Electronic Properties of Flexible Systems • Introduction •UNO-CAS Tim Clark •FRET •SHG in membranes •Very large scale MO •SAMFETs Computer-Chemie-Centrum and Excellence Cluster “Engineering of Advanced Materials” Friedrich-Alexander-Universität Erlangen-Nürnberg Tim.Clark@chemie.uni-erlangen.de Centre for Molecular Design University of Portsmouth Tim.Clark@port.ac.uk 1 Acknowledgements • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs • • • • • • • Dr. Harry Lanig Dr. Frank Beierlein Dr. Catalin Rusu Dr. Matthias Hennemann Dr. Christof Jäger Dr. Olaf Othersen Pavlo Dral M.Sc. • • • • Prof. Siegfried Schneider (FRET) Prof. Carola Kryschi (SHG) Prof. Nigel Richards (EMPIRE) Prof. Markus Halik (SAMFETs) € Deutsche Forschungsgemeinschaft (DFG) € Bavarian State Government (KONWIHR) 2 Modeling • The Hamiltonian • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs – Force field – no electronics, but good sampling and geometries – Semiempirical MO/CI – CC-DFTB/TD-CC-DFTB No good for charge transfer – DFT/TDDFT – Ab initio Can‘t do large systems • SAMPLING !!!! – Molecular dynamics – QM/MM electronics 3 Semiempirical MO Theory • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs • Is very fast – Can therefore handle either very large systems or very many smaller ones • Generally gives very good one-electron properties – because the semiempirical electron density is good – because the parameterization probably used a related property – Because the MEP is good, solvent effects are also good • Semiempirical CI is good for excited states – Also better for frontier orbital energies than “higher” levels of theory • Is therefore ideal for calculating the properties of many “hot” geometries (snapshots) from MD simulations to obtain ensemble properties 4 Topics • UNO-CAS for Band Gaps • Introduction • Simulating FRET in Biological Systems •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs • Simulating SHG in Biological Membranes • EMPIRE – Very Large massively parallel Semiempirical MO calculations • Self-Assembled Monolayer Field-Effect Transistors (SAMFETs) 5 • Introduction •UNO-CAS •FRET Semiempirical UNO-CAS for Optical Band Gaps Pavlo Dral •SHG in membranes •Very large scale MO •SAMFETs 6 UNO-CAS • UHF Natural Orbital – Complete Active Space configuration interaction • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO – J. M. Bofill and P. Pulay, J. Chem. Phys. 1989, 90, 3637. – Semiempirical UNO-CAS and UNO-CI: Method and Applications in Nanoelectronics, P. O. Dral and T. Clark, J. Phys. Chem. A, 2011, 115, asap (DOI: 10.1021/jp204939x). •SAMFETs P U U T 7 UHF Natural Orbitals (UNOs) • Diagonalize the total ( + ) UHF density matrix • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs • The eigenvectors are the UHF Natural orbitals and the Eigenvalues are the UNO occupation numbers (0 or 2 for RHF, partial values between 0 and 2 for UHF) • Significant Fractional Occupation Numbers (SFONs) between 0.02 and 1.98 define the active space 8 Advantages • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs • The active space defined by the SFONs is usually small enough to allow a full CI calculation (UNO-CAS) • A CI-Singles (CIS) or CISD approach can be used for larger active spaces • The active space is defined automatically • UNOs contain some multi-reference information derived from the components of the UHF wavefunction 9 Disadvantages • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO • It is sometimes very difficult to find the correct UHF wavefunction (there may be many solutions close in energy) • Only applicable for systems that exhibit RHF/UHF instability (symmetry breaking) •SAMFETs 10 Calculated Band Gaps: Polyynes • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs 11 Polyacene band gaps • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs 12 Optical Properties • Two examples – Fluorescence resonant energy transfer (FRET) in TetR (S. Schneider) – Second-harmonic generation (SHG) by dyes in biological membranes (C. Kryschi) • Introduction •UNO-CAS •FRET • A Numerical Self-Consistent Reaction Field (SCRF) Model for Ground and Excited States in NDDO-Based Methods, G. Rauhut, T. Clark and T. Steinke, J. Am. Chem. Soc., 1993, 115, 9174. • NDDO-Based CI Methods for the Prediction of Electronic Spectra and Sum-Over-States Molecular Hyperpolarizabilities, T. Clark and J. Chandrasekhar, Israel J. Chem., 1993, 33, 435. • A Semiempirical QM/MM Implementation and its Application to the Absorption of Organic Molecules in Zeolites, T. Clark, A. Alex, B. Beck, P. Gedeck and H. Lanig, J. Mol. Model. 1999, 5, 1. •SHG in membranes •Very large scale MO •SAMFETs • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO FRET in the Tetracycline Repressor Frank Beierlein, Prof. Siegfried Schneider, Harry Lanig, Olaf Othersen Simulating FRET from Tryptophan: Is the Rotamer Model Correct? , •SAMFETs F. R. Beierlein, O. G. Othersen, H. Lanig, S. Schneider and T. Clark, J. Am. Chem. Soc. , 2006 , 128 , 5142-5152. 14 FRET (SFB 473) Tetracycline Tryptophan • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs One monomer of the Tetracycline Repressor (TetR) Protein The Experimental Problem • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs • Fluorescence decay in the protein is biexponential • Usually treated using the “rotamer model” – Each individual exponential decay process can be attributed to a corresponding tryptophan rotamer – Differences in distance and, above all orientation, relative to the acceptor (tetracycline) give different decay rates (Förster theory) – Is this model correct? Chromophores Tryptophan Two low-lying excited states • Introduction 1L , a polar, solvent sensitive, usually the emitting state (~350nM) •UNO-CAS •FRET 1L , b •SHG in membranes •Very large scale MO •SAMFETs Tetracycline:Mg2+ “BCD” Chromopohore Absorption overlaps with tryptophan emission, making FRET possible non-polar Glycyltryptophan Absorbance Spectra (H2O) • Introduction - Experimental •UNO-CAS - SCRF ( = 78.36) •FRET - QM/MM (explicit water) •SHG in membranes •Very large scale MO •SAMFETs Tryptophan Transition Dipoles • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs From above the ring In the ring plane 10% of the calculated snapshots shown Rotamer Distribution • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs Einstein Coefficients (no FRET) • Introduction - Total •UNO-CAS - Rotamer 1 •FRET - Rotamer 2 •SHG in membranes •Very large scale MO •SAMFETs FRET Rate Constants (Förster theory) • Introduction - Total •UNO-CAS - Rotamer 1 •FRET - Rotamer 2 •SHG in membranes •Very large scale MO •SAMFETs Exponential Fits Total without FRET • Introduction •UNO-CAS No. of Exponentials •FRET •SHG in membranes Rotamer 1 Rotamer 2 with with FRET FRET Total with FRET 1 2 2 2 (ns) 4.65 4.03, 1.76 3.65, 1.70 3.94, 1.74 Coefficient(s) (%) 100 57, 43 66, 33 59, 41 •Very large scale MO •SAMFETs Fit for the total is approximated well by the weighted average of the parameters for the individual rotamers, not as two individual decay components. FRET Conclusions • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO • Individual rotamers with significant lifetimes can be identified in the MD simulations • Including FRET makes the decay curves biexponential for each rotamer • Biexponentiality is caused by the distribution of the FRET rates, rather than by individual rotamers • “Spectroscopic Ruler” distances may be in error by as much as 6 Å if the orientation factor is not considered explicitly •SAMFETs • Simulating FRET from Tryptophan: Is the Rotamer Model Correct?, F. R. Beierlein, O. G. Othersen, H. Lanig, S. Schneider and T. Clark, J. Am. Chem. Soc., 2006, 128, 5142-5152. • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO SHG in Biological Membranes Catalin Rusu, Prof. Carola Kryschi, Harry Lanig Monitoring Biological Membrane-Potential Changes: a CI QM/MM Study •SAMFETs C. Rusu, H. Lanig, T. Clark and C. Kryschi, J. Phys. Chem. B , 2008 , 112 , 2445-2455 25 SHG in Membranes • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs • Second-harmonic generation (SHG) has been used recently to monitor action potentials (AP) in cardiomyocytes or neurons • The intensity of the SHG (ISHG) is monitored as a function of the transmembrane potential • Di-8-ANEPPS was used as a typical lipophilic dye that is incorporated into the membrane • The simulation system consisted of one dye molecule, 63 DPCC lipid molecules and 3,840 water molecules SO3- N+ (H2C)7 CH3 N (CH2)7 CH3 The Simulation System • Introduction •UNO-CAS • Water: blue • Lipids: green (head groups bold) • Dye: red •FRET •SHG in membranes •Very large scale MO •SAMFETs • GROMOS force field with optimized Lennard-Jones parameters for lipids • Periodic boundary conditions • PME electrostatics, NPT ensemble • 10 ns equilibration + 10 ns production MD • 700 snapshots per trajectory (last 7 ns of the production phase) QM-CI/MM Snapshots • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs • Di-8-ANEPPS used as the QM-part (chromophore, 91 atoms) • MM surroundings (DCCP + water) consisted of 14,700 atoms • 18 active orbitals • 18 active electrons • Single + pair-double excitations • QM/MM = 4.0 • Excitation energy = 1.17 eV (for sum-overstates ) Trans-Membrane Potential • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs • External potential applied to the QM-CI/MM calculations • Change in dye dipole moment in vacuo used to calibrate the system • External potential then adjusted to give a local potential at the dye of 0.1 V • Three calculations at +0.1, 0.0 and 0.1 V for each snapshot • Total simulated AP is therefore 0.2 V (about twice as large as in the experiment) Dye – Vertical Stability • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs z-coordinate of simulation cell [nm] 8 5 2 3 4 5 6 7 time [ns] 8 9 10 Calculated ISHG (V = 0.2V) 30 25 • Introduction •FRET •SHG in membranes •Very large scale MO •SAMFETs Counts •UNO-CAS 20 15 10 5 0 15 20 25 30 35 40 45 50 55 Delta I SHG [%] MD1 MD2 Simulation 1: ISHG = 41.6 11.1 % Simulation 2: ISHG = 43.2 13.0 % Experiment: ISHG 40 % 60 65 70 75 80 SHG Conclusions • The qualitative picture of the dye in the membrane is correct • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs • The MD simulations give lateral diffusion rates several orders of magnitude higher than those deduced from experiment – Force-field problem (van der Waals)? – Experimental interpretation ? • SHG enhancement of the order found in the experimental studies is also found in the simulations • C. F. Rusu, H. Lanig, O. G. Othersen, C. Kryschi and T. Clark, to be submitted to J. Am. Chem. Soc. (2007) • Introduction •UNO-CAS •FRET EMPIRE: A Very Large Scale Parallel Semiempirical SCF Program Matthias Hennemann •SHG in membranes •Very large scale MO •SAMFETs 33 The Big Hammer Approach • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs Develop a completely new semiempirical MO Program (EMPIRE) ; design specifications: • Neither LMO nor D&C • Need to treat conjugated systems • Massively parallel: • SCF 50,000 Atoms using 1,000 cores • Configuration Interaction (CI) 5,000 Atoms using 1,000 cores • Program • Direct on-the-fly calculation of the 2-electron integrals and the one-electron matrix • Avoid matrix diagonalization 34 Comparison with VAMP 910 Atoms 1,960 Orbitals VAMP 11 Cycles 59 Seconds (1 Core) • Introduction •UNO-CAS •FRET •SHG in membranes EMPIRE 16 Cycles 58 Seconds (1 Core) 7.8 Seconds (12 Cores) •Very large scale MO •SAMFETs 35 Scaling on one Node • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs Dual-Hex-Core Xeon 5650 “Westmere” 2.66 GHz (@ 2.93 GHz) with 12 MB cache per chip und 24 GB RAM. 36 Benchmark results: Adamantane 666 11,232 Atoms 24,192 Orbitals • Introduction 412 Cores: 78.4 Minutes •UNO-CAS •FRET •SHG in membranes 812 Cores: 44.3 Minuten •Very large scale MO 1612 Cores: 25.6 Minuten •SAMFETs 22 Cycles 37 Benchmark-Results: HLRB II • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs HLRB II: 9,728 Cores - 512 per Partition: 1.6 GHz dual core Itanium 2 “Montecito”, 4 GB RAM per Core, NUMAlink 4 with 6,4 GByte/s per link und direction 38 Hard Scaling (LiMa) • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs LiMa 500 Dual-Hex-Core Xeon 5650 “Westmere” 2,66 GHz (@ 2.93 GHz) 12 MB Cache per Chip 24 GB RAM per Node Infiniband with 40 Gbit/s per link and direction 39 Application: Organic Field-Effect Transistors • Introduction •UNO-CAS 0 •FRET •SHG in membranes •Very large scale MO •SAMFETs • molecular scale electronic devices with pure and mixed SAMs • relation of device characteristics on molecular structure and SAM composition • SAMs as important dielectric and bifunctional layers in condensers and FETs 40 Application: Organic Field-Effect Transistors • Constructed of self-assembled monolayers (SAMs) • Introduction •UNO-CAS •FRET • Head groups such as fullerenes can function as the semiconductor • No additional semiconductor layer necessary •SHG in membranes • Properties vary widely •Very large scale MO • Can an adequate permanent semiconductor layer be attained? •SAMFETs C10PA + C60C18PA • Classical MD simulations with AM1 single-points on snapshots • Prof. Marcus Halik C60C18PA 41 C10PA + C60C18PA - Monolayer 6,050 Atoms 15,950 Orbitals • Introduction 25 Minutes (812 Cores) •UNO-CAS •FRET 36 Cycles •SHG in membranes At the moment: 50 Snapshots •Very large scale MO •SAMFETs 42 Local Electron Affinity (EAL) • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs 43 Section through the SAM (EAL) • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs 44 Section through the SAM (EAL) • Introduction •UNO-CAS •FRET •SHG in membranes •Very large scale MO •SAMFETs 45