We prepared cross sections of P3HT:PCBM bulk heterojunction (BHJ) organic solar cells (OSCs) for the characterization of their potential distribution with scanning Kelvin probe microscopy. We compared results of samples obtained by microtome cutting of OSCs on plastic substrates, cleaving of OSCs on glass substrates, and milling with a focused ion beam. Their potential distributions were in good agreement with each other. Under short circuit conditions, potential gradients were detected in vicinity of the electrode/organics interfaces, with negligible electric fields within the bulk. We contacted the OSCs in a defined manner and studied their potential distribution under operating conditions.
Polymer organic solar cells (OSCs) are object of intensive research as they promise to combine high power conversion efficiency with cheap fabrication processes based on solution processing.1,2 Highest power conversion efficiencies of up to 8.6 percent3 are achieved by employing a so-called bulk heterojunction (BHJ) structure.4 Despite many efforts undertaken in the research on BHJ OSCs, for example by transmission electron microscopy,5 optical microscopy6 or with scanning probe microscopy methods,7 the principles that govern the device operation are not completely understood yet. The nanoscopic phase segregation of the materials in the BHJ does not allow a clear distinction between bulk and interface phenomena. The understanding of both the morphological as well as the electronic structure by itself does not guarantee predictive power, when discussing complete devices. It is important to reveal the structure-function relationship, which includes collective effects that can hardly be predicted. Studies of the local charge transport within the OSC with scanning Kelvin probe microscopy (SKPM)8 can help to gain a deeper understanding of the structure-function relationship and bridge the gap from nanoscopic to meso- and macroscopic understanding. Unlike the above mentioned methods,5–7 SKPM allows for mapping of the electrical field distribution within the OSC under operating conditions. Its results can give detailed information on how charge is transported in the cell. SKPM was already successfully used to study microscopic phenomena of many electronic devices.9–11 We present a SKPM study of P3HT:PCBM organic solar cells.
The material combination of poly(3-hexylthiophene) and phenyl-C61-butyric acid methyl ester (P3HT:PCBM) as active layer, transparent PEDOT:PSS coated ITO bottom anode and LiF/Al top cathode is an often studied model system for BHJ OSCs,12 and was also used in our experiments. The layer structure of these OSCs is parallel to the substrate surface, so that the electronic transfer in the cell occurs in vertical device direction. Therefore the solar cell cross section has to be exposed to gain information about the microscopic transport phenomena within the cell. In this work we present three methods for the preparation of OSC cross sections and discuss the SKPM results obtained from the different methods.
We prepared OSC cross sections by milling with a focused ion beam (FIB)13 under high vacuum conditions (about 10−6 mbar), by cleaving of OSCs on glass substrates as well as by microtome cutting of OSCs on plastic substrates in ambient air. We chose these different methods to study the surface potentials of the differently prepared OSCs and to thus learn about the influence of the preparation process on the obtained cross sections. It was an open question, whether the contamination through ambient air after cleaving/microtome cutting or the doping with Ga+ ions while the FIB preparation do suppress surface studies with SKPM. In our experiments we could show that all the presented methods delivered meaningful SKPM results.
The BHJ OSCs on glass substrates used in our experiments were processed in the conventional way of spincoating P3HT and PCBM solved in chlorobenzene in the weight ratio 1:1 onto the PEDOT:PSS coated ITO bottom anode. LiF and Al applied by thermal evaporation formed the top cathode. As the microtome cutting requires soft substrate materials, we fabricated BHJ OSCs on plastic substrates. We used Thermanox© Cover Slips coated with an 11 nm thick gold layer acting as a semi-transparent bottom anode. The further fabrication process of the OSCs on plastic substrates was equal to that of the conventional ones on glass substrates. The samples fabricated for the different preparation processes showed significant variations in terms of active layer thicknesses (see Fig. 2(a)–2(c)). This was due to the applied spincoating process. While the spincoating, the plastic substrates were mechanically deformed, what led to thin (and inhomogenous) active layers. Small substrates of 5 × 5 mm2 had to be used for the FIB milling preparation in order to fit the AFM sample holder geometry. Also this led to variations in the thickness of the active layers in comparison to the conventional glass cells used in the cleaving process, especially because of solvent accumulation on the substrate edges.
The SKPM measurements were carried out using a Zeiss Auriga© Cross Beam System, which is upgraded with a scanning probe microscope from DME©. This setup allows for applying scanning electron microscopy (SEM), focused ion beam (FIB) and SKPM simultaneously under high vacuum conditions (the cross beam geometry is depicted in Figure 1(a)). We can position the cantilever tip under SEM observation very precisely at the OSC cross section and cross-check the SKPM data with SEM images. For that, we determine the OSC layer thicknesses with SEM images of high resolution after the SKPM measurements (because of the destructive effect of the electron irradiation) to be sure to interpret the layer structure obtained from SKPM measurements properly. To expose the solar cell cross section, we used the FIB to mill a small trench of about 40 μm2 into the OSC (Fig. 1(a) and 1(d)). To cleave the solar cells on glass substrates, they were scratched on their back side with a glass cutter and cleaved over an edge (Figure 1(b) and 1(e)). The plastic solar cells were cut with a Leica EM FC 7© ultramicrotome, equipped with a diamond knife (Figure 1(c) and 1(f)). We cleaved/cut the OSC samples in ambient air and transferred them afterwards into the vacuum of the workstation for SKPM measurements. The SEM images 1(d)-1(f)) show the OSC cross sections during SKPM measurements. The approached cantilever tip scans the cross section in amplitude modulated AFM mode line-by-line, leading to two-dimensional surface potential maps14,15 (see results in the insets of Fig. 2). The FIB milling (Fig. 1(d)) leads to well-defined edges, the (plane corrected) height fluctuations are smaller than five nm across the cross section. Cleaved as well as cut cross sections are partly damaged (Fig. 1(e) and 1(f)), but also on these samples we could find small areas where a complete cross section was accessible and smooth enough to be scanned with SKPM. Nevertheless, the latter samples exhibited roughnesses of more than 50 nm in one scan line, also in the smoothest areas. Monitoring the IV characteristics of the OSCs before and after cleaving/cutting/FIB milling as well as during the measurements assured that the OSC samples did not undergo significant changes in terms of device performance (see Fig. 1(g)–1(i)). Within the measurement setup we illuminated the samples with a white LED. Contrary to established results from literature,16 but in line with standard solar cell physics, we found a rather strong light intensity dependence of the open circuit voltage Voc. Compared to AM 1.5 illumination (Voc = 550 ± 50 mV), the weaker illumination led to smaller open circuit voltages of about 300 ± 100 mV during our measurements (see Fig. 2(d)).
Surface potential profiles of OSC cross sections obtained by the three different preparation methods under short circuit conditions (bias voltage of 0 V applied) are shown in Fig. 2(a)–2(c). These profiles are derived from the two-dimensional surface potential maps that are displayed in the respective inset. Arrows indicate the location of the exhibited profile. The profiles have qualitatively similar shapes: The surface potential of the Al contact is 500 ± 100 mV higher than that of the BHJ, and the potential of the ITO (or gold) contact is 200 ± 50 mV above the latter. Contrary to the results of Lee et al.,11 which show a negative slope towards the ITO anode, we observed this convex shape in all measurements. Fig. 2(d) shows the potential distributions of a cleaved OSC under illumination. We could show that disconnecting a contact under illumination goes along with a potential shift of the removed (floating) contact in the SKPM signal. In the presented case of the floating Al cathode there is a negative shift (of about −300 mV), in the case of the floating ITO anode there is a positive shift. The magnitude of these shifts matches the open circuit voltage Voc that we determined from IV curves taken within our measurement setup right before and after the SKPM measurement (data not shown here). The same results were obtained also for samples from the other preparation methods (data from FIB milled samples published in Saive et al.).17 In a further analysis (not presented here) we were able to prove that this photovoltage drops at the BHJ-Al/LiF interface only, no matter if the top cathode or the bottom anode was floating.17 Also these results could be reproduced in all of the measurements and on samples from all of the three preparation methods. With the presented methods we investigated OSC cross sections under bias voltages too (presented in Saive et al.).17
The high degree of agreement between the SKPM results proves that with all of the three presented methods it is possible to prepare samples that are suitable for meaningful SKPM studies. The modification of the surface's electrical properties through contamination in ambient air or the FIB treatment seems not to have a significant influence on the SKPM results. The fact that we could detect a potential shift by the full (macroscopically measured) photovoltage Voc directly on the cleaved OSC cross section (see Fig. 2(d)) with SKPM proves that there is no significant screening effect through contamination, although the cross section was exposed to air before. On the other hand, the Al top cathode seems to shield the FIB prepared cross section, especially of interest here is the organic BHJ, from a significant deposition of Ga+ ions (see the geometry of the setup in Fig. 1(a)). Ga+ ions implanted in an organic matrix cause very strong doping effects that can even lead to metallic behavior on the sample's surface.18 We did not observe a significant doping effect in our measurements on FIB prepared OSC cross sections. We assume that this is caused by the shielding of the organic layer through the Al on top of it.
Even if there is a rather good agreement with respect to the potential distributions observed, there are huge differences in terms of (lateral and energy) resolution as well as in terms of sample reliability for the three different preparation methods. As proved in many measurements, the FIB milling is the most reliable preparation technique and allows SKPM measurements with the highest lateral and energy resolution. The high reliability of the FIB method is caused by the high precision and the easy handling of the FIB milling process, which does not produce shortcuts in the device, an often observed issue for the cleaving and microtome cutting methods. We assume that the superiority of the SKPM data from FIB milled OSC cross sections over those from the two other methods is caused by their high degree of smoothness. A FIB prepared OSC cross section fluctuates in the range of less than five nm only, whereas cleaved and microtome cut samples show height variations of more than 50 nm also on the smoothest areas. This roughness causes cross-speaking between the different OSC layers and a smearing of the potential profiles at the interfaces (see Fig. 2(a)–2(c)), so a decrease of lateral resolution.
In conclusion, we showed that all the presented preparation techniques allow for the preparation of OSC cross section samples that are accessible for SKPM studies and deliver meaningful results. We were able to map the internal potential distribution of the OSCs. Furthermore we could attribute macroscopic device characteristics as the open circuit voltage Voc to changes in the nanoscopic potential distribution.
We acknowledge the German Federal Ministry of Education and Research (BMBF) for generous financial support (FKZ 13N10794, FKZ 13N10723).