🔬 physics

CVD Multilayer Boron Nitride-encapsulated graphene Solution-Gated Field-Effect Transistors for wearable and implantable biosensors

This study demonstrates that encapsulating graphene Solution-Gated Field-Effect Transistors with commercially available, lower-quality multilayer hexagonal boron nitride effectively overcomes stability and mobility limitations, thereby enabling high-performance neural recordings for wearable and implantable biosensors.

Original authors: Carole-Anne Lernoud, Lionel Rousseau, Vincent Bouchiat, Blaise Yvert, Clément Hébert

Published 2026-10-30

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Carole-Anne Lernoud, Lionel Rousseau, Vincent Bouchiat, Blaise Yvert, Clément Hébert

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a world where the human body could be monitored not by bulky wires and rigid sensors, but by soft, flexible sheets that hug the skin or nestle gently against the brain. This is the promise of wearable and implantable biosensors, devices designed to listen to the body's electrical whispers, from the rhythm of a heartbeat to the complex firing of neurons. To make this possible, scientists have turned to a remarkable material: graphene. This is a single layer of carbon atoms, so thin it is essentially two-dimensional, yet it conducts electricity with incredible speed and sensitivity. Because it is so flexible and biocompatible, it can be integrated into devices that move with the body, reducing the risk of rejection by living tissue. However, there is a catch. When graphene is placed directly into the salty fluids of the body, its performance drops. The liquid environment creates electrical interference and traps that slow down the flow of electrons, making the sensor less sensitive and noisier than it theoretically should be.

To solve this, researchers have long suspected that wrapping the graphene in a protective shell could restore its high performance. The ideal shell is a material called hexagonal boron nitride, which acts like a pristine, invisible shield, keeping the graphene clean and allowing electrons to zip through without obstruction. The problem is that the highest quality versions of this shield are incredibly difficult and expensive to produce in large sizes. This leaves a gap between the theoretical potential of these sensors and what can actually be built in a lab. The question remained: could a more accessible, lower-quality version of this protective material still make a difference?

A team of researchers set out to answer this by testing a commercially available, multi-layer version of boron nitride that is easier to obtain. They built two types of sensors: one where the graphene was left exposed to the liquid, and another where the graphene was sandwiched between layers of this commercial boron nitride. They tested these devices on both rigid silicon wafers and flexible plastic films, mimicking the conditions of a real medical implant. The results were clear and encouraging. Even with the lower-quality, commercially sourced material, the encapsulated sensors performed significantly better than the exposed ones. The electrical signal flowing through the protected graphene was stronger, and the background static noise was noticeably quieter. In practical terms, this means the sensors could detect much fainter electrical signals, a crucial requirement for listening to the subtle activity of brain cells.

The researchers dug deeper to understand why this happened. They found that the protective layer did two important things. First, it improved the connection between the graphene and the metal wires that read the signal, reducing resistance at the contact points. Second, and perhaps more importantly, it shielded the graphene from the chaotic electrical environment of the liquid, allowing the electrons to move more freely. This resulted in a measurable increase in the speed at which the material could respond to changes in voltage. When the team calculated how fast the electrons moved, they found that the protected graphene was significantly more mobile than the unprotected version, confirming that the shield was doing its job despite not being the highest purity material available.

The study did not stop at rigid test boards. The team also fabricated these sensors on flexible polyimide films, the kind of material used in soft medical implants. Here, the protective layer proved to be a lifesaver for the manufacturing process itself. On the flexible substrates, only about 40 percent of the unprotected sensors survived the fabrication process and worked correctly. In contrast, 75 percent of the sensors with the boron nitride shield remained functional. This suggests that the encapsulation not only improves electrical performance but also makes the devices more robust and reliable to build.

Finally, the team tested whether these improved sensors could actually hear the electrical signals of the brain. They applied a synthetic signal that mimicked the sharp spikes of a neuron firing, with amplitudes as small as 150 microvolts. The sensors with the boron nitride shield successfully detected these tiny spikes, distinguishing them clearly from the background noise. One specific sensor, which had the highest sensitivity, could even pick up the smallest of these signals repeatedly. This demonstration proved that the technology is ready to move beyond the lab bench toward real-world applications, capable of recording neural activity with a clarity that was previously difficult to achieve with commercially available materials.

While the study used a lower-quality version of the protective material, the results suggest that the path forward is promising. The researchers noted that using even higher purity materials could further enhance performance, and that the protective layer might also help the sensors last longer in the body by preventing oxidation. For now, the work confirms that we do not need perfect, laboratory-grade materials to build the next generation of neural sensors. By using accessible, commercial materials to shield the graphene, scientists have taken a significant step toward creating wearable and implantable devices that are sensitive enough to listen to the human body, yet robust enough to survive the journey inside it.

Technical Summary: CVD Multilayer Boron Nitride-encapsulated Graphene Solution-Gated Field-Effect Transistors for Wearable and Implantable Biosensors

Problem Statement
Graphene Solution-Gated Field-Effect Transistors (gSGFETs) are promising candidates for wearable and implantable biosensors, particularly for neural recording, due to graphene's high carrier mobility, biocompatibility, and ability to be integrated into flexible substrates. However, current gSGFET performance often falls short of theoretical limits. When the graphene channel is in direct contact with liquid media, carrier mobility is reduced, and sensitivity is compromised by charge traps and mobility fluctuations caused by environmental impurities. Furthermore, stability issues arise from oxidation at the liquid interface. While encapsulation with high-quality, exfoliated hexagonal Boron Nitride (hBN) has been shown to mitigate these issues, the production of large-scale, high-purity hBN remains challenging and inaccessible for commercial applications. The central problem addressed is whether commercially available, lower-quality Chemical Vapor Deposited (CVD) multilayer BN can effectively improve gSGFET performance for neural recording applications.

Methodology
The study employed a comparative fabrication and characterization approach using two device types: bare gSGFETs and gSGFETs encapsulated with commercial multilayer CVD-BN.

Fabrication: Devices were fabricated on both rigid SiO2 substrates and flexible polyimide (PI) substrates. The process involved wet transfer of CVD-BN and single-layer graphene onto copper foils, followed by transfer to the target substrates. Metal contacts (Ti/Au) were defined via photolithography and evaporation. Graphene channels were etched using ICP plasma, and a final BN layer was transferred to fully encapsulate the channel. A SU8 layer was used to insulate tracks and define the liquid-contact area.
Statistical Scope: The study analyzed a large dataset comprising 96 BN-encapsulated and 96 bare gSGFETs distributed across a 4-inch silicon wafer to ensure statistical relevance.
Characterization: Devices were tested in Phosphate Buffered Saline (PBS). Key metrics included transfer curves ( $I_{ds}$ vs. $V_{gs}$ ), transconductance ( $G_m$ ), and intrinsic noise (measured as root mean square current noise, $I_{ds,rms}$ , and equivalent gate voltage noise).
Analysis Techniques:
- Transfer Length Method (TLM): Used to extract sheet resistance ( $R_{sh}$ ), specific contact resistivity ( $\rho_c$ ), and mobility ( $\mu$ ) by varying channel lengths.
- Direct Transconductance Method (DTM): Used as a comparative method for mobility extraction.
- Neural Recording Simulation: Flexible devices were tested with synthetic hippocampal action potentials (spikes of 300 µV and 150 µV) applied to an Ag/AgCl gate electrode to evaluate recording capability.

Key Results

Performance Enhancement: Encapsulation with commercial multilayer BN significantly improved device performance compared to bare devices.
- Transconductance: Mean conductance increased from 1.2 mS/V (bare) to 1.8 mS/V (encapsulated) on rigid substrates. On flexible substrates, normalized transconductance improved from 0.5 mS/V/sq to 1.0 mS/V/sq.
- Noise Reduction: The equivalent gate voltage noise decreased. For square devices on rigid substrates, noise dropped from 200 µV $_{rms}$ to 150 µV $_{rms}$ . On flexible substrates, it decreased from 250 µV $_{rms}$ to 130 µV $_{rms}$ .
- Mobility: Mobility values extracted via TLM increased from ~2100 cm²V⁻¹s⁻¹ (bare) to ~2200 cm²V⁻¹s⁻¹ (encapsulated) on SiO2. DTM analysis showed a similar trend, with encapsulated devices reaching 402 cm²V⁻¹s⁻¹ versus 210 cm²V⁻¹s⁻¹ for bare devices on flexible substrates.
Mechanism of Improvement: The performance gains were attributed to two factors identified through TLM analysis:
- Reduced Contact Resistance: Contact resistivity decreased from ~22 kΩ·µm (bare) to ~17.5 kΩ·µm (encapsulated).
- Improved Channel Conductivity: The sheet resistance ( $R_{sh}$ ) decreased, indicating higher channel mobility. This is attributed to the screening of Coulomb scattering from the liquid medium and a reduction in charge traps.
Fabrication Yield: Encapsulation proved beneficial for device robustness. On flexible polyimide substrates, the functional yield of encapsulated devices was 75%, compared to only 45% for bare devices, suggesting the BN layer protects the graphene during fabrication and handling.
Neural Recording Capability: Encapsulated flexible gSGFETs successfully recorded synthetic neuronal signals. Channels with high transconductance detected 300 µV spikes clearly above background noise. Notably, the most sensitive channel (B-023) could also detect 150 µV spikes, demonstrating the potential for detecting low-amplitude neural events.

Significance and Claims
The paper claims that even using commercially available, lower-quality CVD multilayer BN, it is possible to significantly enhance the sensitivity and noise performance of gSGFETs for neural recording applications. The study demonstrates that encapsulation is a viable strategy to overcome the limitations of direct graphene-liquid contact without requiring the difficult-to-obtain, high-purity exfoliated hBN.

The authors highlight that the improvement in performance (higher transconductance and lower noise) enables the detection of neuronal signals as low as 150 µV. Furthermore, the increased fabrication yield on flexible substrates suggests that BN encapsulation improves the reliability of wearable and implantable biosensors. The paper modestly concludes that while the current results are promising, further improvements in contact quality (e.g., via ozone treatment) and the use of higher-purity hBN could further enhance sensitivity. Additionally, the protective nature of BN suggests potential for increased device longevity and stability over time, though this was not explicitly tested in the current study. The work establishes a foundation for using accessible CVD materials to advance the practical application of graphene-based biosensors in clinical and wearable settings.

Technical Summary: CVD Multilayer Boron Nitride-encapsulated Graphene Solution-Gated Field-Effect Transistors for Wearable and Implantable Biosensors

More like this