1.021, 3.021, 10.333, 22.00 : Introduction to Modeling and Simulation : Spring 2012 Part II – Quantum Mechanical Methods : Lecture 3 From Many-Body to Single-Particle: Quantum Modeling of Molecules Jeffrey C. Grossman Department of Materials Science and Engineering Massachusetts Institute of Technology 1 Part II Topics 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 2 It’s a Quantum World:The Theory of Quantum Mechanics Quantum Mechanics: Practice Makes Perfect From Many-Body to Single-Particle; Quantum Modeling of Molecules Application of Quantum Modeling of Molecules: Solar Thermal Fuels Application of Quantum Modeling of Molecules: Hydrogen Storage From Atoms to Solids Quantum Modeling of Solids: Basic Properties Advanced Prop. of Materials:What else can we do? Application of Quantum Modeling of Solids: Solar Cells Part I Application of Quantum Modeling of Solids: Solar Cells Part II Application of Quantum Modeling of Solids: Nanotechnology Motivation Last time: 1-electron quantum mechanics to describe spectral lines ? Image of NGC 604 nebula is in the public domain. Source: Hubble Space Telescope Institute (NASA). Via Wikimedia Commons. Image adapted from Wikimedia Commons, http://commons.wikimedia.org. . Today: many electrons to describe materials. ? 3 Courtesy of Bernd Thaller. Used with permission. Lesson outline • Review • The Many-body Problem • Hartree and Hartree-Fock • Density Functional Theory • Computational Approaches • Modeling Software • PWscf 4 Review: Schrödinger equation H time independent: in f˙(t) f (t) = 1(⌃ r , t) = 1(⌃ r ) · f (t) H1(r r) 1(r r) H (⇧ r ) = E (⇧ r) time independent Schrödinger equation stationary Schrödinger equation 5 = const. = E /(⌃ r , t) = /(⌃ r) · e - i Et Review:The hydrogen atom stationary Schrödinger equation Hl = El T +V 2 e v l o ts jus 2m ☺ 2 2m 6 ⇥ 2 ⇥2 + V e2 4⇥ 0r ⇥ ⇥ ⇥ =E (⌃ r ) = E (⌃ r) ⇤(r) = E⇤(r) Radial Wavefunctions for a Coulomb V(r) R10 r2R210 2 0.4 1 Z 0 0.2 0 1 2 3 4 R20 2 6 Thickness dr y R30 r rR 2 0 1 2 3 4 r 2 6 10 r 0 2 6 10 r 0 2 6 10 r 2 21 0.15 0.1 0.05 0 rR 2 0.4 2 30 0.08 0.2 4 0 -0.1 8 12 16 0.04 r R31 X 0.15 0.1 0.05 0 10 R21 r 0 r2R220 0.6 0.4 0.2 0 -0.2 0.12 0.08 0.04 0 0 r 0 0 4 8 12 16 r 0 4 8 12 16 r 0 4 8 12 16 r r2R231 0.08 0.8 0.04 4 0 -0.04 8 12 16 r 0.4 0 r2R232 R32 0.04 0.8 0.02 0 0.4 0 4 8 12 16 r 0 Radial functions Rnl(r) and radial distribution functions r2R2nl(r) for atomic hydrogen. The unit of length is aµ = (m/µ) a0, where a0 is the first Bohr radius. 7 Images by MIT OpenCourseWare. Angular Parts Image by MIT OpenCourseWare. 8 Review:The hydrogen atom quantum numbers n 1 2 2 2 l 0 0 1 1 ml 0 0 0 ±1 a0 = 9 Atomic Orbital 1s ψn l m (r, θ, φ) l 1 -r/a0 e π a03/2 2s 1 r -r/2a0 2 e 3/2 4 2π a0 a0 2p 1 r -r/2a0 e cos θ 3/2 a 4 2π a0 0 2p 1 r -r/2a0 ±iφ e sin θ e 8 π a03/2 a0 2 h = .0529 nm = first Bohr radius 2 me Image by MIT OpenCourseWare. 10 Courtesy of David Manthey. Used with permission. Source: http://www.orbitals.com/orb/orbtable.htm. Review:The hydrogen atom 11 © R. Nave. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/. Review:The hydrogen atom © unknown. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/. 12 Review: Next? Helium e- r1 + r2 r12 e- H1 = E1 H 1 + H2 + W T1 + V1 + T2 + V2 + W 2 2m ⇥21 e2 4⇥ 0 r1 2 2m ⇥22 cannot be solved analytically 13 e2 4⇥ 0 r2 + ⇥ ⇥ (⌥ r1 , ⌥ r2 ) = E (⌥ r1 , ⌥ r2 ) (r1 , r2 ) = E (r1 , r2 ) e2 4⇥ 0 r12 ⇥ ⇤(r1 , r2 ) = E⇤(r1 , r2 ) problem! Review: Spin new quantum number: spin quantum number for electrons: spin quantum number can ONLY be up 14 down Everything is spinning ... Stern–Gerlach experiment (1922) ⇧ F = ⇤E ⇧ = ⇤m ⇧ ·B Image courtesy of Teresa Knott. 15 Everything is spinning ... In quantum mechanics particles can have a magnetic moment and a ”spin” magnetic moment 16 m ⇥ spinning charge Everything is spinning ... conclusion from the Stern-Gerlach experiment for electrons: spin can ONLY be up 17 down Spin History Discovered in 1926 by Goudsmit and Uhlenbeck Part of a letter by L.H. Thomas to Goudsmit on March 25 1926 (source: Wikipedia). © Niels Bohr Library & Archives, American Institute of Physics. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/. 18 Pauli’s exclusions principle Two electrons in a system cannot have the same quantum numbers! hydrogen quantum numbers: main n: 1,2,3 ... orbital l: 0,1,...,n-1 magnetic m: -l,...,l spin: up, down 19 ... 3s 2s 1s ... ... 3p 3d 2p ... Pauli Exclusion Principle “Already in my original paper I stressed the circumstance that I was unable to give a logical reason for the exclusion principle or to deduce it from more general assumptions. I had always the feeling, and I still have it today, that this is a deficiency.” W. Pauli, Exclusion Principle and Quantum Mechanics, Nobel prize acceptance lecture, Stockholm (1946). Image via Wikimedia Commons. License: CC-BY-SA. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/. 20 Periodic table This image is in the public domain. Source: Wikimedia Commons. 21 The many-body problem helium: 2e iron: 26e er1 r 12 + er2 = (⇧ r1 , . . . , ⇧ rn ) Image © Greg Robson on Wikimedia Commons. License: CC-BY-SA. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/. 22 Dirac Quotes Year 1929… The underlying physical laws necessary for the mathematical theory of a large part of physics and the whole of chemistry are thus completely known, and the difficulty is only that the exact application of these laws leads to equations much too complicated to be soluble. P.A.M. Dirac, Proc. Roy. Soc. 123, 714 (1929) ...and in 1963 If there is no complete agreement […] between the results of one’s work and the experiment, one should not allow oneself to be too discouraged [...] P.A.M. Dirac, Scientific American, May 1963 23 The Multi-Electron Hamiltonian Hl = El Remember the good old days of the 1-electron H-atom?? 2 2m e2 ⇥2 4⇥ 0r They’re over! 2 ⇥ ⇤(r) = E⇤(r) 1 h Zi e 1 e h 2 2 ∇ Ri + ∑∑ − ∇ ri − ∑∑ + ∑∑ H = −∑ ∑ 2M i 2 i=1 j=1 ri − r j 2 i=1 j =1 Ri − R j 2m i=1 i=1 i=1 j=1 R i − r j N N N 2 Zi Z j e 2 n N n n 2 i≠ j kinetic energy of ions n 2 i≠ j kinetic energy of electrons potential energy of ions electron-ion interaction electron-electron interaction Multi-Atom-Multi-Electron Schrödinger Equation H (R1 ,..., R N ;r1 ,..., rn ) Ψ (R1 ,..., R N ;r1 ,..., rn ) = E Ψ (R1 ,..., R N ;r1 ,..., rn ) Born-Oppenheimer Approximation Source: Discover Magazine Online 25 © Discover Magazine Online. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/. Born-Oppenheimer Approximation (skinless version) • mass of nuclei exceeds that of the electrons by a factor of 1000 or more • we can neglect the kinetic energy of the nuclei • treat the ion-ion interaction classically • significantly simplifies the Hamiltonian for the electrons: Images in public domain. Born Oppenheimer This term is just an external potential V(rj) Zi e 1 e h 2 ∇ ri − ∑∑ + ∑∑ H =− ∑ 2m i=1 2 i=1 j=1 ri − r j i=1 j=1 R i − r j 2 26 n N n 2 n n i≠ j 2 Solutions quantum chemistry Moller-Plesset perturbation theory MP2 27 density functional theory coupled cluster theory CCSD(T) Hartree Approach Write wavefunction as a simple product of single particle states: Ψ(r1,...,rn ) = ψ1 (r1 )ψ 2 (r2 )...ψ n (rn ) Hard Product of Easy Leads to an equation we can solve on a computer! 28 ( , 2 2 n 2 e ψ j (r ) * * h 2 ∇ +Vext (r) + ∑ ∫ dr )− -ψi (r) = εiψi (r ) rj − r * * 2m j=1 j≠i + . Hartree Approach ( , 2 n e2 ψ j (r ) * * h2 2 ∇ +Vext (r) + ∑ ∫ dr )− -ψi (r) = εiψi (r ) rj − r * * 2m j=1 j≠i + . The solution for each state depends on all the other states€ (through the Coulomb term). 29 • we don’t know these solutions a priori • must be solved iteratively: - guess form for { Ψiin(r) } - compute single particle Hamiltonians - generate { Ψiout(r) } - compare with old in(r) } = { Ψ out(r) } and repeat if different set { Ψ i i - if same, you are done • obtain the self-consistent solution Simple Picture...But... Interacting - - Non-Interacting - After all this work, there is still one major problem: the solution is fundamentally wrong The fix brings us back to spin! 30 Symmetry Holds the Key Speculation: everything we know with scientific certainty is somehow dictated by symmetry. The relationship between symmetry and quantum mechanics is particularly striking. 31 Exchange Symmetry • all electrons are indistinguishable • electrons that made da Vinci, Newton, and Einstein who they were, are identical to those within our molecules a bit humbling... • so if • I show you a system containing electrons • you look away • I exchange two electrons in the system • you resume looking at system • there is no experiment that you can conduct that will indicate that I have switched the two electrons 32 Mathematically • define the exchange operator: χ12ψ1 (r1 )ψ2 (r2 ) = ψ1 (r2 )ψ2 (r1 ) • exchange operator eigenvalues are ±1: € suppose χ 12 φ = χφ ^! χ 12 χ 12 φ = χ φ = φ ^! ^! 2 χ = 1, or χ = ±1. 2 33 Empirically • all quantum mechanical states are also eigenfunctions of exchange operators - those with eigenvalue 1 (symmetric) are known as Bosons - those with eigenvalue -1 (antisymmetric) are known as Fermions • profound implications for materials properties - wavefunctions for our many electron problem must be anti-symmetric under exchange - implies Pauli exclusion principle 34 Hartree-Fock • Employing Hartree's approach, but - enforcing the anti-symmetry condition - accounting for spin • Leads to a remarkable result: ( 2, 2 n n e ψ j (r ) * * h2 2 ∇ +Vext (r) + ∑ ∫ dr )− -ψi (r) − ∑δsi ,s j rj − r * * 2m j=1 j=1 j≠i j≠i + . e2 * ∫ dr0 r0 − r ψ j (r0)ψi (r0)ψ j (r) = εiψi (r) • Hartree-Fock theory is the foundation of molecular 35 orbital theory. • It is based upon a choice of wavefunction that guarantees antisymmetry between electrons. it’s an emotional moment.... But...Hartree-Fock • neglects important contribution to electron energy (called “correlation” energy) • difficult to deal with: integral operator makes solution complicated • superceded by another approach: density functional theory 36 Solutions quantum chemistry Moller-Plesset perturbation theory MP2 37 density functional theory coupled cluster theory CCSD(T) Solving the Schrodinger Equation No matter how you slice it, the wavefunction is a beast of an entity to have to deal with. For example: consider that we have n electrons populating a 3D space. Let’s divide 3D space into NxNxN=2x2x2 grid points. To reconstruct Ψ(r), how many points must we keep track of? 38 Solving the Schrodinger Eq. N divide 3D space into NxNxN=2x2x2 grid N N points © source unknown. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/. Ψ= Ψ(r1, ... , rn) 39 # points = N3n n=# electrons Ψ(N3n) ρ(N3) 1 8 8 10 109 8 100 1090 8 1,000 10900 8 Working with the Density The electron density seems to be a more manageable quantity. Wouldn’t it be nice if we could reformulate our problem in terms of the density, rather than the wavefunction? Energy Electron Density E0=E[n0] 40 Walter Kohn (left), receiving the Nobel prize in chemistry in 1998. Why DFT? computational expense for system size N: Quantum Chemistry methods; MP2, CCSD(T) O(N5-N7) Density Functional Theory silicon 2 atoms/unit cell DFT: 0.1 sec 41 Example CCSD(T): 0.1 sec MP2: 0.1 sec O(N3); O(N) silicon 100 atoms/unit cell DFT: 5 hours CCSD(T): 2000 years!!! MP2: 1 year Why DFT? 100,000 10,000 e Number of Atoms n Li c s ar 1000 g n ali T F D DFT MP2 100 QMC CCSD(T) 10 1 42 Exact treatment 2003 2007 Year 2011 2015 Image by MIT OpenCourseWare. Density functional theory = (⇧ r1 , ⇧ r2 , . . . , ⇧ rN ) n = n(⇤ r) : n o i t c n u f e wav licated! comp ele de ctron nsi eas ty: y! Walter Kohn DFT 1964 © unknown. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/help/faq-fair-use/. 43 Density functional theory electron density n ion Total energy is a functional of the electron density. E[n] = T [n] + Vii + Vie [n] + Vee [n] kinetic ion-ion 44 ion-electron electron-electron © source unknown. All rights reserved. This content is excluded from our Creative Commons license. For more information, see http://ocw.mit.edu/fairuse.. Density functional theory E[n] = T [n] + Vii + Vie [n] + Vee [n] kinetic ion-ion ion-electron electron density n(⌅r) = i electron-electron r )|2 |ci (⌅ Eground state = min E[n] c Find the wave functions that minimize the energy using a functional derivative. 45 Density Functional Theory Finding the minimum leads to Kohn-Sham equations ion potential Hartree potential exchange-correlation potential equations for non-interacting electrons 46 Density functional theory Only one problem: vxc not known! approximations necessary local density approximation LDA 47 general gradient approximation GGA Self-consistent cycle Kohn-Sham equations Iterations to selfconsistency sc f lo n(⌅ r) = i |ci (⌅ r )| 2 3.021j Introduction to Modeling and Simulation - N. Marzari (MIT, May 2003) 48 op Modeling software name 49 license ABINIT free ONETEP pay Wien2k pay VASP pay PWscf free basis functions pro/con plane very waves structured Wannier linear functions scaling Ylm + very plane waves accurate plane fast waves plane fast waves Basis functions ⇥= Matrix eigenvalue equation: i expansion in orthonormalized basis functions H1 = E1 ci ci = E H ⇥ i d⌥ r jH ci i i =E ci ci ⇥i Hji ci = Ecj i H⇤ c = E⇤ c 50 ci ci d⌥ r ci j i i Plane waves as basis functions plane wave expansion: ⇧ j ·⇧ iG r (⇧ r) = cj e j plane wave Cutoff for a maximum G is necessary and results in a finite basis set. Plane waves are periodic, thus the wave function is periodic! periodic crystals: Perfect!!! (next lecture) 51 atoms, molecules: be careful!!! Image by MIT OpenCourseWare. Put molecule in a big box unit cell Make sure the separation is big enough, so that we do not include artificial interaction! Images by MIT OpenCourseWare. 52 scf loop DFT calculations total total total total total total total energy energy energy energy energy energy energy = = = = = = = -84.80957141 -84.80938034 -84.81157880 -84.81278531 -84.81312816 -84.81322862 -84.81323129 At the end we get: 53 Ry Ry Ry Ry Ry Ry Ry exiting loop; result precise enough 1) electronic charge density 2) total energy • structure • bulk modulus • shear modulus • elastic constants • vibrational properties • sound velocity 54 binding energies reaction paths forces pressure stress ... Convergence g i b x o b y m s a ? h W g u o en ? Was m y basis big enoug h? f c s e h t t i x e I t d h i g i D r e h t t a loop ? t n i po 55 PWscf input What atoms are involved? Where are the atoms sitting? How big is the unit cell? At what point do we cut the basis off? When to exit the scf loop? All possible parameters are described in INPUT_PW. 56 water.input &control pseudo_dir = ''” / &system ibrav =1 celldm(1) = 15.0 nat =3 ntyp =2 occupations = 'fixed' ecutwfc = 60.0 / &electrons conv_thr = 1.0d-8 / ATOMIC_SPECIES H 1.00794 hydrogen.UPF O 15.9994 oxygen.UPF ATOMIC_POSITIONS {bohr} O 0.0 0.0 0.0 H 2.0 0.0 0.0 H 0.0 3.0 0.0 K_POINTS {gamma} Review • Review • The Many-body Problem • Hartree and Hartree-Fock • Density Functional Theory • Computational Approaches • Modeling Software • PWscf 57 Literature • Richard M. Martin, Electronic Structure • Kieron Burke,The ABC of DFT chem.ps.uci.edu/~kieron/dft/ • wikipedia,“many-body physics”,“density functional theory”,“pwscf”, “pseudopotentials”, ... 58 MIT OpenCourseWare http://ocw.mit.edu 3.021J / 1.021J / 10.333J / 18.361J / 22.00J Introduction to Modelling and Simulation Spring 2012 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.