- Atlas
- Natural Sciences
- Theoretical Chemistry
Theoretical Chemistry
Theoretical chemistry is the branch of chemistry that tries to explain and predict chemical behaviour from physical laws and mathematics rather than from experiment alone.
10 frameworks, from 1870 to today · Chemistry · Natural Sciences
- Chemical thermodynamics 1870-Present
- Statistical mechanics 1900-Present
- Valence Bond Theory 1927-Present
- Post–Hartree–Fock 1928-Present
- Molecular Orbital Theory 1929-Present
- Molecular mechanics 1946-Present
- Density Functional Theory 1964-Present
- Molecular dynamics 1964-Present
- QM/MM 1976-Present
- Machine-Learned Interatomic Potentials 2007-Present
Overview
Theoretical chemistry is the branch of chemistry that tries to explain and predict chemical behaviour from physical laws and mathematics rather than from experiment alone1. Its practitioners ask what a chemical bond actually is, why molecules have the shapes and energies they do, how fast a reaction proceeds, and how a solvent or a protein changes all of this1. What is at stake is enormous: a theory that can compute the energy of a molecule from scratch, without making it in a flask, lets chemists design drugs, catalysts, and materials on a computer before anyone touches a beaker. The field's history is a long argument about how far that ambition can be pushed—and what must be sacrificed to get an answer at all.
From the Mechanical Philosophy to Thermodynamics
The roots lie in the seventeenth-century mechanical philosophy, which held that matter consists of particles whose motions obey natural laws. Robert Boyle's The Sceptical Chymist (1661) argued that chemical phenomena should be explained by the arrangement and motion of corpuscles rather than by the four Aristotelian elements. This was a programme rather than a working theory, but it set the question that theoretical chemistry would pursue for centuries: can chemistry be reduced to mechanics?
The first genuinely predictive theory came from thermodynamics. In the 1870s Josiah Willard Gibbs developed the phase rule and the chemical potential, showing that the direction of chemical change could be determined from a single quantity—the free energy—without knowing anything about molecules. Thermodynamics told chemists whether a reaction could happen and how far it would go, but it was silent on mechanism and on the nature of the bond. It remains the backbone of chemical engineering and of the modern theory of reaction equilibria.
The Quantum Revolution and the Chemical Bond
The decisive break came with quantum mechanics. In 1927 Walter Heitler and Fritz London applied the new wave mechanics to the hydrogen molecule and showed that a covalent bond arises from the exchange of electrons between two atoms—a purely quantum effect with no classical analogue. This was the birth of what became valence bond theory: the idea that a molecule is built from localised bonds, each formed by pairing electrons on neighbouring atoms, with hybridisation invoked to explain geometries such as the tetrahedral carbon of methane23. Linus Pauling's The Nature of the Chemical Bond (1939) systematised this picture and made it the working language of organic chemistry; resonance structures, hybrid orbitals, and electronegativity all belong to it.
A rival appeared almost immediately. In 1929 John Lennard-Jones and others developed molecular orbital theory, in which electrons occupy orbitals spread over the whole molecule rather than being confined to bonds between pairs of atoms. The two approaches are mathematically related but conceptually different: valence bond theory thinks in local bonds and resonance, molecular orbital theory in delocalised orbitals and their symmetries34. Molecular orbital theory won the computational argument because its orbitals are easier to calculate systematically, and it became the basis of the Woodward–Hoffmann rules for predicting reaction stereochemistry in the 1960s. Valence bond theory survived as an interpretive framework and, in modern forms, as a computational method—but the mainstream of quantitative work moved to orbitals4.
Hartree–Fock and the Wavefunction Programme
The central computational problem is the many-electron Schrödinger equation, which cannot be solved exactly for any molecule larger than hydrogen. In the late 1920s Douglas Hartree and Vladimir Fock independently developed an approximation in which each electron moves in the average field of all the others—the Hartree–Fock method. It gives a single determinant wavefunction and captures most of the energy, but it neglects electron correlation: the tendency of electrons to avoid one another beyond the average field. The difference between Hartree–Fock and the exact energy is the correlation energy, and recovering it became the defining challenge of the field5.
From the 1950s onward a hierarchy of methods was built to do this: configuration interaction, in which the wavefunction is written as a mixture of excited determinants; Møller–Plesset perturbation theory (MP2 and beyond); and coupled-cluster theory, which by the 1980s had become the gold standard for small molecules. These methods are systematically improvable—you can, in principle, approach the exact answer—but their cost grows steeply with system size, which is why they remain confined to tens of atoms. The wavefunction programme thus established the benchmark against which every cheaper method is judged.
Density Functional Theory: The Cheap Route
A different route opened in 1964, when Pierre Hohenberg and Walter Kohn proved that the ground-state energy of a system is a unique functional of its electron density—a function of three spatial coordinates rather than the 3N coordinates of a wavefunction. In 1965 Kohn and Lu Jeu Sham turned this into a practical scheme: a set of single-particle equations, like Hartree–Fock, but with an exchange-correlation functional that in principle contains everything missing from the mean field67. The catch is that the exact functional is unknown, so practitioners must approximate it6.
The approximations came in waves. The local density approximation and the generalised gradient approximations of the 1980s and 1990s made density functional theory (DFT) accurate enough for many chemical problems at a fraction of wavefunction cost. Hybrid functionals such as B3LYP, introduced in the early 1990s, mixed in a portion of exact Hartree–Fock exchange and became the workhorse of computational chemistry. DFT's rise is the dominant story of late-twentieth-century theoretical chemistry: it made routine calculations on molecules of fifty to a few hundred atoms feasible, and it is now the most widely used electronic-structure method in the field. Its weakness is that no systematic improvement is guaranteed—results depend on the chosen functional, and some problems (strong correlation, charge-transfer states, dispersion) remain difficult6.
Statistical Mechanics and the Bridge to Bulk Matter
Electronic structure describes a single molecule at zero temperature. To connect to real chemistry—liquids, solutions, proteins, reaction rates at finite temperature—the field turned to statistical mechanics, the theory that derives bulk thermodynamic properties from the statistical behaviour of large numbers of molecules1. Gibbs and Boltzmann had laid its foundations in the nineteenth century; in the twentieth century it became a computational tool. Monte Carlo methods, developed at Los Alamos in the 1940s, sample configurations by random moves; transition-state theory, refined by Henry Eyring and Michael Polanyi in the 1930s, gives reaction rates from the energy barrier between reactants and products89. Together these let theorists predict free energies, equilibrium constants, and rate constants—the quantities chemists actually measure.
Molecular Mechanics and the Classical Approximation
For systems too large for quantum methods—proteins, polymers, liquids—the field developed molecular mechanics, which treats atoms as balls and bonds as springs governed by a force field: a set of parameterised equations for bond stretching, angle bending, torsional rotation, and non-bonded attraction and repulsion1011. The approach dates to the 1940s–1960s work of researchers such as F. H. Westheimer and Norman Allinger, whose MM series of force fields (MM2, MM3) became standard for organic molecules. Force fields are fast and can handle millions of atoms, but they cannot break or form bonds, so they describe structure and dynamics rather than reactivity. Their accuracy depends entirely on the parameters, which must be fitted to experiment or quantum calculation11.
Molecular Dynamics: Watching Molecules Move
Molecular dynamics simulation, introduced by Aneesur Rahman in 1964 for liquid argon and extended to biomolecules from the 1970s, integrates Newton's equations of motion for every atom in a force field, producing a trajectory over time. It turned theoretical chemistry from a static, energy-focused discipline into a dynamic one: one could watch a protein fold, a membrane flex, or a solvent reorganise. The limitation is the timescale—femtosecond steps mean that biologically relevant events taking milliseconds are far out of reach without enhanced sampling techniques, which became a major research area from the 1990s.
Multiscale Methods: QM/MM
The split between quantum accuracy and classical speed suggested a hybrid. In 1976 Arieh Warshel and Michael Levitt introduced QM/MM (quantum mechanics/molecular mechanics), in which a small region—the active site of an enzyme, say—is treated quantum mechanically while the surrounding protein and solvent are treated with a force field12. This made it possible to study reactions in realistic biological environments, and it earned Warshel, Levitt, and Martin Karplus the 2013 Nobel Prize in Chemistry12. Multiscale embedding has since expanded to include continuum solvation models and machine-learned environments, and it remains the standard way to bring quantum chemistry to bear on large systems.
Machine Learning and the New Surrogate Models
The most recent turning point is the use of machine learning to approximate quantum-mechanical potential energy surfaces. The idea—train a flexible model on a set of quantum calculations, then use it to predict energies and forces for new configurations at a fraction of the cost—was demonstrated in the 2000s and 2010s with neural-network potentials and Gaussian process regression. By the 2020s, machine-learned interatomic potentials were reaching near-quantum accuracy for many systems while running orders of magnitude faster, and they are now used for molecular dynamics, materials screening, and reaction discovery. The open questions are familiar ones in new dress: how to guarantee that a model extrapolates safely outside its training data, how to represent the quantum state itself rather than just the energy, and how to keep the physics—symmetries, conservation laws—built in rather than learned.
The Present
Theoretical chemistry today is a layered enterprise. Wavefunction methods and DFT provide electronic structure; statistical mechanics and molecular dynamics provide the bridge to bulk and to time; force fields and machine-learned potentials provide speed; QM/MM and embedding provide the multiscale link. The field's central tension—between systematic accuracy and computational reach—has not been resolved but has been pushed outward, and the newest methods blur the old boundaries between quantum and classical, physics-based and data-driven. What remains open is the same question that animated Boyle: how much of chemistry can be derived from first principles, and what must still be measured rather than calculated.
References
- Theoretical chemistry (Overview, introduction, Branches of theoretical chemistry), Wikipedia
- Quantum chemistry (History), Wikipedia
- Valence bond theory (introduction, History), Wikipedia
- Modern valence bond theory, Wikipedia
- Post–Hartree–Fock, Wikipedia
- Density functional theory (Origins, introduction), Wikipedia
- Kohn–Sham equations (History), Wikipedia
- Transition state theory, Wikipedia
- Monte Carlo method (History), Wikipedia
- Molecular mechanics, Wikipedia
- Force field (chemistry), Wikipedia
- QM/MM, Wikipedia
The frameworks of Theoretical Chemistry
Each framework is a way of working the field has used: when it held, and the question it answers.
- 1870-Present
Chemical thermodynamics
What determines whether a reaction occurs and how far it goes?
- 1900-Present
- 1927-Present
- 1928-Present
Post–Hartree–Fock
How can we recover the energy missing from a mean-field picture?
- 1929-Present
Molecular Orbital Theory
How are electrons arranged across a whole molecule?
- 1946-Present
Molecular mechanics
Can molecular shape and motion be described by classical mechanics?
- 1964-Present
Density Functional Theory
Can the energy of a molecule be computed from its electron density alone?
- 1964-Present
- 1976-Present
- 2007-Present
Machine-Learned Interatomic Potentials
Can machine learning replace expensive quantum calculations of energy and forces?
How the ideas connect
Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.
Molecular Orbital Theory rival of Valence Bond Theory
Valence bond theory and molecular orbital theory offered competing pictures of the chemical bond from the late 1920s, one built on localised electron pairs and the other on delocalised orbitals.
Carried by Walter Heitler; Fritz London; John Lennard-Jones.
Molecular dynamics grew out of Molecular mechanics
Molecular dynamics integrates Newton's equations using the force fields developed in molecular mechanics, so the trajectory method grew directly out of the classical force-field approach.
Carried by Aneesur Rahman.
Statistical mechanics influenced by Chemical thermodynamics
Gibbs's formulation of chemical potential and free energy provided the thermodynamic quantities that statistical mechanics later derived from molecular ensembles, as in Gibbs's own statistical mechanics and its chemical applications.
Carried by Josiah Willard Gibbs.
Molecular dynamics influenced by Statistical mechanics
Molecular dynamics simulations sample the microstates whose statistical averages yield thermodynamic properties, applying the ensemble framework of statistical mechanics to trajectories of atoms.
Carried by Aneesur Rahman.
Density Functional Theory influenced by Post–Hartree–Fock
Hybrid functionals such as B3LYP incorporated a fraction of exact Hartree–Fock exchange from wavefunction theory, improving density functional theory's accuracy for thermochemistry.
Carried by Axel Becke.
Density Functional Theory influenced by Molecular Orbital Theory
The Kohn–Sham formulation of density functional theory borrowed the molecular orbital picture, using single-particle orbitals to construct the electron density and kinetic energy.
Carried by Walter Kohn; Lu Jeu Sham.
QM/MM influenced by Molecular Orbital Theory
The quantum region in QM/MM is typically treated with molecular orbital methods such as semiempirical or Hartree–Fock calculations, so the hybrid scheme depends on the orbital framework.
Carried by Arieh Warshel; Michael Levitt.
QM/MM influenced by Molecular mechanics
QM/MM combined a quantum-mechanically treated region with a molecular-mechanics force field for the surroundings, taking the classical force-field description as its starting point.
Carried by Arieh Warshel; Michael Levitt.
Machine-Learned Interatomic Potentials influenced by Density Functional Theory
Machine-learned interatomic potentials are trained on energies and forces computed with density functional theory, using DFT as the reference data source.
Carried by Jörg Behler; Michele Parrinello.
Machine-Learned Interatomic Potentials influenced by Molecular dynamics
Machine-learned potentials are used to drive molecular dynamics simulations, replacing the empirical force field with a learned surrogate for quantum energies and forces.
Carried by Jörg Behler; Michele Parrinello.
Beyond the field
Where Theoretical Chemistry's ideas came from in other fields, and what they led to there: the strongest 12 of 90 links.
Chemical thermodynamics influenced by Classical Thermodynamics of Heat Engines in Chemical Thermodynamics
Carnot's analysis of heat-engine cycles, generalised by Clausius and Kelvin into the second law and entropy, supplied the thermodynamic laws that Gibbs then applied to chemical equilibrium in 1876.
Carried by Sadi Carnot; Rudolf Clausius; Josiah Willard Gibbs.
Chemical thermodynamics led to Fugacity and Activity Coefficient Framework in Chemical Thermodynamics
The fugacity and activity coefficient framework is a branch of chemical thermodynamics, extending its equilibrium theory to real solutions and gases.
Carried by Gilbert N. Lewis.
Chemical thermodynamics influenced Thermoeconomics in Ecological Economics
Chemical thermodynamics supplied the entropy and free-energy concepts that thermoeconomics applies to economic processes. The later field's central claim, that economic activity is an entropy-generating transformation of matter and energy, is a direct application of chemical thermodynamics.
Carried by Nicolas Georgescu-Roegen.
Chemical thermodynamics led to Electrochemical thermodynamics in Electrochemistry
The thermodynamic treatment of cell EMF and electrode potentials is a direct extension of chemical thermodynamics into electrochemical systems. The free-energy formalism was carried over essentially unchanged.
Carried by Walther Nernst.
Chemical thermodynamics influenced Geochemical modeling in Geochemistry
Geochemical modeling applies chemical thermodynamics, especially Gibbs free energy minimization and equilibrium constants, to natural waters and rocks. The thermodynamic formalism developed by Gibbs and later physical chemists supplied the equations and equilibrium concepts that geochemical speciation and reaction-path models compute.
Carried by Josiah Willard Gibbs; Harold C. Helgeson.
Statistical mechanics led to Nonequilibrium statistical mechanics in Statistical Mechanics
Theoretical-chemistry statistical mechanics rests on the same Gibbs ensemble foundation that nonequilibrium theory extends; Onsager's reciprocal relations and later linear response theory were developed within and beyond that chemical-physics tradition.
Carried by Lars Onsager.
Statistical mechanics influenced Molecular Simulation in Physical Chemistry
Statistical mechanics supplied the ensemble averages and Boltzmann weighting that molecular simulation computes by sampling.
Carried by Josiah Willard Gibbs.
Statistical mechanics influenced Polymer Chain Statistics and Statistical Mechanics in Polymer Science
Flory and Huggins applied the methods of statistical mechanics, particularly lattice models and partition functions, to derive the thermodynamics of polymer solutions and chain conformations.
Carried by Paul Flory; Maurice Huggins.
Statistical mechanics led to Soft matter and polymer statistical mechanics in Statistical Mechanics
Soft matter and polymer statistical mechanics is a branch of statistical mechanics, applying its ensemble and scaling methods to polymers, colloids and liquid crystals.
Carried by Pierre-Gilles de Gennes.
Statistical mechanics influenced Monte Carlo method in Computational Physics
Statistical mechanics supplied the ensemble averages and Boltzmann weights that Monte Carlo methods estimate by sampling. Metropolis and Ulam's 1949 paper and the 1953 Metropolis algorithm were framed explicitly as computing statistical-mechanical averages.
Carried by Nicholas Metropolis; Stanislaw Ulam.
Statistical mechanics led to Computational statistical mechanics in Statistical Mechanics
Computational statistical mechanics is the numerical implementation of the statistical-mechanical framework of theoretical chemistry; Metropolis and colleagues' 1953 Monte Carlo method was devised to compute equilibrium properties of many-body systems within that tradition.
Carried by Nicholas Metropolis; Metropolis et al. (1953).
Statistical mechanics influenced Transition-State Theory in Physical Chemistry
Statistical mechanics supplied the partition-function machinery by which Eyring and others converted the activated complex into an absolute rate expression.
Carried by Henry Eyring; Michael Polanyi.

Theoretical Chemistry is a planet in Nousfarer
Its city has a tower for each of these 10 frameworks. Land, walk in, and climb one concept at a time.
Land on Theoretical ChemistryRelated fields
Fields its ideas are linked to
- Chemical Thermodynamics 14 links
- Physical Chemistry 14 links
- Computational Chemistry 12 links
- Atomic Molecular Physics 7 links
- Statistical Mechanics 6 links
- Chemistry 5 links
- Computational Materials Science 5 links
- Computational Physics 5 links
- Separation Processes 2 links
- Organic Chemistry 2 links
- Biophysics 2 links
- Thermodynamics 2 links
More in Chemistry
- Analytical Chemistry
- Catalysis
- Chemical Biology
- Chemical Kinetics
- Chemistry
- Computational Chemistry
- Electrochemistry
- Green Chemistry
- Inorganic Chemistry
- Materials Chemistry
- Medicinal Chemistry
- Nuclear Chemistry
- Organic Chemistry
- Organometallic Chemistry
- Physical Chemistry
- Spectroscopy
- Supramolecular Chemistry
- Surface Chemistry
See where Theoretical Chemistry sits on the Map of Human Knowledge