dimensions are of the order of the exciton diffusion length of organic materials, making it a promising technique for interpenetrated organic solar cells. We present imprinted P3HT layers with feature size down to 50 nm and aspect ratios up to 2. The subsequent step of applying the acceptor was done in the vapor phase rather than from solution, to avoid dissolution of the extremely fine nanoimprinted features of the donor layer. Direct evaporation of the cathode resulted in very poor yield, therefore we introduce a planarization layer on the donor���acceptor structure, consisting of an acetone-soluble metal-oxide nanoparticle ink. We show that the increase in photocurrent is proportional to the increase in the effective area of the folded heterojunction. 2. Experimental details Glass substrates with patterned indium tin oxide (ITO, Merck Display Technology, �� 20 / ) were cleaned with soap, deionized water, isopropanol and acetone. A 15 min UV���ozone clean directly prior to spin coating of subsequent layers com- pleted the cleaning procedure. In order to reduce the injection barrier [23], and hence the resistance of the device [24], the substrates were covered with a spin-coated 0.45 ��m filtered poly(3,4-ethylenedioxythiophene):poly(styrenesulfonate) (PE- DOT:PSS) solution, purchased from H C Starck, at 3000 rpm for 30 s to produce a 30 nm thick layer. The substrates were subsequently heated on a hotplate in air at 120 ���C for 10 min to remove excess water. Different mass-to-volume ratios (10��� 20 mg ml���1 solvent) of the donor material P3HT (Rieke) were dissolved in chlorobenzene. The solution was spin-coated in- side a nitrogen atmosphere to form layers ranging from 20 to 80 nm thick. A differential scanning calorimeter (T. A. Instruments 1920) was used to provide information on the glass transition temperature of P3HT. The material was loaded in an aluminum crucible. The heat flux was measured as a function of temperature, which was first increased from 0 to 250 ���C at a rate of 10 ���C min���1 and was lowered after a stabilizing period of 3 min from 250 to 0 ���C at the same rate. The imprint molds used were created by e-beam lithography. The molds, with a layer stack of silicon, silicon oxide and silicon nitride, have a pitch ranging from 200 to 100 nm, with a trench ratio of 50% and a trench depth of 100 nm. The resulting aspect ratios (depth versus width of the trench) are in between 1 and 2. Prior to use, the molds were cleaned with a mixture of hydrogen peroxide and sulfuric acid (H2O2:H2SO4 3:7) to remove possible contamination. On the clean surface of the mold, a silane (1H,1H,2H,2H- perfluorodecyltrichlorosilane, or FDTS) was deposited from the vapor phase to create an anti-sticking layer [25, 26]. Nanoimprinting was done with a manual two-column hydraulic press, equipped with temperature controllable plates (P/O/Weber), and located in a nitrogen atmosphere. The sample and mold, sandwiched between two thermally conductive soft-silicone pads to enhance the thermal contact and to reduce local stress effects, were placed on a hot chuck, preheated to 120 ���C, and a constant pressure of 14 MPa was applied for 15 min. Subsequently, the whole system was cooled down, and the pressure was released when T 40 ���C. The mold could easily be removed after use. The perylene derivative N , N -ditridecyl-3,4,9,10-perylen- etetracarboxylic diimide (PTCDI-C13, Aldrich) was used as an acceptor molecule, deposited by organic vapor phase deposi- tion (OVPD) [13, 27, 28]. The source cell temperature was set to 360 ���C and a preheated nitrogen carrier gas flow of 200 sccm (cubic centimeters per minute at STP) was blown through the source cell. The source flow was diluted with a supplemental preheated nitrogen carrier gas flow so that the total flow was 1000 sccm. At the entrance of the deposition chamber, the gas was evenly distributed onto the substrate by use of a shower- head. The deposition rate was 1.3 �� A s���1 at a growth pressure of 5 Torr. Zinc oxide (ZnO) nanoparticles dispersed in acetone (10 mg ml���1) were spin-coated (3000 rpm for 60 s) in order 2