Flexible radiofrequency carbon nanotube transistors operating at frequencies above 100 GHz

This paper reports the development of flexible radiofrequency transistors based on aligned carbon nanotube arrays that utilize electro-thermal co-design to achieve cutoff frequencies exceeding 100 GHz, thereby meeting the high-frequency, low-power, and flexibility requirements for 6G human-centric applications.

The Gist

The Big Picture: 6G Needs a Flexible Brain

Imagine the future of wireless internet (called 6G). It won't just be faster; it will be so fast that it can transmit data at the speed of light, allowing for things like holographic calls or remote surgery. To make this happen, our devices need to talk at incredibly high speeds—frequencies over 100 GHz.

But there's a catch: for these devices to be truly useful for humans (like smart bandages, skin sensors, or foldable phones), they need to be flexible (bendable) and low-power.

The problem? Making flexible electronics that run this fast is like trying to build a Formula 1 race car out of rubber. Usually, when you make electronics flexible, they get slow and overheat. This paper presents a breakthrough: a new type of transistor that is both super-flexible and super-fast, capable of handling 6G speeds without melting.

---

The Hero: The Carbon Nanotube "Highway"

The secret ingredient here is Carbon Nanotubes (CNTs).

• The Analogy: Imagine a standard silicon chip (like in your current phone) is a busy city street with lots of traffic jams. Now, imagine a Carbon Nanotube is a perfectly straight, empty, high-speed highway where cars (electrons) can zoom through without stopping.
• The Challenge: While these "highways" are amazing on rigid, hard surfaces (like a silicon wafer), they struggle on flexible plastic. Plastic is like a soft, squishy mattress; it doesn't conduct heat well. When the electrons zoom fast, they generate heat. On a mattress, that heat gets trapped, causing the device to overheat and break.

The Solution: "Electro-Thermal Co-Design"

The researchers didn't just build a fast transistor; they built a smart cooling system around it. They call this "electro-thermal co-design." Think of it as designing a race car and its cooling system at the exact same time, rather than building the car first and then trying to bolt on a radiator.

Here is how they solved the heat problem using clever engineering:

1. The "Heat Sinks" (Thick Metal Contacts):

   • The Problem: The flexible plastic substrate (the base) is a terrible heat conductor. It's like trying to cool a hot pan by putting it on a wool blanket.
   • The Fix: They made the metal contacts (the source and drain) very thick.
   • The Analogy: Imagine the transistor is a hot stove. Instead of relying on the wool blanket (plastic) to cool it, they attached thick copper pipes (the metal contacts) directly to the stove. These pipes act as a highway for the heat to escape sideways, away from the hot spot, before it can damage the device.

2. The "Thermal Bridge" (Gate Stack):

   • The Problem: The top part of the transistor (the gate) also gets hot.
   • The Fix: They used a special stack of materials (Aluminum, Titanium, Gold) and a specific type of glass (Aluminum Oxide) that conducts heat better than the usual materials.
   • The Analogy: It's like replacing a wooden roof on a hot attic with a metal roof that instantly vents the heat out the chimney.

3. The "Air Gap" Trick:

   • They left tiny air gaps between the gate and the wires. Air is an insulator for electricity (which is good for speed) but they managed the heat so well that the device didn't overheat despite this.

The Results: Breaking Records

Because of this smart design, the results are incredible:

• Speed: The transistors can switch on and off at 152 GHz (current frequency cutoff). This is the first time a flexible transistor has broken the 100 GHz barrier.
• Power: They do this while using very little electricity (low power), which is crucial for battery-powered wearable devices.
• Durability: They bent the devices like a piece of paper and even bent them back and forth 1,000 times. The performance barely dropped. It's like bending a race car into a pretzel, and it still runs perfectly.

The Real-World Test: The Amplifier

To prove this isn't just a lab trick, they built a Radio Frequency Amplifier (a device that boosts signals) using these transistors.

• They made it flexible.
• They ran it at 18 GHz (a common frequency for high-speed data).
• It successfully boosted the signal power significantly.
• The Metaphor: It's like taking a whispering microphone, making it bendable, and turning it into a megaphone that can shout clearly across a stadium, all while being light enough to wear on your wrist.

Why This Matters

This paper is a giant leap for 6G technology. It proves that we don't have to choose between "fast" and "flexible." By treating heat management as a core part of the design (not an afterthought), we can now imagine:

• Smart Skin: Sensors that stick to your skin to monitor your heart rate or blood sugar at super-fast speeds.
• Foldable 6G Phones: Phones that can fold in half but still download movies in a split second.
• Human-Integrated Tech: Devices that can be woven into clothes or attached to the body to communicate with the world instantly.

In short: The researchers took a material that is naturally fast (carbon nanotubes), put it on a material that is naturally slow and hot (plastic), and used clever engineering to make them work together like a perfectly tuned engine. The result is a flexible electronic component that is ready for the future of the internet.

Technical Summary

1. Problem Statement

The development of sixth-generation (6G) wireless communication technology requires terminals capable of operating at frequencies exceeding 100 GHz (sub-THz and THz ranges) while maintaining low power consumption and flexibility for human-centric applications (e.g., wearable health monitors).

• Current Limitations: Existing flexible radiofrequency (RF) transistors based on organic materials, amorphous oxides, 2D materials (like graphene), or III-V compounds struggle to exceed 100 GHz. Graphene-based flexible transistors have reached ~56 GHz, while previous flexible Carbon Nanotube (CNT) transistors were limited to ~1.6 GHz.
• The Bottleneck: A primary challenge for flexible devices is the poor thermal conductivity of polymer substrates (e.g., Polyimide/PI). High-frequency operation generates significant self-heating, which can exceed the breakdown temperature of the CNTs or the glass transition temperature of the substrate, leading to performance degradation or device failure.

2. Methodology

The authors developed a high-performance flexible RF transistor using aligned semiconducting carbon nanotube (CNT) arrays on a flexible Polyimide (PI) substrate, employing a strategy of electro-thermal co-design.

• Device Architecture:
  • Substrate: 2 µm thick PI substrate with low surface roughness (0.27 nm) and low dielectric loss.
  • Channel: Aligned CNT arrays (~150 CNTs/µm) transferred from a silicon wafer to the PI substrate.
  • Gate Structure: Top-gated architecture with air gaps between the gate and source/drain to minimize parasitic capacitance.
  • Scaling: Gate lengths ($L_g$) scaled down to 75 nm, with channel lengths ($L_{ch}$) of 120 nm, achieving a width-to-length ratio of 133:1.
• Electro-Thermal Co-Design Strategies:
  • Contact Engineering: Used a 20 nm Palladium (Pd) / 20 nm Gold (Au) stack. Pd ensures ohmic contact with CNTs, while the thicker Au layer enhances thermal conductivity and heat spreading through the contacts.
  • Gate Stack Optimization: Implemented an $Al_2O_3$ / Al / Ti / Au stack.
    • $Al_2O_3$ was chosen over $HfO_2$ for higher thermal conductivity.
    • An Al layer directly on $Al_2O_3$ creates a high thermal boundary conductance (TBC) interface (~200 MWm⁻²K⁻¹).
    • A Ti layer prevents unstable Au-Al intermetallic formation.
    • The total gate thickness (190 nm) was optimized to provide near-bulk metal thermal conductivity while minimizing gate resistance ($R_g$).
• Fabrication: Utilized dimension-limited self-alignment for CNT transfer, electron-beam lithography for patterning, and capillary-assisted electrochemical delamination to release the flexible chip from the temporary silicon carrier.

3. Key Contributions

• Record-Breaking Frequencies: First demonstration of flexible RF transistors with both extrinsic current-gain cutoff frequency ($f_T$) and maximum oscillation frequency ($f_{max}$) exceeding 100 GHz.
• Electro-Thermal Co-Design Framework: Established a comprehensive design methodology that balances RF performance (parasitic minimization) with thermal management (heat dissipation) specifically for flexible polymer substrates.
• Device Stability: Demonstrated that effective thermal management prevents self-heating-induced breakdown, a common failure mode in previous flexible high-frequency CNT devices.

4. Key Results

• DC Performance:
  • On-state current ($I_{on}$): 0.947 mA/µm.
  • Transconductance ($g_m$): 0.728 mS/µm.
  • Operated at low supply voltages ($V_{ds} \le 0.65$ V for peak $f_T$).
• RF Performance:
  • Peak Extrinsic $f_T$: 152 GHz (at $L_g = 75$ nm).
  • Peak Extrinsic $f_{max}$: 102 GHz (at $L_g = 160$ nm).
  • Intrinsic Performance: After de-embedding, intrinsic $f_T$ reached 310 GHz and intrinsic $f_{max}$ reached 168 GHz, entering the THz regime.
  • Power Consumption: Low DC power density of 199 mW/mm (for peak $f_T$) and 147 mW/mm (for peak $f_{max}$).
• Thermal Management:
  • Finite-element simulations showed that without co-design, CNT temperatures would exceed 600°C (breakdown). With co-design, the maximum temperature rise ($\Delta T$) was limited to 331°C for CNTs and 168°C for the PI substrate, keeping the substrate safely below its glass transition temperature (~350°C).
  • Devices showed excellent reliability, with no breakdown after repeated RF measurements, unlike control devices without thermal optimization which failed immediately.
• Flexibility and Amplifier Demonstration:
  • Mechanical Robustness: After 1,000 bending cycles (3 mm radius), $f_T$ degraded by only ~9.8%.
  • RF Amplifier: A flexible power amplifier (6 fingers, $L_g=160$ nm) operating at 18 GHz (K-band) achieved:
    • Output Power ($P_{out}$): 64 mW/mm.
    • Power Gain: 11 dB.
    • Power-Added Efficiency (PAE): 10%.

5. Significance

This work represents a major milestone for 6G technology and flexible electronics:

1. Breaking the 100 GHz Barrier: It proves that flexible electronics can operate in the sub-THz regime, a prerequisite for high-speed, low-latency 6G communications.
2. Material Superiority: It validates aligned CNTs as the superior channel material for flexible high-frequency applications, outperforming graphene, organic semiconductors, and 2D materials by a significant margin (e.g., ~3x higher $f_T$ than graphene, ~100x higher than previous flexible CNTs).
3. Integration Potential: The compatibility of CNTs with CMOS-like processes and the ability to create both digital and analog/RF circuits on the same material platform pave the way for System-on-Chip (SoC) solutions for wearable and implantable devices.
4. Thermal Management Paradigm: The study provides a critical blueprint for designing high-power flexible electronics, demonstrating that thermal engineering is as crucial as electrical scaling for achieving high performance on polymer substrates.

In conclusion, the authors have successfully bridged the gap between theoretical potential and practical application for flexible high-frequency electronics, enabling the next generation of human-centric, high-speed wireless communication terminals.