introduction to computational quantum chemistryintroduction to computational quantum chemistry...

30
Introduction to Computational Quantum Chemistry Lesson 8: Population analysis Martin Nov ´ ak & Pankaj Lochan Bora Population Analysis October 13, 2015 1 / 30

Upload: others

Post on 24-May-2020

17 views

Category:

Documents


1 download

TRANSCRIPT

Introduction to Computational Quantum Chemistry

Lesson 8: Population analysis

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 1 / 30

Importance

Pictures of orbitals are informative, however numerical values aremuch easier to quantified and compared. For example σ vs π bondingin organic molecules.

What does a population analysis deliver?Determination of the distribution of electrons in a moleculeCreating orbital shapeDerivation of atomic charges and dipole ( multiple ) moments

Methods of calculationBased on the wave function ( Mulliken, NBO)Based on the electron density (Atoms in Molecules)Fitted to the electrostatic potential (CHELPG, MK)

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 2 / 30

Mulliken Population Analysis

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 3 / 30

Mulliken Population Analysis

AdvantagesMost popular methodStandard in program packages like GaussianFast and simple method for determination of electron distributionand atomic charged

DisadvantageStrong dependance of the results from the level of theory (basisset or kind of calculation)

Example: Li-charge in LiF

Population basis set q(Li,RHF) q(Li,B3LYP)Mulliken STO-3G +0.227 +0.078

6-31G +0.743 +0.5936-311G(d) +0.691 +0.558

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 4 / 30

Natural Bond Orbital Analysis

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 5 / 30

Natural Bond Orbital Analysis

Based on the theory of Natural Orbitals by Lowdin.

Two parts of the methodsNPA→ Natural population analysis to identify the populationnumbersNBO→ Analysis of the bond order based on the electronpopulation obtained by NPA

AdvantagesSmaller dependence on the basis setbetter reproducibility for different moleculesOrientates itself at the formalism for Lewis formulas

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 6 / 30

Practical task

Draw HF molecule, optimize the geometry and generate G09input.Use pop=(nbo,savenbo) for NBO or Pop=Full for MullikenAfter Pop command, add a space and type ”FormCheck”Run the calculation

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 7 / 30

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 8 / 30

Visualizing the orbitals

Open the *FChk file in AvogadroClick on Extensions→ Create SurfaceSelect ”Molecular Orbital” as surface typeChoose the MO you want to visualize and calculate

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 9 / 30

You should be able to see something like these that shows the HOMOand LUMO of HF molecules

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 10 / 30

Lesson 8: Solvation models

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 11 / 30

Solvent effects

The solvent environment influences structure, energies, spectraetcShort-range effects

Typically concentrated in the first solvation sphereExamples: h-bonds, preferential orientation near an ion

Long-range effectsPolarization

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 12 / 30

Imlicit vs. Explicit solvation

Implicit solvationDielectric continuumNo water molecules per seWavefunction of solute affected by dielectric constant of solventAt 20 °C: Water - ε = 78.4; benzene: ε = 2.3 ...

Explicit solvationSolvent molecules included (i.e. with electronic & nuclear structure)Used mainly in MM approachesMicrosolvation: only few solvent molecules placed around soluteCharge transfer with solvent can occur

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 13 / 30

Implicit Models

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 14 / 30

Basic assumptions

Solute characterized by QM wavefunctionBorn-Oppenheimer approximationOnly interactions of electrostatic originIsotropic solvent at equilibriumStatic model

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 15 / 30

Cavity

Solute is placed in a void of surrounding solvent called “cavity”Size of the cavity:

Computed using vdW radii of atoms (from UFF, for example)Taken from the electronic isodensity level (typically ˜0.001 a.u.)

The walls of cavity determine the interaction interface (SolventExcluded Surface, SES)Size of the solvent molecule determines the Solvent AccessibleSurface (SAS)

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 16 / 30

Visualizing cavity

Geomview software (in the modules)SCRF=(read) in the route section of the jobUse G03Defaults in SCRF command“geomview” in the SCRF specificationVisualize the “tesserae.off” file

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 17 / 30

Electrostatic Interactions

Self-consistent solution of solute-solvent mutual polarizationsSolute induces polarization at the interface of cavityThis polarization acts back on the solute changing its wavefunctionVarious solvation models use different schemes for evaluation ofsolvation effectsProblems arise when electrostatics do not dominate solvent-soluteinteractions

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 18 / 30

Polarizable Continuum Model (PCM)

Treats the solvent as polarizable dielectric continuumInduced surface charged represent solvent polarizationImplemented in Gaussian, GAMESS

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 19 / 30

Solvation Model “Density” (SMD)

Full solute density is used instead of partial chargesLower unsigned errors against experimental data than othermodels

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 20 / 30

COnductor-like Screening MOdel (COSMO)

Solute in virtual conductor environmentCharge q on molecular surface is lower by a factor f(ε):

q = f(ε)q∗ (1)

where f(ε) = (ε− 1)/(ε+ x); x being usually set to 0.5 or 0Implemented in Turbomole, ADF

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 21 / 30

Beyond basic models

Anisotropic liquidsConcentrated solutions

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 22 / 30

Explicit Models

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 23 / 30

Two models:

MicrosolvationFew solvent molecules (1 to 3) put at chemically reasonable placeWater close to exchangeable protons (OH, NH2...)

MacrosolvationFirst (sometimes second) solvent layer around the whole moleculeUsually snapshots from MD

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 24 / 30

Pros & Cons

+++ Modelling of real interactions with solvent (this can be crucialfor exchangeable protons in protic solvents)- Microsolvation lacks sampling- Computationally more demanding- For macrosolvation only single point calculations - the geometryis as good as forcefield

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 25 / 30

Practical task

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 26 / 30

Reaction

Model the Cl− + CH3Br→ CH3Cl + Br−

Find the energy barrier for the reactionSelect any solvent from Gaussian library (be not concerned aboutsolubility of species or chemical relevance)Assume Sn1 and Sn2 reaction pathwaysUse “SCRF=(solvent=XY)” in the route section of the calculation

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 27 / 30

Procedure

Use B3LYP 6-31++g(d,p) methodUsage of difuse functions when dealing with anions is crucial!Use ultrafine integration gridUse Frequency calculations to be sure where on PES you areFor the scan use the distance between C and Cl as RCNegative value of step defines two atoms approaching

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 28 / 30

Module “qmutil”

Extraction of values from gaussian runs:extract-gopt-ene logfileextract-gopt-xyz logfileextract-gdrv-ene logfileextract-gdrv-xyz logfileextract-xyz-str xyzfile framenumberextract-xyz-numstr xyzfile

Values ready for plotting in your favorite software

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 29 / 30

Turbomole

Prepare job using define module (see presentation 6 for help)Setup COSMO using cosmoprep moduleSet epsilon to 78.4 and rsolv to 1.93

Leave all other values at their defaultDefine radii of atoms using “r all o” for optimized valuesOptimize all geometries

Martin Novak & Pankaj Lochan Bora Population Analysis October 13, 2015 30 / 30