Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Maria Chan Project Name: HybridPerov Division: CNM Project title: Atomistic simulations of electrical properties and electronic structure of hybrid perovskites for intermediate band photovoltaics Associated funding: LDRD Other Systems: NERSC (320,000 core hours) Science: Intermediate band photovoltaics (IBPVs) are a class of multi-junction devices designed to surpass the Shockley-Queisser (S-Q) limit of solar conversion efficiency by providing large photocurrents while maintaining a high output voltage [1-3]. IBPVs consist of an IB material sandwiched in between an n-type and a p-type semiconductor, which act as selective contacts to the conduction band (CB) and valence band (VB) respectively. In the IB material, sub-bandgap photons can be absorbed through transitions from the VB to the IB and from the IB to the CB, which means the solar cell can harvest a greater portion of the solar spectrum. The theoretical efficiency of an IB solar cell has been calculated to be nearly 20% higher than a single-gap solar cell under the same conditions [2]. Further, the optimal bandgap of an IB material has been calculated to be around 2.0 eV, which is split by the mid-gap level into two sub-bandgaps of approximately 0.7 eV and 1.3 eV. Lead halide perovskite semiconductors have emerged as attractive candidates for photovoltaic applications owing to their large absorption coefficients, simple synthesis and easy tailoring of properties by composition engineering. Using ABX3 perovskites where lead occupies the B site, X is a halogen atom and A is Cs, Methylammonium (MA = CH3NH3) or Formamidinium (FA = C2H5NH3), solar efficiencies as high as 22% have been reported, which is on par with CdTe. Thus, APbX3 perovskites have been the subject of recent studies dealing with partial substitution of Pb to obtain stable mid-gap states with tunable energy level [4]. Recently, our team substituted Pb in MAPbBr3 (an ideal parent semiconductor with a bandgap ~2.2 eV) with various transition metals [5] and discovered that Co was able to create a mid-gap density of states; this was later confirmed by experimentally measured absorption spectra. First principles density functional theory (DFT) calculations, widely used to study atomic configurations, electronic structure and properties of materials, were used for this work. 1/8th substitution of Pb by Co showed a thermodynamic penalty similar to the thermal energy, and an examination of the electronic density of states revealed the presence of an intermediate band below the conduction band of the parent perovskite material [3]. Based on the above calculations, our team at Argonne successfully synthesized a new class of halide perovskites with the general formula CH3NH3Pb1-xCoxBr3-yCly, where x = 0, 0.125 and 0.25 and y = 0 and 1.5 [3]. While absorption spectra calculated from DFT using the frequency-dependent dielectric function showed additional peaks around 1.8 eV, UV-vis-NIR spectroscopy revealed sub-gap absorption features between 1.65 eV and 2 eV for a number of thin films. The sub-gap absorption was composed of multiple peaks, some of which are located at the energy expected for the absorption from [CoBr4]2- or [CoBr6]2- ions. This is consistent with the DFT calculations where the mid-gap states are expected to be mainly composed of Co d orbitals with a small amount of mixing from Br p orbitals. Moreover, the decrease of bandgap energy upon Co substitution and the decrease in absorption intensity of the parent gap indicate that Co is chemically incorporated in the crystal lattice of PbBr3. How ever, the absence of second sub-gap absorption, which is otherwise expected to be located at 0.5 – 0.7 eV, indicates the mid-gap states are unfilled at the current level of doping. If the filling of the IB can be controlled, the identified new density of states in Co-substituted PbBr3 systems could lead to generation of greater photocurrent in solar cells. To this end, we would like to address the following questions: 1) Where is the location and what is the filling status of the IB formed by Co substitution? Not only is the position of the IB inside the band gap important, but it must ideally be half-filled to allow transitions from VB to IB and from IB to CB [1]. 2) What are the possible substituents besides and along with Co, and their optimum concentrations for ideal IB formation? While some transition metals have already been tested [5], it is interesting to study if substituting Pb by other candidates such as Mn, Cr or Bi can lead to intermediate bands as well. Further, could co-substituents be added along with Co (which already produces an IB) to move the IB to a more desirable position and make it a half-filled band? 3) What really determines the electrical properties of MAPbBr3? Various intrinsic defects in semiconductors determine the non-radiative recombination behavior, and thus the performance of solar cells made from them. A look at the formation energies of vacancy, interstitial and substitutional defects in MAPbBr3 can provide a picture of where the equilibrium fermi level lies and the particular defects that determine it. The relative influence of an external metal substitution (for example CoPb, i.e. Co substituting Pb) towards the fermi level can then be ascertained [6]. 4) What other parent perovskite semiconductors can be used? While MAPbBr3 was a suitable parent semiconductor because its bandgap is close to 2 eV, the strong s-p antibonding between Pb and Br leads to the low formation energy of lead and bromine vacancies, which in turn affects the equilibrium fermi level [6]. Thus, it may be worthwhile to explore other parent perovskite semiconductors, such as MAPbCl3, MAPbBr1.5Cl1.5, CsPbBr3, double perovskites, 2D hybrid perovskites, etc., where vacancy formation might be difficult and the substituent atom could determine the equilibrium fermi level. 5) How do the intermediate bands and charge transition levels relate to one another? A general impurity-level problem in semiconductors is determining the difference between the location of the IB as obtained from the density of states and the (q+1, q) charge transition level between the substituent charge states q and q+1. In other words, the quasi fermi level obtained from the formation energies of different charged states can help us comment on the reliability of equating the IB level as the quasi fermi level. REFERENCES [1] A. Luque, A. Marti and C. Stanley, “Understanding intermediate-band solar cells”, Nat. Photonics. 6, 146–152 (2012). [2] A. Luque and A. Marti, “Increasing the efficiency of ideal solar cells by photon induced transitions at intermediate levels”, Phys. Rev. Lett. 78, 5014-5017 (1997). [3] M.D. Sampson, J.S. Park, R.D. Schaller, M.K.Y. Chan and A.B.F. Martinson, “Transition metal-substituted lead halide perovskite absorbers”, J. Mater. Chem. A. 5, 3578 (2017). [4] M.T. Klug, A. Osherov, A.A. Haghighirad, S.D. Stranks, P.R. Brown, S. Bai, J.T-W. Wang, X. Dang, V. Bulović, H.J. Snaith and A.M. Belcher, “Tailoring metal halide perovskites through metal substitution: influence on photovoltaic and material properties” Energy & Environ. Sci. 10, 236 (2017). [5] J.S. Park, M.D. Sampson, N. Jeon, A.B.F. Martinson, M.K.Y. Chan, “First-principles screening of metal substitution in perovskite halides for intermediate band solar cell applications” (In preparation) [6] Y. Yan, W-J. Yin, T. Shi, W. Meng and C. Feng, “Defect Physics of CH3NH3PbX3 (X = I, Br, Cl) Perovskites”, Springer International Publishing, Switzerland (2016). [7] C. Freysoldt, B. Grabowski, T. Hickel and J. Neugebauer, “First-principles calculations for point defects in solids”, Rev. Mod. Phys. 86, 253 (2014). Project description: The calculations will involve the items below. (1) Intrinsic point defects (native defects) in MAPbBr3, MAPbBr1.5Cl1.5 and MAPbCl3: We will perform DFT optimization (without volume relaxation) on structures (charged and neutral) with various vacancy, interstitial and substitutional defects, such as lead vacancy (VPb), bromine in an interstitial site (Bri), methylammonium substituting a lead atom (MAPb), etc. With 12, 20 and 12 total defects each for the three perovskites respectively, and with the consideration of 2 or 3 charged states for each defect, this leads to a total of approximately 100 DFT calculations. By calculating the formation energies of each defect as a function of the fermi level, a formation energy vs fermi level picture will be constructed for every parent perovskite, and the positions of equilibrium fermi levels will be determined. (2) Various substituents and co-substituents in MAPbBr3, MAPbBr1.5Cl1.5 and MAPbCl3: We will perform 1/8th substitution of Pb atoms by various selected atoms, namely Co, Mn, Cr, Fe, Al, Sb, Bi and Y, as well as substitute an additional Pb atom by the latter 7 substituents following 1/8th substitution of Co. For each of the 15 resulting structures, DFT will be used to obtain a) relaxed geometries, lattice parameters and formation energies, b) projected density of states, especially to probe the location of the impurity atom in the parent DOS, and c) absorption spectra, employing the frequency-dependent dielectric function. (3) 15 substituents in MAPbBr3, MAPbBr1.5Cl1.5 and MAPbCl3: DFT relaxation will be performed for each substituent M in different charged states to obtain the lowest formation energy states at different fermi levels. Following this, every impurity MPb (M substituting Pb) will be added to the formation energy vs fermi level diagram in addition to all the native defects. This enables us to ascertain whether the equilibrium fermi level in MAPbX3 (and consequently, its electrical properties) is determined by the substituent or by the native defects. Further, the relationship between the charge transition levels in M-substituted MAPbX3 (i.e., the fermi energy at which one charged state has the same formation energy as another) and the position of the intermediate band in the DOS (from calculations in (2)) will be studied. The DFT calculations will be performed using the plane wave code VASP, with GGA-PBE exchange correlation functional and the projector augmented wave (PAW) treatment of core electrons. For the systems containing transition metals such as Co and Mn, the on-site Coulomb correction (+U) will be used in addition, with a U value of 3 eV. All the DFT relaxation calculations are based on the MAPbBr3 2x2x2 supercell (containing 96 atoms), using a KPOINTS mesh of 3x3x3. For the density of states and absorption spectra calculations, a dense grid of 2000 points and a KPOINTS mesh of 4x4x4 are used. In the case of charged defects (native or substituent), a charge correction based on the technique of Freysoldt et al. [7] is applied to correct for the formation energy values. The time estimates for the various parts above, based on benchmark results with VASP running on Blues and Bebop, are listed below. (1) (36 cores/node) x (4 nodes) x (24 hours) x (100 calculations) = 345,600 core-hours (2) (a) (36 cores/node) x (4 nodes) x (24 hours) x (45 calculations) = 155,520 core-hours (b) (36 cores/node) x (4 nodes) x (6 hours) x (45 calculations) = 38,880 core-hours (c) (36 cores/node) x (4 nodes) x (6 hours) x (45 calculations) = 38,880 core-hours (3) (64 cores/node) x (4 nodes) x (6 hours) x (6 charged states) x (45 systems) = 414,720 core-hours The total request is 993,600 core-hours. Dr. Arun Mannodi-Kanakkithodi and Dr. Maria Chan will be responsible for carrying out the simulations. Industry partnership: N/A Project URL: Current FY Hours Used: undetermined amount New FY Requested allocation: 993600 Q1: 250000 Q2: 250000 Q3: 250000 Q4: 243600 Justification: The VASP code is tested on NERSC with a 350 electron system and 65%-80% scaling efficiency is obtained 64 to 128 cores compared to 16 cores. Given this performance, the VASP code has the efficiency to simulate large number configurations in reasonable amount of time. Further, the increased efficiency of VASP simulations with Haswell36 nodes has been well documented [8]. REFERENCE 8. C. H. Chang, H. Long, S. Sides, D. Vaidhynathan, and W. Jones, “Parallel Application Performance on Two Generations of Intel Xeon HPC Platforms”, National Renewable Energy Laboratory, Technical Report NREL/TP-2C00-64268 (2015) Storage requirements: 1 TB Thank You, The LCRC Accounts System