Verifying Reactant and Product Connectivity

Introduction

Mechanism diagrams usually put a reactant on the left and a desired product on the right before any calculation begins. A computed transition structure may look compatible with that arrow, yet connect different minima. Endpoint verification asks a narrower, testable question: which structures lie downhill on the two sides of this particular saddle? Its answer can reveal an unexpected intermediate, a rotated conformer, or a complex that must dissociate before the named product appears.

Core explanation

A first-order saddle has one unstable local direction. A frequency calculation identifies its Hessian index and suggests the immediate atomic motion, but the eigenvector is only a local straight-line approximation. Further downhill, the path can curve. Therefore one follows both IRC branches, saves a sequence of geometries, and optimises the final structures to stable minima. Each endpoint must be identified using atomic composition, formal charge or electron count, spin state, bond lengths, stereochemistry and conformer geometry as appropriate.

The correct comparison is more careful than a single drawing. Bond perception from distances is useful, but a stretched bond in a loose complex can be ambiguous. Compare several internal coordinates and electronic structure indicators when necessary. A purported proton-transfer endpoint, for example, needs a clear donor–H and acceptor–H comparison and a plausible charge assignment. For organometallic steps, the coordination environment and oxidation-state interpretation may matter. For radical reactions, spin and electronic-state consistency are essential; an IRC on one electronic surface cannot silently jump to another.

Reactants and products in an experimental equation may be separated molecules, while the IRC ends at pre-reaction or post-reaction complexes . Suppose A and B associate weakly, rearrange through a saddle, and yield a bound C···D complex. The IRC can verify the bond-changing step between the two complexes. Association into the reactant complex and dissociation into separate C and D require their own energetic and kinetic treatment. Reporting a single saddle as the full reaction barrier without accounting for the correct reference state can distort an activation free energy.

Stereochemical identity also matters. A computed path may lead to an enantiomer, diastereomer or conformer different from the named target. Atom labels should be tracked across the path, because a symmetrically equivalent atom can make a naive bond-change report misleading. If the branch terminates on a plateau or repeatedly fails numerical integration, its endpoint is unresolved rather than proven. Try smaller steps, altered algorithms, a displaced-geometry optimisation or a related path method; report remaining ambiguity honestly.

Finally, verify that both paths were calculated at a consistent electronic level and with appropriate solvent and charge settings. Comparing a reactant from one model to a saddle from another is not endpoint verification. Even once connectivity is established, barrier accuracy, competing paths and measured observables remain separate questions. Local connectivity is a necessary link in a mechanism, not the entire mechanism.

Step-by-step reasoning

Check that the candidate geometry is a first-order saddle and inspect its unstable mode. Follow both downhill IRC branches at the same computational level. Optimise each endpoint and confirm that no imaginary internal frequencies remain. Compare the two optimised structures with the proposed species using atom maps, bond lengths, charges and stereochemistry. Distinguish bound complexes from free molecules. If one endpoint differs from the proposed scheme, redraw the elementary step before searching for any missing subsequent step.

Visual explanation

Draw a saddle as a central pass with two winding descending paths. Put a question mark at each far end until the endpoint minimisations are complete. Show a proposed B structure above one end and an actual C structure below it, with the differing proton location circled. A second panel can show A + B approaching to form an A···B complex before crossing the saddle and C···D separating afterward.

Real-world analogy

A road sign at a mountain pass may point generally toward a city, but following each road identifies the actual valleys and intermediate towns. A snapshot at the pass cannot establish the end of a winding route. Likewise, an animated imaginary mode points initially downhill but does not tell the complete chemical destination. The analogy is about path verification, not travel time.

Real-world example

In a substitution model, the reaction drawing shows an incoming nucleophile replacing a leaving group. A computed saddle's reverse IRC reaches the expected reactant ion pair, while the forward branch reaches a tight contact pair of product and leaving ion. The desired free products are not themselves the immediate IRC endpoint. If solvent stabilises separated ions, dissociation may be favorable, but that step must be considered separately when relating computed energies to solution kinetics.

Why?

Why not infer endpoints from the imaginary mode? It is a local linear motion, whereas the route can bend across many coordinates. Why optimise the terminal geometry? A numerical path may stop close to a minimum without reaching its stationary point. Why preserve atom mapping? Distinct substitutions or hydrogen positions can look alike in an unlabeled sketch. Why check charge and spin? Geometry alone cannot establish electronic identity.

Common misconception

Seeing the expected bond length change in a transition structure does not establish the expected reaction. A saddle can connect two conformers or lead to an intermediate. Another error is treating a post-reaction complex as identical to infinitely separated products when calculating a barrier: association and separation change both energy and entropy references.

Worked example

Question: A candidate saddle for A + B → C + D has one imaginary mode involving bond exchange. The reverse IRC ends at A···B and the forward IRC at C···D. Does this prove the barrier from separated A + B to separated C + D?

Reasoning: The path supports local connectivity between the two associated complexes. It says nothing by itself about the free-energy cost of bringing A and B together or of separating C and D. Those steps affect a rate constant and depend strongly on concentration standard state and environment. The single electronic saddle-to-complex energy difference is one part of the full profile.

Answer: No. It verifies the complex-to-complex elementary step; association, dissociation and free-energy corrections still need evaluation.

Quick check

1. Why must both IRC branches be followed from a proposed transition structure? Answer: Only both branches identify the two minima that the saddle actually connects on the calculated surface.

Exam focus

Use the words “local connectivity” precisely. Give an ordered verification workflow: first-order saddle, both downhill paths, optimised endpoint minima, structural and electronic identity. Distinguish a complex from separated species and state what further evidence is needed for kinetics.

Advanced insight

Automated molecular-graph assignment can fail for stretched transition structures and weakly bound ion pairs, so a binary bond table should not overrule geometry and electronic analysis. In bifurcating landscapes, one nominal IRC can enter a valley that does not capture most finite-energy trajectories. For nonadiabatic processes, an electronic-state crossing complicates any assertion that a single-surface path reaches the experimental product. These cases expand the verification problem rather than eliminating its need.

Summary

Follow both sides of a proposed saddle to determine its actual adjacent minima. Optimise those endpoints, inspect structural and electronic identity, and distinguish complexes from free reagents or products. If the result differs from the original reaction drawing, revise the elementary step. Connectivity verification supports a mechanistic assignment but does not determine rate dominance or experimental branching by itself.

Practice questions

1. An IRC reaches an unexpected tautomer on one side. What should be claimed? Answer: The calculated saddle connects to that tautomer; the originally proposed product needs another pathway or subsequent step.

2. Why can a distance-based bond table mislabel a weak product complex? Answer: Noncovalent contacts and stretched bonds do not have a unique distance cutoff, so multiple structural indicators are needed.

3. What must remain consistent between a saddle and its IRC endpoint calculations? Answer: Electronic method, charge, spin state and relevant environmental model should be consistent for the same surface.

4. An IRC branch stops without a clear minimum. Is its product verified? Answer: No. Continue or use another path and optimise an endpoint before assigning connectivity.

Sources: IUPAC Gold Book, intrinsic reaction coordinate; Journal of Chemical Theory and Computation, transition-state characterisation; Journal of Chemical Information and Modeling, reaction-path endpoint verification.