Geometry Optimization and Normal Coordinates Minimize EFF by varying the geometry. (Find minimum energy geometry.) Consider the 3N Cartesian coordinates, {x1, y1, z1, x2, y2, z2 … xN, yN, zN} or {x1, x2 , x3 … x3N}, trying to find geometry for which (d EFF / d xi) = 0 for all i = 1…3N As all the components of EFF depend either on the coordinates of individual atoms or on coordinates of pairs of atoms EFF = ∑i ∑j EFF (xi, xj) i and j = 1 … 3N Could write the energy as a matrix: E x , x E x , x E x , x E x , x E x , x FF FF FF i 1 1 2 FF 1 2 E x , x FF 1 3N 3N E x , x j FF 1 1 FF 3N 3N and then the total energy is the sum of all the terms. (Note There is a set of coordinates {q1, q2 , q3 … q3N} which are related to the Cartesian coordinates {x1, x2 , x3 … x3N} such that: E q , q 0 E q , q 0 FF FF 1 1 1 0 E q , q FF 2 2 1 0 E q , q FF 3N 3N these are the Normal Coordinates. EFF = ∑i EFF (qi, qi) 6 of the EFF (qi, qi) = 0 (translation and rotation) Any square, real symmetric matrix such as (EFF (xi, xj)) can by made ‘diagonal’ by a matrix A: (Aij-1) (EFF (xi, xj)) (Aij) = (EFF (qi, qj)) and this matrix relates the normal coordinates to the Cartesian coordinates: A A q A 11 q q 1 21 2 3N 3N1 A A A 12 13 N 3N 3N x x x 1 2 3N Even better is that the minimum energy geometry is given by (d EFF / d qi) = 0 for all i, which becomes: (d EFF(qi ,qi) / d qi) = 0. (d EFF(qi ,qi) / d qi) = the force along the coordinate qi , indicates how much the geometry should be moved in the direction of qi Read in geometry – convert to Cartesian Coordinates { xi } Use the Molecular Force Field model to calculate all EFF (xi, xj) Diagonalize matrix (EFF (xi, xj)) to get normal coordinates and total energy Calculate forces along normal coordinates – if all forces are zero output this geometry and energy and then STOP 5) Change geometry to reduce the forces – new { xi } 6) Return to 2 1) 2) 3) 4) Molecular Potential Energy Surfaces A plot of EFF vs {qi} is (apart from being physically impossible if there are more than 2 normal coordinates) is called a Potential Energy Surface. It carries considerable information about the molecule and its chemistry. Example: O3, fixed angle (to reduce number of normal coordinates to 2), plot of energy against the two bond lengths: O3 O2 + O ‘channel’ minimum energy geometry H2O, energy plotted against bond angle and both bond lengths – the symmetric stretch motion (no dissociation evident) A typical diatomic + atom reaction PES AB +C ABC A + BC E vs R(AB) and R(BC) reaction coordinate Characterization of PES In the last example, AB +C ABC A + BC, could plot E vs. reaction coordinate plot, Usual diagram where ABC may be an intermediate or a transition state. A minimum, (d EFF / d qi) = 0 for all i, corresponds to reactants, products, intermediates and transition states. Can distinguish intermediates and transition states the ‘double differential’ (d2 EFF / d qi d qj) > 0 for all ij pairs then the geometry is a stable minimum. If one just one of the (d2 EFF / d qi d qj) < 0 then it is necessary that i = j, the geometry is a ‘saddle point’ and qi is the reaction coordinate. Chemically the interesting positions on the PES are: global minimum, local minima, saddle points. A real molecular problem involves E vs (3N-6) qi – cannot plot diagrams, have to rely on mathematical methods to search the PES to find the important positions. (Applied mathematics and Computer Science – optimization theory .)