Extended States

In the realm of materials science and condensed matter physics, one of the most fundamental questions is what determines whether a material conducts electricity or not. The answer lies in the quantum mechanical nature of electrons, which can exist in one of two profoundly different conditions: they can be free to roam across the entire material in ​​extended states​​, or they can be trapped within a tiny region in ​​localized states​​. This simple-sounding duality is the key to understanding the behavior of metals, insulators, and semiconductors, and it reveals a world where perfection and chaos give rise to astonishing physical phenomena.

This article addresses the transition from the ideal world of perfect crystals, where all states are extended, to the real world of disordered materials, where quantum mechanics enables electrons to become caged by randomness. We will explore how this competition between extension and localization governs the properties of materials. You will learn about the core principles differentiating a conductor from an insulator at the quantum level, the pivotal concept of a "mobility edge" that separates free from trapped electrons, and the experimental signatures used to detect these invisible quantum states. Following that, we will see how these ideas have profound real-world consequences. We will begin our exploration by delving into the fundamental principles and mechanisms that govern this quantum duality, before moving on to see how these concepts are applied across various scientific and technological domains.

Imagine an electron wandering through a material. What does it see? What path does it take? The answer, it turns out, is one of the most profound and beautiful stories in modern physics, a tale that takes us from the perfect order of crystals to the chaotic anarchy of randomness, and finally to a new, breathtaking kind of order born from that very chaos. Our journey is about understanding the dual nature of electronic states: their ability to be either ​​extended​​—stretching across the entire universe of the material—or ​​localized​​—trapped forever in a tiny quantum cage.

Let's begin in an idealized world: a perfect, flawless crystal. Here, the atoms are arranged in a perfectly repeating lattice, like an infinite, three-dimensional chessboard. An electron moving in this pristine environment isn't a tiny marble bouncing off atoms. It's a wave, and thanks to the perfect periodicity, this wave can propagate endlessly without scattering. These are the quintessential ​​extended states​​, described by the famous ​​Bloch's theorem​​. The electron's wavefunction isn't located somewhere; its probability is spread evenly throughout the entire crystal. It is, in a very real sense, everywhere at once.

This wave-like nature leads to the formation of energy ​​bands​​—allowed highways of energy for the electrons—separated by ​​band gaps​​, which are forbidden zones. If the highest-energy electrons (those at the ​​Fermi energy​​, $E_F$) happen to live in a partially filled band, they can easily move and conduct electricity. This gives us a metal.

But what if the Fermi energy falls squarely within a band gap? In this case, even though the underlying states are extended Bloch waves, all the energy bands are either completely full or completely empty. A full band can't conduct current—it's like a completely packed parking garage where no car can move. With no available states at the Fermi energy to carry a current, the material is a ​​band insulator​​. Its conductivity at absolute zero temperature is precisely zero because the density of available states at the Fermi energy, $D(E_F)$, is zero. All the doors are open, but there's no one standing at the threshold to walk through.

The real world, however, is never so perfect. Every material has flaws: a missing atom here, an impurity there. This is ​​disorder​​. Classically, we think of disorder as a nuisance that causes electrons to scatter and creates electrical resistance. But quantum mechanics reveals a far stranger and more dramatic possibility.

Imagine dropping a pebble into a perfectly still pond. The circular waves expand outwards indefinitely. This is our extended state in a perfect crystal. Now, imagine a pond littered with random rocks and poles. The waves scatter, reflect, and interfere. If the disorder is strong enough, the wave can become trapped, with its energy confined to a small region. The interference patterns conspire to create a cage of destructive interference from which the wave can never escape. Its amplitude decays exponentially away from a central point. This is the essence of ​​Anderson localization​​, named after the physicist Philip W. Anderson who first predicted it. The resulting state is a ​​localized state​​.

This brings us to a fundamentally new kind of insulator, the ​​Anderson insulator​​. Unlike a band insulator, an Anderson insulator can have plenty of available electronic states at the Fermi energy ($D(E_F) > 0$). The problem is that every single one of these states is a quantum cage. At absolute zero temperature, an electron placed in one of these localized states is stuck. It cannot travel across the material to conduct electricity. The doors are there, but they all lead to tiny, locked rooms. In a perfect crystal, this simply cannot happen; the absence of disorder guarantees that all states are extended Bloch waves.

So, we have two extreme possibilities: the perfectly extended states of a flawless crystal and the hopelessly localized states of a very messy one. But what about the vast landscape in between? Is every state in a disordered material localized?

The answer is a resounding no, and this is where the picture gets truly interesting. It turns out that an electron's fate—whether it is free or caged—depends on its energy. A great analogy is swimming in the ocean. Near the shore, in the shallow water with lots of rocks and complicated currents (the "band tails" where the density of states is low), it's easy to get trapped or pushed back. But out in the deep, powerful water (the center of the band), you can swim freely across vast distances.

In a disordered material, there exists a critical energy that acts as a sharp boundary between these two regimes. This is the ​​mobility edge​​, denoted as $E_c$.

• If an electron's energy $E$ is below the mobility edge ($E E_c$), it is localized.
• If an electron's energy $E$ is above the mobility edge ($E > E_c$), it is extended.

The mobility edge is the great divide. The ultimate nature of the material—whether it's a metal or an insulator—depends on which side of this divide the Fermi energy $E_F$ falls. If $E_F$ lies in the sea of localized states, the material is an insulator. If we can somehow "add more water" (by adding more electrons, for instance) and raise $E_F$ above $E_c$, the material undergoes an ​​Anderson metal-insulator transition​​ and becomes a conductor.

This transition isn't always abrupt. As a metallic system becomes more disordered, quantum interference effects begin to build up. Even before states become fully localized, a phenomenon called ​​weak localization​​ emerges. It's a precursor, a quantum echo of the coming confinement. Here, an electron scattering through a random path has a chance of following the exact same path in reverse. These two time-reversed paths interfere constructively, which has the subtle effect of making it more likely for the electron to end up back where it started. This enhances backscattering, increasing resistance and acting as a quantum correction to the classical diffusion of extended states. As disorder increases, this effect grows, pushing the system towards the full localization transition, which often occurs around the point where the electron mean free path $\ell$ becomes comparable to its wavelength $\lambda$, a condition known as the Ioffe-Regel criterion, $k\ell \sim 1$ (where $k=2\pi/\lambda$).

This is a beautiful theory, but how do we know it's true? We can't take a "photograph" of an electron's wavefunction. Instead, physicists act like detectives, looking for clues and signatures left behind by these extended and localized states.

One set of clues comes from ​​transport measurements​​. At zero temperature, an Anderson insulator doesn't conduct. But what if we heat it up a little? An electron in a localized state can borrow a bit of thermal energy from the vibrating atoms (phonons) to "hop" to a nearby localized state. At very low temperatures, it's most efficient for the electron to make long hops to find a state that's close in energy, a process called ​​variable-range hopping​​. This gives rise to a very specific, and experimentally verified, temperature dependence of conductivity, the Mott law, where $\ln \sigma \propto -T^{-1/4}$ in three dimensions. If we supply more heat, some electrons gain enough energy to jump all the way over the mobility edge into the extended states, leading to a conductivity that increases exponentially with temperature, following an Arrhenius law.

We can also probe the system with an alternating current (AC) field. An extended state in a metal behaves like a flowing river of charge. A localized state, however, is more like a charge on a spring. It can't travel far, but it can be "sloshed" back and forth by an AC field. This sloshing motion allows it to absorb energy and contribute to conductivity at finite frequency, $\omega$. A hallmark of this hopping conduction in the localized regime is that the AC conductivity grows with frequency, typically as $\sigma(\omega) \propto \omega^2$, in stark contrast to the Drude model for metals where conductivity is highest at zero frequency.

A second, more subtle, signature is found in the "music" of the quantum system: the statistics of its energy levels. If you list all the possible energy eigenvalues, the spacing between them tells a profound story.

• In the localized regime, wavefunctions are spatially separated and don't interact. Their energies are uncorrelated, like random numbers thrown on a line. The distribution of their spacings follows a ​​Poisson distribution​​, $P(s) = \exp(-s)$, where finding two levels very close together is quite probable.
• In the metallic regime, the extended states overlap and interact strongly. Thisquantum mechanical "interaction" causes the energy levels to repel each other—it's very unlikely to find two levels with nearly the same energy. This phenomenon of ​​level repulsion​​ is a deep signature of quantum chaos and is described by ​​Wigner-Dyson statistics​​, where the probability of small spacing goes to zero, $P(s) \to 0$ as $s \to 0$. The exact form depends on the fundamental symmetries of the system, like the presence or absence of a magnetic field. Listening to the level spacings is like distinguishing a random sequence of notes from a complex symphony.

What, then, is the nature of a state exactly at the mobility edge, $E = E_c$? This is a critical point, a place of transition, and like many critical points in physics, it's where the most bizarre and beautiful structures emerge.

A state at the mobility edge is neither fully localized nor fully extended. It is a ​​critical state​​. Its wavefunction doesn't decay away exponentially like a localized state, nor does it fill space uniformly like an extended state. Instead, it forms a ​​multifractal​​. Imagine a coastline. From a satellite, it looks jagged. Zoom in with an airplane, it's still jagged. Stand on the beach, and the line of the water on the sand is still jagged. It exhibits structure on all length scales; it is self-similar.

The probability distribution $|\psi(\vec{r})|^2$ of a critical state is just like that: it is highly intermittent, with sparse regions of high probability and vast voids of near-zero probability. But the pattern of this clumpiness looks statistically the same no matter how closely you zoom in. To quantify this, physicists use tools like the ​​Inverse Participation Ratio (IPR)​​, which measures how "spread out" a state is. For an extended state in a system of size $L$, the IPR shrinks like $L^{-d}$ (where $d$ is the dimension). For a localized state, it's a constant. For a critical multifractal state, it follows a unique power law, $L^{-D_2}$, where $D_2$ is a fractal dimension that is less than the dimension of space itself. This confirms that the state is an infinitely intricate object, more than a point but less than a fully filled space.

Our story culminates in one of the most stunning phenomena in all of science: the ​​Integer Quantum Hall Effect​​. Consider a two-dimensional gas of electrons. A remarkable result of scaling theory is that in 2D, in the absence of a magnetic field, any amount of disorder, no matter how weak, localizes all electronic states. The 2D world should be universally insulating!

But then, we apply a strong magnetic field perpendicular to the 2D plane. This does two things. First, it breaks time-reversal symmetry, which kills the weak localization effect that is so potent in 2D. Second, it quantizes the electron motion into circular orbits called ​​Landau levels​​.

What happens when you have both disorder and these Landau levels? The result is magical. Each sharp Landau level broadens into a band of states. And within each of these bands, the mobility edge re-emerges in a spectacular fashion. Most of the states in the band—those in the tails—are localized by the disorder. But, as an act of profound topological necessity, there must exist at least one ​​extended state​​ at the very center of each Landau level.

These rare, critical extended states act as superhighways for charge transport across the sample. The vast majority of localized states act as reservoirs, trapping electrons and keeping them out of the flow. When the Fermi level lies within the sea of localized states, the longitudinal conductivity vanishes, but a perfect, quantized Hall conductivity is carried by the system's edge states. As the Fermi level is swept through the center of the band, it crosses the single extended state, causing the Hall conductivity to jump precisely to the next integer multiple of $e^2/h$.

The existence of extended states, a concept that began with simple questions about crystals and disorder, becomes the key to unlocking a macroscopic quantum phenomenon of unprecedented precision. It shows us that even in a world riddled with randomness, deep principles of quantum mechanics and topology can conspire to create a perfect, incorruptible order. The journey from the perfect crystal to the disordered landscape leads us not to ruin, but to a new and more subtle form of perfection.

After our journey through the fundamental principles of extended and localized states, you might be left with a delightful and nagging question: "This is all very interesting, but what is it for?" It is a wonderful question, the kind that drives science forward. You see, the distinction between an electron being trapped in a small region versus being free to roam across a whole universe of atoms is not some abstract bit of quantum bookkeeping. It is, in fact, one of the most powerful and practical concepts in modern science. It is the secret behind the silicon in your computer, the reason some materials perform quantum mechanical miracles, and even a profound challenge for the very computers we use to simulate reality.

Let us now explore this landscape of applications. We will see how this simple-sounding dichotomy—stuck versus free—is the key that unlocks a stunning variety of phenomena, from the mundane to the truly magical.

When we think of a semiconductor, we often picture a perfect, crystalline lattice, a repeating, orderly city of atoms. But many of the most important semiconductors in our daily lives—the stuff of solar panels and the large flat-screen displays in our living rooms—are not crystalline at all. They are amorphous, a jumble of atoms frozen in place, more like a sprawling, chaotic city than a neat grid. In this disorder, we find the first, and perhaps most commercially important, application of our ideas.

In a perfect crystal, there is a clean "band gap," an energy range that is strictly forbidden to electrons. To get conduction, an electron must be given enough energy to jump Clean across this gap, from the full "valence band" to the empty "conduction band." But what happens in a disordered, amorphous material? The structural chaos creates a messy energy landscape. It's as if the sharp cliffs of the band gap have crumbled, creating slopes of rubble on either side. These slopes are populated by a zoo of "tail states" that extend into what would have been the forbidden gap.

Now, an electron in one of these tail states is localized. It is trapped in a small pocket of the disordered landscape, a quantum pothole from which it cannot easily escape. Imagine you shine light on such a material. A photon might have just enough energy to kick an electron from a valence state into one of these localized tail states. We have absorbed the light, yes, but has it helped us create an electrical current? Not at all. The electron is trapped. It cannot contribute to conduction.

To generate a current, the electron must be promoted all the way up to the extended states—the quantum superhighways that run through the material. The energy required to do this, to get from the highest extended state in the valence band to the lowest extended state in the conduction band, is called the ​​mobility gap​​. This gap is larger than the "optical gap," which only measures the energy to create any excitation, even to a useless, trapped state. Therefore, if you want to design a good solar cell, it is the mobility gap that you must pay attention to. It tells you the true energy cost for creating a mobile charge carrier that can be harvested as electricity. In lower-quality materials with more disorder, the tails are more extensive, and the difference between the optical gap and the mobility gap becomes even more critical. This simple distinction is at the heart of engineering the electronic devices that power our world.

We have seen that disorder creates localized states. But can we go the other way? Can we start with a collection of purely localized states and, by some trick, coax them into forming a highway of extended states? The answer is a resounding yes, and it describes a beautiful phenomenon known as the ​​metal-insulator transition​​.

Imagine an insulating crystal, like silicon, into which we introduce a few impurity atoms—say, phosphorus. Each phosphorus atom has one more electron than the silicon atoms it replaces. This extra electron is loosely bound to its parent phosphorus atom, like a small moon in orbit. It is in a localized hydrogen-like state, described by a characteristic size, the effective Bohr radius $a_B^*$. If the impurity atoms are very far apart, their little electron-moons do not interact. The material remains an insulator, because each electron is tied to its own atom.

Now, let’s start adding more phosphorus atoms. The average distance between them shrinks. Eventually, the fuzzy quantum "bubbles" of the electron wavefunctions begin to overlap. At first, it's just a slight touch, and an electron might occasionally "hop" from one atom to a neighbor. But as we continue to increase the concentration, $n$, a magical moment occurs.

At a certain critical concentration, $n_c$, the overlapping bubbles merge into a continuous, sample-spanning ocean. A path of extended states has formed throughout the entire crystal! An electron is no longer tied to any single atom; it is free to cruise through this newly formed "impurity band." The material has abruptly transformed from an insulator into a metal. This transition, known as a Mott transition, follows a wonderfully simple and predictive rule of thumb first proposed by Sir Nevill Mott. It occurs when the average distance between impurities, $n_c^{-1/3}$, becomes a specific multiple of the electron's orbital size, $a_B^*$. For a vast range of materials, this relationship is found to be shockingly consistent: $n_c^{1/3} a_B^* \approx 0.25$ This is a beautiful example of a percolation phenomenon. The emergence of an extended state is the moment when isolated islands of electron probability connect to form a continent. By simply tuning the density of impurities, we can create extended states on demand, flipping a switch that turns a material from an insulator into a wire.

Perhaps the most profound and astonishing consequence of the interplay between extended and localized states is found in the ​​Integer Quantum Hall Effect​​. Here, a seemingly simple experiment—measuring the electrical resistance of a two-dimensional electron gas in a strong magnetic field—reveals a quantum mechanical world of exquisite perfection, a perfection made possible only by the presence of imperfection.

The experimental discovery was bizarre. As the magnetic field was increased, the Hall resistance (the voltage measured perpendicular to the current flow) did not change smoothly. Instead, it formed a series of incredibly flat plateaus, quantized to fantastically precise values of $h/ie^2$, where $i$ is an integer. How could such perfection emerge from a real, messy, disordered sample?

The answer lies in the beautiful dance between localization and extension. In a strong magnetic field, electrons are forced into circular cyclotron orbits, quantizing their energy into discrete "Landau levels." If the sample were perfect, the density of states would be a series of infinitely sharp spikes. But in a real sample, the ever-present disorder potential—the same kind of messiness we saw in amorphous silicon—broadens these sharp levels into bands.

And here is the crucial insight: according to the scaling theory of localization, in two dimensions with a strong magnetic field, almost all the states within these broadened bands become ​​localized​​. We can picture this using a semiclassical analogy: the disorder creates a random landscape of hills and valleys. An electron's guiding center drifts along the equipotential contours of this landscape. For most energies, corresponding to being near the top of a hill or the bottom of a valley, these contours are closed loops. An electron on such a path is localized; it just goes around in circles forever, a private little eddy that never contributes to a net flow across the sample.

However, at exactly one special, "critical" energy at the center of each broadened Landau band, something remarkable happens. Here, the equipotential contour percolates, snaking its way from one end of the sample to the other. This is our ​​extended state​​. It is the river that flows through the landscape of eddies.

Now the existence of the plateaus becomes clear. The Hall resistance plateaus occur whenever the Fermi energy—the "water level" of the electron sea—lies within a region of localized states. Since all the states at the water's surface are just spinning in place in their private eddies, they cannot respond to a small change in voltage or magnetic field. They form an incompressible quantum fluid. The transport properties are "stuck," frozen. The Hall conductivity is determined only by the number of extended-state "rivers" that are fully submerged below the Fermi level. And since this is an integer number, the conductivity is perfectly quantized. The plateaus are wide precisely because there's a large range of energies populated by these localized states that act as a buffer.

The transition between one plateau and the next occurs only when the Fermi level crosses one of the critical energies harboring an extended state. For a brief moment, a conducting river is at the water's surface, allowing the system's topological character to change, and the Hall resistance jumps to the next quantized value. The localized states are not just bystanders; they are essential actors. They are the reason for the stability and width of the plateaus. Using an elegant thought experiment involving adiabatically threading a magnetic flux quantum through the sample, one can rigorously show that only the extended states can carry charge across the system; the localized states are immune to this process and contribute nothing to the Hall current. The Quantum Hall Effect is thus a macroscopic quantum phenomenon in which the chaos of disorder gives birth to an unprecedented level of precision, all thanks to the stark contrast between localized and extended states.

We have seen how the physics of extended states governs the flow of electrons in materials. But in a final, mind-bending turn, this same concept haunts the very computational tools we use to understand quantum mechanics itself. The field is quantum chemistry, and the tool is Density Functional Theory (DFT), a Nobel-prize-winning method used to calculate the properties of molecules and materials.

One of the great challenges in modern DFT is a subtle but pernicious problem known as ​​delocalization error​​. Many of the most common and practical approximations to DFT have an inherent bias: they incorrectly favor situations where an electron's charge is smeared out, or delocalized, over many atoms.

Let's consider a simple, yet profound, thought experiment. Take two hydrogen molecules and place them very far apart. Now, add one extra electron to this system. Where does it go? Physics tells us the answer is obvious: the electron will localize on either the first molecule or the second one. The ground state is degenerate, with two equally valid localized possibilities. But if you ask a standard approximate DFT calculation, it will often give you a bizarre answer. It will claim the lowest energy state is one where the electron is delocalized, with half a charge on the first molecule and half a charge on the second.

This is a failure to describe a fundamental aspect of electron correlation. The theory incorrectly finds a delocalized, extended state to be more stable than the true, localized one. This error arises because these approximate theories have a flaw in their mathematical structure; they exhibit a "convex" behavior for the energy as a function of electron number, which always favors spreading things out. The exact theory, in contrast, is "piecewise linear"—it correctly understands that you can't have half an electron in two places at once.

This is not just a theorist's curiosity. This delocalization error leads to very real problems, for instance, in predicting the energy required to remove an electron from a molecule (the ionization potential). For states that are naturally delocalized, like in conducting polymers, this error causes the calculated orbital energies to be unreliable predictors of the true ionization potential. What a beautiful, unifying irony! The very same physical principle—the competition between an electron being localized versus being extended—that dictates whether a piece of silicon conducts electricity also manifests as a "ghost in the machine," a deep challenge for the computational methods we rely on to explore the quantum world.

From the glowing screen of your phone to the frontiers of quantum computation and theoretical chemistry, the simple question of whether an electron is free or trapped echoes through science. It is a testament to the fact that in nature, the most profound consequences often spring from the simplest of ideas.