Silicon's Successor

How InGaAs Chips Are Forging the Future of Computing

The relentless march of Moore's Law is meeting silicon's physical limits—and a revolutionary compound semiconductor is stepping into the breach.

Introduction: The Silicon Scaling Crisis

For over five decades, silicon transistors have powered the digital revolution. But as chip features shrink below 7 nanometers, silicon's diminishing returns threaten to stall computing progress. Enter indium gallium arsenide (InGaAs), a III-V compound semiconductor whose exceptional electron mobility—up to 50× faster than silicon—positions it as the leading candidate to extend Moore's Law into the next decade. Recent breakthroughs presented at the 2025 VLSI Symposium reveal how scientists are overcoming integration hurdles to unlock InGaAs' potential for next-generation CMOS technology 2 .

Electron Mobility

InGaAs offers electron mobility up to 50 times faster than silicon, enabling significantly higher performance at lower power.

Moore's Law Extension

As silicon approaches physical limits, InGaAs emerges as the most promising successor to continue transistor scaling.

Key Concepts: Why InGaAs?

1 The Speed Advantage

InGaAs' crystal structure enables electrons to zip through channels with minimal resistance. This translates to:

  • Lower power consumption at high frequencies
  • Higher drive currents for computational density
  • Enhanced high-frequency performance (critical for 5G/6G and AI accelerators)
Did You Know?

InGaAs' electron velocity remains high even at nanoscale dimensions due to ballistic transport effects 2 .

2 The Scaling Paradox

While planar InGaAs MOSFETs have demonstrated record transconductance (3.45 mS/µm), their gate lengths struggle to drop below 40 nm. To transcend this, researchers pivoted to 3D architectures like finFETs—the same structures used in cutting-edge silicon nodes 2 .

Semiconductor wafer
Modern semiconductor wafer with nanoscale features.
FinFET Revolution

The transition to 3D finFET architectures has enabled InGaAs to overcome planar scaling limitations while maintaining performance advantages 2 .

3 Silicon Integration Imperative

The ultimate goal is monolithic integration of InGaAs channels on silicon wafers. Aeluma's recent demonstration of 200 mm CMOS-compatible InGaAs photodetectors proves large-scale manufacturing feasibility, bridging III-V performance with silicon's economies of scale 5 7 .

Industry Insight: Combining InGaAs performance with silicon manufacturing infrastructure is key to commercial viability.

In Focus: MIT's Record-Breaking InGaAs FinFET Experiment

How researchers at MIT tamed fin defects and contact resistance to achieve unprecedented performance.

Methodology: The Dual-Step Fin Sculpting

1. Reactive Ion Etching
  • BCl₃/SiCl₄/Ar plasma mixture
  • Initial fin widths ~15 nm
  • Sidewall roughness issues
2. Digital Etch Refinement
  • 3+ cycles of atomic-scale sculpting
  • Final fin width: 8 nm
  • Aspect ratio >21
3. Contact Engineering
  • Molybdenum contacts
  • Contact resistivity: ~10⁻⁹ Ω·cm²
  • Minimized signal loss

Results & Analysis: Closing the Gap with Silicon

Electron Mobility vs. Fin Width
Fin Width (nm) Electron Mobility (cm²/V·s)
30 3,500
20 2,900
8 1,200

Mobility declined below 15 nm but remained superior to silicon. Critically, electron velocity (the true performance metric) stayed high due to ballistic transport 2 .

Transconductance Benchmarking
Normalization Method InGaAs finFET (MIT) Si finFET (Intel 14nm)
By gate periphery ~1.8 mS/µm ~2.0 mS/µm
By fin width 0.65 mS/µm 1.2 mS/µm

While trailing silicon in footprint-normalized metrics, MIT's devices set a 50% improvement over prior InGaAs finFETs—a pivotal leap toward competitiveness 2 .

Sidewall Interface Quality

The digital etch process slashed interface defects by 16×, enabling effective gate control 2 .

Achieved interface state density of 3×10¹² eV⁻¹cm⁻²

The Scientist's Toolkit: Key Research Reagents

Material/Reagent Function
In₀.₅₃Ga₀.₄₇As substrates Matches InP lattice constant; optimal electron mobility
Molybdenum (Mo) Low-resistivity ohmic contacts for source/drain regions
BCl₃/SiCl₄ plasma Anisotropic etching of III-V fins with vertical sidewalls
Cyclic digital etchants Atomic-scale fin width control and surface passivation
High-κ dielectrics Al₂O₃/HfO₂ stacks for enhanced gate control with minimal leakage
200-300mm Si wafers CMOS-compatible substrates for heterogeneous integration (Aeluma's approach) 5 7

Beyond the Lab: Industry's Roadmap

Aeluma's upcoming presentations at CS Mantech 2025 and IEEE OIP 2025 will spotlight:

Quantum Dot Lasers

Monolithic quantum dot lasers for silicon photonics

SWIR Photodetectors

On 300mm silicon wafers for mass production

High-speed APDs

For optical interconnects in AI clusters 7

Simultaneously, Semiconductor Today reports massive investments in GaN-on-SiC and InP HBTs, confirming the industry's pivot toward compound semiconductors .

Conclusion: The Path to Commercialization

MIT's finFET breakthroughs and Aeluma's wafer-scale integration mark a watershed in the InGaAs revolution. As Dr. Jesús del Alamo (MIT) notes: "We're not just extending Moore's Law—we're enabling new synergies between logic, sensing, and quantum computing." Challenges remain in defect density and cost, but with prototypes already matching silicon's electrostatic control at 20 nm gates, InGaAs channels could enter high-volume production by 2030—powering everything from neuromorphic AI chips to ultra-sensitive LiDAR 2 7 .

Further Reading

Covered in Semiconductor Today's July 2025 issue: www.semiconductor-today.com

Timeline to Commercialization
2025

Lab-scale prototypes demonstrate viability

2027

First commercial test chips

2030

High-volume production expected

References