Chemists used to learn about molecules only by doing experiments in a laboratory. Today a large part of chemical research also happens on a computer: we build a model of a molecule, apply the laws of physics to it and calculate its shape, energy and properties. This first module explains what computational chemistry is, what it can and cannot do, why it has become so important in modern research and industry, and how a typical computational investigation is planned and carried out. It maps to course outcome CO1 and sets up the words (model, level of theory, basis set, geometry optimization, single point) that you will meet in every later module.
Summary
Definition: Computational chemistry is the branch of chemistry that uses computers, together with mathematical models based on quantum mechanics, classical mechanics and statistical mechanics, to calculate the structures, energies and properties of atoms, molecules and materials, and to study how they react. In simple words, it is chemistry done by solving equations on a computer instead of (or along with) mixing chemicals in a flask. It is closely linked to theoretical chemistry: theoretical chemistry develops the equations and approximations, and computational chemistry turns them into computer programs and applies them to real chemical problems.
Computational versus experimental chemistry: An experiment measures a property of a real sample, which may contain many molecules, impurities and a solvent, and the result always has some measurement error. A calculation works on a model, usually a single molecule or a small group of molecules, and its error comes from the approximations made in the model. Experiments are the final judge of what is true, but calculations can study things that are difficult to measure, such as short-lived intermediates, transition states, toxic or explosive compounds and molecules that have not yet been made. The two approaches support each other: experiments test and improve the models, and calculations explain and predict experimental results.
Scope: Computational chemistry can predict (1) molecular geometry: bond lengths, bond angles and the most stable conformation; (2) energies: heats of formation, bond energies, energies of reaction and relative stability of isomers; (3) reaction mechanisms: the path a reaction follows, its transition state and its activation energy; (4) spectra: IR and Raman vibrational frequencies, NMR chemical shifts and UV-visible absorption; (5) electrical properties: dipole moment, charge distribution, polarizability, ionization energy and electron affinity; (6) behaviour of large systems: proteins, DNA, polymers, liquids, surfaces and crystals, through simulations. Its applications reach drug design, catalyst design, new materials (batteries, solar cells, semiconductors), polymers, environmental chemistry and chemical engineering.
Main families of methods: The methods differ in how much physics they include. Molecular mechanics treats atoms as balls and bonds as springs and uses classical physics; it is very fast and can handle thousands to millions of atoms. Semi-empirical methods solve a simplified quantum mechanical problem and use parameters fitted to experimental data. Ab initio methods (Latin for "from the beginning") such as Hartree-Fock solve the Schrödinger equation using only physical constants, without experimental parameters. Density functional theory (DFT) calculates the energy from the electron density and is the most widely used quantum method today because it gives good accuracy at moderate cost. Molecular dynamics and Monte Carlo simulations follow the motion or the many possible arrangements of molecules to obtain properties at a given temperature. Module 2 discusses each of these tools.
Accuracy versus cost: Every method is a compromise between accuracy and computer time. The time needed grows with the size of the system, usually written as a power of N, the number of atoms or basis functions. Molecular mechanics scales roughly as N squared (better with cut-offs), DFT roughly as N cubed, Hartree-Fock as N to the power 4, and very accurate correlated methods such as CCSD(T) as N to the power 7. So doubling the size of a molecule makes a Hartree-Fock calculation about 16 times slower and a CCSD(T) calculation about 128 times slower. Choosing the right method for the size of the problem and the accuracy needed is the first skill of a computational chemist.
Importance in modern chemical research: Calculations save time, money and chemicals, because thousands of candidate molecules can be screened on a computer and only the most promising are made in the laboratory. They are safe for studying toxic, radioactive or explosive substances. They give information that experiments cannot give directly, such as the structure of a transition state that lives for about a femtosecond, or the shape of a molecular orbital. They help interpret experimental spectra and data. Faster computers, better algorithms and cheap cloud and GPU computing have made calculations a routine part of research in universities and industry. The importance of the field is shown by Nobel Prizes in Chemistry: 1998 (Walter Kohn for density functional theory and John Pople for computational methods in quantum chemistry), 2013 (Martin Karplus, Michael Levitt and Arieh Warshel for multiscale models of complex chemical systems) and 2024 (David Baker for computational protein design, and Demis Hassabis and John Jumper for protein structure prediction).
Computational investigations: A computational study follows clear steps. First define the chemical question (for example, which of two isomers is more stable). Then build the starting structure using a molecule builder, a SMILES string, Cartesian coordinates or a Z-matrix (internal coordinates: bond lengths, angles and dihedral angles). Next choose the model: the method (MM, semi-empirical, HF, DFT and so on), and for quantum methods a basis set, together with charge, spin multiplicity and whether to include a solvent model. Then choose the type of job: single point energy, geometry optimization, frequency calculation, potential energy scan, transition state search or molecular dynamics. Run the calculation, check that it finished properly (converged, no imaginary frequencies for a minimum), analyse the results and compare them with experiment or with more accurate calculations. The combination of a method and a basis set, such as B3LYP/6-31G(d), is called the level of theory or model chemistry.
Common calculation types: A single point calculation finds the energy and properties of one fixed geometry. A geometry optimization moves the atoms step by step downhill on the potential energy surface until the forces on all atoms are nearly zero, giving the equilibrium structure. A frequency calculation computes the vibrational frequencies; all real frequencies confirm a minimum, while exactly one imaginary frequency indicates a transition state; it also gives zero-point energy and thermodynamic quantities such as enthalpy and Gibbs energy. A scan changes one coordinate (for example a dihedral angle) in steps and records the energy. Molecular dynamics follows the atoms over time at a set temperature.
Units and limitations: Quantum chemistry programs report energies in hartree (Eh) and distances in bohr (a0). 1 Eh = 27.211 eV = 2625.5 kJ/mol = 627.5 kcal/mol, and 1 bohr = 0.5292 Å. Results within about 1 kcal/mol (4.2 kJ/mol) of experiment are called chemical accuracy. Limitations: every model has approximations; many calculations describe an isolated molecule in the gas phase at 0 K, while experiments are done in solution at room temperature; large accurate calculations need much computer time; and a wrong input (wrong charge, spin or starting structure) gives a confident but wrong answer. Results must therefore always be checked.
Mind map
Introduction to Computational Chemistry
- Definition
- Computers + physical models
- Structure, energy, properties
- Link with theoretical chemistry
- Complements experiment
- Scope
- Geometry and conformation
- Energies and stability
- Reaction mechanisms, transition states
- Spectra (IR, NMR, UV-Vis)
- Large systems: proteins, polymers, materials
- Methods
- Molecular mechanics
- Semi-empirical
- Ab initio (HF and beyond)
- DFT
- Molecular dynamics, Monte Carlo
- Importance
- Saves time, money, chemicals
- Safe for toxic species
- Sees transition states
- Drug and materials design
- Nobel Prizes 1998, 2013, 2024
- Computational investigation
- Question
- Build structure
- Choose level of theory
- Job type: SP, opt, freq, scan, TS, MD
- Analyse and validate
Steps of a computational investigation
- Define the problem clearly: what property or comparison is needed, and how accurate must it be?
- Build the model system: draw the molecule or enter its coordinates (Cartesian, Z-matrix or SMILES); set the total charge and spin multiplicity.
- Choose the method according to size and accuracy: MM for very large systems, semi-empirical for quick estimates, DFT for most medium molecules, high-level ab initio for small molecules needing high accuracy.
- For quantum methods choose a basis set (for example 6-31G(d)); decide whether to include a solvent model.
- Optimize the geometry to reach the nearest energy minimum on the potential energy surface.
- Run a frequency calculation to confirm a true minimum (no imaginary frequencies) and get zero-point energy, enthalpy and Gibbs energy.
- Compute the required properties (energies, dipole moment, charges, spectra, orbitals).
- Analyse the results and compare with experiment or higher-level calculations; repeat with a better model if the error is too large.
- Report the method, basis set, software and all settings so that others can reproduce the work.
Theory, computation and experiment
Think of three corners of a triangle. Theory provides the laws (the Schrödinger equation, Newton's laws, statistical mechanics). Computation applies these laws with approximations to a specific molecule. Experiment measures what really happens. When a calculation agrees with experiment we gain confidence in the model and can use it to predict new results; when it disagrees, either the model or the interpretation of the experiment must be improved. This loop is how computational chemistry contributes to research.
Important definitions
| Term | Meaning |
|---|---|
| Computational chemistry | Use of computers and mathematical models based on physical laws to calculate structures, energies and properties of chemical systems. |
| Theoretical chemistry | Development of the mathematical theories and approximations that describe chemical systems. |
| Model | A simplified mathematical description of a chemical system used in a calculation. |
| Level of theory (model chemistry) | The combination of method and basis set used, written as method/basis, for example HF/6-31G(d). |
| Ab initio | "From first principles": methods that use only physical constants and no experimental parameters. |
| Semi-empirical method | A simplified quantum method in which some integrals are replaced by parameters fitted to experiment. |
| Molecular mechanics | A classical method that calculates energy from bond stretching, angle bending, torsion and non-bonded terms using a force field. |
| Single point calculation | Calculation of energy and properties at one fixed geometry. |
| Geometry optimization | Repeated adjustment of atomic positions until the energy reaches a minimum (forces nearly zero). |
| Frequency calculation | Calculation of vibrational frequencies at a stationary point; confirms a minimum or transition state and gives thermochemistry. |
| Transition state | The highest-energy point on the lowest-energy path between reactants and products; a first-order saddle point. |
| Scaling | How computer time grows with system size N, written as N to some power. |
| Chemical accuracy | An error of about 1 kcal/mol (4.2 kJ/mol) or less compared with experiment. |
| Hartree (Eh) | Atomic unit of energy; 1 Eh = 27.211 eV = 2625.5 kJ/mol = 627.5 kcal/mol. |
Formulas
| Formula | Meaning |
|---|---|
| \( \hat{H}\Psi = E\Psi \) | Time-independent Schrödinger equation, the central equation of quantum chemical methods. |
| \( E_{\text{MM}} = E_{\text{str}} + E_{\text{bend}} + E_{\text{tors}} + E_{\text{vdW}} + E_{\text{elec}} \) | Molecular mechanics energy as a sum of simple terms (Module 4). |
| \( \text{time} \propto N^{p} \) | Cost scaling: p is about 2 for MM, 3 for DFT, 4 for HF, 5 for MP2, 7 for CCSD(T). |
| \( \dfrac{t_2}{t_1} = \left(\dfrac{N_2}{N_1}\right)^{p} \) | Ratio of computer times when the system size changes from N1 to N2. |
| \( 1\ E_h = 27.211\ \text{eV} = 2625.5\ \text{kJ mol}^{-1} = 627.5\ \text{kcal mol}^{-1} \) | Energy unit conversions. |
| \( 1\ a_0 = 0.5292\ \text{Å} \) | Bohr to angstrom. |
| \( \dfrac{N_B}{N_A} = e^{-\Delta E/RT} \) | Boltzmann ratio of two conformers or isomers (equal degeneracy) differing by energy ΔE per mole. |
Key points to remember
- Computational chemistry calculates chemical properties from physical laws using computers; it complements, not replaces, experiment.
- It can study transition states, unstable intermediates, dangerous compounds and molecules not yet made.
- Methods in order of increasing accuracy and cost: MM, semi-empirical, DFT and HF, correlated ab initio (MP2, CCSD(T)).
- Ab initio means no experimental parameters; semi-empirical uses fitted parameters; MM uses a classical force field.
- Level of theory = method/basis set, for example B3LYP/6-31G(d).
- A minimum has all real frequencies; a transition state has exactly one imaginary frequency.
- Cost grows steeply with size: HF as N⁴, CCSD(T) as N⁷.
- 1 Eh = 27.211 eV = 2625.5 kJ/mol; chemical accuracy is about 4 kJ/mol.
- Nobel Prizes: 1998 Kohn and Pople; 2013 Karplus, Levitt and Warshel; 2024 Baker, Hassabis and Jumper.
- Always validate results against experiment or a higher level of theory.
Solved examples
Example 1. Two conformers of a molecule differ in energy by 0.0050 Eh. Express this difference in kJ/mol, kcal/mol and eV.
Using 1 Eh = 2625.5 kJ/mol: \(\Delta E = 0.0050\times2625.5 = 13.1\) kJ/mol. In kcal/mol: \(13.1/4.184 = 3.14\) kcal/mol (or \(0.0050\times627.5 = 3.14\)). In eV: \(0.0050\times27.211 = 0.136\) eV.
Example 2. Using the result of Example 1, estimate the ratio of the number of molecules in the higher conformer to the lower conformer at 298 K (assume equal degeneracy).
\(N_B/N_A = e^{-\Delta E/RT}\) with \(\Delta E = 13.1\times10^{3}\) J/mol, R = 8.314 J/(mol K), T = 298 K. \(\Delta E/RT = 13126/(8.314\times298) = 5.30\). \(N_B/N_A = e^{-5.30} = 0.0050\). So only about 0.5 percent of molecules are in the higher-energy conformer; nearly all are in the more stable one.
Example 3. A Hartree-Fock calculation on a molecule with 50 basis functions takes 1 hour. Estimate the time for a molecule with 150 basis functions. How long would a CCSD(T) calculation take if its 50-function run took 1 hour?
HF scales as \(N^4\): \(t_2 = 1\times(150/50)^4 = 3^4 = 81\) hours. CCSD(T) scales as \(N^7\): \(t_2 = 3^7 = 2187\) hours, about 91 days. This is why very accurate methods are used only for small molecules.
Example 4. The calculated bond length of H2 is 1.40 bohr. Convert it to angstrom.
\(1.40\times0.5292 = 0.741\) Å, which agrees with the experimental bond length of H2 (0.741 Å).
Example 5. If the computer time of a DFT calculation grows as N³, by what factor does it grow when the molecule is doubled in size?
\((2N/N)^3 = 2^3 = 8\). The calculation takes about 8 times longer. For HF (N⁴) the factor would be 16.
Important questions and answers
Q1. Define computational chemistry. Explain its scope. (5 marks)
Computational chemistry is the branch of chemistry that uses computer programs based on quantum mechanics, classical mechanics and statistical mechanics to calculate the structure, energy and properties of molecules and materials and to study chemical reactions. Scope: (1) geometry: bond lengths, angles and stable conformations; (2) energetics: heats of formation, reaction energies, relative stability of isomers; (3) reaction mechanisms: transition states and activation energies; (4) spectroscopy: IR, Raman, NMR and UV-Vis spectra; (5) electronic properties: dipole moment, charges, ionization energy, electron affinity, molecular orbitals; (6) large systems: proteins, DNA, polymers, liquids and crystals by molecular mechanics and molecular dynamics. It is applied in drug design, catalysis, materials science, energy storage and environmental chemistry.
Q2. Distinguish between computational chemistry and experimental chemistry. (5 marks)
Experimental chemistry works with real substances in a laboratory, needs chemicals, instruments and safety measures, measures properties of bulk samples and has measurement errors. Computational chemistry works with a model on a computer, needs only software and computing power, usually describes individual molecules, and has errors due to the approximations in the model. Experiments give the true behaviour but cannot easily observe very short-lived species; calculations can study transition states, unstable intermediates, hazardous or not-yet-made compounds, and can give details such as orbital shapes and charge distributions. Experiment is used to validate computation, and computation is used to explain and predict experimental results, so the two are complementary.
Q3. Explain the importance of computational chemistry in modern chemical research. (5 marks)
(1) Cost and time: thousands of candidate molecules (drugs, catalysts, materials) can be screened on a computer before synthesis. (2) Safety: toxic, explosive or radioactive systems can be studied without risk. (3) Insight: it shows transition states, reaction paths, orbitals and charge distribution that cannot be seen directly. (4) Interpretation: it helps assign IR, NMR and UV spectra. (5) Prediction: it can predict properties of new molecules and materials. (6) Large biological systems: protein folding, enzyme action and drug binding are studied by molecular mechanics and dynamics. (7) Growing power: faster computers, GPUs and better algorithms make calculations routine. Its importance is recognized by Nobel Prizes in 1998, 2013 and 2024.
Q4. What is meant by a computational investigation? Describe the steps involved. (8 marks)
A computational investigation is a planned study in which a chemical question is answered by calculations. Steps: (1) define the question and required accuracy; (2) build the molecular model (coordinates, charge, spin multiplicity); (3) select the method according to system size and accuracy (MM, semi-empirical, HF, DFT, correlated ab initio); (4) select a basis set for quantum methods and decide on solvent treatment; (5) perform geometry optimization to locate the energy minimum; (6) perform a frequency calculation to confirm a minimum (no imaginary frequency) or a transition state (one imaginary frequency) and obtain zero-point and thermal energies; (7) calculate the needed properties; (8) analyse results and compare with experiment or higher-level calculations; (9) report all details for reproducibility. Explaining each step in one or two lines and drawing a flow chart earns full marks.
Q5. Compare the main methods used in computational chemistry. (7 marks)
Molecular mechanics: classical, balls and springs, force field parameters, very fast, up to millions of atoms, cannot describe bond breaking or electronic properties. Semi-empirical: simplified quantum mechanics with fitted parameters (AM1, PM3, PM7), fast, hundreds to thousands of atoms, accuracy depends on parameterization. Ab initio (HF, MP2, CCSD(T)): solves the Schrödinger equation without experimental parameters, systematically improvable, expensive (N⁴ to N⁷), small to medium molecules. DFT: uses electron density, includes electron correlation approximately, good accuracy at about N³ cost, the most widely used quantum method. Molecular dynamics: follows atomic motion with time using Newton's laws, gives properties at finite temperature, usually uses a force field.
Q6. What is the difference between a single point calculation, a geometry optimization and a frequency calculation? (5 marks)
A single point calculation computes the energy and properties at one fixed geometry without moving the atoms. A geometry optimization calculates forces on the atoms and moves them repeatedly downhill on the potential energy surface until the forces are nearly zero, giving an equilibrium geometry. A frequency calculation, done at an optimized geometry, computes the second derivatives of energy to give vibrational frequencies; all real frequencies mean a true minimum, one imaginary frequency means a transition state; it also gives zero-point energy, enthalpy, entropy and Gibbs energy, and a predicted IR spectrum.
Q7. Why is a balance between accuracy and computational cost needed? (3 marks)
More accurate methods include more physics, so they need far more computer time, which increases steeply with system size (HF as N⁴, CCSD(T) as N⁷). A very accurate method may be impossible for a large molecule, while a cheap method may be too inaccurate for a small one. So the method is chosen to give enough accuracy for the question within the available computer time.
Exam tips
- Start every definition answer with a one-line definition, then list scope or applications as numbered points.
- A neat flow chart of the steps of a computational investigation is quick to draw and fetches marks.
- For comparison questions use a table with rows such as basis, speed, system size, accuracy, parameters and examples.
- Remember the scaling powers (MM ~N², DFT ~N³, HF ~N⁴, MP2 ~N⁵, CCSD(T) ~N⁷) for short numericals on computer time.
- Know the conversion 1 Eh = 27.211 eV = 2625.5 kJ/mol = 627.5 kcal/mol.
- Mention the Nobel Prizes (1998, 2013, 2024) when asked about the importance of the field.
FAQs
Will computational chemistry replace laboratory experiments?
No. Calculations are based on models with approximations, so their results must be checked against experiment. They reduce the number of experiments needed and explain results, but experiments remain the final test.
Do I need to be good at programming for this subject?
At first-year level you mainly need to understand the ideas and use ready-made software. Basic computer skills and a willingness to learn simple commands are enough; programming helps later but is not required.
Which software is used in computational chemistry?
Common programs include Gaussian, ORCA, GAMESS, NWChem, Psi4 and Quantum ESPRESSO for quantum calculations, GROMACS, AMBER and LAMMPS for molecular dynamics, and Avogadro, GaussView and VMD for building and viewing molecules. ORCA, Psi4, GAMESS, GROMACS and Avogadro are free for academic use.
Why do programs report energy in hartree?
Hartree is the atomic unit of energy, in which the equations of quantum mechanics take their simplest form. Energies are converted to kJ/mol or eV afterwards for chemical use.
These notes follow the University of Mumbai first-year engineering syllabus (NEP 2020, Introduction to Computational Chemistry) and were prepared with AI help and checked.