Recent Advances in mm-Wave and Sub-THz/THz Oscillators for FutureG Technologies

Abstract: This paper provides a concise yet comprehensive review of recent advancements in millimeter-wave (mm-wave) oscillators below 100 GHz and sub-terahertz (sub-THz/THz) oscillators above 100 GHz for next-generation computing and communication systems, including 5G, 6G, and beyond. Various design approaches, including CMOS, SiGe, and III-V semiconductor technologies, are explored in terms of performance metrics such as phase noise, output power, efficiency, frequency tunability, and stability. The review highlights key challenges in achieving high-performance and reliable oscillator designs while discussing emerging techniques for performance enhancement. By evaluating recent design trends, this work aims to offer valuable insights and design guidelines that facilitate the development of robust mm-wave and sub-THz/THz oscillators for future communication, computing, and sensing applications.

• The paper demonstrates multi-core oscillator architectures that achieve up to 195–200 dBc/Hz FoM with significantly reduced phase noise.
• It details harmonic extraction and waveform shaping techniques that extend tuning range and enable sub-THz/THz signal generation beyond device fmax limits.
• Innovative passive resonator designs and multi-modal coupling are shown to enhance spectral purity and power efficiency in FutureG systems.

Advances in mm-Wave and Sub-THz/THz Oscillator Design for FutureG Systems

---

Context and Motivation

The exploration and utilization of mm-wave (30–100 GHz) and sub-THz/THz (>100 GHz) frequency bands underpin the ongoing evolution of 5G/6G and emerging FutureG communication, sensing, and computing systems. The pursuit of higher spectral efficiency, robust phase noise (PN) characteristics, wide frequency tunability, and power-efficient architectures is driven by novel use cases including massive-MIMO, ISAC, wireline transceivers (Tbps data rates), automotive radars, metaverse/IoT, and pervasive edge-AI. Contemporary oscillator design is challenged by the conflicting requirements for ultra-low PN, broad tunability, scalable integration, and area efficiency, especially as frequency generation blocks become system-critical for both communication and sensing modalities.

---

Fundamental Concepts and Performance Metrics

Oscillator performance is conventionally quantified by PN, tuning range (TR), efficiency, and stability. The figure of merit (FoM), defined as:

$$\text{FoM} = -\mathcal{L}(\Delta\omega) + 20\log_{10}\Delta\omega - 10\log_{10}(10^3 \cdot P)$$

captures PN at a specified offset, normalized for power consumption. However, FoM does not embody TR, leading to alternative metrics (e.g., FoMT), though lacking rigorous theoretical foundations. Theoretical models, notably Leeson's formula and Impulse Sensitivity Function (ISF) theory, guide PN analysis and optimization, accounting for tank Q, thermal and flicker noise sources, upconversion phenomena, and waveform shaping (2604.26903).

---

mm-Wave Oscillators: Fundamental-Tone and Harmonic-Tone Architectures

Fundamental-Tone Oscillator Design

Recent designs emphasize multi-core, coupling, and geometric innovations to reduce PN and enhance TR:

• Multi-Core Architectures: Bilateral coupling of N oscillator cores leverages PN reduction $\sim10\log_{10}N$. Resonant mode engineering (quad-core, mesh, circular XFMR) strategically augments Q, suppresses undesired oscillation modes, and fortifies synchronization against frequency mismatches. Experimental realizations have achieved FoM exceeding 195 dBc/Hz and octave-spanning TR (2604.26903).
• Waveform Shaping/ISF Engineering: Techniques including class-F, class-C, and superharmonic coupling manipulate waveform symmetry and harmonic content, reallocating noise sensitivity to less-critical phases. Integrated multi-resonant tanks and ISF-tailoring circuits achieve PN suppression, with reported reductions up to 3 dB.
• Passive Resonator Innovation: Inductor sharing, folded/mesh topology, and distributed slow-wave structures optimize area and Q, delivering low PN at high frequencies. Series-resonance tanks theoretically offer $10\log(Q^2)$ dB PN improvement, validated by BiCMOS prototypes displaying -138 dBc/Hz PN at 1 MHz offset.
• Wideband/Mode-Switching Techniques: Octave TR and dual-mode operation are realized via switchable cores, magnetic/capacitive mode switching, and transformer-based varactor coupling. These approaches mitigate switch-induced Q degradation, mode ambiguity, and spurious tone penalties.

Harmonic-Tone and Frequency Multiplier-based Oscillators

Harmonic oscillators, including push-push, triple-push, and class-F variants, are pivotal for sub-THz/THz signal generation:

• Harmonic Extraction: Strong PN and output power enhancement are demonstrated through symmetry-enforced harmonic extraction (e.g., triple-push for third harmonic, push-push for second harmonic), harmonic-rich tank design, and auxiliary gm feedback.
• Waveform Shaping/Noise Suppression: Gate-drain phase shifts, common-mode resonance expansion, and flicker noise filtering/acoustic LC resonance circuits are increasingly adopted to limit noise upconversion, maintaining PN profiles within stringent FutureG targets.
• Passive Filtering and Power Combining: Self-mixing topology, cross-coupled harmonic filtering networks, and capacitive/inductive neutralization mechanisms enable robust harmonic purity and improved power efficiency.

---

Oscillator Architectures Above 100 GHz

Fundamental-Tone and Harmonic Oscillators

Direct oscillation at THz frequencies contends with device $f_{max}$ limitations; hence, harmonic extraction and injection-locked multiplication predominate:

• Fundamental Oscillators: Power-combining and phase-shifter networks use large transistors and minimized parasitics for DC-to-RF efficiency gains. Mode-switching and capacitive splitting feedback boost resonance frequency and TR.
• Harmonic Oscillators: Triple-push, class-F, and ring oscillator topologies are extensively employed for third-harmonic extraction, often validated with output frequencies exceeding 600 GHz and FoM up to 196 dBc/Hz. Harmonic extraction is also combined with frequency multipliers and coherent oscillator arrays to realize high radiation efficiency and scalable output power.

Non-Silicon and Non-Trivial Approaches

Non-silicon technologies (e.g., GaN HEMT, InP HBT) expand the reach of THz oscillators, employing coplanar waveguide-based resonators, PIN diode comb radiators, and frequency quadruplers for robust output power and efficiency under high-frequency operation. Self-oscillating harmonic mixers and radiative oscillator arrays address frequency pulling, matching, and phase-locking challenges with EM-aware layouts and co-designed buffer/amplifier interfaces.

---

Strong Numerical Results and Contradictory Claims

• Recent oscillators routinely meet or exceed FoM targets of 195–200 dBc/Hz, with PN as low as -138 dBc/Hz at 1 MHz offset and TR approaching or surpassing 100% in carefully engineered multi-core and mesh configurations.
• Series-resonance designs achieve FoM up to -190 dBc/Hz rivaling III-V devices, despite substantial supply currents, suggesting CMOS/BiCMOS can match or outpace legacy compound semiconductors in spectral purity.
• Harmonic extraction arrays and triple-push oscillators demonstrate output frequencies well above device $f_{max}$ (e.g., >600 GHz) with coherent combining and efficient radiation, challenging long-standing assumptions about CMOS frequency scaling.

---

Practical and Theoretical Implications

This review signals a paradigm shift in oscillator co-design, where phase noise, spectral agility, and power consumption are balanced through advanced multi-core coupling, harmonic engineering, and EM-aware layout. For FutureG networks and ISAC applications, oscillator spectral purity directly governs throughput, Doppler resolution, detection sensitivity, and reliability under dense spectral conditions. Ongoing device modeling, passive network innovation, and integration with radiators/antennas will define the scalability and practical deployment of mm-wave/sub-THz systems.

Open theoretical questions concern the joint optimization of FoM, TR, multi-core scaling (spurs and mode control), temperature and aging modeling near $f_{max}$, and the co-design of oscillator/multiplier/radiator/package for robust EMI suppression and efficient radiation.

---

Future Directions

The trajectory of oscillator research points toward:

• Systematic co-optimization of high-Q resonator/passive networks with digital control, exploiting emerging FinFET, FDSOI, and compound material nodes.
• Robust many-core and harmonic array scaling for integrated THz sources and radiators in imaging, sensing, and high-capacity wireless links.
• Further reduction of flicker noise corner through advanced waveform shaping and harmonic resonance expansion, targeting ultra-low jitter for Tbps wireline/wireless transceivers.
• Improved device and passive/noise modeling frameworks to minimize over-design and ensure robust behavior across thermal/aging process corners.

---

Conclusion

Significant advances in oscillator design across mm-wave and sub-THz/THz bands have yielded remarkable improvements in phase noise, tuning range, and power efficiency via multi-core, harmonic extraction, waveform shaping, and innovative passive circuit engineering. These innovations facilitate the deployment of FutureG communication, sensing, and computing systems, while challenging established limits and informing the co-design requirements for subsequent generations of high-frequency integrated technologies. Open challenges in spectral purity, efficiency, scalability, and system-level integration remain focal points for future research in high-performance oscillator architectures (2604.26903).