Sven Erik Matzen

Software Architect | Cloud & Security Expert | AI-enabled Solutions

The Most Exact Number in Physics: The Quantum Hall Effect and the Topology of Matter

🎧 Listen to this article

Quantum Physics · 2026-10-01

EU label: fully AI-generated content Fully AI-generated article (no prior review).

The Hook: A Number That Holds to Nine Digits

Physics has precise measurements, and then it has the quantum Hall effect.

Take a semiconductor chip, cool it in liquid helium, switch on a strong magnetic field, and push a current through it. Then measure the voltage that appears perpendicular to the current and divide it by the current. The result is an electrical resistance. And this resistance is not approximate, not roughly right, not correct within the error bars, but equal — to a relative uncertainty of about one part in a billion — to

$$R_H = \frac{1}{\nu}\cdot\frac{h}{e^2}$$

where \(\nu\) is an integer, \(h\) is Planck's constant and \(e\) is the elementary charge. Nothing else. No material parameter. No sample thickness. No electron mobility. No doping concentration. You can build the sample out of silicon or gallium arsenide or a single layer of carbon atoms, you can take it from a sloppy batch and tear its edges with tweezers — the value stays the same.

That is an absurd claim. Solid-state physics is the discipline of dirt effects. Real crystals are full of foreign atoms, lattice defects, strain and thermal disorder; almost every measurable property of a solid depends sensitively on how good the sample is. And yet it is precisely this dirty, disorderly world that delivers a constant of nature with an accuracy that competes with atomic clocks. Since the revision of the International System of Units in May 2019 the value is even fixed exactly: the von Klitzing constant is \(R_K = h/e^2 = 25{,}812.807\,459\,304\,5\ldots\ \Omega\), because \(h\) and \(e\) now have fixed numerical values by definition. Every national metrology institute in the world realizes the ohm from it today.

The question this article answers is: why does this work? Why is a measurement on an imperfect piece of matter so exact that you can found an SI unit on it?

The answer, in one word, is topology. The quantized Hall resistance is not the measurement of a material property but the counting of an integer — a topological invariant that simply cannot change continuously, because integers have no values in between. This insight, worked out between 1980 and 1988, did not merely explain an experiment. It introduced a new ordering principle into condensed matter physics, produced three Nobel Prizes (1985, 1998, 2016), founded an entire class of materials called "topological insulators," and predicted particles that are neither fermions nor bosons. And to this day it supplies what is perhaps the most serious — and so far most contested — candidate for a quantum computer that is error-resistant by nature.

Part 1: From Edwin Hall to the Quantum Staircase

The Classical Hall Effect, 1879

The story starts a hundred years earlier, with a graduate student. Edwin Hall, then at Johns Hopkins University, wanted to settle a question in 1879 that Maxwell had left open: does a magnetic field act on the current itself, or only on the wire that carries it? Hall placed a magnetic field perpendicular to a thin gold leaf carrying a current, and found a small voltage transverse to the direction of flow.

The classical explanation takes two sentences. The charge carriers drifting through the conductor experience the Lorentz force in the magnetic field and are deflected to one side. There they pile up until the resulting transverse electric field exactly compensates the Lorentz force. The associated Hall voltage is proportional to the current and to the magnetic field, and inversely proportional to the carrier density. It follows that the Hall resistance grows linearly with the magnetic field.

That was a wonderful tool. The Hall effect tells us the density of charge carriers in a material and even their sign — this is how physics learned that in some semiconductors the mobile charges are positive (holes). To this day every other rotation sensor contains a Hall probe. After the first few decades, however, the effect was no longer a sensation. It was textbook material, first semester, settled.

Flatness as a Prerequisite

For that textbook material to become a revolution, two technical prerequisites were needed that only became available in the 1960s and 1970s.

The first is the two-dimensional electron gas (2DEG). In a silicon MOSFET — the standard transistor of microelectronics — the electrons gather at the interface between silicon and silicon dioxide in a channel so thin that their motion perpendicular to the interface is quantum-mechanically frozen out. Only motion within the plane remains. Physically speaking, the electrons live in a world with two spatial dimensions. Later, molecular beam epitaxy with gallium arsenide–aluminium gallium arsenide heterostructures delivered far cleaner 2DEGs, in which electrons fly hundreds of micrometres without being scattered.

The second prerequisite is low temperatures and strong magnetic fields. Typical conditions for the standard are a few kelvin or less and flux densities of 10 to 20 tesla — values reachable only in dedicated high-field laboratories.

Landau Levels: When Circular Orbits Are Quantized

What happens to a two-dimensional electron gas in a strong magnetic field? Classically the electrons would run on circular orbits. Quantum-mechanically those orbits are not arbitrary: their energies form a discrete ladder, the Landau levels, with spacings proportional to the magnetic field. The continuous energy band of the free two-dimensional electron gas therefore breaks up, in a magnetic field, into a sequence of sharp energy steps with no states in between — in other words, into bands and gaps whose positions can be shifted by tuning the field.

Each Landau level can accommodate a certain number of electrons, proportional to the magnetic field. The ratio of electrons present to places available per level is called the filling factor \(\nu\). If \(\nu\) is exactly an integer, then exactly \(\nu\) Landau levels are completely occupied and all higher ones completely empty. In its interior the system is then an insulator: there are no nearby free places into which an electron could move without crossing the energy gap.

It is in precisely this state that the miracle happens.

The Night of 4–5 February 1980

In February 1980 Klaus von Klitzing was working at the high magnetic field laboratory in Grenoble on silicon MOSFETs supplied by Gerhard Dorda (Siemens) and Michael Pepper (Plessey/Cambridge). His actual goal was unspectacular: he wanted to study electron mobility in a magnetic field.

What he found in the traces was something else. The Hall resistance did not grow smoothly and linearly with the field. It formed broad, perfectly flat plateaus with steep transitions between them — like a staircase. And on the plateaus, precisely there, the longitudinal resistance vanished almost completely: the current flowed without dissipation. The plateau values were not arbitrary numbers but fractions of a universal quantity with integer denominators.

The decisive night shift was that of 4–5 February 1980. Von Klitzing satisfied himself that the plateau height did not depend on the sample. The paper appeared the same year: K. v. Klitzing, G. Dorda and M. Pepper, "New Method for High-Accuracy Determination of the Fine-Structure Constant Based on Quantized Hall Resistance," Physical Review Letters 45 (1980), 494–497. The title reveals where von Klitzing himself saw the point: here was a way to measure a constant of nature on a semiconductor chip. Five years later, in 1985, he received the Nobel Prize in Physics for it — alone, for the discovery of the quantized Hall effect.

Part 2: Why the Quantization Is So Unreasonably Exact

The Real Puzzle

That there are steps at all is not, by itself, entirely surprising — Landau levels are quantized, so structure is to be expected. The puzzle is the exactness and the width of the plateaus.

Exactness: the plateau height agrees with \(h/(\nu e^2)\) to parts in a billion. No solid-state effect does that. In ordinary solid-state physics the rule of thumb is that a theory should be glad to get factors of two right.

Width: the plateau is not a sharp point at exactly integer filling but an extended range of magnetic field. You can detune the field by percent, change the electron density, vary the temperature — and the value does not budge.

Most baffling of all: the plateaus get better when the sample gets worse. Disorder — normally the enemy of every precise solid-state experiment — is here the precondition for the robustness.

Edge Channels: The Current Flows Along the Rim

A first, intuitive part of the explanation lies in the geometry. At integer filling the interior of the sample is an insulator with an energy gap. At the edge, however, the confining potential bends the Landau levels upward; there they inevitably cross the Fermi energy, and conducting channels appear that run along the sample boundary.

These edge channels have a decisive property: they are chiral. At a given edge the electrons move in one direction only — the magnetic field fixes the sense of circulation. An electron that hits an impurity cannot be backscattered, because a backward-running state does not exist at that edge at all. It would have to tunnel straight across the insulating interior to the opposite edge. For a macroscopic sample that probability is astronomically small.

This makes it clear why the longitudinal resistance vanishes: there is no mechanism to remove momentum from the flow. And it makes it clear why disorder helps. The impurities trap the surplus electrons in localized states inside the energy gap. These localized states carry no current, but they pin the Fermi energy inside the gap over a wide range of magnetic field. Disorder therefore widens the plateau without touching its value — it is the buffer that preserves the quantization across a whole field range.

Topology: The TKNN Number

The edge-channel picture explains the robustness qualitatively. The deeper reason why the numerical value is exact and not merely almost right was supplied in 1982 by David Thouless, Mahito Kohmoto, Peter Nightingale and Marcel den Nijs — known in the literature as TKNN (Physical Review Letters 49 (1982), 405).

Their result, in words: the Hall conductance of a filled band is not some integral over material details but proportional to an integer describing a global, geometric property of the quantum-mechanical wavefunctions in momentum space. That number is today called the Chern number, and it is a concept from differential geometry, not from solid-state physics.

The analogy that helps me most is this. If you continuously deform a doughnut — squeeze it, stretch it, bend it — it remains an object with exactly one hole. The number of holes is a topological invariant: it does not change under continuous deformation, only if you tear the object or glue it. There is no "doughnut with 1.03 holes."

That is exactly what happens in the quantum Hall state. Impurities, strain, temperature changes and sample shape are continuous deformations of the system — as long as the energy gap is not closed. They cannot shift the Chern number by a small amount, because integers have no neighbourhood. The quantization is therefore not imprecise and accidentally good; it is enforced by mathematics. The only way to leave the plateau is to close the gap — and that is precisely the steep transition between two plateaus, a genuine phase transition.

Here lies the real conceptual force of the result. Before 1980 condensed matter physics essentially knew one ordering principle for phases: Landau's, in which phases differ by broken symmetries and a local order parameter (a magnet breaks rotational symmetry, a crystal breaks translational symmetry). The quantum Hall state breaks no additional symmetry and has no local order parameter. It differs from an ordinary insulator only by a global integer. That was a new kind of order.

Part 3: The Unit That Sits on a Chip

For metrology, von Klitzing's discovery was a gift whose value is hard to overstate.

Until then the ohm was embodied in material artefacts — precisely manufactured wire resistors kept in safes, compared against one another at regular intervals, and drifting all the same, because metal ages. That is the classic weakness of any artefact unit: you cannot rebuild it, only copy it, and copies diverge.

The quantum Hall effect replaced the artefact with a law of nature. Anyone with a 2DEG, a helium cryostat and a strong magnet can produce the ohm themselves, without ever having seen a reference resistor from another laboratory. From 1990 the metrology institutes used a stipulated conventional value, \(R_{K\text{-}90} = 25{,}812.807\ \Omega\). With the SI revision of 20 May 2019, in which \(h\) and \(e\) were assigned exact numerical values, \(R_K = h/e^2\) itself became an exactly defined quantity; the German PTB gives it as \(25{,}812.807\,459\,304\,5\ldots\ \Omega\) and achieves relative reproducibilities of about one part in a billion in practice.

The practical hurdle remained the overhead for a long time: liquid helium, fields of 10 to 20 tesla, a high-field magnet — nothing that fits into an ordinary calibration lab. This is where graphene has changed the picture. Because the Landau gap in graphene is far larger at the same field than in gallium arsenide, the quantum Hall state can be held stable there under considerably milder conditions. He and colleagues demonstrated in Nature Communications in 2022 an array of 236 Hall elements on epitaxial graphene on silicon carbide realizing \(R_K/236 \approx 109\ \Omega\) — at about 2 kelvin, 5 tesla and currents of several milliamperes, with the two subarrays agreeing to 0.2 parts in a billion. That is the road to a quantum-accurate resistance standard that needs no high-field laboratory.

To me this is the most beautiful practical payoff of the whole story: a unit that used to be a piece of metal in a safe is today a property of the universe, callable up in any sufficiently well-equipped laboratory.

Part 4: The Fractional Effect — When Electrons Lose Their Identity

Two years after von Klitzing, Daniel Tsui and Horst Störmer at Bell Labs studied the behaviour at still lower temperatures and still stronger fields, using extremely clean gallium arsenide heterostructures grown by Arthur Gossard. They found a plateau where, by the theory of the day, none was allowed to be: at filling factor \(\nu = 1/3\) (Physical Review Letters 48 (1982), 1559).

This was no minor addendum but a break with the theory. Integer quantization follows from completely filled Landau levels of independent electrons. One third of a level cannot be "full." A plateau at \(\nu = 1/3\) means the system develops an energy gap at that filling — and a gap at an only partially filled level cannot come from individual electrons. It must arise from their interaction.

Laughlin 1983 and the One-Third Charge

Robert Laughlin supplied the answer in 1983 in the form of a wavefunction of remarkable simplicity (Physical Review Letters 50 (1983), 1395). He described the state at \(\nu = 1/3\) as a new kind of incompressible quantum fluid in which the electrons get out of each other's way with maximal cleverness — a correlated liquid, neither a gas of independent particles nor a crystal.

The genuinely radical consequence concerned the excitations of this fluid. Laughlin's state possesses quasiparticles with charge \(e/3\). Not because the electron breaks apart — the electron remains indivisible — but because the collective excitation of the many-body state carries a charge that is one third of the elementary charge. An emergent property that resides in no single building block.

This one-third charge was measured directly in 1997 by two groups independently, by analysing the shot noise of the current — the noise reveals the charge of the packets carrying the current. In 1998 the Nobel Prize in Physics went to Laughlin, Störmer and Tsui, "for their discovery of a new form of quantum fluid with fractionally charged excitations."

A particularly elegant way of ordering the thicket of observed fractions (\(1/3\), \(2/5\), \(3/7\), \(2/3\), \(5/2\) and many more) came from the picture of composite fermions: one thinks of each electron as "dressed" with an even number of magnetic flux quanta. These composite objects feel a strongly reduced effective magnetic field — and the fractional states of the electrons then appear as integer states of the composite fermions. The complicated fractional effect becomes the simple integer effect in a different description.

Anyons: Neither Fermion nor Boson

Stranger still than the charge is the statistics of these quasiparticles. In three spatial dimensions there are only two kinds of particle: fermions, whose many-body wavefunction changes sign when two particles are exchanged, and bosons, for which it stays unchanged. In two dimensions, however, a third possibility is allowed: the wavefunction may pick up an arbitrary complex phase on exchange. Such particles are called anyons (from "any phase").

The fractional quantum Hall quasiparticles are anyons — at \(\nu = 1/3\) with an exchange phase of one third of the fermionic one. Two experiments from 2020 demonstrated this directly: Bartolomei and colleagues observed the signatures of fractional statistics in anyon collisions in Science, and Nakamura, Liang, Gardner and Manfra showed in Nature Physics 16 (2020), 931, in a Fabry–Pérot interferometer, the direct observation of anyonic braiding statistics: lead one anyon around another, and the wavefunction remembers that loop in a measurable phase.

The reason this goes far beyond curiosity value is hidden in that word "remembers." If the state of a system depends on how particles have been led around one another — and not on where exactly they were or how fast — then that information is immune to local disturbances. A jittering anyon stays on the same path topology. This is precisely the founding idea of topological quantum computing.

Part 5: From Laboratory Curiosity to a Class of Materials

Until the mid-1980s the quantum Hall effect was a spectacular but narrowly bounded phenomenon: two-dimensional electrons, millikelvin, tesla. The next decades consisted of recognizing that topology is the more general principle and the magnetic field merely one way of producing it.

Haldane 1988. Duncan Haldane constructed a lattice model that shows a quantized Hall conductance even though the mean magnetic field vanishes and there are no Landau levels. This proved that the Chern number is a property of the band structure, not a property of the magnetic field. All you need is broken time-reversal symmetry — an external field is one way to get it, intrinsic magnetism another.

Quantum spin Hall effect, 2005–2007. Charles Kane and Eugene Mele showed in 2005 that a topological classification exists even without broken time-reversal symmetry — a \(\mathbb{Z}_2\) invariant that takes only the values "trivial" and "non-trivial." The corresponding state has pairs of counter-propagating, spin-polarized channels at its edge. Bernevig, Hughes and Zhang proposed mercury telluride quantum wells as a concrete material in 2006; Laurens Molenkamp's group confirmed it experimentally in 2007. The name for this class of materials — topological insulators — became one of the most productive terms in condensed matter physics of the following fifteen years: materials that insulate in their interior and necessarily conduct at their surface, because the topology of the interior compels it.

Quantum anomalous Hall effect, 2013. The group of Cui-Zu Chang and Qi-Kun Xue observed quantized Hall conductance for the first time without any external magnetic field, in thin films of magnetically doped \((\mathrm{Bi},\mathrm{Sb})_2\mathrm{Te}_3\) (Science 340 (2013), 167; DOI 10.1126/science.1234414). Haldane's theoretical model of 1988 had become reality.

Nobel Prize 2016. The Royal Swedish Academy awarded the physics prize to David J. Thouless (1/2), F. Duncan M. Haldane (1/4) and J. Michael Kosterlitz (1/4) "for theoretical discoveries of topological phase transitions and topological phases of matter." Topology was now officially recognized as an ordering principle of matter in its own right.

Part 6: The Present — Fractions Without a Magnetic Field, and One Open Dispute

Fractional Chern Insulators, 2023/2024

The obvious question suggested itself: if the integer effect exists without a magnetic field (2013) and the fractional effect exists with one (1982) — does the fractional effect also exist without a field? Theoretically such a state is called a fractional Chern insulator, and it was hunted for more than a decade.

In 2023 it was found. Cai, Anderson and colleagues reported in Nature 622 (2023), 63–68, signatures of fractional quantum anomalous Hall states in twisted molybdenum ditelluride (\(\mathrm{MoTe}_2\)): in a bilayer twisted by about 3.8° a moiré superlattice forms in which ferromagnetic, fractionally quantized states appear at fillings of \(\nu = -2/3\) and \(\nu = -3/5\) — at around 1.6 kelvin.

Shortly afterwards Lu, Han and colleagues showed the same phenomenon in rhombohedral pentalayer graphene on hexagonal boron nitride, in Nature 626 (2024), 759–764, and with a whole series of plateaus (\(\nu = 1\), \(2/3\), \(3/5\), \(4/7\), \(4/9\), \(3/7\), \(2/5\)) at exactly zero magnetic field and temperatures up to about 400 millikelvin. That the same physics appears in two completely different material systems is the strong argument that this is not an artefact but a robust phenomenon.

The appeal is not merely academic. A fractional state without a magnetic field would be a far more practical platform for anyonic qubits than a high-field magnet — and at suitable fillings (classically discussed at \(\nu = 5/2\)) theory expects non-Abelian anyons, for which the order of the braiding operations changes the result. Such anyons would be directly usable as quantum gates.

The Open Dispute: Microsoft's Topological Qubit

Honesty requires a look here at the murkiest part of the field. The most heavily promoted application of the topological idea is the topological qubit based on Majorana states in semiconductor–superconductor nanowires, which Microsoft has been working on for over a decade.

In February 2025 Microsoft Azure Quantum published a paper in Nature 638 (2025), 651–655, on interferometric single-shot parity measurement in InAs–Al hybrid devices, and in parallel announced the "Majorana 1" chip. The criticism followed promptly and has since been formally published: Henry Legg (University of St Andrews) argues in a Matters Arising contribution in Nature of 24 June 2026 that the "topological gap protocol" used by Microsoft yields contradictory classifications depending on the choice of parameters, and that the conductance data available from the public repository show no clear energy gap in the relevant regions, but rather disorder and quantum-dot behaviour. Microsoft published a peer-reviewed reply in the same issue, in which the protocol is reframed as a mere tuning tool and the actual evidence is located in the radio-frequency capacitance measurements.

There is no retraction and no formal correction. The dispute is open and unsettled in the peer-reviewed literature. I am of the opinion that this is an instructive contrast: the integer quantum Hall effect was independently reproduced to parts in a billion by several laboratories within a few years — which is why an SI unit rests on it today. For the topological qubit, that independent, reproduced signature is precisely what is still missing. Topology as an ordering principle is secure; its use as quantum computer hardware is not.

Frameworks to Take Away: The Key Terms

Term What it means Why it matters
Hall resistance Transverse voltage divided by longitudinal current Classically linear in field, quantum-mechanically stepped
2DEG Electron gas with only two directions of motion Prerequisite for everything that follows
Landau level Quantized circular orbit in a magnetic field Creates energy gaps that the field can shift
Filling factor \(\nu\) Electrons per available place per level Integer = integer effect; fraction = fractional effect
\(R_K = h/e^2\) Von Klitzing constant, \(25{,}812.807\,459\ldots\ \Omega\) Exact since 2019; realizes the ohm
Edge channel Chiral conduction channel along the sample boundary No backscattering → no longitudinal resistance
Chern number (TKNN) Topological invariant of the band structure Integer, hence immune to perturbations
Localized states Electrons trapped by disorder Widen the plateau without changing its value
Anyon Particle with arbitrary exchange phase (2D only) Information resides in the braiding history
Fractional Chern insulator Fractional state without a magnetic field More practical platform for anyonic qubits

The Central Takeaway

The transferable lesson of the quantum Hall effect is a lesson about the right kind of robustness.

We are used to seeking reliability through precision in the details: tighter tolerances, purer materials, better calibration, more decimal places. The quantum Hall effect shows a different route. Its accuracy does not come from anything being particularly well manufactured — it comes from the fact that the measurable quantity can take only discrete values. An integer cannot drift. Where a continuous parameter is allowed to wander by a part per thousand, an integer can only jump all the way or not at all; and as long as the gap persists, it cannot jump.

This principle is immediately familiar in computing, even if it is rarely named this way. A digital signal is more robust than an analogue one precisely because only two states are permitted and small noise reaches neither — the "gap" between 0 and 1 plays the role of the energy gap. A hash is either equal or unequal. A consensus is either reached or not. In every one of these cases we buy robustness by leaving the continuum behind.

The practical prompt: the next time you design a system whose reliability worries you, do not ask first "how do we tighten the tolerances?" but "can the critical quantity be reformulated so that it takes only discrete values — and is there a sufficiently large gap between those values?" A state machine with five named states is more robust than a progress percentage for the same reason that a quantum Hall plateau is exact. Discretization is not a simplification; it is a robustness strategy.

Cross-References in the Vault

This topic connects closely with several articles already covered. Superconductivity, which like the quantum Hall effect is a macroscopic quantum phenomenon with an energy gap as the cause of its zero resistance — and whose type-II vortices are likewise topological objects — is treated in Resistance Is Futile: Superconductivity from Onnes to BCS to the Room-Temperature Dream. The idea of storing quantum information non-locally, so that local disturbances cannot touch it, is the thread running through Order from Noise: Quantum Error Correction and the Road to a Fault-Tolerant Quantum Computer; the surface code described there is itself a topological code. That quantum states can carry non-local correlations at all, on which anyonic braiding statistics rests, is shown in Spooky Action at a Distance: Quantum Entanglement from Einstein to the Quantum Internet and The Wager Against Reality: Bell's Theorem and the End of Local Realism. Why we see these states only at millikelvin temperatures while the everyday world appears classical is clarified in Why the World Turns Classical: Decoherence, Einselection, and Quantum Darwinism. The tunnelling between edge channels that ultimately limits the plateaus is the subject of The Leap Through the Wall: Quantum Tunneling from Gamow's Alpha Decay to the Attosecond Riddle. And how hard the road is from an artefact standard to a unit defined by a law of nature can be read in the history of navigation: The Clock That Measured the Ocean: John Harrison, the Longitude Problem, and How a Carpenter Saved Navigation.

Something to Think About

The quantum Hall effect is exact because its measured quantity can take only integer values — and because an energy gap ensures that small perturbations are not enough to jump from one value to the next. Where in your own work do you measure a continuous quantity when what actually matters is a discrete distinction? And how large would the "gap" between the permitted states have to be for you to rely on that distinction the way a metrologist relies on \(h/e^2\)?

Sources

← All articles