Appendix E — Computational Chemistry Software Setup

Reference for the computational exercises across the book, plus enough method-selection guidance to extend your work beyond what we assigned. Computational chemistry is now standard infrastructure in modern organic — treat it as a sixth spectroscopic tool.

1. Why computational chemistry matters in modern organic

Three concrete uses appear throughout this book:

• Predicting reactivity — HOMO/LUMO surfaces (Ch 2, 19), electrostatic potential maps (Ch 3), Fukui indices for site selectivity in EAS (Ch 21).
• Transition state energies — locating TS structures to rationalize selectivity (Ch 10 SN2 vs SN1 partitioning, Ch 19 endo/exo Diels-Alder, Ch 39 sigmatropic stereospecificity).
• NMR shift prediction — DFT GIAO calculations to assign ambiguous diastereomers and natural product structures (Ch 6, Ch 38). Modern GIAO/DP4+ analysis frequently distinguishes regio- and stereoisomers when experimental NMR is ambiguous.

Other routine uses: IR frequency assignment, conformational scanning, pKa estimation, dipole moment, partial charges, NBO bonding analysis, NCI (non-covalent interaction) surfaces.

Limits to remember: computational chemistry estimates. Errors are typically ±2 kcal/mol for DFT on organic systems, ±0.1-0.3 ppm for ¹H NMR shifts, ±5 ppm for ¹³C. Reliability scales with system size and method cost — pick the method to match the question.

2. Tool tiers

Free / open source

Academic (license required)

Web / cloud

• WebMO — browser front-end to Gaussian/GAMESS/Q-Chem; departmental install
• Chemcraft — Windows visualization for output files (free for non-commercial)
• IQmol — free GUI tied to Q-Chem
• Chemcompute.org — free shared GAMESS/PSI4 access for students

For this book's exercises, the free stack (Avogadro + ORCA + RDKit) covers every problem.

3. Setting up Avogadro

avogadro.cc. Versions referenced here: Avogadro 1.2.x (classic, stable) and Avogadro 2.0+ (rewrite, faster, plugin-based). The book's screenshots use 1.2.x; 2.0 has identical menus for the operations we need.

Windows

1. Download the .exe installer from avogadro.cc.
2. Run installer. Default install path: C:\Program Files\Avogadro.
3. Launch from Start menu.

macOS

1. Download .dmg.
2. Drag Avogadro.app to /Applications.
3. On first launch: right-click → Open (bypasses Gatekeeper warning).

Linux

sudo apt install avogadro          # Debian/Ubuntu
sudo dnf install avogadro          # Fedora
flatpak install flathub cc.avogadro.Avogadro2   # Avogadro 2 via Flatpak

First-run sanity check

1. Click empty canvas → places C atom.
2. Drag → forms bond to new C.
3. Add Hs: Build → Add Hydrogens (or toolbar H button).
4. Optimize: Extensions → Optimize Geometry (default UFF).
5. Energy reported in status bar.

Force-field choice within Avogadro

• UFF — Universal; covers full periodic table; less accurate for organics.
• MMFF94 — Best default for closed-shell organics. Use this.
• GAFF — Better for biomolecules and ligands.
• Ghemical — Legacy; avoid.

4. Installing ORCA

orcaforum.kofo.mpg.de. Free for academic and personal non-commercial use; registration required before download. Commercial users need a license. Versions current at time of writing: ORCA 6.x.

Download

1. Register on the ORCA forum.
2. Download platform binary (Windows: .zip; Linux: .tar.xz; macOS: ARM/x86 builds).
3. Extract to a permanent path, e.g. C:\orca\ or /opt/orca/.

Environment variables (Windows PowerShell)

$env:Path = "C:\orca;$env:Path"
$env:OMP_NUM_THREADS = "4"

Add the same Path entry permanently in System Properties → Environment Variables.

Linux/macOS .bashrc or .zshrc

export PATH=/opt/orca:$PATH
export LD_LIBRARY_PATH=/opt/orca:$LD_LIBRARY_PATH
export OMP_NUM_THREADS=4

First input file — water DFT optimization

File water.inp:

! B3LYP def2-SVP Opt
%pal nprocs 4 end
* xyz 0 1
O   0.000000   0.000000   0.117790
H   0.000000   0.755453  -0.471161
H   0.000000  -0.755453  -0.471161
*

Run:

orca water.inp > water.out

Output appears in water.out. Optimized geometry in water.xyz; orbitals in water.gbw; thermochemistry in the .out tail.

Opt + Freq + single-point (typical workflow)

! B3LYP def2-SVP Opt Freq
! B3LYP def2-TZVP

%pal nprocs 8 end
%maxcore 3000

* xyz 0 1
[coordinates]
*

The second ! line is read as a single-point on the optimized geometry using a larger basis — the standard "optimize cheap, refine energy" pattern.

5. Setting up RDKit in Python

Conda (recommended)

conda create -n rdkit-env python=3.11
conda activate rdkit-env
conda install -c conda-forge rdkit

Pip

pip install rdkit

The pure-pip wheel works on Linux/macOS/Windows since RDKit 2022.09.

Minimum example — SMILES to 3D to FF-optimized

from rdkit import Chem
from rdkit.Chem import AllChem, Draw

# 2-bromobutane
mol = Chem.MolFromSmiles('CCC(C)Br')
mol = Chem.AddHs(mol)

# Embed 3D coords using ETKDG (Riniker-Landrum)
AllChem.EmbedMolecule(mol, AllChem.ETKDGv3())

# Optimize with MMFF94
AllChem.MMFFOptimizeMolecule(mol)

# Write .xyz for ORCA / Avogadro
Chem.MolToXYZFile(mol, 'butane.xyz')

# 2D depiction
Draw.MolToFile(mol, 'butane.png', size=(400, 400))

Conformer search (ETKDG)

from rdkit import Chem
from rdkit.Chem import AllChem

mol = Chem.AddHs(Chem.MolFromSmiles('CC(C)(C)OC1CCCCC1'))
params = AllChem.ETKDGv3()
params.numThreads = 4
cids = AllChem.EmbedMultipleConfs(mol, numConfs=50, params=params)

# Optimize each conformer, collect energies
results = AllChem.MMFFOptimizeMoleculeConfs(mol, numThreads=4)
energies = [(cid, e) for cid, (status, e) in zip(cids, results) if status == 0]
energies.sort(key=lambda x: x[1])
best_cid, best_e = energies[0]
print(f"Lowest-E conformer: id={best_cid}, E={best_e:.3f} kcal/mol")

6. Method selection guide

DFT functional choices

For 90% of organic questions in this book: B3LYP-D3(BJ)/def2-SVP for geometry + frequencies, ωB97X-D/def2-TZVP for single-point energies. For NMR: mPW1PW91/6-311+G(2d,p) GIAO is a widely cited recipe.

7. Basis set guide

Heuristic: def2-SVP for geometry, def2-TZVP for energy, add diffuse functions when treating anions, hydrogen-bonded clusters, or excited states.

8. Common calculation types

TS validation: always run Freq after OptTS and check that exactly one imaginary frequency is present and its mode visually corresponds to the bond making/breaking motion. Then run IRC to verify it connects the intended reactant and product.

9. Computational exercise solutions (cross-reference)

For each: build in Avogadro → preoptimize MMFF94 → export .xyz → wrap with ORCA input → run → analyze.

10. Sanity-checking results

Computational chemistry rewards skepticism. Run through this checklist on any new result before trusting it:

Always visualize the result. A geometry that converges to the right energy but looks distorted is usually a bug.

11. Reading the literature

When citing computational results, papers report at minimum: - Method — functional + basis set, e.g., "B3LYP-D3(BJ)/def2-TZVP//B3LYP-D3(BJ)/def2-SVP" (energy basis // geometry basis). - Software + version — "ORCA 5.0.4" or "Gaussian 16, rev C.01." - Solvation model — "SMD(toluene)" or "gas phase." - Thermal corrections — "Gibbs free energies at 298.15 K, 1 atm."

Typical error bars for organic chemistry DFT: - Bond lengths: ±0.01-0.02 Å - Bond angles: ±1-2° - Reaction enthalpies: ±2-3 kcal/mol (good functional); ±5+ (poor choice) - Activation barriers: ±1-3 kcal/mol - ¹H NMR shifts: ±0.1-0.3 ppm after referencing - ¹³C NMR shifts: ±2-5 ppm - IR frequencies: scale factor ~0.96-0.97 needed; ±20 cm⁻¹ residual

Foundational papers to cite for method validation: - Becke 1993; Lee, Yang, Parr 1988 (B3LYP) - Grimme et al. 2010 / 2011 (D3, BJ damping) - Weigend & Ahlrichs 2005 (def2 basis sets) - Zhao & Truhlar 2008 (M06 family) - Chai & Head-Gordon 2008 (ωB97X-D) - Riniker & Landrum 2015 (ETKDG) - Grimme et al. 2017 (GFN-xTB)

Computational chemistry is fast, cheap, and increasingly trustworthy — but only when you understand what you asked the computer to do.