Atomic machines are functional devices built from individual molecules or small clusters of atoms, engineered to perform mechanical work at scales a hundred thousand times smaller than the width of a human hair. The concept dates back to Richard Feynman’s famous 1959 speculation that nothing in physics prevents us from maneuvering things atom by atom, but it took decades of chemistry and nanotechnology before anyone could actually build one. Today the field spans synthetic molecular motors that spin when hit with light, tiny ring-shaped molecules that slide along a track like beads on an abacus, and DNA structures folded into working hinges and drills. The 2016 Nobel Prize in Chemistry went to researchers who pioneered this work, and the science has accelerated since then into territory that sounds like science fiction but runs on real chemistry.
Nature Built Them First
Long before chemists assembled their first molecular motor, biology had been running atomic-scale machines for billions of years. Your cells are full of them. ATP synthase, the enzyme that produces the energy currency your body runs on, is a rotary motor embedded in cell membranes. Protons flowing across the membrane spin one part of the enzyme like a turbine, and that rotation drives the chemical reaction that generates ATP.1PubMed. Molecular architecture of the rotary motor in ATP synthase It is, by any reasonable definition, a machine: it has moving parts, converts one form of energy into another, and does useful work.
Kinesin is another example. This protein literally walks along structural filaments inside your cells, carrying cargo from one place to another in a hand-over-hand gait, like a person climbing a rope.2PubMed Central. Kinesin rotates unidirectionally and generates torque while walking on microtubules Bacteria, meanwhile, swim using flagella powered by rotary motors that spin at hundreds of revolutions per second. These motors are driven by ions flowing across the bacterial cell membrane, and researchers have measured their torque output directly.3PubMed Central. Study of the torque of the bacterial flagellar motor using a rotating electric field The flagellar motor even has the equivalent of a gearbox: it can switch direction, and models of its mechanics explain properties like backstepping and varying step sizes.4PubMed Central. Mechanics of torque generation in the bacterial flagellar motor
These biological machines set the benchmark. They operate in water at body temperature, they are extraordinarily efficient, and they self-assemble from genetic instructions. The challenge for synthetic chemistry has been to build something even remotely comparable from scratch.
How Chemists Build a Machine from Molecules
The most successful synthetic atomic machines are built from a class of structures called mechanically interlocked molecules. Think of two rings linked together like a chain, or a ring threaded onto a dumbbell-shaped rod so it can slide back and forth but never fall off. These architectures, known as catenanes and rotaxanes, are the backbone of the field and were central to the 2016 Nobel Prize awarded to Jean-Pierre Sauvage and Fraser Stoddart.5PubMed Central. Simplicity in the Design, Operation, and Applications of Mechanically Interlocked Molecular Machines
A molecular shuttle is one of the simplest devices. A ring-shaped molecule sits on a rod and can be driven between two resting positions by changing the chemistry around it. In one well-studied system, a positively charged ring shuttles between two electron-rich “stations” on the rod. When a voltage is applied or light hits the system, the electronic properties of one station change, making it less attractive, and the ring slides to the other station.6PubMed. A redox-driven multicomponent molecular shuttle The motion is real, measurable, and reversible. These shuttles have been described as candidates for artificial muscles because their sliding motion resembles the way filaments slide past each other in biological muscle fibers.7PubMed. Rotaxane-based molecular muscles
Molecular motors take the concept further. Instead of sliding back and forth, they rotate continuously in one direction. Ben Feringa, the third 2016 Nobel laureate, reported the first light-driven molecular rotor in 1999. It was a specially designed molecule with a carbon-carbon double bond at its center. Ultraviolet light triggers rotations around that bond in four discrete steps, and the molecule’s built-in chirality ensures rotation only goes one way, preventing the motor from just wobbling randomly.8PubMed. Light-driven monodirectional molecular rotor Each full 360-degree turn requires two bursts of light and two temperature-induced shape changes, but the result is genuine unidirectional rotation at the molecular level.
What Powers Them
Energy input is a fundamental challenge. At the molecular scale, random thermal jostling from surrounding molecules is enormous relative to any deliberate force. A molecular machine that is not actively driven by energy will simply be buffeted around by Brownian motion and accomplish nothing useful. Different energy sources have been developed to overcome this.
Light is the most common. Ultraviolet or visible light can trigger the shape changes that make molecular motors rotate or shuttles slide. A recent advance pushed this into the near-infrared range by pairing molecular motors with tiny upconversion nanoparticles that absorb near-infrared light and re-emit it at the shorter wavelengths the motor needs. The motors driven this way reached performance comparable to direct UV-powered systems, which matters because near-infrared light penetrates biological tissue much more deeply than UV.9PubMed Central. Highly Efficient Near-Infrared Light-Driven Molecular Motor Rotation Enabled by Upconversion Nanoparticles as Nanoscale Light Sources
Chemical fuel is the alternative. One system demonstrated a small ring being continuously transported around a circular molecular track, powered by a chemical fuel that is consumed in the process, much like gasoline in an engine. The motor keeps running as long as unreacted fuel is present.10PubMed. An autonomous chemically fuelled small-molecule motor This is closer to how biological motors work: ATP synthase runs on a proton gradient, kinesin burns ATP. In synthetic systems, chemical fueling has the advantage of not requiring an external light source, but the disadvantage that fuel molecules are consumed and waste products accumulate.
Electric fields and oscillating voltages offer a third approach. Researchers have created directed transport in nanoscale channels by combining asymmetric surface features with oscillating electric fields, producing what is called a rocking Brownian motor.11PubMed. Nanofluidic rocking Brownian motors Instead of fighting Brownian motion directly, these systems harness it: the random jiggling of molecules is rectified into net directional transport by the asymmetry of the environment, with the oscillating field providing the energy to keep the process going.
Molecular Pumps and Controlled Assembly
One of the most impressive recent developments involves molecular pumps that thread rings onto a track in a controlled sequence. In a 2025 study, researchers built a dual-pump system that accumulates ring-shaped molecules on one segment of a molecular thread, then drives them onto a different segment with a second pumping cycle. By controlling which cycle runs and whether free rings are available, they could produce different spatial arrangements of the rings along the thread.12PubMed Central. Controlled assembly of rotaxane translational isomers using dual molecular pumps This is significant because it moves beyond simply making something move. It demonstrates positional control: putting components where you want them, not just making them slide around.
The ability to position molecular components deliberately echoes a much older technique. Since the early 1990s, scanning tunneling microscopes have been able to push individual atoms around on surfaces, placing them at selected positions to build structures atom by atom.13PubMed. Atomic and molecular manipulation with the scanning tunneling microscope A related method uses voltage pulses to nudge adsorbed atoms into desired locations, and this works even at room temperature.14PubMed. Manipulation of adsorbed atoms and creation of new structures on room-temperature surfaces with a scanning tunneling microscope These scanning-probe techniques are exquisitely precise but painfully slow, which is why self-assembling molecular machines that organize themselves are so attractive as an alternative path to building complex nanoscale structures.
DNA as Construction Material
DNA might seem like an odd material for engineering, but it turns out to be one of the best building blocks for nanoscale devices. DNA origami, a technique where a long strand of DNA is folded into a desired shape using hundreds of short “staple” strands, can create custom nanoscale objects with remarkable precision. These structures can be designed with moving parts that respond to triggers like specific DNA sequences, pH changes, temperature shifts, or light.15PubMed Central. Dynamic DNA Origami Devices: from Strand-Displacement Reactions to External-Stimuli Responsive Systems
Researchers have built DNA origami devices that function as rotors and hinges. A DNA nano-rotor consists of a rotating arm anchored to a platform by a flexible joint, while a DNA nano-hinge has two arms connected by flexible single-stranded linkers that allow them to open and close like a book. By attaching magnetic nanoparticles to these structures, researchers can control them with external magnetic fields, steering the rotor or adjusting the hinge angle in real time.16Nature Communications. Real-time magnetic actuation of DNA nanodevices via modular integration with stiff micro-levers The appeal of DNA-based machines is programmability: you can design the structure on a computer, order the DNA strands from a supplier, and mix them together. The machine assembles itself.
Medical Applications
If you can build a molecular machine that drills through a cell membrane, you have a potential weapon against cancer cells. This is not hypothetical. Researchers have built molecular nanomachines that, when activated by near-infrared light using a two-photon excitation technique, physically bore through cell membranes and kill the cells. Because the two-photon approach is inherently focused on a very small spot, this offers high precision. The nanomachines can also be decorated with targeting peptides so they home in on specific cell types.17ACS Nano. Near-Infrared Light Activates Molecular Nanomachines to Drill into and Kill Cells
DNA origami has been used for similar purposes. A DNA origami nanodrill was engineered to attach to the surface of a cell, penetrate its membrane, and induce cell death. In tests against tumor cells, these nanodrills caused mechanical damage to the membrane, triggered a flood of calcium ions into the cell, and released intracellular proteins, all signs of the cell being physically ripped open.18PubMed. Predesigned DNA Origami Nanodrills Mediate Controlled Pore Formation on a Plasma Membrane and Cell Death
These are still laboratory demonstrations, not clinical treatments. But the precision of targeting, combined with the physical rather than purely chemical mechanism of action, is what excites researchers. A drilling nanomachine does not rely on a drug that cancer cells could develop resistance to. It mechanically destroys the membrane, and there is no straightforward way for a cell to evolve resistance to being punctured.
From Nanoscale to Human Scale
A single molecular motor produces an almost unimaginably small amount of force. To do anything visible, you need trillions of them working together. The challenge of amplifying molecular motion to the macroscopic scale has been one of the field’s central puzzles, and recent progress has been encouraging.
In one demonstration, light-driven molecular motors were embedded in a polymer material with their orientations aligned. When activated by light, the coordinated rotation of the motors at the nanoscale translated into a visible, repeated back-and-forth swaying motion of the polymer at the macroscopic scale. Each cycle of molecular rotation produced one cycle of macroscopic movement.19Chem. Macroscopic motion from synchronized molecular power strokes This is conceptually similar to how muscle works: individual myosin molecules produce tiny power strokes, but because billions of them are organized in parallel and synchronized, the result is the contraction of a whole muscle fiber.
Getting molecular machines to cooperate at large scales is an active area of research. The key insight is that both spatial alignment and temporal coordination matter. Motors pointing in random directions will cancel each other out, and motors firing at random times will produce only noise. The field of collective molecular machines focuses on strategies for organizing large numbers of molecular devices so their individual motions add up rather than cancel.20PubMed Central. Collective Molecular Machines: Multidimensionality and Reconfigurability
Putting Machines Inside Frameworks
One promising strategy for organizing molecular machines is to embed them inside porous crystalline materials called metal-organic frameworks. These are scaffold-like structures built from metal ions and organic linker molecules, and they can be designed so that the linker molecules themselves are functional machines. A metal-organic framework was built using a rotaxane as the linker, with the ring component of the rotaxane free to rotate rapidly inside the pores of the framework once the material was activated and space was cleared.21PubMed. Metal-organic frameworks with dynamic interlocked components
Simulations of these framework-embedded machines reveal complex behavior. Depending on the local environment, the moving rings inside the framework can oscillate quickly, move cooperatively with their neighbors, or even undergo directed shuttling. The interactions between neighboring machines within the framework modulate their behavior in ways that do not happen when the machines operate in isolation.22PubMed Central. Dynamic network of intermolecular interactions in metal-organic frameworks functionalized by molecular machines In another approach, dipolar molecular rotors embedded in a fluorinated framework could be triggered by carbon dioxide gas, which caused the rotors to reorient and change their collective behavior.23PubMed. Benchmark Dynamics of Dipolar Molecular Rotors in Fluorinated Metal-Organic Frameworks This kind of responsiveness to chemical stimuli hints at materials that could sense their environment and reconfigure themselves.
Toward Molecular Computing
If you can build switches at the molecular scale, you can in principle build logic gates, and if you can build logic gates, you can compute. This idea has been explored since the late 1990s. Researchers fabricated AND and OR logic gates from arrays of configurable switches, each consisting of a single layer of rotaxane molecules sandwiched between metal electrodes. The rotaxanes could be switched between two states electronically, and combining several switches reproduced the logic functions used in conventional computer circuits.24PubMed. Electronically configurable molecular-based logic gates
A different approach uses optical transitions in single molecules to perform logic operations. Light at different wavelengths serves as the input signals, and the molecule’s fluorescence serves as the output. A single molecule can act as a logic gate equivalent to at least two conventional switches.25PubMed. A molecular logic gate Molecular computing has not displaced silicon, and realistic assessments suggest it may never compete head-to-head with conventional electronics for general-purpose computation. Where it could find a niche is in specialized contexts: computation embedded in materials, diagnostic devices that process chemical signals directly, or hybrid systems where molecular switches interface with biological environments that silicon cannot easily reach.
Watching Machines That Are Too Small to See
One persistent challenge is simply observing these machines in action. You cannot watch a molecular motor spin under an ordinary microscope. The wavelength of visible light is hundreds of nanometers, and these machines are a few nanometers across. Researchers rely on indirect methods. Fluorescence techniques can measure distances between specific points on a molecule in the range of roughly 2.5 to 10 nanometers, allowing researchers to infer when parts of a machine have moved relative to each other.26ACS Nanoscience Au. Through the Eyes of Creators: Observing Artificial Molecular Motors
High-speed atomic force microscopy offers a more direct view. This technique scans a physical probe across a surface rapidly enough to produce movies of nanoscale objects. It has been used to watch DNA walkers taking steps of about 7.4 nanometers along a DNA origami track, captured in real time at a rate of one frame every 10 seconds.27ACS Nanoscience Au. Through the Eyes of Creators: Observing Artificial Molecular Motors These imaging advances matter because you cannot optimize a machine you cannot watch. Much of the early progress in the field was limited by the difficulty of confirming whether a molecular device was actually doing what the design intended.
Why Friction Does Not Work the Way You Expect
One of the most counterintuitive aspects of atomic machines is that friction, the great enemy of conventional engineering, is not really the problem everyone assumes it is. At the nanoscale, machines operate in a regime where thermal fluctuations dominate, and the thermodynamics of this regime are different from what we experience in daily life. An isothermal nanomotor, one that operates at a constant temperature, can in principle achieve thermodynamic efficiency approaching 100% despite strong dissipation, because at thermal equilibrium the net heat exchange between the motor and its surroundings is zero. The biological F1-ATPase motor, for instance, has been shown experimentally to operate near this theoretical limit.28Beilstein Journal of Nanotechnology. Molecular machines operating on the nanoscale: from classical to quantum
The practical implication is that designing nanoscale machines by trying to minimize friction, as a mechanical engineer would for a car engine, is misguided. The real design challenge is controlling directionality: ensuring that random thermal energy gets converted into net directional motion rather than pointless jiggling. This is why asymmetry in molecular structures, chirality in molecular motors, and carefully designed energy landscapes matter so much. The machine does not need to be slippery. It needs to be biased.
Safety Questions Nobody Expected
As molecular nanomachines become more effective at drilling through cell membranes, an uncomfortable question emerges: what happens when they encounter healthy tissue, or whole organisms? A study tested light-activated molecular nanomachines on multicellular organisms and found that fast-rotating machines caused roughly 70% mortality in roundworms and reduced movement and heart rate in water fleas. When applied topically to mouse skin and activated by light, the machines caused ulceration and microlesions in the skin tissue, penetrating into deeper layers of the epidermis.29PubMed Central. Molecular Nanomachines Can Destroy Tissue or Kill Multicellular Eukaryotes
These results are not surprising in hindsight. If a machine is designed to destroy cell membranes, it will destroy healthy cell membranes as readily as cancerous ones unless it is precisely targeted. The targeting problem, making sure the machines only attack the intended cells, is likely to be the bottleneck for any medical application. The peptide-targeting approach used in some studies helps, but biological targeting is never perfectly selective. And unlike a drug that is metabolized and cleared, a physical drilling mechanism raises questions about what happens to the machine after it has done its work. These are early-stage concerns, but they illustrate that atomic machines powerful enough to be useful are also powerful enough to cause harm.
Swarms and Self-Organization
If scaling up coordinated molecular machines sounds difficult enough with motors embedded in a fixed polymer or framework, consider the even wilder prospect of free-roaming swarms. Inspired by the collective behavior of ant colonies and fish schools, researchers are exploring how large numbers of active nanoscale agents could self-organize through local interactions alone, without any central controller. The idea is that simple rules governing how individual agents interact with their nearest neighbors can produce complex emergent behavior at the group level.30PubMed Central. Swarm Autonomy: From Agent Functionalization to Machine Intelligence
Synthetic swarm research is still in its early stages, bridging the gap between fabricating individual agents and getting them to do something collectively useful. Potential applications range from environmental sensing, where a swarm of nanoscale machines could map chemical gradients in a fluid, to materials science, where swarms could assemble structures from the bottom up. The field sits at the intersection of chemistry, robotics, and complex systems theory, and its progress depends partly on advances in the individual machines described throughout this article and partly on a better understanding of how simple local rules generate large-scale order.