The effect of three-dimensional morphology on the efficiency of hybrid solar cells

The effect of three-dimensional morphology on the efficiency of hybrid solar cells

(Parte 1 de 3)


The effect of three-dimensional morphology on the efficiency of hybrid polymer solar cells

Stefan D. Oosterhout1, Martijn M. Wienk1, Svetlana S. van Bavel2,3, Ralf Thiedmann4, L. Jan Anton Koster1, Jan Gilot1, Joachim Loos2, Volker Schmidt4 and René A. J. Janssen1*

The efficiency of polymer solar cells critically depends on the intimacy of mixing of the donor and acceptor semiconductors used in these devices to create charges and on the presence of unhindered percolation pathways in the individual components to transport holes and electrons. The visualization of these bulk heterojunction morphologies in three dimensions has been challengingandhashamperedprogressinthisarea.Here,wespatiallyresolvethemorphologyof2%-efficienthybridsolarcells consisting of poly(3-hexylthiophene) as the donor and ZnO as the acceptor in the nanometre range by electron tomography. The morphology is statistically analysed for spherical contact distance and percolation pathways. Together with solving the three-dimensional exciton-diffusion equation, a consistent and quantitative correlation between solar-cell performance, photophysical data and the three-dimensional morphology has been obtained for devices with different layer thicknesses that enables differentiating between generation and transport as limiting factors to performance.

Efficient organic solar cells use a bulk heterojunction of two semiconductors with offset energy levels to create charges. By intimately mixing electron donor and acceptor materials, the intrinsic limitations related to the low mobility and lifetime of excitons in organic semiconductors can be overcome, resulting in effective carrier generation at the extended donor–acceptor interface. The most efficient organic solar cells combine p-type conjugated polymers with n-type fullerenes1, which have recently reached efficiencies of 6% (refs 2, 3). Control over the morphology of the blend by the proper choice of processing conditions is generally essential to reach this level of performance and one of the traditional challenges in the field of organic solar cells lies in optimization of the morphology of the mixed layer. As an alternative to fully organic solar cells, hybrid solar cells use a combination of organic and inorganic materials. This concept has been demonstrated by combining semiconducting polymers as the donor with different inorganic materials, including CdSe

(refs 4, 5), TiO2 (refs 6, 7) and ZnO (refs 8–10), as the acceptor. Potential advantages of the inorganic semiconductors are a high dielectric constant, which facilitates carrier generation processes, a high carrier mobility and thermal morphological stability of the blended materials. Ultimately, hybrid cells offer the prospect of actual control over the morphology of the blend by first constructing the inorganic scaffold with the proper layout and dimensions11–13, but until now the best hybrid solar cells were made by simultaneous deposition of the two components. This often involves precarious processing, owing to the rather different nature of the materials involved. These drawbacks can largely be circumvented by the in situ generation of the inorganic semiconductor inside the organic material14,15. In this process, a well-soluble organometallic precursor is deposited from solution together with the semiconducting polymer. During and after this deposition the precursor is converted, by reacting with

1Molecular Materials and Nanosystems, Eindhoven University of Technology, PO Box 513, 5600 MB Eindhoven, The Netherlands, 2Laboratory of Materials and Interface Chemistry and Soft Matter Cryo-TEM Research Unit, Eindhoven University of Technology, PO Box 513, 5600 MB Eindhoven, The Netherlands, 3Dutch Polymer Institute, PO Box 902, 5600 AX Eindhoven, The Netherlands, 4Institute for Stochastics, Ulm University, 89069 Ulm, Germany. *

moisture from the surrounding atmosphere, to an inorganic network inside the polymer film. This has provided 1.4%-efficient photovoltaic cells, using diethylzinc as the ZnO precursor and poly(3-hexylthiophene)(P3HT)asthesemiconductingpolymer16.

The reasons for the reduced efficiency of hybrid solar cells compared with the more efficient polymer/fullerene cells are only partly understood and need to be analysed to further increase the performance. Here, we describe and analyse in situ P3HT/ZnO solar cells that reach a power conversion efficiency of 2%. Photoinduced absorption spectroscopy and three-dimensional (3D) electron microscopy of the active layers, combined with statistical analysis of percolation paths and solving the exciton diffusion equation are used to provide an unprecedented detailed insight into the role of the nanoscale morphology in creating and transporting charges in these bulk heterojunctions. Such analysis enables a unique and quantitative insight into the loss mechanisms associated with morphology that control, and ultimately limit, the power conversion of these rather efficient P3HT/ZnO hybrid solar cells.

The active layer of the photovoltaic cells was applied by spin-coating a blend of diethylzinc and P3HT from a mixture of chlorobenzene, toluene and tetrahydrofuran onto a transparent electrode, consisting of poly(3,4-ethylenedioxythiophene)/ poly(styrenesulfonate) (PEDOT/PSS) on an indium tin oxide (ITO)-covered glass substrate. During spin-coating, the diethylzinc was exposed to humidity, causing hydrolysis and the formation of Zn(OH)2. The subsequent condensation reaction is completed by annealing the film at 100◦C to form an interpenetrating net- work of ZnO inside the P3HT. Electron diffraction experiments demonstrate the presence of crystalline P3HT and ZnO domains in these active layers (see Supplementary Information). These mixed films do not visibly scatter light, indicating that no coarse phase separation between the organic and inorganic component takes

818 NATURE MATERIALS | VOL 8 | OCTOBER 2009 | © 2009 Macmillan Publishers Limited. All rights reserved.


Current density (mA cm

Voltage (V) 0

80 Fraction absorbed photons (%)

Wavelength (nm)

Absorbed photons (10

J SC (mA cm

Thickness (nm)IQE (%) Thickness (nm)

Figure 1 | Performance of P3HT/ZnO solar cells. a, Current density/voltage (J–V) characteristics of the best P3HT/ZnO device (film thickness is 225nm) in the dark (black line) and under illumination with simulated solar light (red line). b, EQE of the same device, corrected for the nonlinear response of the current with light intensity. Convolution of this spectral response with the AM1.5G solar spectrum provides an estimated current density of 5.6mAcm−2. The fraction of absorbed light, as calculated by optical modelling, using ellipsometry data is included. c, Evolution of the current density of P3HT/ZnO solar cells with the thickness of the active layer. For comparison, the amount of photons absorbed in the active layer is included. d, IQE of solar cells versus thickness of the active layer. The scatter in the data is attributed to differences in the morphology for films of similar thickness, but manufactured independently.

place during the deposition process. An aluminium top electrode completes the device.

Several mass ratios of P3HT versus ZnO were tested. A 1:1 ratio (w/w) was found to give the best performance. This translates into a ZnO volume fraction of about only 20%, noticeably smaller than what is commonly used in polymer/fullerene solar cell devices, where the optimal fullerene content generally exceeds 50%. In addition, the thickness of the P3HT/ZnO active layer was systematically varied between 50 and 250nm, by adjusting the spin rate during the coating process, while keeping the same P3HT and diethylzinc concentration. Despite some spread in the results, a clear trend is observed: the performance of the devices improves with increasing photoactive layer thickness, mainly as a result of increasing current densities. The overall best performance was obtained for a device with an active layer thickness of 225nm. Under illumination with 100mWcm−2 simulated solar light, this device delivers an open-circuit voltage (VOC) of 0.75 V, a short-circuit current density (JSC) of 5.2 mA cm−2 and a fill factor of 0.52, resulting in an estimated device efficiency of 2.0%.

Spectralresponsemeasurements(Fig. 1b)revealamaximalexternal quantum efficiency (EQE) of 4% at 520nm. The broad, rather flat shape of the spectral response is related to the thickness of the device. Convolution of the EQE with the AM1.5G spectrum affords a current density of 5.6mAcm−2, which is in reasonable agreement withthevalueobtainedfromJ–V measurements.

It is quite uncommon for polymer solar cells to have an optimal layer thickness significantly larger than 100nm, because many

Energy (eV)

Figure 2 | PIA spectra of P3HT/ZnO layers. The normalized PIA spectra of P3HT/ZnO layers used in solar cells recorded at T =80K. In the devices, the active layers had thicknesses of 44nm (black), 54nm (red) and 151nm (blue). Spectra were normalized at the 2.2eV bleaching band.

material combinations suffer from charge collection limitations at larger thicknesses. P3HT/PCBM ([6,6]-phenyl-C61-butyric acid methyl ester) is one notable exception17, demonstrating that in P3HT carrier mobilities are sufficiently high to allow thick photoactive layers. The good performance of the relatively thick P3HT/ZnO layer is a strong indication that electron collection

NATURE MATERIALS | VOL 8 | OCTOBER 2009 | 819 © 2009 Macmillan Publishers Limited. All rights reserved.




Oxide layer 100 nm

c d b




Figure 3 | Electron tomography of P3HT/ZnO solar cells. a, Transmission electron micrograph of a cross-section of a typical P3HT/ZnO photovoltaic cell. The 80–100-nm-thick sample was obtained using a FIB. The different layers present in the device are indicated in the image; the platinum layer is deposited later to allow preparation of the TEM sample. b, Reconstructed volumes of P3HT/ZnO layers, obtained by electron tomography. Three samples with thicknesses of 57nm (left), 100nm (centre) and 167nm (right) are shown. The size of all samples is 700×700nm; ZnO appears yellow, P3HT transparent. c, Reconstructed volume of a cross-section of the active layer of a completed P3HT/ZnO device. d, The green arrow indicates an isolated ZnO domain. The red arrow indicates a ZnO domain that is connected to the top, but not through a strictly rising path.

820 NATURE MATERIALS | VOL 8 | OCTOBER 2009 | © 2009 Macmillan Publishers Limited. All rights reserved.


Table 1|Calculated volume fraction, exciton quenching and percolation connectivity of P3HT:ZnO bulk heterojunction layers inferred from electron tomography.

Without electrodes With electrodes

Film thickness(nm)

ZnOvolume fraction (%)

Quenchedby ZnO(%)

Quenchedby ZnO(%)

Quenchedby electrodes (%)

ZnO connected to top (%)

ZnO monotonously connected to top (%)


Distance (nm)

(Parte 1 de 3)