Reducing non-linear effects in Kelvin Probe Force Microscopy of back-gated 2D semiconductors

This paper demonstrates that using a suitably thin hBN back-gate dielectric in Kelvin probe force microscopy (KPFM) effectively mitigates non-linear electrostatic doping effects, enabling accurate measurement of Fermi levels and bandgaps in back-gated 2D semiconductors like WSe2.

The Gist

The Big Picture: Listening to the "Voice" of 2D Materials

Imagine you have a tiny, ultra-thin sheet of material (like a single layer of atoms) that acts as a semiconductor. Scientists want to know exactly how much "electricity" is flowing through it or what its energy level is at any given spot. To do this, they use a super-sensitive tool called Kelvin Probe Force Microscopy (KPFM).

Think of KPFM as a super-sensitive microphone that listens to the "voice" (voltage) of the material. By listening to this voice, scientists can map out the material's properties, find defects, and understand how it works.

The Problem:
Usually, this microphone works great on metals. But when you try to use it on these thin semiconductor sheets, it starts to distort the sound. Why? Because the microphone itself is a metal tip that hovers just above the material. When the tip vibrates to "listen," its own electric field accidentally pushes electrons around in the material, changing the very thing it's trying to measure. It's like trying to measure the temperature of a cup of coffee with a hot spoon; the spoon changes the temperature before you can read it.

In the past, this distortion made the measurements unreliable, especially when the material wasn't heavily "doped" (filled with extra electrons).

The Solution: The "Heavy Gate" Trick

The researchers in this paper discovered a clever way to fix this distortion. They realized the problem was that the "listening tip" was too strong compared to the "gate" (the control knob that sets the material's state).

The Analogy: The Tug-of-War
Imagine a tug-of-war between two teams:

1. Team Tip: The vibrating microscope tip trying to pull the electrons around.
2. Team Gate: The back-gate electrode (a control knob underneath the material) trying to hold the electrons in place.

In old setups, the gate was separated from the material by a thick layer of insulation (like a thick blanket). This made the gate weak. The Tip was strong and could easily yank the electrons, causing the measurement to be wrong.

The Fix:
The researchers used a very thin layer of insulation (hexagonal boron nitride, or hBN) between the gate and the material.

• Thinner insulation = Stronger Gate.
• Now, the Gate is a giant, muscular team that can easily overpower the Tip.
• Even though the Tip tries to wiggle the electrons, the Gate holds them firmly in place. The Tip can finally "listen" without changing the conversation.

What They Did

1. Built Better Devices: They created tiny transistors using Tungsten Diselenide (WSe2), a popular 2D material. They made some with thin insulation (the new, strong gate) and some with thick insulation (the old, weak gate).
2. The Test: They measured the "voice" (voltage) of the material while changing the gate knob.
3. The Result:
   • Thick Insulation (Old Way): The measurement was messy and didn't match the theory. The Tip was messing things up.
   • Thin Insulation (New Way): The measurement was perfect! It matched the mathematical models exactly. They could accurately measure the material's "bandgap" (the energy gap between conducting and non-conducting states) and found it matched what other scientists had calculated theoretically.

Why This Matters

Before this, scientists were often guessing when using KPFM on these materials because the tool was "breaking" the sample.

Now, with this Thin Insulation Trick:

• Accuracy: Scientists can finally trust their measurements.
• New Discoveries: They can now map out tiny defects, see how electricity flows at the edges of materials, and understand how these devices work in real-world electronics.
• Simplicity: They didn't need a brand-new, expensive machine. They just needed to change the thickness of the insulation layer, which is a simple tweak to existing technology.

The Takeaway

Think of it like trying to take a photo of a shy animal.

• The Old Way: You used a bright, loud camera flash (the Tip). The animal got scared and ran away (the electrons moved), so you couldn't get a good picture.
• The New Way: You used a powerful, quiet telephoto lens (the Thin Insulation Gate) that kept the animal calm and in place. Now, you can take a clear, perfect photo of the animal exactly as it is.

This paper shows us how to take that perfect photo of the microscopic world, opening the door to better, faster, and more efficient 2D electronic devices in the future.

Technical Summary

1. Problem Statement

Kelvin Probe Force Microscopy (KPFM) is a powerful tool for mapping surface potentials and Fermi levels ($E_F$) in 2D field-effect transistors (2DFETs). However, its application to 2D semiconductors (2DSCs) faces a critical ambiguity:

• Non-linear Response: In standard KPFM, an AC bias ($\tilde{V}_{tip}$) is applied to the tip to measure the contact potential difference. In 2DSCs, this AC voltage electrostatically "dopes" the channel, causing local band bending.
• Measurement Distortion: Because 2DSCs lack a bulk surface depletion region (unlike bulk semiconductors), they respond to the tip's field via in-plane charge flow. If the gate dielectric is thick (e.g., standard 90 nm SiO$_2$), the gate capacitance ($C_g$) is small compared to the tip capacitance ($C_{tip}$). This allows the tip's AC field to significantly modulate the channel potential ($V_{CH}$), creating a non-linear, asymmetric response.
• Consequence: This non-linearity causes the measured KPFM signal ($V_{KP}$) to deviate from the true Fermi level, particularly in low-doping regimes. Previous literature often reported slopes ($dV_{KP}/dV_g$) significantly less than 1 in low-doped regions, leading to ambiguous interpretations of band gaps and contact barriers.

2. Methodology

The authors employed a combination of theoretical modeling and experimental validation to isolate and mitigate these non-linear effects.

Theoretical Approach:

• Quasistatic Charge-Balance Model: The authors developed a model balancing charges on the tip ($Q_{tip}$), gate ($Q_g$), and channel ($Q_{CH}$).
• Capacitance Ratio ($R$): They defined a critical parameter $R = C_g / C_{tip}$. The model predicts that if $R \gg 1$, the gate effectively "shields" the channel from the tip's AC field, minimizing non-linear distortions.
• Iterative Simulation: They simulated the time-varying channel voltage ($\tilde{V}_{CH}$) over an AC cycle. By averaging the response, they calculated the DC component ($\langle V_{CH} \rangle$) to determine the error introduced in $V_{KP}$ for various $R$ values.
• Defect Modeling: To match experimental data, they incorporated an exponentially distributed "tail" of electron acceptor states near the conduction band edge, consistent with sulfur vacancy defects in similar materials.

Experimental Setup:

• Sample Fabrication: They fabricated back-gated 2DFETs using mechanical exfoliation and dry-transfer techniques.
  • Samples A-C: Monolayer WSe$_2$ with thin hexagonal Boron Nitride (hBN) gate dielectrics ($W_g \approx 13\text{--}22$ nm).
  • Sample E: Few-layer (3-layer) WSe$_2$ with a thick hBN gate dielectric ($W_g \approx 115$ nm), mimicking standard SiO$_2$ conditions.
• Measurement Technique: Dual-pass Frequency Modulation KPFM (FM-KPFM) was used under ambient conditions.
  • First Pass: Tapping mode to map topography.
  • Second Pass ("Nap" scan): The tip follows the topography at a controlled lift height ($W_T$) with reduced oscillation amplitude (attractive regime) to measure the KPFM signal.
• Variable Parameters: For Sample E, they varied the tip lift height ($W_T = 40, 60, 80$ nm) to modulate $C_{tip}$ and thus the ratio $R$.

3. Key Contributions

• Identification of the Root Cause: The paper definitively attributes the sub-linear KPFM response in 2DSCs to the non-linear interaction between the tip's AC field and the channel when $C_g$ is not sufficiently larger than $C_{tip}$.
• Design Rule for Accurate KPFM: The authors establish that using a thin gate dielectric (specifically hBN) to maximize $C_g$ relative to $C_{tip}$ ($R \ge 8$) is the necessary condition for obtaining quantitative, linear KPFM data in 2DSCs.
• Quantitative Modeling: They provide a validated model that accounts for quantum capacitance ($C_Q$) and defect states, allowing researchers to extract accurate band gaps and defect densities even when non-ideal conditions exist.

4. Key Results

• Thin Dielectric Success (Samples A-C):
  • With $W_g \approx 13\text{--}22$ nm, the ratio $R$ was $\ge 8$.
  • The measured $V_{KP}$ tracked the gate bias linearly with a slope $dV_{KP}/dV_g \approx 1$ (ranging 0.93–1.06) even in low-doping regimes.
  • The data matched the theoretical model perfectly, yielding a band gap ($E_g$) of 1.65 eV for monolayer WSe$_2$, consistent with literature.
  • The inclusion of band-tail states in the model improved the fit, revealing the presence of defect states.
• Thick Dielectric Failure and Recovery (Sample E):
  • With $W_g = 115$ nm, $R$ was low ($\approx 1.4\text{--}4$).
  • The KPFM signal showed significant non-linearity and a slope $< 1$, confirming the theoretical prediction of distortion.
  • However, by varying the tip height to increase $R$, the authors could model the non-linear response. The iterative model successfully reproduced the experimental data, extracting an $E_g$ of 1.1–1.2 eV for few-layer WSe$_2$.
• Validation of Shielding: The results confirmed that when $C_g \gg C_{tip}$, the tip's AC field is effectively screened, and the channel potential remains stable, allowing $V_{KP}$ to directly reflect the Fermi level.

5. Significance

• Reliability of KPFM: This work resolves a major ambiguity in the literature regarding KPFM measurements on 2D materials. It proves that KPFM can yield quantitative data on Fermi levels, band gaps, and contact barriers, provided the device geometry is optimized.
• Device Characterization: The technique enables the direct measurement of:
  • Spatial non-uniformities in 2D channels.
  • Sub-gap densities of states (defects).
  • Contact barriers and built-in potentials.
• Practical Guidelines: The paper provides a clear design guideline for future experiments: Use thin hBN gate dielectrics rather than thick SiO$_2$ to ensure $C_g \gg C_{tip}$. This minimizes the need for complex post-processing corrections and allows for more accurate, real-time characterization of 2D electronic devices.
• Broader Impact: By validating the quasistatic charge-balance model, the study extends the utility of KPFM from simple metallic surfaces to complex, gated 2D semiconductor systems, facilitating the development of next-generation 2D electronics.