QuantumPatch (QP) is a parameter-free quantum chemistry software package, which performs DFT calculations to predict molecular properties from first principles, using just the chemical structure as input. QP is an efficient tool for the calculation of microscopic electronic structures and corresponding properties of molecules embedded in organic thin films, such as the HOMO LUMO energies, ionization energy (IP) and electron affinity (EA), intermolecular electronic couplings, reorganisation energy, or distributions thereof. The computation of these quantities can help to streamline time-consuming and costly experimental R&D efforts by efficient pre-screening of material candidates or by establishing structure-function relationships. Further, they serve as input for device simulations with lightforge(link). Electronic properties of organic molecules are modified, when embedded in thin films, as illustrated in Figure 1.
Figure 1: Schematical view of the shell setup of the QuantumPatch calculation. The figure is taken from https://publikationen.bibliothek.kit.edu/1000069385 and slightly modified.
The molecular properties should therefore not be computed either for molecules in vacuum, but for molecules embedded in thin films, taking into account their unique electrostatic environment and structure on a full quantum-mechanical level. To this end QP computes the electronic structure of molecules by self-consistent equilibration of the charge densities of all molecules using coupled single molecule-DFT level calculations. Electronic properties are computed on a representative subset of molecules near the center of atomistic morphologies, hence, surrounding molecules have to be included to account for polarization effects.
Figure 2: Interativ calculation protocol of a QuantumPatch run.
In the following, the documentation of QP, several tutorials and use cases will explain how to use and setup QP calculations, to predict the requested material properties. The general setup of the calculation is quit similar and therefore briefly outlined here and visualized in Figure 2 and 3. In Figure 2, the setup and chronological order of the individual calculations in a QP run are visualized. The charges of the molecules are optimized iteratively, to predict the requested molecular properties exactly. Firstly, the molecules will be calculated in vacuo, displayed in the first row 1 of Figure 2. Then, the point charges of each molecule are recalculated separately, considering the surrounding point charges of the other molecules. In these calculations of the molecules the charges of the surrounding molecules are considered directly in reoptimizing the charges of the molecules. The charge clouds of each molecule are color coded (red-orange-yellow-green), which illustrates the accuracy of the charge distribution. After a defined number of iteravtive calculations of the individual molecules, the process is finished and the final charge distribution is established, displayed in the rows 2-4 of Figure 2. In the end, the calculation of the electronic structure of the molecules can be done to get the property of interest, displayed in the 5. row.
The previously described calculation, uses a flexible shell setup as illustrated below in Figure 3. Thus, different levels of computational accuracy can be applied easily(e.g. DFT or semi-empirical methods, such as DFTB or XTB). This is called hybrid-Quantum Patch , optimizing runtime without reducing the accuracy significantly. Generally, the system can be divided into as many shells as required. Then, the radius of each shell (r1, r2, r3, ...) and level of theory, used for the calculation of the molecules in the shell, need to be defined. The size of the whole system, included in the QP calculation, is equal the radius for the last defined shell, r3 in the example of Figure 3. Thus, the individually defined shells always start at the core of the system and range in the periphery depending on their radii. In the center of the morphology and therefore the origin of the shells, a number of molecules needs to be fined. Then, for each of these molecules the defined amount of shells are drawn around them and the molecules, located in the shells, are considered in the prediction of this molecule. As illustrated in Figure 3, one molecule in the core results into a spherical quantum chemical calculation space of the morphology. Taking two molecules in the core, the total space of the quantum chemical calculations belongs to two closely overlapping spheres. For three molecules the space belongs to three overlapping spheres and so on.
Figure 3: Schematical view of the shell setup of the QuantumPatch calculation.
- Coordinates of pristine or mixed molecular systems
Site energies (HOMO and LUMO)
Absolute transport energy levels (electron affinity (EA) and ionization potential (IP) )
Energy disorder (local and global)
Pairwise electronic coupling matrix elements (HOMO and LUMO)
- Python 2.7
The Quantum Patch method [1-4] analyses atomistically resolved small molecule systems in a full quantum-mechanical way (DFT-based). It self-consistently calculates the energies of the frontier orbitals of each molecule in the amorphous system. After equilibration of the electronic structure of the thin film, quantum chehmistry methods are used to compute molecular properties on a subset of molecules in the thin film, taking into account environmental effects. For example, pairwise energy differences can be extracted and used to define the molecule specific energy disorder, or pairwise electronic coupling matrix elements. The energy disorder values and electronic couplings obtained from the Quantum Patch method lead to an accurate prediction of the charge carrier mobility of different relevant organic materials such as α-NPD, Alq3 and Pentacene.  Further details on the method can be found in the references below.
Besides the references below, there is a variety of use cases and tutorials that illustrate potential applications of QuantumPatch:
- Computation of transport levels, e.g. needed for transport simulations of mixed systems, doping activation simulations or Multilayer OLED simulations. Tutorials can be found here for IP levels and here for EA levels.
- Computation of input for charge transport simulations, e.g. for the computation of charge carrier mobilities
- Computation of coulomb binding energies between charges of an activated host dopant pairs (tutorial available here), required for the simulation of doped injection layers
The Quantum Patch method [1-4] analyses atomistically resolved small molecule systems in a full quantum-mechanical way (DFT-based). It self-consistently calculates the energies of the frontier orbitals of each molecule in the amorphous system. Pairwise energy differences can be extracted and used to define the molecule specific energy disorder. Furthermore, it calculates the pairwise electronic coupling matrix elements. The energy disorder values and electronic couplings obtained from the Quantum Patch method lead to an accurate prediction of the charge carrier mobility of different relevant organic materials such as α-NPD, Alq3 and Pentacene. 
The computational cost of the Quantum Patch scales linearly with the system size. The parallel mpi-based implementation also scales linearly to up to several hundred CPUs . This allows the calculation of microscopic electronic properties of structures on the 10 nm scale (e.g. from Deposit) for the analysis of bulk systems and interfaces.
Polaron Quantum Patch
In order to calculate accurate values of the electron affinity (EA) and the ionization potential (IP) of amorphous structures, polarization effects due to explicit charges have to be taken into account. In the Polaron Quantum Patch method, this polarization effect is modeled explicitly .
- J. Chem. Theory Comput., 2014, 10 (9), Pages 3720-3725
- J. Chem. Theory Comput., 2015, 11 (2), Pages 560-567
- Procedia Comput. Sci., 2016, 80, Pages 1244-1254
- Adv. Functional Mater. 2016, 26 (31), Pages 5757-5763
- Phys. Rev. B 91, 2015, 155203
- Adv. Mater. 2017, 29, 1703505.
- Adv. Mater. 2019, 31, 1808256.
- Phys. Rev. B 93, 2016, 195209
- ACS Nano, 2020, doi: 10.1021/acsnano.0c00384
- arXiv:1908.11854 [cond-mat.soft]
The results of the search are