One of the most crucial parameters determining photophysical processes and performance of organic solar cells is their electronic structure. This is indirectly related to intrinsic materials properties, e.g. energy levels or crystallization behavior, of the donor and acceptor constituents. Importantly, the electronic structure is directly related to the type and distribution of phases within a donor:acceptor blend, which forms a cell’s photoactive layer. Parameters to consider here are large vs small phase domains, crystalline vs amorphous etc. These indirect and direct properties lead to a highly complex parameter space of a functional morphology. For instance, an acceptor material that tends to form crystallites in pure form might be “forced” into amorphous domains by the thermodynamic or kinetic behavior when combined with a specific donor material. Another aspect is that at domain interfaces or within an intermixed phase, local entropic effects or charge transfer can lead to subtle changes in the electronic structure [1].

Since the electronic processes within a specific morphology are strongly interdependent, we call the components that determine this morphology functional phases. Typically, the size of the functional phases is in the range of only a few to several ten nanometers, which creates great demands on a characterization technique. There are many valuable tools from spectroscopy and microscopy available that are used to decipher a subset of morphological parameters. Here, we focus on the capabilities of analytical transmission electron microscopy (ATEM). It is well known that TEM provides high resolution. However, we show that only by adding analytical, or spectroscopic, modes in a spatially resolved manner, the various functional phases can be identified [2]. The reason is that the electronic structure of a functional phase is reflected in a fingerprint-like signal within low-energy electron loss spectra (EEL spectra). Subtle signal differences are recovered by machine learning algorithms. An example of a computed phase distribution based on spatially resolved EEL spectra (spectrum image = stack of images at different energy losses) is given in Figure 1a for the prominent model system P3HT:PC60BM. Figure 1b shows a small part (i.e. one image at a specific energy loss) of the spectrum image with reliable contrast – based on material properties - between the phases. This type of functional imaging provides insights into the two-dimensional (2D) phase distribution. Information is based on projection data through the three-dimensional (3D) sample object. Therefore we extent the method by spectral tomography to map the electronic structures in 3D (see Figure 1c) [3]. It shows that phase assignments from 2D imaging are valid.

It will be discussed how ATEM is used to identify the functional phases in fullerene and non-fullerene acceptor blends. Non-fullerene acceptors in conjunction with polymers form a particularly challenging system for visualization since their electronic structures are exceedingly similar. However, successful mapping of the phases is highly rewarding since a detailed nanoscale visualization is lacking so far and the most recent record efficiency exceeds 16% in a single junction cell [4]. In addition, novel concepts of spectral imaging with analytical electron microscopy will be introduced. These methods promise enhanced energy resolution or assignment of functional signals to the underlying electronic blend structure based on secondary or back-scattered electron spectra.