Condensed Matter
The Barkhausen Effect: Magnetism That Moves in Jumps
The Barkhausen effect is the discovery that a piece of iron does not magnetize smoothly. When you slowly ramp up the field around it, the magnetization climbs in a staircase of tiny, sudden jumps — and if you wire the sample to a coil and a loudspeaker, you literally hear those jumps as a burst of crackling static. In 1919 Heinrich Barkhausen turned that hiss into the first direct evidence that the interior of a magnet is carved into domains, and each click is a microscopic wall breaking free and lurching forward.
- Discovered1919, H. Barkhausen (Dresden)
- Bloch-wall width (iron)~40 nm
- Wall energy density (iron)~3 mJ/m²
- Saturation field (iron)μ₀Mₛ ≈ 2.15 T
- Avalanche exponentP(s) ∝ s⁻ᵀ, τ ≈ 1.5
- Curie point (iron)1043 K (770 °C)
Interactive visualization
Press play, or step through manually. The visualization is yours to drive — try it before reading on.
Watch the 60-second explainer
A condensed visual walkthrough — narrated, captioned, under a minute.
A smooth ramp that answers in steps
Wind a coil of wire around an iron rod, connect the coil through an amplifier to a loudspeaker, and slowly bring a permanent magnet toward the rod. Intuition says the iron's magnetization should rise smoothly, so the coil should be silent. Instead you hear a rush of crackling static — a sound like frying bacon or crumpling foil. That is the Barkhausen effect, first heard by Heinrich Barkhausen at Dresden Technical University in 1919.
Each click is a real physical event. By Faraday's law the coil's voltage is ε = −N dΦ/dt, so the coil is silent only while the enclosed flux changes smoothly. The crackle proves that the magnetization is not climbing continuously: it advances in a rapid staircase of discrete jumps, each jump a sudden burst of dM/dt that induces a sharp voltage pulse. Plot magnetization M against field H under high magnification and the steep flank of the hysteresis loop resolves into thousands of tiny irreversible steps rather than a clean curve.
Barkhausen's jumps were the first direct experimental evidence for magnetic domains, the regions of uniform magnetization that Pierre Weiss had proposed theoretically in 1907. Barkhausen himself first imagined that each click was an entire domain flipping over at once. The later, correct picture — established through the 1930s — is subtler: on the steep part of the loop the dominant process is irreversible domain-wall motion, the boundaries between domains breaking loose from defects and lurching forward, not whole domains reversing in unison.
Domains, walls, and the four competing energies
Why does a magnet break into domains at all? A ferromagnet's configuration is set by the competition of four energy terms:
- Exchange energy — the quantum coupling that aligns neighbouring spins. It is minimized when the whole sample points one way, favouring a single domain.
- Magnetostatic (demagnetizing) energy — the energy stored in the stray field a uniformly magnetized body throws into space. It is enormous, and it is minimized by breaking the sample into oppositely oriented domains that close their flux internally. This term is why domains exist.
- Magnetocrystalline anisotropy — the crystal has easy axes along which magnetization prefers to lie (the <100> directions in iron). Anisotropy constant K ≈ 4.8×104 J/m3 for iron.
- Magnetoelastic energy — magnetization couples to strain (magnetostriction), so stress and dislocations reshape the domain pattern.
The compromise is a mosaic of domains, each magnetized to the saturation value Ms (for iron Ms ≈ 1.7×106 A/m, or μ0Ms ≈ 2.15 T), separated by domain walls. In a bulk metal these are Bloch walls, thin slabs in which the spin rotates gradually from one domain's direction to the next. The wall width scales as δ ≈ π√(A/K), balancing exchange stiffness A ≈ 10−11 J/m (which wants a wide, gentle twist) against anisotropy K (which wants a narrow wall so spins spend little time off-axis). For iron this gives δ ≈ 40 nm and a wall energy density σ ≈ 4√(AK) ≈ 3 mJ/m2. Magnetizing the sample means sweeping these walls so that favourably oriented domains grow at the expense of the rest.
Pinning and depinning: the anatomy of one jump
Here is the mechanism, force by force. An applied field H exerts a real pressure on a 180° domain wall. The domain aligned with the field has lower Zeeman energy density (−μ0MsH) than the anti-aligned one (+μ0MsH); moving the wall to grow the favourable domain releases energy, so the field pushes on the wall with a pressure p = 2μ0MsH. If the material were a perfect defect-free crystal, the wall would glide smoothly and reversibly and there would be no Barkhausen noise.
Real metals are full of pinning sites: nonmagnetic inclusions, voids, precipitates, grain boundaries, dislocation tangles, and frozen-in stress. Each defect is a local energy minimum for the wall — passing it costs wall area or lets the wall shed magnetostatic energy — so the wall clings there. As H rises, the wall does not move; it bows elastically like a soap film pinned at its edges, storing energy while its pinned segments hold fast. When the field pressure finally exceeds the maximum restoring force the defect can supply, the segment depins and snaps forward at high speed to the next stable configuration. That abrupt jump — a volume of material suddenly flipping its magnetization — is a single Barkhausen event.
Because the transition is a fall from a metastable state, it is irreversible and dissipative: the energy stored in the bowed wall and the Zeeman release are radiated away as eddy currents (in a conductor) and spin waves, and are converted to heat. Summed over the whole magnetization cycle, these irreversible jumps are the area of the hysteresis loop — the hysteresis loss. That is why Barkhausen noise is loudest on the steep flank of the loop near the coercive field, where irreversible wall motion dominates, and fades toward saturation where the remaining process is smooth, reversible rotation of magnetization out of its easy axis.
Avalanches, power laws, and crackling noise
Depinning events are not independent. When one wall segment jumps, it locally changes the demagnetizing field and the elastic tension felt by its neighbours, which can push them past their own thresholds. The result is an avalanche: a cascade of correlated jumps that can involve anything from a single defect to a large connected patch of the sample. This is why the crackle sounds like static rather than a metronome — the events come in bursts of wildly different sizes.
Measure the distribution of avalanche sizes s (the flux swept per event) and it is scale-free: P(s) ∝ s−τ over several decades, with a cutoff set by the sample or the demagnetizing field. There is no characteristic jump size, only a power law. For many soft ferromagnets τ ≈ 1.5, the value predicted by mean-field theory; short-range-dominated materials cluster nearer τ ≈ 1.27. The avalanche durations and energies follow their own power laws, and the noise power spectrum runs roughly as 1/f.
Two frameworks capture this. The ABBM model (Alessandro, Beatrice, Bertotti and Montorsi, 1990) treats the wall as a single elastic interface driven at constant rate through a random pinning landscape; it reproduces the avalanche statistics, the τ = 3/2 mean-field exponent, and the measured spectra. More fundamentally, Barkhausen noise is a nonequilibrium critical phenomenon. Sethna, Dahmen and coworkers showed it maps onto the random-field Ising model tuned near a disorder-induced critical point — a genuine phase transition in the shape of the hysteresis loop — which explains why microscopically messy, chemically different magnets share the same exponents (universality). The same statistics describe earthquakes, fracture, crumpling paper and superconducting vortex avalanches; the field calls it crackling noise (Sethna, Dahmen & Myers, Nature, 2001), close kin to Bak, Tang and Wiesenfeld's self-organized criticality, though here the criticality is disorder-tuned rather than strictly self-organized.
How we listen, and how we measure
The detector is Faraday induction. A pickup coil of N turns and cross-section A hugging the sample senses ε = −N dΦ/dt ≈ −Nμ0A dM/dt. A slow, smooth ramp of the applied field gives negligible signal, but each avalanche is a sub-microsecond spike of dM/dt, so the coil delivers a train of sharp voltage pulses. Amplified into a loudspeaker they become the audible crackle; on an oscilloscope they resolve into discrete steps; through a spectrum analyzer they yield the power-law statistics. Modern rigs record from roughly the audio band up to hundreds of kHz or beyond, filtering the slow drive from the fast jumps.
Independent techniques confirm that the jumps really are walls moving. Francis Bitter's colloid technique (1931) decorates walls with magnetic particles so they can be seen under a microscope; the magneto-optical Kerr effect (MOKE), Lorentz transmission electron microscopy, and magnetic force microscopy (MFM) all image domain patterns directly, and time-resolved versions catch walls jumping between pinning sites in step with the electrical Barkhausen pulses. The typical flux per jump is minute — a single event may flip a volume of order 10−15–10−9 m3 — and in bulk conductors the wall's speed during a jump is throttled to the order of metres to hundreds of metres per second by the eddy currents its own motion induces.
When it happens, when it does not, and why we care
The Barkhausen effect needs three ingredients: a material that is ferromagnetic or ferrimagnetic and below its Curie temperature (for iron, 1043 K / 770 °C — above it thermal agitation destroys the domain order and the crackle vanishes), a domain structure with mobile walls, and disorder to pin them. Remove any one and it disappears. A defect-free single crystal magnetizes almost silently by smooth reversible wall glide; paramagnets and diamagnets have no domains and never crackle; and the reversible bowing at very low fields, or the coherent rotation near saturation, produce no discrete jumps. Ultra-soft amorphous ribbons, with weak, delocalized pinning, crackle in a distinctly different statistical regime from hard, defect-rich steels.
Because the noise is a direct readout of the pinning landscape, it has become a practical tool. Magnetic Barkhausen Noise (MBN) analysis is a nondestructive-testing method: since stress, hardness, grain size, and phase content all change how strongly walls are pinned, they change the amplitude and spectrum of the crackle. Industry uses MBN to map residual stress, detect grinding burn, gauge case-hardening depth, and track fatigue in camshafts, gears, rails, bearings and turbine components — reading a part's mechanical history from the sound its domains make. In the opposite role, Barkhausen noise is a fundamental limit: it sets a floor on the resolution of fluxgate and magnetoresistive sensors and injects noise into magnetic recording read heads. A century after Barkhausen first pressed headphones to a coil, his crackle remains both a window into the physics of disordered systems and a working diagnostic on the factory floor.
| System | The avalanching unit | Slow drive | Size distribution |
|---|---|---|---|
| Barkhausen noise (ferromagnet) | Domain-wall segment depinning | Ramped external field H | P(s) ∝ s⁻τ, τ ≈ 1.5 |
| Earthquakes (crust) | Fault slip patch | Tectonic plate loading | Gutenberg–Richter, b ≈ 1 |
| Fracture / crumpling paper | Micro-crack or crease pop | Applied strain | Acoustic-energy power law |
| Superconductor vortex avalanches | Flux-line bundle jump | Rising applied field | Power-law jump sizes |
| Martensitic 'twin' transitions | Interface jump between phases | Cooling / stress | Acoustic-emission power law |
Frequently asked questions
What exactly makes the crackling sound in the Barkhausen experiment?
Nothing mechanical is making noise inside the iron. Each time a domain wall depins and jumps, the sample's magnetization changes abruptly, and by Faraday's law that sudden dM/dt induces a voltage spike in a surrounding pickup coil. Fed through an amplifier to a loudspeaker, the train of spikes becomes the audible crackle. The 'sound' is an electrical signal, not an acoustic one from the metal.
Why does magnetization jump instead of rising smoothly?
Domain walls are held by defects — inclusions, dislocations, grain boundaries and stress that act as pinning sites. As the field rises, a wall bows elastically but stays pinned until the field pressure (p = 2μ₀MₛH) overcomes the defect's grip, then snaps forward to the next site. That abrupt, irreversible release is the jump; a perfect defect-free crystal would magnetize smoothly with no Barkhausen noise.
What is an avalanche and why is the jump-size distribution a power law?
When one wall segment jumps it changes the local field and elastic tension on its neighbours, which can trigger them too, cascading into an avalanche. Because the system sits near a disorder-induced critical point, there is no characteristic avalanche size: the sizes follow a scale-free power law P(s) ∝ s⁻τ with τ ≈ 1.5. The same math describes earthquakes, fracture and crumpling paper, a family called crackling noise.
Did Barkhausen's experiment prove that magnetic domains exist?
It gave the first direct experimental evidence for them, in 1919, consistent with the domain theory Pierre Weiss had proposed in 1907. Barkhausen initially thought each click was a whole domain flipping; later work showed the jumps on the steep part of the loop are mainly domain walls breaking free from pinning sites. Direct imaging (Bitter patterns, MOKE, MFM) later confirmed the domains and their moving walls outright.
Where on the magnetization curve is the effect strongest?
On the steep flank of the hysteresis loop near the coercive field, where the magnetization process is dominated by irreversible domain-wall motion. It is weak at very low fields, where walls only bow reversibly, and near saturation, where the remaining change comes from smooth, reversible rotation of the magnetization out of its easy axis rather than jumping walls.
What is Barkhausen noise actually used for?
Magnetic Barkhausen Noise analysis is a nondestructive test. Because residual stress, hardness, grain size and phase composition all change how tightly domain walls are pinned, they alter the noise amplitude and spectrum, so measuring the crackle reveals a steel part's microstructure and stress state. It is used industrially to check residual stress, grinding damage, case-hardening depth and fatigue in gears, camshafts, bearings and rails.