Asianometry Podcast: How Moore’s Law Revolutionized RF-CMOS

Host: Jon Y
Date: April 3, 2025

Overview

This episode delves into how continuous advances in Moore's Law completely transformed the landscape of radio frequency (RF) integrated circuits—which went from a world of disparate, costly, and bulky radio modules to today's sleek, compact, and affordable RF CMOS (Complementary Metal Oxide Semiconductor) chips. Host Jon Y explains the science behind RF front end modules, traces the industry’s shift from specialized compounds to ubiquitous silicon, and discusses the technical and economic forces that made RFCMOS the backbone of wireless connectivity. The episode concludes with a thoughtful look at the future—especially as 6G and subterahertz technologies challenge CMOS’ dominance.

Key Discussion Points & Insights

1. Fundamentals of RF Chips and Signal Transmission

• RF Frequency Definition: RF chips operate in frequencies from ~100 MHz to 100 GHz.
• RF Front End Module Structure (00:52):
  • Starts with a low-frequency baseband signal (e.g., human voice: 80–200 Hz).
  • Low-frequency signals are upconverted to higher frequencies using a mixer and a local oscillator.
  • The upconverted (passband) signal is amplified and sent via the antenna.
  • Receiving employs the same process in reverse, using a low noise amplifier and mixer(s) to bring the signal back to baseband.
  • Superheterodyne transceiver concept: Combines transmit/receive in a single device.

"Trying to send such a signal as is, however, would need a large antenna. Since a low frequency means a long wavelength... an antenna for a 200 Hz signal would be hilariously impractically big."
—Jon Y, [01:02]

2. Materials and Semiconductor Technologies in RF

• Early Devices: RF circuits once heavily relied on BJTs (Bipolar Junction Transistors) and compound semiconductors (III-V materials, e.g., Gallium Arsenide).
• Why Not Silicon, Initially?
  • Silicon MOSFETs couldn't switch fast enough for RF (GHz-range) applications.
  • Superior electron mobility in Gallium Arsenide (8,500 cm²/V·s vs. silicon’s 1,400).
• Technical Milestones:
  • 1958: Germanium BJT amplifies GHz signals.
  • 1965: Carver Mead fabricates first GaAs MESFET.
  • 1971: MESFETs reach 10 GHz cutoff frequency, allowing 1 GHz operation.

“The MOSFET’s key advantage over the MESFET is that charge carriers can very quickly speed through the channel, in part because it is made from gallium arsenide.”
—Jon Y, [11:57]

3. CMOS and Its Advantages (16:08)

• Origins: Invented in 1963 by Frank Wanlass & Tom Sa, CMOS has both PMOS (positive) and NMOS (negative) transistors side by side.
• Early Adoption: Japanese calculator companies leveraged CMOS for power savings in the 1970s. US industry followed as power efficiency became critical.
• Dominance: Intel used CMOS from the 386 CPU onward; it became the workhorse of digital logic by the late 1980s.

"They noticed that their CMOS arrangement drew close to zero power when in standby, just whatever luggage power there might be."
—Jon Y, [17:49]

4. The Obstacles to RF-CMOS Adoption

Perceptions of Silicon’s Limitations

• Belief in Scaling Limits:
  • 1980s experts believed MOSFET gate length could not go below 100nm, limiting cutoff frequency (~200 MHz), thus making CMOS unfit for RF.
• Anecdote:
  • “One unnamed MIT professor said RF is a solved problem, and using an inferior technology like CMOS to solve it yet again is stupid squared.”
    —Jon Y, [25:47]
• Reality Check:
  • Shrinking via Moore’s Law increased cutoff frequencies.
  • By early 1990s: 100 GHz chips speculated by IBM (mainly for CPUs).

Technical Barriers in Passive Components

• On-Chip Inductors (Spiral Inductors):
  • Considered impossible to fabricate at scale—quality problems due to substrate interaction.
  • Breakthrough: Suspended inductor innovation in early 1990s (selective acid etch), enabling practical RF CMOS ICs.
• Role of Professor Asad Abidi:
  • Co-author of the first RF CMOS integrated amplifier paper (1993); key figure in RFCMOS commercialization.

5. The Boom of RF-CMOS Integration

Economic and Technical Breakthrough (36:12)

• Silicon Labs’ RF CMOS GSM/GPRS Transceiver (2001):
  • Integrated all front-end components onto one chip (80% fewer parts, 50% less board space).
  • Led to massive cost/space savings; stock tripled after adoption by Samsung & others.

"The CMOS version not only cost half as much as its predecessors, it took just a fifth of the board space."
—Jon Y, [43:16]

• Industry Follows:
  • Companies like Infineon, Philips, Lucent rapidly introduced their own single-chip solutions.
  • By 2002, entire front-ends were implemented in silicon CMOS.

Resulting Impact

• RFCMOS ended most RF research on BJTs, GaAs MESFETs.
• Today: Used in RFID, WLAN, GPS, Wi-Fi, Bluetooth, automotive radar—where size, cost, integration trump per-component performance.

6. Limits of Scaling and the Future (49:33)

• Scaling Walls:
  • Sub-100nm: Issues with power leakage, thin oxides, short channels.
  • FinFETs help but aren’t a panacea.
• RF-CMOS in 5G:
  • Used for mmWave (24–48 GHz) radios; also forms basis of 77 GHz radar.
• Looking to 6G & Beyond:
  • 6G may require sub-terahertz (100–300 GHz) operation.
  • Current CMOS cutoff record: 485 GHz (since 2007)—likely insufficient for 6G.
  • Significant challenge: High-frequency operation requires not just cutoff frequency but dramatically higher power output with limited cooling in handsets.

“If that is indeed the case, then perhaps the future of RFCMOS will involve going back to the past, meaning the use of heterogeneous integration to connect RFCMOS transceivers with power amplifiers made with III-V semiconductors. In other words, RF chiplets.”
—Jon Y, [52:44]

Reflection

• Industry may return to hybrid RF modules (RF chiplets)—combining CMOS with III-V materials for new frequency bands.
• 6G’s future is yet unclear—but if it happens, it might close the Moore’s Law era of RFCMOS scaling.

Notable Quotes & Memorable Moments

• On Technological Skepticism:
  "One unnamed MIT professor said RF is a solved problem, and using an inferior technology like CMOS to solve it yet again is stupid squared."
  —Jon Y, [25:47]
• On the Impact of RFCMOS:
  "The rise of RFCMOS was so significant that it ended a great deal of research being done on items like silicon BJTs and gallium arsenide MESFETs."
  —Jon Y, [44:16]
• On the Future:
  "If that is indeed the case, then perhaps the future of RFCMOS will involve going back to the past, meaning the use of heterogeneous integration to connect RFCMOS transceivers with power amplifiers made with III-V semiconductors."
  —Jon Y, [52:44]

Timestamps for Key Segments

| Segment | Timestamp | |-----------------------------------------------|------------| | What is RF, the signal chain explained | 00:02–05:00| | Early RF semiconductors: BJT, MESFET, GaAs | 08:42–15:20| | Origins and rise of CMOS | 16:08–21:10| | Silicon’s limitations & breakthrough moments | 21:13–31:27| | Fabricating passive devices on CMOS | 33:00–37:45| | Asad Abidi and the suspended inductor | 38:34–41:00| | Silicon Labs’ integrated RF CMOS chip | 43:05–45:44| | The RFCMOS boom and industry-wide adoption | 45:48–48:02| | 5G, future challenges, and heterogeneous ICs | 49:33–54:11|

Episode Takeaway

Jon Y makes clear that Moore’s Law didn’t just make computers faster and cheaper—it revolutionized the very circuits connecting people wirelessly, by bringing the benefits of scaling and integration to RF chips. As we hit physical and performance barriers at ever-higher frequencies, the next leap may require mixing CMOS with old rivals—heralding a new age of hybrid RF technologies.