AI summary by TwinMind

LK-99 Wouldn't Have Changed Semiconductors Anyway

Asianometry

14m 27s

LK-99 Superconductor Status Update

  • Chinese scientists tests indicate LK-99 likely not room temperature superconductor
  • Hype died down, vanished from social feeds and science news
  • Even if real room temperature superconductor found tomorrow, wouldn't revolutionize semiconductor industry

Interconnects in Integrated Circuits

  • Two types: local and global interconnects
  • Local interconnects: lowest chip levels, connect local circuit blocks, short wires, less concerned about resistance than heat
    • Room temperature superconductor wouldn't make sense here due to heat sensitivity
  • Global interconnects: higher stratum levels, connect distant chip parts, deliver clock signals/power
    • Made from low resistance metals (aluminum/copper)
    • Thick, solid construction due to distance requirements

Power Consumption Analysis

  • Advanced AI accelerators (Nvidia H100s for ChatGPT): 80-90% power budget spent on data transfer vs actual computation
  • MIT PhD Alex Sleds consultation on superconducting interconnects:
    • Short wire lengths would significantly decrease energy consumption for transmitting data
    • Energy won't become free due to capacitance and inductance
    • Wire creates electric field (capacitance) and magnetic field (inductance) causing energy storage/loss
  • Power consumption gains especially for global interconnects transmitting power
  • Gains might not be as dramatic as initially expected

Computing Speed Limitations

  • Substantial portion of computer cycle time spent waiting for data arrival
  • Fundamental limits on signal speed:
    • Absolute ceiling: speed of light
    • Signals can't travel at full light speed (wire not vacuum)
  • Resistance-capacitance (RC) delay considerations:
    • Interconnects store charge, delay signal transmission
    • Capacitance depends on dielectric material surrounding wire (K value)
    • Even with zero line resistance, capacitance delay remains
  • Speed gains may not be as significant as initially thought

Current Density Limitations

  • Superconductivity requires: low temperature + low magnetic field + low current density
  • Current density = electric current ÷ superconductor cross-sectional area
  • High-temperature cuprate superconductors investigation for interconnects:
    • Critical current density limits not very high
    • Interconnects would need to be very wide (perhaps centimeter wide)
    • Unfeasible for most digital circuits
  • This prevented cuprates from entering integrated circuit arena
  • LK-99 may suffer same issues, making it scientific curiosity while copper retains dominance

Manufacturability Concerns

  • Dylan Patel (Semi-Analysis chief analyst) insights:
    • LK-99 shows promise as low resistivity material or potential superconductor
    • Manufacturability still questionable
    • Simple process for low purity/low volume, but scaling at high volume/high purity with <1 part per billion defects = tremendous challenge
  • TSMC statistics: single IC uses >100 kilometers copper wiring, billions of chips produced monthly
  • Room temperature superconductor = complex alloy, tend not to scale well
  • Integration into existing process flow challenges
  • IBM took 15 years to solve copper interconnect materials engineering issues
  • Copper known for 10,000 years vs any room temperature superconductor known much shorter

Josephson Effect and Junctions

  • 1962: Brian Josephson (22-year-old Cambridge grad student) predicted Josephson effect
  • Effect: two superconductors with thin insulating barrier (few atomic layers)
  • Produces supercurrent: direct current flows through insulator without resistance, little energy loss
  • Contrast with traditional transistor gates generating heat
  • Quantum tunneling observable effect
  • Josephson won Nobel Prize (superconductivity research won 1/10 of all physics Nobel prizes since 1911)

Josephson Junction Implementation

  • 1963: PW Anderson and Jay Rowell (Bell Labs) created experimental circuit
  • Popular implementation: niobium superconductor + aluminum oxide insulator sandwich
  • Switch on/off in few picoseconds (trillionth of second)
  • Data encoded in rapid pulses enabling 100+ gigahertz processing speeds

Josephson Junction Limitations

  • IBM worked on Josephson junction computers for two decades, closed efforts 1983
  • Major issues:
    • Struggled producing proper superconducting memory circuit
    • Memory basic/essential to computers
    • Largest Josephson memory unit: 4 kilobit (NEC, 1990)
    • Extremely difficult producing junctions in large densities
  • Most successful systems: analog systems (few components, easier competition)
  • Superconducting nanowire single photon detectors for sensitive photon detection
  • Digital systems: millions/billions consistent transistors, far harder to produce competitive superconducting digital system

Hybrid Solutions

  • Semiconductor-superconductor hybrid modules (like AMD chiplets)
  • Combine semiconductor-based memory with superconductor-based processing systems

Temperature Operating Requirements

  • BCS theory rule of thumb: keep operating temperature at half transition temperature
  • Existing niobium Josephson circuits: kept at 4-5 Kelvin despite 9.3K transition temperature
  • LK-99 claimed transition temperature: ~400 Kelvin
  • Practical working temperature: ~200 Kelvin (-99.67°F, -73°C)
  • For true room temperature superconducting chip without freezing: need ~600 Kelvin transition temperature (620°F)
  • Refrigeration remains single biggest limiting factor, adding big energy usage cost

Final Assessment

  • Even if LK-99 real: historic scientific landmark, might benefit power transmission/energy industries
  • Superconductors not serious candidates for disrupting existing semiconductor technologies
  • Prediction: no superconductor-based computer will become commercially competitive in future
  • Hype was enjoyable for about a week

TwinMind

Get summaries like this for your own meetings

TwinMind takes notes in the background during meetings, lectures, and calls, then writes the summary for you.

Free plan to get started

No bot joins your calls

Phone, desktop & Chrome apps

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome

Your next meeting, summarized like this

TwinMind listens in the background and turns meetings, lectures, and calls into searchable notes you can chat with. Start free.

Available on phone, desktop & Chrome