ORGAnic solar cell VOLTage by numerical computation. – ORGAVOLT
«ORGAVOLT« - optimizing organic photovoltaics by predictive computation
Cheaply printable organic electronics and solar cells are major growth areas. Their optimization remains empirical and it could benefit from modeling of their fundamental processes. <br /><br />Although first-principle computations provide band gaps of inorganic semiconductors with an accuracy of a few percent, their CPU time grows as N4, where N=100...1000 for organic semiconductors. <br /><br />Therefore, a key aim of our project is to extend our N³ GW algorithm from molecules to bulk organic semiconductors.
First-principles, predictive computation of the bulk properties of organic semiconductors and a semi-empirical study of photocurrent generation at the donor-acceptor interface.
Bulk photo emission spectroscopy (PES) confirms the existence of bands in crystalline organic semiconductors. However the same data also show a broadening of the quasi particle peaks with increasing temperature. <br /><br />Photons incident on the donor (D) acceptor (A) interface can be considered to pump electrons from the valence band of the donor (D) to the conduction band of the acceptor (A) ignoring, in a first approach, the exciton binding energy and surface band shifts. <br /><br />In the simplest picture, the photon creates a spatially limited excited state (exciton) which diffuses to the D/A interface where the exciton is separated into an A electron (e) and a D hole (h) which are then responsible for the observed photo current. <br /><br />This picture has been criticized by Alan Heeger as too slow to produce time-resolved photo currents, but it is consistent with typical photo-chemical time scales which are of the order of 100 fs.<br /><br />To implement Heegers idea requires describing creation and separation of excitons at the interface which is somewhat beyond the current state of the art in the description of processes with external electrodes. Our team member Remi Avriller at Bordeaux is working on this. <br /><br />Our Grenoble team member, Mark E. Casida, views the D/A interface in terms of the theory of photochemical reactions. Thus charge separation leads in the first instance to an accumulation of locally-bound excited states (excimers) at the D/A interface which then decay (e.g., because of thermal effects or excimer-excimer interactions) to delocalized conduction states by passing through photochemical funnels (such as conical intersections). <br /><br />A key challenge he is trying to address, by semi-empirical methodology, is to identify candidate photochemical pathways which can then be investigated in detail with more resource-intensive first-principle calculations.
This project consists of (i) first-principles electronic structure calculations of bulk phases of organic semi conductors and of (ii) semi-empirical modeling of their donor-acceptor interfaces.
First-principles electronic structure calculations for solids are rooted in parallel work by Lars Hedin and Walter Kohn in the 1960s. Hedin's approach was first applied to semiconductors in the 1980s when sufficiently powerful computers became available and it gives bands and band gaps within about 0.2 eV of experiment.
Standard ways of solving the GW equations scale as N4, where N is the number of atoms in the unit cell. Thus, even with increasing computer power, new algorithms are needed to treat organic crystals once N becomes of the order of a few hundred.
A method to reduce the complexity of TD-DFT computations from N4 to N³ by using locality was introduced by some of us in the earlier ANR project Nossi (2008-2010). Similar techniques resulted in a reduction of the complexity of GW calculations for large molecules from N4 to N³ and the extension of this algorithm is one of the key aims of this project.
Part of our mandate is to obtain a better understanding of the basic phenomenology underlying organic photocells and to validate this understanding by testing the results of our calculations against experiment for simple organic solar cells. This also involves building better models of the underlying photocell mechanism which is as yet insufficiently understood.
Using the relatively simple semi-empirical approaches will help us to identify the key processes and the resultant models can then be studied by more demanding first-principle methods for final quantitative validation. This is a standard strategy in molecular and solid-state theory and may be considered a multi scale modeling scheme. Once validated in simple systems, the theory will also provide descriptors for building quantitative structure property relations (QSPR) in more complex systems.
The N³ scaling property of the dominant-product algorithm for molecular GW calculations depends on the locality of the atomic orbital description in the SIESTA code of which Daniel Sanchez-Portal is one of the principal authors.
We have now worked out an extension of the dominant-product algorithm to periodic bulk and this algorithm is currently being implemented. A reduction of the response basis was found by Peter Koval in San Sebastian and it will lead to significant reduction of the time needed to carry out periodic GW calculations.
A BSE code with N³ scaling has been implemented by Mathias Ljungberg, a postdoc at Bordeaux in 2013 who has now joined the San Sebastian-Marburg collaboration on “Internal Interfaces« the aim of which is very close to the «ORGAVOLT« project. A semi empirical model of the photo current generation has been developed by Remi Avriller.
The doctoral student originally hired in Grenoble had to suddenly quit his studies due to family problems in China, but he will be replaced by a post doc (recruitment underway).
The Bordeaux post doc has also left and must be replaced (recruitment underway). These external constraints have lead to significant delays in our project and the first mile stones are only beginning to be attained now.
However the advancement of the field has also lead to an apparent simplification of the modeling task. When we submitted our project, we were unaware of any systematic investigation of organic crystals and we were very concerned that most work in this field was done on polymers, rather than on the organic semiconductors based upon small molecules for which our models are most applicable.
Since we submitted our project, the current world record in (tandem) organic solar cells was achieved by Peter Baeuerle of Heliatek using organic semiconductor based on small molecules, i.e. precisely the type of materials to which the San Sebastian-Bordeaux computer code will apply.
Also, Alejandro Briseno from Amherst, Massachusetts, and visitor at Bordeaux, has made a very detailed study of a large variety of organic crystals of precisely the type where we can apply our code. He has also discovered a novel type of D/A interface that is (i) ordered and (ii) of large projected area, in a significant departure from the bulk hetero junction device due to Heeger and collaborators. But this work is still unpublished and no details are available.
Although external constraints have seriously slowed down the ORGAVOLT project, we find prospects to be bright for the application of the GW and BSE computer codes we are developing as well as for building more realistic models for describing the overall mechanism of photo current generation using semi-empirical models.
None, so far.
The last 10 years have seen increasingly efficient organic bulk heterojunction cells and there is a promising effort to print them. But the mechanism of the generation of the photocurrent is not yet understood. The implementation of an improved algorithm from our team will permit a realistic modeling of the donor-acceptor interface where the photocurrent is generated. This will allow the open-circuit voltage VOC of future organic bulk heterojunction cells to be predicted before their synthesis and so aid in their design and optimization.
Context, social and economic issues. The context of the present project is the strong growth of photovoltaics in general (with total peak power doubling every 3 years). The younger field of organic bulk heterojunctions is also growing, but--for the moment--mainly in terms of research effort. After two decades, organic bulk heterojunctions have now reached 9% efficiency and there are promising efforts of cheaply printing large area devices.
Many start-up companies have been founded that make organic cells or modules. Among these, the company Konarka of Alan Heeger, a Nobelist for conducting plastics, is one of the largest.
The efficiency of the comparatively young bulk heterojunction organic solar cells is critical for them to become viable. Currently, efficiency grows mainly by trial and error because the crucial process of photocurrent generation at the donor acceptor interface is not yet well understood.
This is somewhat reminiscent of the status of heterogeneous catalysis a decade or two ago when knowledge of surface physics and chemistry was too rudimentary to allow the simulation of catalytic processes. Today that situation is changing in catalysis, with in silico design becoming increasingly important. As with catalysis, we expect that an improved understanding of the fundamental processes taking place in organic photocells will help in their design and hence in rendering them more competitive compared to other technologies.
Our techniques will allow the prediction of the open-circuit voltage of such cells which is an important step towards improving this type of solar cell and thus render such cells more competitive.
Dietrich FOERSTER (Laboratoire Ondes et Matière d'Aquitaine) – firstname.lastname@example.org
The author of this summary is the project coordinator, who is responsible for the content of this summary. The ANR declines any responsibility as for its contents.
LOMA Laboratoire Ondes et Matière d'Aquitaine
DCM Laboratoire de Chimie Théorique
University of San Sebastian Centro Mixto CSIC-UPV/EHU et DIPC
Help of the ANR 295,678 euros
Beginning and duration of the scientific project: September 2012 - 36 Months