IRC / Reaction Path

Simulate > IRC / Reaction Path... follows the reaction path downhill from a transition state, in both directions, towards the reactants and the products. It is the natural next step after a Transition state search (Baker).

The method is a steepest-descent reaction path. It requires a converged first-order saddle point: the Hessian is computed once at the starting structure and its single negative-eigenvalue mode is followed downhill in both directions, producing the reactant-side and product-side branches one after the other. If the Hessian of the starting structure does not have exactly one negative eigenvalue, the structure is rejected.

The window (Figure 1)

• Coordinates: the transition state structure.

• Log Frequency (default 1): one line is written to the log every N steps.

• Max. iterations (per branch) (default 50): the number of steps taken in each direction.

• Function step (default 0.2): the size of the very first displacement away from the saddle point, along the negative-eigenvalue eigenvector (larger values give a bigger initial jump off the saddle).

• Path step (default 0.025): the step taken along the gradient direction at every iteration after the initial displacement. Smaller values trace the path more finely but need more iterations to cover the same distance.

• Mass-weighted: uses mass-weighted coordinates, matching the classic (Fukui) definition of the intrinsic reaction coordinate. Off by default (plain Cartesian steepest descent).

• Save Trajectory / Working Folder / Trajectory Frequency: both branches are written into a single trajectory, in the order they are produced. Saving cannot be disabled: the trajectory is the only meaningful output of this tool.

Reading the result

Each branch always takes the full number of iterations, at a fixed step size, with no convergence check. The final structure is therefore not a meaningful chemical end-point by itself. Always inspect the saved trajectory (the RMS Gradient column of its log) and pick, for each branch, the frame where the gradient reaches its minimum. Those frames can then be optimized with Geometry Optimization to obtain the reactant and product minima connected to the transition state.