An 8-Inch PZT Photonics Platform Reaches 3.2 Tbps Across 16 Channels
Researchers integrated a 16-channel optical transmitter on an 8-inch thin-film PZT platform, with modulators reaching 70 GHz and up to 224 Gbps per test. The 3.2 Tbps-class figure is aggregate laboratory capacity, not a deployed link or commercial-yield result.
Quick summary
- A peer-reviewed study reports an 8-inch thin-film PZT photonics platform made with wafer-scale processes.
- Its Mach-Zehnder modulator reached a 70 GHz electro-optic response and data rates up to 224 Gbps.
- A monolithically integrated 16-channel transmitter supported a 3.2 Tbps-class aggregate optical link.
- The result is a laboratory platform, not evidence of commercial yield, field reliability or lower total system cost.
Researchers at the Chinese Academy of Sciences and partner institutions have demonstrated a thin-film lead zirconate titanate, or PZT, photonics platform on 8-inch silicon dioxide on silicon wafers. The peer-reviewed paper, published in Nature Communications on November 20, reports a 16-channel wavelength-division multiplexing transmitter designed to combine many optical carriers on one chip.
The most important result is not a single record number. It is the combination of high-speed modulation, wavelength routing and an 8-inch fabrication flow on one material platform.
What the team measured
The researchers formed uniform PZT films using chemical solution deposition followed by magnetron sputtering, then patterned devices with deep-ultraviolet stepper lithography. Their Mach-Zehnder modulators showed a half-wave voltage-length product of 1.19 V·cm and an electro-optic response extending to 70 GHz. Individual data experiments reached up to 224 Gbps.
The integrated transmitter used 16 wavelength channels and supported what the authors describe as 3.2 Tbps-class optical interconnects. That figure is an aggregate capacity across channels. It should not be read as one electrical lane running at 3.2 Tbps, nor as a deployed network result.
Why PZT matters
PZT is ferroelectric: an applied electric field can organize its internal domains and activate a strong Pockels effect, allowing voltage to change how light propagates through a waveguide. The material can therefore supply fast modulation while also carrying passive routing structures such as arrayed waveguide gratings.
That integration is relevant to data centers and high-performance computing, where traffic between processors increasingly strains electrical interconnects. Moving more modulation and wavelength-routing functions onto the same photonic platform could reduce packaging complexity. For a related example of how photonic chips manipulate mixed optical signals, see NewTqnia's report on a nine-input silicon photonic mode separator.
Reality check
The paper establishes a research platform, not a production line. It does not demonstrate commercial wafer yield, long-duration reliability, packaging cost or operation inside a deployed data-center link. The article is also an early version of record that the journal says may receive editorial changes. The 3.2 Tbps-class number combines 16 optical channels, so comparisons with single-lane or end-to-end system benchmarks require care.
What to watch next
The practical test will be whether the process can maintain low optical loss and consistent modulator performance across large production batches, then survive packaging, temperature variation and long operating periods. Independent demonstrations will also need to compare energy per bit and total link cost with silicon photonics, thin-film lithium niobate and lithium tantalate platforms.
Verified topics and entities
Sources and citations3 sources
External references used to support the reporting in this article.
Published by
NewTqnia Science Desk
An institutional editorial team within NewTqnia