Why can a partly filled electronic band fail to conduct?
A band is a range of allowed electron energies in a crystal. A simple band picture associates a partly filled band with electrical conduction: nearby empty states allow electrons to change their motion when a field is applied. Some transition-metal compounds exposed the limits of that reasoning. Mott’s 1949 discussion used nickel oxide as a central challenge. The missing ingredient was that moving one electron changes the energy of the other electrons around it.
Moving charge rearranges several electrons at once. The key question is how much energy the whole electronic system must pay for that rearrangement. A crystal can be periodic and still resist conduction because electrons strongly avoid occupying the same local region. This interaction-driven obstruction is the idea behind a Mott insulator. It is different from both a filled-band insulator and a wave localized solely by static disorder.
Sources: [1][2]
One hop can create a doubly occupied site
Consider a lattice with roughly one electron per site. If an electron moves onto a neighbouring occupied site, it creates a doubly occupied site and leaves an empty one behind. Repulsion makes this charge rearrangement costly. Delocalization lowers kinetic energy, so the state is determined by a competition between the gain from hopping and the cost of local double occupation.
The Hubbard model expresses that competition with two energies: a hopping strength and a repulsion cost for double occupation. It includes two spin components, the quantum alternatives that allow two electrons to share one orbital. Half filling means an average of one electron per site in that two-component band. Strong repulsion can suppress fluctuations away from this local number even though the noninteracting band would be partly empty. A real oxide also has several electronic orbitals, states on neighbouring ligand atoms and lattice motion, so its measured charge gap reflects more ingredients than this one local cost.
Sources: [2][1]
Spins still interact when charge motion is suppressed
When repulsion greatly exceeds hopping, a real hop that leaves double occupation behind may be energetically suppressed. Even without a completed charge transfer, double-occupation states can enter the quantum state with a small amplitude and shift the energy. This virtual hopping affects the relative energies of different spin arrangements. In the simple half-filled Hubbard model, this generates an exchange energy approximately four times the hopping energy squared divided by the repulsion energy, provided repulsion greatly exceeds hopping. Charge and spin can therefore remain active on very different energy scales.
For an illustrative choice of hopping of one tenth of an electronvolt and repulsion of two electronvolts, the simple exchange estimate is about twenty millielectronvolts. Those are invented model parameters, not measurements of nickel oxide. The example explains why an insulating sample can nevertheless have magnetic correlations and low-energy spin excitations. Magnetic order then concerns a lower energy scale than the cost of moving charge, and the two observations reveal different parts of the electronic motion.
Sources: [2]
Pressure offered a way to change the competition
Changing pressure alters interatomic distances and orbital overlap. It can therefore change electronic bandwidth and interactions without simply adding more carriers. McWhan, Rice and Remeika reported a first-order metal–insulator transition in chromium-doped vanadium sesquioxide in 1969, controlled by chromium concentration and pressure. Their phase diagram distinguished metallic, insulating and antiferromagnetic insulating regions.
This made pressure a physical tuning variable, not merely a way to warm or cool the same resistance curve. To interpret such measurements, one follows resistivity with temperature and pressure, checks reversibility or hysteresis, and compares magnetic and structural information. A jump in resistance supports a transition but does not uniquely identify the microscopic cause. Lattice changes and electronic correlations can influence one another. The value of the phase diagram is to test which changes occur together and which can be separated.
Sources: [3]
Cold-atom measurements asked directly about double occupation
A later experiment separated some of the material complications by trapping a two-component gas of fermionic atoms in an optical lattice. In 2008, Jördens and colleagues varied repulsive interactions without changing the lattice tunnelling and confinement in the same way as pressure changes a solid. They aimed to suppress motion through repulsion while controlling these ingredients separately.
The evidence included strongly suppressed double occupation, reduced compressibility inferred from the response to atom number, and a gapped excitation mode. Each observation addressed a consequence of costly charge rearrangement. A low double-occupation fraction alone would not suffice: a dilute cloud also has few pairs on the same site. The density response and excitation measurement made that simple alternative less plausible. The experiment did not measure electrical resistance, because the particles were neutral. It tested the underlying many-body mechanism using observables appropriate to atoms.
Sources: [2]
Electrons in glass posed another localization problem
Mott also studied conduction in noncrystalline materials. There the absence of a periodic lattice changes the character of the electronic states. Some states can be spatially localized, and an energy separating localized from extended states is called a mobility edge. An occupied localized state may contribute to a spectrum without carrying charge through the entire sample.
This is distinct from the interaction-driven Mott obstruction just discussed. Mott’s 1975 account of electrons in glass explicitly addressed localized states, hopping and debates over the mobility edge. The shared surname can obscure that difference when every poorly conducting material is called a Mott insulator. A useful explanation specifies whether it invokes disorder, repulsion, a filled band or several ingredients at once. In doped and amorphous solids these mechanisms can coexist, each affecting a different part of the measured current.
Sources: [4]
A hopping curve needs more than a straight-line fit
A localized electron can sometimes move with help from a vibration. At low temperature, the best hop need not reach the nearest site: a farther site can have a more favourable energy. Balancing spatial overlap against thermal energy leads to variable-range hopping. In the simplest constant-density-of-states picture, the characteristic exponent depends on dimension; the three-dimensional version predicts that conductivity falls exponentially with the fourth root of characteristic temperature divided by sample temperature.
Plotting logarithmic conductivity against inverse temperature to the one-quarter power can therefore test one candidate mechanism. It does not prove it over a narrow temperature interval. A fixed activation gap, a Coulomb-induced depletion of accessible states, contact resistance or mixed conduction paths can change the interpretation. One should compare plausible plots over a meaningful range and check sample geometry and low-field behaviour. The characteristic hopping temperature summarizes the spatial extent and number of available states; the charge gap in a periodic, strongly repelling system measures a different energy cost.
Sources: [4][5]
The essential advance was to ask what moving charge costs
Mott’s questions forced electronic structure to account for processes that a list of one-electron levels can hide. Interactions can make a local rearrangement expensive; disorder can localize states; thermal motion can assist hops between them. Experiments distinguish these situations by combining transport with occupancy, excitation and structural information.
Pressure experiments and atomic lattices approach the same question from opposite directions. One changes a real solid with all its coupled degrees of freedom; the other creates a simplified landscape and measures occupancy directly. Their comparison keeps the local price of a hop connected to a macroscopic loss of conduction. That connection is Mott’s enduring insight: a list of empty electron states becomes meaningful for transport only after the energetic response of the other electrons is included.
Sources: [1][2][4]