Tip-Induced Band Bending

Scanning Tunneling Microscopy (STM) is a revolutionary tool that allows us to visualize and probe the world at the atomic scale. When applied to semiconductors—the foundation of modern electronics—it promises to unlock secrets of their electronic structure with unprecedented detail. However, this powerful technique harbors a subtle but profound complication: the very act of measurement can alter the object of study. The sharp, biased STM tip generates a strong local electric field that perturbs the delicate electronic landscape of a semiconductor, making it difficult to discern the material's true, unperturbed properties. This leads to a critical knowledge gap: how can we trust our measurements when our probe fundamentally changes what we are trying to measure?

This article delves into the physics of this observer effect, known as tip-induced band bending (TIBB). Across the following chapters, you will gain a comprehensive understanding of this phenomenon. The first section, "Principles and Mechanisms," will unpack the fundamental electrostatics that govern why a semiconductor's energy bands bend under the influence of the tip, how this bending distorts experimental results, and how specific surface conditions can alter this behavior. Following this, the "Applications and Interdisciplinary Connections" section will pivot from viewing TIBB as a mere problem to a powerful feature. You will discover how understanding TIBB enables the accurate characterization of semiconductor devices and how this concept links STM to other analytical techniques and physical principles, transforming a potential pitfall into a sophisticated diagnostic tool.

Imagine you want to measure the precise height of a person, but they are standing on a soft mattress. As you press down with your measuring tape, the mattress compresses. The harder you press, the more it sinks, and the measurement you get is wrong. The reading on your tape depends not just on the person's height, but on the firmness of the mattress and how hard you push. In the world of scanning tunneling microscopy (STM), when we try to "measure" the electronic properties of a semiconductor, we face a remarkably similar problem. The electric field from our STM tip, the very tool we use to probe the surface, inadvertently "presses down" on the semiconductor's electronic structure, distorting the very thing we wish to observe. This distortion is a beautiful and subtle piece of physics known as ​​tip-induced band bending (TIBB)​​.

To understand why this happens, let's think about how materials respond to an electric field. Picture the electric field from the STM tip as an uninvited guest trying to barge into the sample.

In a ​​metal​​, the situation is simple. A metal is swimming in a dense sea of mobile electrons. As the tip's electric field approaches, this sea of electrons instantly surges to the surface to form a perfect, infinitesimally thin shield. This shield creates an opposing field that completely cancels the tip's field, preventing it from penetrating the bulk of the metal. It’s like a perfect electric mirror. For this reason, the electronic structure inside a metal is largely unperturbed by the tip, and significant band bending does not occur.

A ​​semiconductor​​, on the other hand, is a different story. It has far fewer mobile charge carriers—perhaps millions of times fewer than a metal. When the tip's field arrives, the available carriers rush to the rescue, but there just aren't enough of them to form a perfect shield right at the surface. They do their best, but the electric field inevitably leaks past them, penetrating a significant distance into the material. This region where the field exists and charge has been redistributed is called the ​​space-charge region​​.

And here is the heart of the matter. According to one of the fundamental laws of electrostatics, ​​Poisson's equation​​, the curvature of the electrostatic potential ($\phi$) is dictated by the local charge density ($\rho$). In the space-charge region, charge has been moved around, creating a net charge density ($\rho \neq 0$). Poisson's equation, $\nabla^2 \phi = -\rho/\varepsilon_s$, tells us that the potential must therefore curve.

In a semiconductor, the allowed energy levels for electrons are not continuous; they are grouped into bands, most famously the ​​valence band​​ (normally full of electrons) and the ​​conduction band​​ (normally empty). The energies of these bands are not fixed; they follow the local electrostatic potential energy, which is given by $-q\phi$, where $q$ is the elementary charge. If the potential $\phi$ curves, the potential energy $-q\phi$ must also curve. This means the energy bands themselves must bend up or down, following the profile of the potential. This is tip-induced band bending.

The effectiveness of this screening depends on the carrier concentration. The more charge carriers available (e.g., from a higher doping concentration), the better the semiconductor can screen the tip's field. This is like making the mattress firmer. A higher doping density leads to a narrower space-charge region and, for a given tip bias, a smaller amount of band bending at the surface.

So, the bands bend. But do they bend up or down? The answer depends on a simple principle: like charges repel, opposites attract. It hinges on the type of semiconductor (n-type, with mobile electrons, or p-type, with mobile "holes") and the polarity of the voltage $V$ applied to the sample relative to the tip.

Let's consider an ​​n-type semiconductor​​, where the majority carriers are negatively charged electrons.

• If we apply a ​​positive sample bias ($V > 0$)​​, the sample becomes more positive than the tip. This positive potential repels the mobile electrons, pushing them away from the surface and deeper into the bulk. This leaves behind a region near the surface that is stripped of its mobile carriers, containing only the positively charged, immobile donor atoms. This is called a ​​depletion region​​. The net positive charge in this region causes the bands to bend ​​upwards​​ in energy.
• If we apply a ​​negative sample bias ($V 0$)​​, the sample becomes more negative than the tip. This negative potential now attracts the mobile electrons, pulling them towards the surface. This creates an excess of electrons right at the surface, a situation known as ​​accumulation​​. This layer of negative charge causes the bands to bend ​​downwards​​ in energy.

For a ​​p-type semiconductor​​, where the majority carriers are positively charged holes, the logic is simply reversed. A negative sample bias ($V 0$) repels the holes, causing depletion and downward band bending, while a positive sample bias ($V > 0$) attracts them, causing accumulation and upward band bending.

This bending has a profound consequence: it can push the band edges across the ​​Fermi level​​, the energy level that marks the boundary between occupied and unoccupied states at zero temperature. For instance, in an n-type sample under a sufficiently negative bias, the downward bending can push the conduction band minimum $E_C$ below the Fermi level $E_F$ at the surface. This fills the bottom of the conduction band with a dense layer of electrons, creating a two-dimensional electron gas. When the STM probes these states, it is seeing a direct tunneling contribution from this ​​accumulation layer​​.

TIBB is not just an academic curiosity; it is a ghost in the machine that can systematically distort our experimental data. One of the most powerful capabilities of STM is performing spectroscopy (STS) by measuring the differential conductance ($dI/dV$), which gives a map of the sample's local density of states (LDOS). From this map, we hope to measure the ​​band gap​​, the energy difference between the valence and conduction bands.

However, because TIBB shifts the bands up or down, the band edges we measure at the surface are not at their true, "flat-band" positions. The voltage onsets we measure for the valence band ($V_V$) and conduction band ($V_C$) are displaced. Typically, the band bending that occurs at positive and negative biases pushes the band edges away from the Fermi level, making the measured voltage difference, $|V_C - V_V|$, appear larger than the true band gap. We are measuring the height of the person on the compressed mattress.

The amount of this distortion depends critically on the tip-sample distance, $z$. The system can be modeled as two capacitors in series: the vacuum gap capacitance ($C_v$) and the semiconductor's space-charge capacitance ($C_s$). The applied bias voltage is divided between these two elements. The band bending is simply the voltage that drops across the semiconductor. Since the vacuum capacitance increases as the tip gets closer ($C_v \propto 1/z$), a smaller tip-sample distance leads to a stronger "push" and thus a larger amount of band bending.

This distance dependence seems like a terrible complication, but in a beautiful twist of scientific reasoning, we can turn this problem into the solution. Since the error (the band bending) depends on distance in a known way, we can systematically eliminate it.

The protocol is as elegant as it is powerful:

1. Acquire a series of $dI/dV$ spectra at the same location on the surface but at different tip-sample distances. This is easily done by changing the "setpoint" current of the STM—a lower current automatically retracts the tip to a larger distance.
2. From each spectrum, extract the apparent onset voltages for the conduction band ($V_C$) and valence band ($V_V$).
3. Plot these measured onset voltages against a quantity that is related to the tip's influence, like the inverse of the distance ($1/z$).
4. Extrapolate the data back to the limit of zero influence—that is, to infinite distance ($1/z \to 0$).

The extrapolated values of $V_C$ and $V_V$ at $1/z=0$ correspond to the true, intrinsic band edge positions, free from the tip's meddling influence. The difference between them gives the true band gap of the material. It is a stunning example of how understanding an artifact allows us to see right through it to the reality underneath.

Of course, sometimes we just want to take a quick image with minimal distortion. The capacitor model gives us a simple rule of thumb: to minimize TIBB, we should work at the ​​largest possible tip-sample distance​​ and use the ​​smallest possible bias voltage​​ that still provides a stable signal. This minimizes the electrostatic "push" on our electronic mattress.

What if the semiconductor surface had its own way of fighting back against the tip's field? This happens if the surface is decorated with a very high density of special electronic states, known as ​​surface states​​, located within the band gap.

Imagine our mattress is topped with a giant, super-absorbent sponge. Now, when you press down, the sponge simply soaks up the force, and the mattress underneath barely moves. These surface states act like an electronic sponge. They can easily absorb or release large amounts of charge to screen the tip's field, requiring only a minuscule change in the surface potential. As a result, the band bending is suppressed, and the Fermi level becomes "stuck" or ​​pinned​​ in place relative to the bands at the surface.

In this pinned regime, the semiconductor's surface behaves electrostatically much like a metal. Almost all the applied voltage drops across the vacuum, and the band bending is negligible and, crucially, independent of the tip-sample distance. A pinned surface thus has a clear experimental signature: its spectroscopic features, like surface-state peaks or band edges, ​​do not shift in energy​​ as the tip-sample distance ($z$) is varied. This provides a definitive way to distinguish a truly pinned surface from a normal, "bendable" one. The combination of measuring the electronic structure ($dI/dV$) with this test of electrostatic response ($z$-dependence) is a powerful diagnostic tool for uncovering the true nature of a semiconductor surface.

This phenomenon can even lead to striking effects in STM images. If the Fermi level is pinned asymmetrically within a band of surface states, a local atomic feature might have many occupied states below the Fermi level but few empty states above it. When imaged at negative bias (probing occupied states), the feature appears bright. But when imaged at positive bias (probing empty states), it appears dark. This bias-dependent ​​contrast reversal​​ is another tell-tale sign that the invisible hand of Fermi-level pinning is at play, clamping the bands and dictating what the microscope is allowed to see.

In our journey so far, we have grappled with the intimate physics of how a sharp scanning probe tip "talks" to a semiconductor surface. We discovered that this conversation, carried on by the flow of tunneling electrons and the language of electric fields, is not always a simple one. The tip, in its very attempt to observe, perturbs the delicate electronic landscape of the sample—it induces band bending. One might be tempted to dismiss this "tip-induced band bending," or TIBB, as a mere nuisance, an experimental artifact that obscures the pristine reality we wish to measure. And in some sense, that is where our story begins. But as is so often the case in physics, what begins as a vexing complication can, with deeper understanding, transform into a powerful tool and a bridge to new insights.

Imagine trying to record the pure sound of a violin. Your microphone, ideally, should just capture the music as it is. But what if the microphone itself produced a loud hum that mixed with the violin's notes? Your first task would be to understand and subtract this hum to hear the true music. This is precisely the challenge faced in Scanning Tunneling Spectroscopy (STS).

The "music" we are after is the electronic density of states (DOS)—a fundamental property that tells us how many quantum states are available for electrons at each energy level. In an ideal world, the differential conductance ($dI/dV$) we measure in STS would be a direct map of the sample's local density of states (LDOS). This beautiful proportionality, however, rests on a few key assumptions: a constant density of states for the metallic tip, an energy-independent tunneling probability, and—most crucially for our discussion—the absence of significant perturbations to the sample's energy levels.

TIBB violates this last assumption in a spectacular way. The electric field from the biased tip forces the semiconductor's energy bands to bend, shifting the very energy levels we are trying to measure. The measured $dI/dV$ spectrum is no longer the pristine melody of the LDOS, but a version distorted by the electrostatic "hum" of the probe. Therefore, the first and most fundamental application of understanding TIBB is learning how to see past it—to deconvolve the measurement and recover the true electronic structure that governs a material's behavior.

Semiconductors are the heart of modern technology, and their function relies on meticulously engineered interfaces, like the junction between $p$-type and $n$-type materials or the contact between a metal and a semiconductor. Understanding the electronic landscape at these interfaces, with nanometer precision, is paramount. This is where STM should shine, but it is also where TIBB can be most deceptive.

Consider a simple $p$-$n$ junction, the fundamental building block of diodes and transistors. Across this junction, the energy bands slope gracefully to create a built-in potential. When we scan an STM tip across it, we expect to see this slope reflected in the onset voltages for tunneling. Indeed, as we move from the $n$-side to the $p$-side, the conduction band moves further from the Fermi level, and the valence band moves closer. We should see the conduction band onset, $V_{\mathrm{CB}}(x)$, increase, and the valence band onset, $V_{\mathrm{VB}}(x)$, become less negative. However, TIBB plays a trick on our eyes. At the biases needed to probe the band edges, the tip's electric field pushes them further away from the sample's Fermi level. This artificially enlarges the measured band gap, $q(V_{\mathrm{CB}} - V_{\mathrm{VB}})$. This effect is most pronounced where the material is least able to screen the tip's electric field—in the depletion region at the heart of the junction. It is as if we are looking at the junction through a funhouse mirror that stretches the very features we want to inspect.

The situation becomes even more critical when we want to measure a precise quantity, like the Schottky barrier height, $\Phi_{Bn}$, which governs the behavior of a metal-semiconductor contact. This barrier is a fixed, equilibrium property of the interface. Yet, when we bring our STM tip nearby to measure it, the TIBB effect again gets in the way, modifying the apparent barrier height compared to its true value. How can we find the true height of this electronic "step"?

Here, physicists employ an elegant strategy. The strength of TIBB depends on the capacitive coupling between the tip and the sample, which in turn depends on their separation, $z$. By performing measurements at several different tip-sample separations and extrapolating the apparent barrier height to the limit of infinite separation ($z \to \infty$), we can computationally "remove" the tip and reveal the intrinsic, TIBB-free Schottky barrier height. It is a beautiful example of using a deep understanding of an artifact to nullify its effects and uncover the underlying truth.

So far, we have treated TIBB as an adversary to be outsmarted. But what if the artifact itself carries information? What if the "distortion" in the funhouse mirror is not random, but a direct consequence of the mirror's own carefully crafted shape?

The degree to which a semiconductor's bands bend under the tip's influence is not a universal constant. It depends critically on the material's ability to screen electric fields. This screening, in turn, is determined by the concentration of mobile charge carriers (set by the dopant density) and the presence of any charge-trapping states at the surface. A heavily doped region screens the tip's field effectively, leading to small TIBB. A lightly doped region will exhibit much more dramatic band bending than one with a high-density of surface states that "pin" the Fermi level.

This realization flips our perspective entirely. Instead of trying to eliminate TIBB, we can embrace it. We can build a complete physical model that includes everything we know: the quantum mechanics of tunneling and the classical electrostatics of band bending as described by Poisson's equation. In this model, the bulk doping concentration and the density of surface states are left as unknown parameters. We can then perform a "fitting" procedure, adjusting these parameters until our simulation of the tunneling current, including all TIBB effects, perfectly matches our experimental data.

By finding the unique parameters that explain the observed, distorted spectra, we can extract these hidden properties of the semiconductor with nanoscale resolution. TIBB is no longer a bug; it's a feature. The electrostatic conversation between tip and sample becomes an interrogation, and by carefully analyzing the sample's response, we can deduce its innermost secrets.

The physics of band bending is not confined to STM. It is a unifying concept that connects different experimental techniques and physical phenomena.

A wonderful example is Kelvin Probe Force Microscopy (KPFM), a technique that uses an oscillating conductive tip to map the surface potential of a sample. Unlike STM, KPFM measures the electrostatic force on the tip, not the tunneling current. It is designed to directly measure the contact potential difference, $V_{\mathrm{CPD}} = (\Phi_{\mathrm{tip}} - \Phi_{\mathrm{sample}})/q$, which depends on the local work function of the sample. On a semiconductor, any spatial variation in the equilibrium band bending—the bending that exists even without a biased tip—will cause a variation in the local work function. KPFM is therefore a direct method for imaging the landscape of surface band bending.

The connection is profound: the static, equilibrium band bending landscape that KPFM directly visualizes is the very same landscape upon which the STM tip superimposes its dynamic, bias-induced perturbation (TIBB). The two techniques provide complementary views of the same underlying surface electrostatics, one static and one dynamic.

Furthermore, these electrostatic effects are deeply intertwined with thermodynamics. The ability of a semiconductor to screen the tip's field depends on its population of mobile carriers. In an intrinsic (undoped) semiconductor, this carrier population, $n_i$, is exquisitely sensitive to temperature. As we raise the temperature, $n_i$ grows exponentially, and the material becomes a much better conductor. This enhanced screening means that the TIBB effect becomes weaker at higher temperatures. Consequently, a temperature-dependent STS measurement of the band gap contains not only the true shrinkage of the gap due to electron-phonon interactions but also an electrostatic artifact that evolves with temperature. Disentangling these two contributions requires a sophisticated model, but it enriches our understanding of the interplay between electronics, electrostatics, and thermodynamics at the nanoscale.

In the end, the story of tip-induced band bending is a microcosm of scientific discovery. We begin with a simple question, encounter a vexing problem, learn to work around it, and finally, learn to harness it. What started as an observer effect that spoiled our measurements becomes a sensitive probe of the system itself. The intimate electrostatic conversation between the tip and the sample, once understood, reveals far more than we initially thought to ask.