Design and Implementation of Full Adder Using Negative Differential Resistance Circuits Based on CMOS Process

Kwang-Jow Gan, Licheng ZhangResearch Square 2024 · DOI

Abstract

We demonstrate a full adder design based on several MOS-NDR circuits connected in series and in parallel. A basic MOS-NDR circuit is made of three standard Si-based metal-oxide-semiconductor field-effect-transistor (MOSFET) devices. By adjusting the appropriate parameters of the gate width of the MOSFET, the Λ-type negative differential resistance (NDR) characteristic can be presented in its combined current-voltage (I-V) curve. The operating principle of this 1-bit full adder is based on the theory of monostable-bistable transmission logic gate (MOBILE). The bias voltage of this circuit is a clock signal, and its output voltage is a low or high logic level determined by the relative current peak conditions of the load and driver components of the MOBILE circuit. Most of the NDR-based applications designed in the general literature use resonant tunnel diode (RTD) devices as the main components. Compared with RTD device, our MOS-NDR circuit can be designed with standard CMOS process and simulated with HSPICE. We implement this full adder using 0.18 µm standard CMOS process.

Introduction

Negative Differential Resistance (NDR) circuits have long been a subject of interest in the field of high-speed, low-power digital logic design. The unique current-voltage (I-V) characteristics of NDR devices—where current decreases as voltage increases over a specific range—allow for the implementation of complex logic functions with significantly fewer components than traditional static CMOS logic. Historically, the primary vehicle for NDR research has been the Resonant Tunneling Diode (RTD). While RTDs offer exceptional switching speeds, they are typically fabricated using compound semiconductors (like GaAs or InP) and require specialized growth techniques such as Molecular Beam Epitaxy (MBE). This creates a significant "integration gap," as these materials and processes are not easily compatible with the standard silicon-based CMOS manufacturing that dominates the semiconductor industry.

To bridge this gap, this research presents a 1-bit full adder designed using MOS-based NDR circuits. By utilizing standard silicon MOSFETs to replicate NDR behavior, the design achieves the architectural benefits of NDR logic while remaining fully compatible with modern VLSI production pipelines.

MOS-NDR Circuit and MOBILE Theory

The foundational element of this design is the MOS-NDR circuit. Unlike a two-terminal RTD, this circuit utilizes three n-channel MOSFETs. By carefully optimizing the gate widths ($W$) and lengths ($L$), the combined I-V curve of these transistors exhibits a Λ-type NDR characteristic. A critical advantage of this MOS-based approach is the presence of a "third terminal" voltage, $V_{GG}$. This terminal allows for post-fabrication tuning of the peak current ($I_P$), providing a mechanism to compensate for process variations—a flexibility not available in traditional RTD devices.

The logic operation of the full adder relies on the Monostable-Bistable Transition Logic Element (MOBILE) theory. A MOBILE gate consists of two NDR devices (a load and a driver) connected in series. The circuit is driven by a clock signal ($V_{bias}$). As the clock voltage rises, the circuit transitions from a monostable state to a bistable state. The final output voltage (high or low) is determined by the ratio of the peak currents of the load and driver components at the moment of switching. By modulating these peak currents through input signals, complex logic can be performed at the rising edge of the clock.

Full Adder Architecture

The 1-bit full adder is implemented by dividing the logic into two primary functional blocks: the Carry circuit and the Sum circuit.

Majority Logic for Carry Generation

The Carry bit in a full adder is essentially a majority function: it outputs a logic '1' if two or more of the three inputs ($V_{in1}, V_{in2}, V_{in3}$) are high. In the MOBILE framework, this is implemented by using three parallel NMOS switches to modulate the current of the load NDR element. This approach is significantly more area-efficient than traditional CMOS majority gates, which require a larger number of transistors to achieve the same result.

XOR Logic for Sum Generation

The Sum bit follows an XOR logic pattern ($V_{in1} \oplus V_{in2} \oplus V_{in3}$). This requires a more sophisticated arrangement of six NMOS switches and multiple NDR elements. The circuit is designed such that the relative peak currents between the driver and load satisfy the XOR truth table across all eight possible input combinations.

Experimental Results

The proposed full adder was simulated using HSPICE and subsequently fabricated using the TSMC 0.18 µm standard CMOS process. The resulting chip occupied an area of approximately $0.875 \times 0.89 \text{ mm}^2$.

Functional verification was conducted at a clock frequency of 2 MHz with a signal amplitude of 1.4V. The experimental measurements confirmed that the fabricated hardware correctly implements the full adder logic.

Truth table for the 1-bit full adder showing Carry and Sum outputs for all input combinations.

As shown in the truth table above, the outputs for both Carry and Sum bits perfectly align with standard binary addition rules. A key highlight of the experimental phase was the successful use of the $V_{GG}$ tuning voltage. By applying voltages ranging from 0.85V to 1.7V, the researchers were able to fine-tune the NDR characteristics to ensure stable logic transitions and overcome minor manufacturing deviations.

Conclusion

This work demonstrates the feasibility of implementing high-density NDR-based logic using standard CMOS technology. By replacing exotic RTDs with a three-transistor MOS-NDR configuration, the researchers have created a path for NDR logic to be integrated into mainstream silicon chips. While the current prototype operates at 2 MHz, the inherent scalability of the CMOS process suggests that significantly higher speeds are attainable with smaller process nodes. Future work will focus on integrating these components into larger Systems-on-Chip (SoC) and optimizing power consumption for large-scale arithmetic units.

Summary updated

Abstract

<title>Abstract</title> We demonstrate a full adder design based on several MOS-NDR circuits connected in series and in parallel. A basic MOS-NDR circuit is made of thee standard Si-based metal-oxide-semiconductor field-effect-transistor (MOSFET) devices. By adjusting the appropriate parameters of the gate width of the MOSFET, the Λ-type negative differential resistance (NDR) characteristic can be presented in its combined current-voltage (I-V) curve. The operating principle of this 1-bit full adder is based on the theory of monostable-bistable transmission logic gate (MOBILE). The bias voltage of this circuit is a clock signal, and its output voltage is a low or high logic level determined by the relative current peak conditions of the load and driver components of the MOBILE circuit. Most of the NDR-based applications designed in the general literature use resonant tunnel diode (RTD) devices as the main components. Compared with RTD device, our MOS-NDR circuit can be designed with standard CMOS process and simulated with HSPICE. We implement this full adder using 0.18 µm standard CMOS process.
BibTeX
@inproceedings{gan2024design,
  author = {Kwang-Jow Gan and Licheng Zhang},
  title = {{Design and Implementation of Full Adder Using Negative Differential Resistance Circuits Based on CMOS Process}},
  booktitle = {Research Square},
  year = {2024},
  doi = {10.21203/rs.3.rs-4739288/v1},
}
Figure