Loading field map
For nearly two centuries, chemists have asked a deceptively simple question: what does a catalyst actually do? The substance itself is not consumed, yet it can accelerate a reaction by factors of millions. Early investigators imagined a mysterious force; later researchers redefined the problem in terms of rates and equilibria. The history of catalysis as a subfield is a sequence of frameworks that replaced, coexisted with, and eventually began to integrate one another. Six major frameworks mark this path: the speculative concept of a catalytic force, Ostwald's kinetic definition, and then three parallel experimental traditions—heterogeneous, homogeneous, and biocatalysis—followed by a computational turn that now reaches across all of them.
The first attempts to explain catalysis were not explanations at all but placeholders. In 1835, Jöns Jacob Berzelius introduced the term "catalytic force" to describe a puzzling set of observations: certain substances could trigger chemical transformations without themselves undergoing permanent change. Berzelius grouped phenomena as diverse as the decomposition of hydrogen peroxide by metals and the action of yeast on sugar under a single, vaguely defined power. His contemporary, the chemist Eilhard Mitscherlich, proposed "contact action" as an alternative label, emphasizing that the mere presence of a surface seemed sufficient. Neither framework offered a mechanism. They were descriptive categories, not theories. The catalytic force concept remained dominant for decades because it named a real problem—why do some materials accelerate reactions?—without providing a testable answer. It was a framework that could not be refined, only replaced.
Wilhelm Ostwald transformed the study of catalysis by grounding it in the emerging science of chemical kinetics and thermodynamics. In the 1890s, Ostwald defined a catalyst as a substance that changes the rate of a chemical reaction without appearing in the final products, and crucially, without altering the equilibrium constant. This definition replaced the vague catalytic force with a measurable, falsifiable claim: a catalyst lowers the activation energy barrier, speeding up both forward and reverse reactions equally. Ostwald's framework drew on the work of Jacobus Henricus van 't Hoff and Svante Arrhenius, who had developed the mathematical tools to describe reaction rates and temperature dependence. Where Berzelius had seen a mysterious power, Ostwald saw a kinetic phenomenon that obeyed the same laws as any other chemical process. The older concept of contact action was not so much disproven as absorbed: Ostwald's definition preserved the observation that surfaces could accelerate reactions, but it reinterpreted that observation as a consequence of energy barriers, not a special force. This framework became the foundation for all later work in catalysis, and it remains the textbook definition today.
Once Ostwald had established catalysis as a kinetic phenomenon, researchers began to ask how different kinds of catalysts worked. The answers diverged sharply, producing three distinct frameworks that developed in parallel for most of the twentieth century.
Heterogeneous catalysis emerged from industrial chemistry. In heterogeneous systems, the catalyst is in a different phase from the reactants—typically a solid surface interacting with gases or liquids. The Haber-Bosch process for ammonia synthesis (developed around 1909) and the catalytic cracking of petroleum (commercialized in the 1930s) drove massive investment in understanding solid surfaces. The distinctive contribution of this framework was its focus on surface chemistry: adsorption, desorption, and the geometry of active sites. Researchers developed concepts such as the Langmuir-Hinshelwood mechanism, which described reactions between adsorbed species, and the Sabatier principle, which stated that an optimal catalyst binds reactants neither too strongly nor too weakly. Heterogeneous catalysis prioritized industrial scalability and thermal stability. Its methods were largely empirical: trial-and-error screening of materials, supported by physical characterization techniques such as X-ray diffraction and electron microscopy. The framework coexisted with homogeneous catalysis by occupying a different domain—large-scale, high-temperature processes where solid catalysts could be easily separated from products.
Homogeneous catalysis took shape later, driven by advances in organometallic chemistry. In homogeneous systems, the catalyst and reactants are in the same phase, usually a liquid solution. The key breakthrough came in the 1950s and 1960s with the discovery of transition-metal complexes that could catalyze reactions such as hydrogenation, hydroformylation, and olefin polymerization. The Wilkinson catalyst (1965) and the Monsanto acetic acid process (1970) demonstrated that molecular catalysts could achieve extraordinary selectivity, including enantioselectivity—the ability to produce one mirror-image isomer over another. Homogeneous catalysis differed from heterogeneous catalysis in its explanatory style: it relied on molecular-level mechanisms, ligand design, and spectroscopic characterization of reaction intermediates. Where heterogeneous catalysis asked "what surface sites are active?", homogeneous catalysis asked "what coordination geometry and electronic structure enable this transformation?" The two frameworks shared Ostwald's kinetic foundation but diverged in methodology and application. Homogeneous catalysis excelled at fine-chemical synthesis and pharmaceutical production, where selectivity mattered more than thermal stability or ease of separation.
Biocatalysis developed alongside heterogeneous catalysis but from a different starting point: biochemistry. In 1897, Eduard Buchner showed that cell-free extracts of yeast could ferment sugar, demonstrating that enzymes—biological catalysts—could function outside living cells. Throughout the early twentieth century, biochemists such as Leonor Michaelis and Maud Menten (1913) developed kinetic models specific to enzymes, introducing concepts like the Michaelis constant and the idea of substrate saturation. Biocatalysis shared Ostwald's kinetic base with the other frameworks, but its methods were distinct. Enzymes are large, chiral proteins (or later, RNA molecules) that operate under mild conditions—aqueous solution, near-neutral pH, ambient temperature—and achieve extraordinary specificity. For much of the twentieth century, biocatalysis remained a separate tradition, studied in biochemistry departments rather than chemistry departments. The framework coexisted with heterogeneous and homogeneous catalysis by occupying a different niche: biological transformations that were difficult to replicate with synthetic catalysts. Only in the late twentieth century did biocatalysis begin to integrate with synthetic chemistry, as techniques such as directed evolution (developed by Frances Arnold in the 1990s) allowed researchers to engineer enzymes for non-natural reactions.
Beginning in the 1990s, computational methods began to transform all three experimental frameworks. Density functional theory (DFT) made it possible to model catalytic cycles on surfaces and in solution with sufficient accuracy to predict reaction barriers and selectivity. Computational catalysis did not replace the experimental frameworks; instead, it provided a new layer of infrastructure that could test mechanistic hypotheses, screen candidate catalysts in silico, and explain experimental observations that were otherwise puzzling. For heterogeneous catalysis, computational models helped identify the structure of active sites on complex surfaces. For homogeneous catalysis, DFT calculations became a standard tool for rationalizing ligand effects and designing new catalysts. For biocatalysis, molecular dynamics simulations and quantum mechanics/molecular mechanics (QM/MM) methods allowed researchers to model enzyme active sites with atomistic detail. The computational framework transformed the subfield by making catalysis more predictive. Where earlier researchers had relied on empirical screening or intuition, computational catalysis offered a way to narrow the search space and to understand why certain catalysts worked. It also acted as a mediating framework, because the same computational tools could be applied to heterogeneous, homogeneous, and biological systems, encouraging cross-talk between traditions that had long developed in isolation.
Today, all four active frameworks—heterogeneous catalysis, homogeneous catalysis, biocatalysis, and computational catalysis—coexist and increasingly overlap. Researchers routinely combine experimental and computational approaches, and the boundaries between the frameworks have become porous. A modern catalysis project might involve designing a molecular catalyst in silico, testing it in solution, and then immobilizing it on a solid support to combine the selectivity of homogeneous catalysis with the separability of heterogeneous catalysis. Biocatalysis has moved from a niche field to a mainstream tool in pharmaceutical synthesis, often integrated with synthetic catalysts in cascade reactions.
Despite this integration, the frameworks retain distinct priorities and assumptions. Heterogeneous catalysis remains focused on industrial scalability, thermal stability, and the complexity of real surfaces. Homogeneous catalysis emphasizes molecular precision, mechanistic understanding, and the rational design of ligands. Biocatalysis prioritizes selectivity, mild conditions, and the evolvability of enzymes. Computational catalysis, while now essential to all three, still faces challenges in accuracy and timescale: DFT calculations are fast but approximate, while more accurate methods are computationally expensive.
The leading frameworks agree on Ostwald's kinetic foundation—no one today disputes that a catalyst lowers an activation barrier without shifting equilibrium. They also agree that understanding mechanism is essential for rational design. The main disagreements are about method and priority. Should the field invest more in high-throughput experimentation or in computational prediction? Is the future of catalysis in synthetic enzymes, solid surfaces, or molecular complexes? These are not settled questions, and the tension between them drives the field forward. The history of catalysis shows that frameworks do not simply replace one another; they differentiate, coexist, and eventually find ways to combine. The subfield's central question—what does a catalyst do?—has been answered many times, and each answer has opened new questions.