SpatiallyResolved InSitu Structural Study of ... - Wiley Online Library

www.advmat.de

COMMUNICATION

www.MaterialsViews.com

Spatially-Resolved In-Situ Structural Study of Organic Electronic Devices with Nanoscale Resolution: The Plasmonic Photovoltaic Case Study B. Paci,* D. Bailo, V. Rossi Albertini, J. Wright, C. Ferrero, G. D. Spyropoulos, E. Stratakis,* and E. Kymakis A major advance in organic electronics technology will be the development of new tools to monitor the internal structure of device components with nanometer resolution. Indeed, diagnostic tools at the nanoscale enable the establishment of structure-property relations that link length scales from local nano/hetero structures and interfaces to large scale devices. Such capability is particularly important in the case of bulk heterojunction (BHJ) organic photovoltaics (OPV), considering that the photovoltaic performance strongly depends on the nanoscale phase separation during the donor/acceptor heterojunction formation[1,2] as well as the nanomorphology of the interfaces formed between the active and charge collection layers.[3–5] Polymer based OPV devices provide low environmental impact and low-cost conversion of solar energy into electricity. Although significant progress has been made due to the introduction of BHJ solar cells, based on blends of a conjugated donor polymer and a soluble fullerene derivative acceptor,[6–8] enhancement of device efficiency, at levels beyond the commercialization threshold, is still a challenging issue. There are two major barriers towards further improvement of OPV cell efficiency. On the one hand, it is the relatively low optical absorption of the ultrathin photoactive layer, leading to incomplete light harvesting; optical absorption is limited by the rather poor spectral match of the polymer blend with the solar Dr. B. Paci, D. Bailo, Dr. V. R. Albertini Istituto di Struttura della Materia C.N.R., Via Fosso del Cavaliere 100. 00133 Roma, Italy E-mail: [email protected] Dr. J. Wright, Dr. C. Ferrero European Synchrotron Radiation Facility 6, Jules Horowitz, 38000 Grenoble, France Mr. G. D. Spyropoulos, Prof. E. Kymakis Center of Materials Technology & Photonics Technological Educational Institute (TEI) of Crete and, Electrical Engineering Department TEI of Crete, Heraklion 71004 Crete, Greece Mr. G. D. Spyropoulos, Dr. E. Stratakis Institute of Electronic Structure and Laser (IESL) Foundation for Research and Technology-Hellas (FORTH) Heraklion, 711 10 Crete, Greece and Department of Materials Science and Technology University of Crete, Greece E-mail: [email protected]

DOI: 10.1002/adma.201301682

4760

wileyonlinelibrary.com

spectrum in the near-IR range. On the other hand, due to the low carrier mobility of the organic materials and the sandwichlike configuration of OPV devices, it is not possible to enhance absorption by thickening of the active layer.[9,10] Therefore, thin photoactive layers are inefficient due to poor absorbance, and thicker ones are inefficient due to bulk recombination losses. In this context, new strategies to enhance light absorption must be explored. A promising approach to meet this requirement relies on the use of efficient light harvesting and coupling mechanisms, by means of plasmonic metallic nanostructures.[11–14] For example, metallic nanoparticles (NPs) of sizes smaller than the visible light wavelengths are known to exhibit a strong absorption band in the UV–visible region, which can be assigned to the localised surface plasmon resonant (LSPR) interaction of the NPs’ surface electrons and the light field.[15] Therefore, incorporation of metallic NPs into the OPV system offers the possibility of enhanced absorption and correspondingly improved carrier photogeneration.[13,16] In this respect, plasmonic NP-based OPV devices are indicated to be promising systems for improved PV performance and stability. The second major limitation is the incomplete understanding of how the electrical performance of OPV cells is correlated to the morphology of their respective components at the nanoscale. In a BHJ OPV cell generation and dissociation of photoexcited excitons occur at the donor/acceptor nano-heterostructures, while the typical exciton diffusion length is limited to sub-10 nm scales. Besides this, the nanomorphology of the interfaces formed between the active and charge collection layers are equally important.[3–5] Considering the key role played by the active layer morphology, much progress has been made in the applicability to OPV materials and devices of different characterization tools at multiple length scales, including nearfield optical electron [i.e., scanning electron microscopy (SEM), cross-sectional transmission electron microscopy (TEM)] and scanning probe microscopy-based techniques.[17] However, in order to gain information on structure-property relations that link length scales from local nanostructures to large scale devices, the structural and/or morphological properties of the different organic layers and interfaces should be probed with nanometer resolution. In this work we propose an original approach to probe, insitu on a single cell, bulk and interface properties of plasmonic NP-based OPV cells at the nanoscale. In particular, by scanning the nanometer sized focal spot of a synchrotron X-ray beam across the cross-sectional area of the OPV device, the structural

© 2013 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Adv. Mater. 2013, 25, 4760–4765

www.advmat.de www.MaterialsViews.com

Figure 1. a) Cross-sectional sketch of the plasmonic OPV device used for the X-ray diffraction experiments. b) Current density-voltage characteristics of the reference (black curves) and NP doped OPVs (red curves) in the slowly grown (SG- solid triangles) and annealed AN1 (solid stars) and AN2 (solid circles) states.

properties of the different organic layers and interfaces as well as the NPs' dispersion configuration can be precisely monitored. In addition to this, thermal annealing performed in-situ and during the X-ray experiment allows monitoring of the respective modifications induced on the structural properties of the organic layers and NP distribution. In all cases, the experimental findings on the local nanostructure and organization were correlated with the photovoltaic performance of the respective OPV cells. Reference OPV cells exhibiting the ITO/PEDOT:PSS/ P3HT:PCBM/Al architecture (undoped) as well as devices with Au NPs embedded into their active layer at different volume ratios were fabricated (Figure 1a). In the following we will present exclusively the results from the devices corresponding to 5% NP volume ratio which showed the optimum photovoltaic performance.[18] The photovoltaic properties in the as-prepared (slowly grown-SG) as well as in two annealed (AN1 and AN2) states were recorded for both doped and undoped devices. AN1 and AN2 correspond to a moderate and the optimized annealed states, respectively. Figure 1b shows the corresponding current density–voltage (J-V) characteristics under illumination, while Table 1 summarizes the respective photovoltaic quantities. Regarding the SG state, the incorporation of Au NPs in the active layer induces a significant improvement of both the device short-circuit current (JSC), the fill factor (FF), and the opencircuit voltage (VOC). Following the annealing process, the photovoltaic performance is further enhanced for both doped and undoped cells.

Adv. Mater. 2013, 25, 4760–4765

Jsc (mA cm−2)

Voc (V)

FF (%)

η (%)

Reference-SG

2.46

0.38

0.21

0.20

Reference-AN1

3.87

0.37

0.22

0.31

Reference-AN2

8.27

0.60

0.61

3.26

Doped-SG

4.63

0.37

0.41

0.72

Doped-AN1

7.29

0.42

0.40

1.19

Doped-AN2

9.77

0.60

0.63

3.71

COMMUNICATION

Table 1. PV parameters of OPV devices studied (related to Figure 1b).

The observed improvement can be firstly attributed to LSPR effects in the vicinity of the small-sized NPs and secondly to multiple scattering by the larger diameter NPs.[16] The most pronounced enhancement is observed in JSC that can be attributed to the enhancement of the hole mobility of poly(3-hexylthiophene-2,5-diyl) (P3HT) in the blend.[18] Accordingly, the device efficiency (η) of the AN devices is almost quadrupled when compared with the SG one. However, we have recently demonstrated that the incorporation of NPs into the photoactive layer gives rise to enhanced structural stability of the blend as well.[3,5] Therefore, the performance enhancement can be also attributed to the improved photoactive layer morphology due to the presence of NPs. To shed more light on the NP induced improvement effect, precise monitoring of the structural properties of the different organic layers as well as the NPs dispersion configuration in the SG and AN states is required. For this purpose a novel diagnostic tool at the nanoscale, we first introduce and propose here, is used to characterize the NP-doped devices. Our approach is based on a high resolution X-ray diffraction (HRXRD) technique developed at the ID11 beamline of the ESRF, Grenoble (France). Using a tightly focused X-ray beam, exhibiting a focal spot of ≈100 nm, a diffraction map of each layer comprising the OPV device cross-section was recorded. At the same time, a fluorescence spectromicrographic study, used to track the elements comprising each device layer (In, Sn in ITO; Au NPs inside the active layer, Al in the top electrode) can be additionally performed. During the diffraction experiment, there is the possibility for in-situ thermal annealing of the device and subsequent monitoring of the evolution of nanoscale structural properties. The X-ray beam size and profile are critical parameters in order to be able to attain a spatial resolution close to 100 nm. For this purpose the sample surface (Scheme 1, x-y plane) should be accurately aligned to be parallel to the direction of propagation of the X-ray beam (x-direction). Furthermore, the sample cross-section (x-z plane) should be properly positioned within the focal depth of the diffraction beam. All details, regarding the samples’ geometry, the procedures adopted for beam focusing, measurement of the beam characteristics and sample alignment, can be found in the Experimental Section. In a typical experiment, the device was vertically translated from the Al top electrode to the ITO layer, while the X-ray diffraction signal was continuously monitored. Considering that the beam size is comparable to the thicknesses of the different device layers, this method allows



4761

www.advmat.de

COMMUNICATION


Scheme 1. Schematic of the setup designed for the X-ray diffraction cross-sectional stratigraphy of plasmonic organic photovoltaic devices.

discrimination of the diffraction signal corresponding to each separate layer. In Figure 2a we report the synchrotron patterns collected on the SG cell during the “stratigraphy” scans. Diffraction patterns for the Al electrode, the Au NPs incorporated in the BHJ layer and the ITO electrode are presented.

Figure 2. XRD patterns acquired during the cross-sectional scan (z-axis of Scheme 1) of the OPV devices: Results for the SG (top) and AN1 states (bottom) are shown. The reflections tracking the different devices constituents are the labeled. SG device: ITO C211 reflection (1); ITO C222 reflection (2); ITO C400 reflection (3); Au C111 reflection (4); Al C111 reflection (5); Al C200 reflection (6); ITO C441 reflection (7). AN1 device: in addition to the previous reflection a further peak is visible, assigned to the P3HT (100) reflection (8).

4762


Figure 3. Spatial evolution of the XRD peaks intensities, obtained from the layer-wise analysis of the XRD patterns of Figure 2; Results from SG (top) and AN1 states (bottom) are shown for comparison.

As expected, the PEDOT:PSS layer is amorphous and lacked any diffraction signal. Surprisingly, the SG BHJ blend showed no diffraction peaks as well, an observation that was not anticipated. This result indicates that the P3HT component inside the blend is essentially X-ray amorphous, or more precisely even if crystallites are present their sizes are too small to be well detected by the probe beam and thus contribute to the overall amorphous signal. This observation is in contrast to what is commonly measured in NP-free samples, where P3HT crystallites embedded in an amorphous matrix are identified via XRD. This suggests that the incorporated Au NPs may disrupt the polymer chain motions and hinder the self-organization and growth of large crystallites.[19] On the contrary, the signal corresponding to Au C111 reflection at q = 2.65 Å−1 was clearly detected. Figure 3 shows the analysis of the diffraction signals obtained after a layer-wise analysis process, described in the Experimental Section. In particular, the total intensity of the XRD reflections corresponding to each layer is plotted as a function of the vertical position of the sample. It is clear that the spatial resolution attained is sufficient to isolate the structural properties of each layer comprising the device, despite the overlap among the different characteristic curves. An important observation is that the Au NP signal maximum is more shifted towards the ITO side and thus the overlap between Au and ITO signals is larger than that between Au and Al ones. This denotes that the Au NP concentration is higher in the lower part of the BHJ layer, i.e., at the ITO side of the cell. The structural changes associated to thermal annealing treatment can be revealed by in-situ heating of the SG sample and simultaneous monitoring of the diffraction signals. It should be noted that the in-situ thermal annealing was performed at the temperature corresponding to the AN1 state (i.e., at lower temperature for longer time), in order to better preserve both the accuracy and spatial resolution of the diffraction technique. Figure 3 (bottom graph) presents the results obtained. The appearance of the P3HT [100] characteristic reflection, at 0.39 A−1, indicates that, due to the annealing treatment, P3HT crystallizes in larger crystals.[20]


Adv. Mater. 2013, 25, 4760–4765


Adv. Mater. 2013, 25, 4760–4765

COMMUNICATION

This result complies with the significant improvement of the photovoltaic performance measured for the AN compared to the SG device, shown in Figure 1. Furthermore, it is interesting that the XRD intensity distribution of Au NPs does not coincide with that of P3HT. Besides this, a comparison of the pre-annealing and the post-annealing analysis in Figure 3 shows that the relative distance between the centroids of Au NPs and ITO signal distributions is reduced by about 60 nm as a consequence of the thermal treatment. This is an indication that during the annealing process the Au NPs migrate towards the bottom of the BHJ, into the PEDOT:PSS layer. As a result the exciton yield, due to the excitation of LSPR in the vicinity of NPs, should be significantly enhanced in the region where the photoactive blend interfaces with the PEDOT:PSS layer. This in turn denotes that, following exciton dissociation, the majority of the excited holes can be readily collected, since the NP/PEDOT:PSS interface may act as a three-dimensional hole collector. The above results indicate that, apart from LSPR and blend morphology improvement effects, an important contribution to the enhanced efficiency of NP doped devices may be the efficient collection of the photoexcited carriers. A final interesting consequence of thermal annealing is a systematic broadening of the ITO reflection peak at ≈30 nm. This is in contrast to the respective Au and Al peaks that did not change within the experimental error. Such broadening may be attributed to In interdiffusion from the ITO into the PEDOT:PSS layer, in agreement with previous reports[21–23] It can be concluded that In diffusion takes places within ≈15 nm from the ITO/PEDOT:PSS interface. The above findings were further supported by fluorescence spetromicrographic measurements performed in combination with the diffraction experiments. Specifically, during the vertical translation of the cell, the fluorescence lines of the various metallic elements present in the device (i.e., In and Sn in the ITO layer; Au NPs inside the BHJ, Al electrode) were continuously detected and recorded. Figure 4 shows a sequence of the fluorescence data for SG and AN1 samples, respectively. The corresponding dependence of the fluorescence line intensities as a function of the vertical step for each metallic component are presented in Figure 5. The fluorescence data were fitted using the PyMca program, developed at ESRF.[24] It is observed that, due to thermal annealing, the relative separation between the centroids of the Au and ITO signal distributions (In and Sn lines) is reduced by ≈30 nm. This result complies with diffraction experiments also suggesting that, during thermal treatment, the incorporated NPs gradually migrate to the lower part of the BHJ. Finally the width of the In fluorescence line was increased by about 25 nm after annealing, which confirms the diffusion of In into the adjacent PEDOT:PSS layer. The present work has introduced the capability of monitoring the structural properties of each separate layer as well as the various interfaces forming an organic electronic device. This is achieved via an original high spatial resolution synchrotron X-ray diffraction technique performed onto an integrated device. Apart from measuring the structural properties of a device in its as-prepared state, our technique can also be applied for in-situ monitoring of the effects of thermal annealing on the nanoscale structural properties. As a case

Figure 4. The fluorescence data acquired during the cross-sectional scan (z-axis of Scheme 1) of the OPV device: Results for the SG (top) and AN1 states (bottom) are shown. The fluorescence lines tracking the different layers’ constituents are the following: AuLα1 (1a) and AuLβ1 (1b); ZrKα1 (2a) and ZrKβ1 (2b); InLα1 (3a) and InLβ1 (3b); SnKα1 (4a) and SnKβ1 (1b); Compton scattering (5). The zirconium lines are due to the detector.

study, we have investigated OPV cells incorporating metallic NPs, representing a state-of-the-art system in plasmonic photovoltaic technology. The experimental findings provide information, with less than 100 nm resolution, about the photopolymer crystallinity, the distribution of the plasmonic NPs into the photoactive layer and the structural properties of the blend/ PEDOT:PSS and PEDOT:PSS/ITO interfacial layer. In all cases the spatially resolved local nanostructure and organization findings were correlated with electrical performance and fluorescence spectromicrographic measurements. Such analysis provided new insight on the role of plasmonic NPs and may help towards elucidation of a structure-property relation of plasmonic bulk organic heterojunctions on the nanometer scale.

Figure 5. Spatial evolution of the fluorescence lines intensities from the layer-wise analysis of the data of Figure 4, for each metallic component of the device; Results from SG (top) and AN1 states (bottom) are shown for comparison.



4763

www.advmat.de

COMMUNICATION


The spatial resolution of the technique developed and presented here can be substantially improved via modern X-ray optics and experiments which are currently in progress. However, based on the present results, this combined X-ray diffraction/fluorescence spetromicrographic study paves the way for the systematic exploration on different types of multilayered thin film electronic devices. In this respect, the universality of the technique and its exploitation for the study of practically any type of multilayer organic electronic device is envisaged.

Experimental Section X-ray measurement setup: The experiments were performed on the beamline ID11 at the ESRF. Due to the constraints imposed by the geometry of the OPV cell (the thickness of each layer ranged from 50 nm to 200 nm), the use of a tightly focused X-ray beam was required to perform spatially-resolved diffraction experiments. In order to probe each layer forming the device cross section the beam direction was set parallel to the cell surface (x-y plane). By progressively shifting the device in the vertical (z) direction, the X-ray beam probed all the layers of the sandwiched specimen, from the top to the bottom electrode. An X-ray energy of 35 keV was selected using the Laue-Laue monochromator of the beamline. A vertical line focus was obtained using planar silicon lenses[25] with a focal distance of about 76 mm (see Figure S1 in the Supporting Information). The physical aperture of the 1D lenses was 50 × 50 μm2 and the effective aperture after including X-ray absorption was about 35 μm in the vertical direction. The X-ray beam was defined by slits in front of the lenses and a 1 mm pinhole was mounted between the lenses and the sample to avoid diffraction from the slit blades. The lenses were mounted on a translation and tilt stage to align them parallel to the centre of the X-ray beam. By focusing the X-ray beam at the entrance of the silicon lenses one may obtain a large gain in flux, creating a virtual source with much less favorable demagnification ratio. This increases the focal spot size at the sample position. For the presented measurements no additional lenses were used in order to have the smallest possible beam size. Behind the sample, a compound detector system allowed the translation among different detectors. A high-resolution Sensicam X-ray camera (1.3 μm pixel size) was used for alignment purposes. A fibre optic coupled CCD area detector was used for the diffraction measurements.[26] Furthermore, a quadrant diode was available for the fine alignment of the sample with respect to the focal depth. In addition, a fluorescence detector, close to the sample surface, allowed a spetromicrographic analysis, to track the position of the various elements composing the device (In, Sn in the ITO layer; Au NPs inside the active layer, Al electrode) and detect possible diffusion processes. Samples were mounted on top of a Linkam heating stage for temperature control, fixed on a y-z piezo scanner. The piezo stage was in turn mounted on a double tilt stage that was fixed to a coarse x/y/z translation stage. In the experiments reported here the vertical step was set to about 25 nm, allowing for oversampling. Sample preparation positioning and alignment procedures: In order to preserve the lateral resolution of the nano-focused X-ray beam to ≈100 nm the probe pathway inside the sample had to be kept as small as possible. For this purpose, the OPV devices used for X-ray experiments were cut in a triangular shape, using a diamond tip pen followed by a brittle fracture (see Figure S1). To ensure that the X-ray beam probes such a small sample area all the experiments were performed with the X-ray beam probing the tip of the triangle acute angle. SEM imaging of the sample prepared verified that the radius of curvature of the area corresponding to the acute angle and in turn to the X-ray path across the device was on the order of 100 μm (see Figure S1). In this way, the alignment of the beam parallel to the device was preserved. Due to this adopted geometry, the parallelism between the sample plane and the beam propagation direction was better than 0.02 degrees, corresponding to 35 nm across

4764


a 100 μm surface. Therefore the layers can be considered adequately extended parallel to the X-ray beam across the whole probe area. Furthermore, cross-sectional SEM images of the device showed that the respective layer thicknesses were uniform over a distance of 100 μm. The requirement of parallelism between the sample plane and the beam direction is crucial to preserve the optimum spatial resolution. For this purpose, the sample was gradually tilted with respect to the beam, while images on a Sensicam X-ray camera placed behind the sample were continuously recorded (Figure S2). Due to the rather long focal length of the X-ray lens, the small divergence of the beam can be neglected on the scale of interest. A faint reflection coming from the sample is visible as a horizontal line at about 210 pixels in the left frame of Figure S2a. The dark band near 125 pixels corresponds to the sample surface. The central and left panels show the result of integrating the detected image across the horizontal rows and stacking plots as a function of the tilt angle. As the tilt is changed, the reflection moves up and down the image crossing the thin film surface when the layer is parallel to the beam. The alignment error via this approach is better than 0.02 degrees, corresponding to 35 nm across a 100 μm surface. The sample could be aligned in the depth of focus (along the beam direction) by looking at the image of the beam far behind the focus, during the scan of the sample through the beam. Indeed, if the sample was positioned between the focal spot and the lenses, as the sample was scanned upwards, the bottom of the beam would appear to be cut first. Conversely, if the sample was located between the lenses and the focus, when it was scanned upwards, the top of the beam would be cut first. This procedure was imaged directly on the Sensicam camera, providing a magnified image of the sample, moving through the beam (Figure S2). For a more quantitative measure, the intensity of the beam was recorded on two quadrants of a four quadrant silicon diode. The diode position was chosen in such a manner that half of the beam was on each of the two vertically separated quadrants and the sample was scanned through the beam recording the upper and lower diode currents. Once the optimal alignment was reached, the corresponding sample position was set for diffraction. For the diffraction experiments, the ITO electrode on top of the BHJ layer, was used as a reference layer. The ITO layer thickness was measured independently by SEM imaging, with 2-3 nm accuracy (Figure S3). Prior to diffraction imaging, this layer was vertically (z-axis) scanned repeatedly and the intensity of the In and Sn fluorescence signals was recorded as a function of height. Both signals were used as reference to confirm the sample alignment. Each vertical scan required around 30 seconds. The sample height was then positioned with respect to the centroid of the In fluorescence peak and the diffraction and full fluorescence spectra were recorded within 60 seconds. This procedure eliminates any drift of the sample or beam height over the course of the measurements. Indeed, any change in sample alignment is evident via the full width at half maximum (FWHM) of the In fluorescence peak. Furthermore, according to the statistical theory, the position of the center of mass in a distribution of experimental data can be defined with a higher accuracy than the FWHM. Indeed, the convolution of the X-ray intensity distribution (in the cross section of the beam) with the interface profile between adjacent layers produces a Gaussian-like distribution of X-ray intensity at the detector. The variance (σ2) of such Gaussian-like peak is the sum of the variances of the two former distributions, so that it is actually expected to be broader than each of them. However, its center of mass can be calculated with a higher accuracy, in the order of σ/N (where N is the number of photons forming the observed peak). Using the above procedure, the positioning error was verified to be below 35 nm. During the annealing process, the HRXRD and XRF measurements were conducted in-situ, without removing the sample. Prior to annealing, the sample was initially aligned and repeatedly scanned while recording the ITO peak positions at each scan. During this initial alignment process, the analysis of the ITO peaks' intensity along the vertical scan provided the relative FWHMs and centroid positions “tracking” the ITO layer location and spatial extension inside the cell. Owing to numerous vertical scans the standard deviation of the intensity distribution positions was practically below 35 nm due to an alignment error.


Adv. Mater. 2013, 25, 4760–4765


Supporting Information Supporting Information is available from the Wiley Online Library or from the author.

Acknowledgements Authors are grateful for the support received from the COST Action MP0902 COINAPO. The authors are greatly indebted to A. Snigirev for providing the nanofocusing optics used for the XRD experiment, to I. Snigireva for performing the SEM analysis of the ITO substrate, to A.

Adv. Mater. 2013, 25, 4760–4765

Generosi for useful discussions and to A. Cerioni for preparing the XRD data graphics. Received: April 15, 2013 Published online: July 24, 2013

[1] A. C. Mayer, S. R. Scully, B. E. Hardin, M. W. Rowell, M. D. Mc Gehee, Mater. Today 2007, 10, 11, 28. [2] K. M. Coakley, M. D. McGehee, Chem. Mat. 2004, 16, 23, 4533. [3] B. Paci, G. D. Spyropoulos, A. Generosi, D. Bailo, V. R. Albertini, E. Stratakis, E. Kymakis, Adv. Funct. Mater. 2011, 21, 3573. [4] M. O. Reese, A. M. Nardes, B. L. Rupert, R. E. Larsen, D. C. Olson, M. T. Lloyd, S. E. Shaheen, D. S. Ginley, G. Rumbles, N. Kopidakis, Adv. Funct. Mat. 2010, 20, 20, 3476. [5] B. Paci, A. Generosi, V. R. Albertini, G. D. Spyropoulos, E. Stratakis, E. Kymakis, Nanoscale 2012, 4, 7452. [6] N.S. Sariciftci, L. Smilowitz, A. J. Heeger, F. Wudl, Science 1992, 258, 1474. [7] S. H. Park, A. Roy, S. Beaupre´, S. Cho, N. Coates, J. S. Moon, D. Moses, M. Leclerc, K. Lee, A. J. Heeger, Nat. Photonics. 2009, 3, 297. [8] L. Dou, J. You, J. Yang, C. Chen, Y. He, S. Murase, T. Moriarty, K. Emery, G. Li, Y. Yang, Nat. Photonics 2012, 6, 180. [9] V. Shrotriya, E. H. E. Wu, G. Li, Y. Yao, Y. Yang, Appl. Phys. Lett. 2006, 88, 064104. [10] G. Yu, J. Gao, J. C. Hummelen, F. Wudl, A. J. Heeger, Science 1995, 270, 1789. [11] E. Stratakis, E. Kymakis, Mater. Today 2013, 16, 133. [12] D. D. S. Fung, L. Qiao, W. C. H. Choy, C. Wang, W. E. I. Sha, F. Xie, S. He, J. Mater. Chem. 2011, 21, 16349. [13] D. H. Wang, D. Y. Kim, K. W. Choi, J. H. Seo, S. H. Im, J. H. Park, O. O. Park, A. J. Heeger, Angew. Chem., Int. Ed. 2011, 50, 5519. [14] E. Stratakis, M. M. Stylianakis, E. Koudoumas, E. Kymakis, Nanoscale 2013, 5, 4144. [15] W. L. Barnes, A. Dereux, T. W. Ebbesen, Nature 2003, 424, 824. [16] G. D. Spyropoulos, M. M. Stylianakis, E. Stratakis, E. Kymakis, Appl. Phys. Lett. 2012, 100, 213904. [17] Wei Chen, Maxim P. Nikiforov, Seth B. Darling, Energy Environ. Sci. 2012, 5, 8045. [18] V. D. Mihailetchi, H. Xie, B. de Boer, L. J. Anton Koster, P. W. M. Blom, Adv. Funct. Mater. 2006, 16, 699. [19] S.-W. Yeh, T.-L. Wu, K.-H. Wei, Y.-S. Sun, K. S Liang, J. Polym. Sci. B Polym. Phys. 2005, 43, 1220. [20] C. J. Brabec, M. Heeney, I. McCulloch, J. Nelson, Chem. Soc. Rev. 2011, 40, 1185. [21] M. P. de Jong, L. J. van IJzendoorn, M. J. A. de Voigt, Appl. Phys. Lett. 2000, 77, 2255. [22] B. Omrane, C. K. Landrock, Y. Chuo, D. Hohertz, J. Aristizabal, B. Kaminska, K. L. Kavanagh, Appl. Phys. Lett. 2011, 99, 263305. [23] D. Bailo, A. Generosi, V. Rossi Albertini, R. Caminiti, R. De Bettignies, B. Paci, Synth. Met. 2012, 162, 808. [24] V. A. Solé, E. Papillon, M. Cotte, Ph. Walter, J. Susini, Spectrochim. Acta Part B 62 2007, 63. [25] A. Snigirev, I. Snigireva, M. Grigoriev, V. Yunkin, M. Di Michiel, G. Vaughan, V. Kohn, S. Kuznetsov, Journal of Physics: Conference Series 2009, 186, 012072. [26] J. C. Labiche, O. Mathon, S. Pascarelli, M. A. Newton, G. G. Ferre, C. Curfs, G. Vaughan, A. Homs, D. F. Carreiras, Rev. Sci. Instrum. 2007, 78, 091301. [27] E. Stratakis, M. Barberoglou, C. Fotakis, G. Viau, C. Garcia, G. A. Shafeev, Opt. Express 2009, 17, 12650. [28] E. Stratakis, V. Zorba, M. Barberoglou, C. Fotakis, G. A. Shafeev, Nanotechnology 2009, 20, 105303. [29] E. Kymakis, E. Stratakis, E. Koudoumas, C. Fotakis, IEEE Trans. Electron. Devices 2011, 58, 860.



COMMUNICATION

X-ray probe beam characteristics: The signal from the ITO thin film was used to determine the X-ray probe characteristics. The spatial profile of the incident beam was measured by recording the indium fluorescence as a function of the position of the ITO film with respect to the beam. A FWHM of 166(3) nm is calculated upon fitting the observed profile with a pseudo-voigt function, without accounting for the film thickness. However, the above-discussed SEM measurements (Figure S3a) indicate that the average thickness of the ITO layer is around 140 nm, thus contributing significantly to the measured profile. If we assume that the profiles add in quadrature, as if they were Gaussian functions, then we disclose a beam FWHM of 89(5) nm. If instead the fit is done with a pseudo-voigt peak convoluted with a 140 nm wide rectangular function, a beam FWHM of 112(5) nm is found. Figure S3b shows the best fit results obtained in the case of the convolution of the pseudo-voigt with a rectangular function. Au NP generation: The NPs were generated by ultrafast laser ablation of metallic targets (Au/99.98%). With this technique, it is possible to produce a large variety of NPs that are free of both surfaceactive substances and counter-ions.[27] The targets were placed into a Pyrex cell and covered by a layer of absolute ethanol. A femtosecond (≈200fs@1kHz) laser beam was focused onto the target through the ethanol layer. The cell was mounted on a computer-driven X-Y stage and translated during laser exposure. More experimental details can be found elsewhere.[28,29] Laser irradiation gives rise to a high temperature gradient in the metal bulk and melts a thin layer of the target. A fraction of this molten layer is dispersed into the liquid as NPs. The related absorption spectra were measured using a Perkin–Elmer UV-VIS spectrophotometer. The respective colloidal solutions exhibit distinct peaks at about 530 nm, close to the theoretically predicted enhanced absorption threshold due to plasmon resonance (Figure S4). OPV device fabrication: The OPV devices were fabricated on 15 × 15 mm pre-patterned indium tin oxide (ITO) glass substrates with a sheet resistance of 10 Ω square−1. To create a buffer layer, poly(ethylene dioxythiophene) doped with poly(4-styrenesulfonate) (PEDOT:PSS), purchased from Bayer AG, was spin-cast from an aqueous solution onto the ITO substrate to give an average thickness of 60 nm, followed by baking for 15 min at 120 °C inside a nitrogen-filled glove box. Regioregular P3HT was purchased from Rieke Metals and [6,6]-Phenyl C61 butyric acid methyl ester (PCBM) was purchased from Nano-C. Regioregular P3HT and PCBM were dissolved in 1,2-Dichlorobenzene (DCB) in a 1:1 ratio and stirred for 24 h at 60 °C. Reference OPV cells exhibiting the ITO/PEDOT:PSS/ P3HT:PCBM/Al architecture (undoped) as well as devices with Au NPs embedded into the active layer (ITO/PEDOT:PSS/P3HT:PCBM-AuNPs/ Al) at different volume ratios were fabricated. Au NPs were blended into the P3HT:PCBM solution at different weight ratios. Composite blends with 3, 4, 5, and 6 wt% Au NPs were prepared. The OPV devices based on the blend ratio showing the best photovoltaic performance (5 wt% Au NPs) were selected for the in-situ XRD experiments. In the slowly grown (SG) state the active layers were exposed to a slow drying (slow growth process) for 1h in a covered Petri dish, followed by evaporation of the Al electrode. In the AN1 state, the complete devices were post annealed for 45 min at 60 °C followed by 15 min at 75 °C in air, while in the AN2 the devices were post-annealed for 15 min at 160 °C.

4765