If you’ve ever wondered why your home Wi-Fi feels slow compared with fiber-optic internet, you’re bumping into a very real bottleneck in modern networks: wireless simply can’t keep up with glass.
Optical fibers already push hundreds of terabits per second, while even advanced microwave and millimeter-wave wireless links struggle to climb into the hundreds of gigabits. As we talk about 6G and “networks beyond 6G,” this gap becomes a serious problem.
A new study in Nature Communications shows a radically different way forward: an all-plasmonic sub-terahertz wireless communication link that uses tiny plasmonic devices, not conventional electronics, to send data at 120 Gbit/s on a 285 GHz carrier across a 5-meter free-space link.
Let’s unpack what that actually means—first in everyday language, then in more technical detail.
Why go to sub-terahertz in the first place?
Today’s 5G systems mostly live below 40 GHz, with some millimeter-wave bands reaching into the 20–70 GHz region. Those bands are crowded and limited.
However, once you climb higher, into the sub-terahertz (sub-THz) band from 0.1 to 0.3 THz, two big things happen:
- Huge new bandwidth opens up. There are “transparency windows” around 100–170 GHz and 200–330 GHz where the atmosphere is relatively kind, and regulators can allocate extremely wide channels.
- Each hertz of bandwidth can carry more data. That’s especially true when you combine wide bandwidth with advanced modulation schemes.
In short, sub-THz behaves like a high-speed express lane for wireless data—perfect for backhaul links, data center interconnects, or fiber-wireless bridges where you want tens or hundreds of gigabits per second over modest distances.
But there’s a catch: generating and detecting those ultra-high frequencies is hard.
The old way: complex III-V electronics and bottlenecks
Traditionally, engineers turn to III-V semiconductor technologies (like InP, GaAs) or specialized RF hardware:
- Frequency multiplier chains and resonant tunneling diodes to generate sub-THz tones
- High-electron-mobility transistors (HEMTs) and Schottky diodes as receivers
These solutions work, but they:
- Often have narrow bandwidths (tens of GHz)
- Show non-linearities and non-flat frequency responses that distort complex modulation formats
- Require extra opto-electronic (OE) and electro-optic (EO) converters when you want to interface with fiber-optic systems
As a result, the optical–wireless–optical chain gets clunky, power-hungry, and expensive.
The new way: go all-plasmonic
The team behind this paper took a very different route. Instead of bolting RF electronics onto optics, they built an all-plasmonic link:
- On the transmitter side, a plasmonic graphene photodetector (PD) performs opto-electronic conversion—it beats two lasers together and directly generates a sub-THz electrical signal.
- On the receiver side, a plasmonic Mach-Zehnder modulator (MZM) performs electro-optic conversion—it takes that sub-THz electrical signal and maps it back onto an optical carrier.
Both devices are:
- Tiny – footprints well under 50 µm²
- Fast – flat frequency response tested up to ~330 GHz and designed for >500 GHz operation
- Highly integrable – compatible with silicon photonics and scalable fabrication
In other words, plasmons—collective oscillations of electrons at metal–dielectric interfaces—become the workhorse, translating between light and high-frequency electrical waves directly, with minimal electronic overhead.
Plain-language picture: how the link works
Think of the system in four steps:
- Data in over fiber
Data starts its life as a standard optical signal around 1550 nm, modulated in a telecom-friendly format and traveling through fiber. - Light turned into a super-high-frequency “radio” wave
At the transmitting remote antenna unit (RAU), the optical data signal is combined with a local-oscillator (LO) laser. When these two optical tones hit the plasmonic graphene photodetector, their beat note appears as a 285 GHz electrical signal that carries the data. - Sub-THz beam across the air
That 285 GHz signal is amplified, fed into a horn antenna and lens, and beams across 5 meters of free space—essentially acting as an ultra-fast wireless “patch cable.” - Super-high-frequency wave turned back into light
At the receiving RAU, the incoming 285 GHz signal directly drives the plasmonic MZM. This device impresses the high-frequency information onto a new optical carrier. The resulting optical sidebands are then received by a standard coherent optical receiver and cleaned up with digital signal processing (DSP).
The whole journey is:
Fiber → plasmonic PD → sub-THz wireless → plasmonic MZM → fiber again
No bulky sub-THz mixers or traditional RF front-end needed at the endpoints.
Under the hood: plasmonic graphene photodetector (OE side)
For readers with more of a science or engineering background, here’s what’s going on in the photodetector.
The plasmonic PD uses a metamaterial perfect absorber design based on a metal–insulator–metal (MIM) stack:
- A gold back mirror
- An aluminum-oxide spacer
- A bilayer graphene sheet near the top
- A patterned top metal layer forming plasmonic dipole resonators
When light hits this structure:
- The resonators excite plasmonic modes and trap light in the stack.
- Graphene absorbs a portion of that intense local field, creating electron–hole pairs.
- Contact doping at the metal–graphene interface creates a built-in electric field.
- That field drives carriers to the contacts, generating a photovoltaic current, even at zero source–drain bias.
Key performance highlights:
- Flat frequency response from 1 GHz to at least 330 GHz (limited by the measurement setup)
- Operation in photovoltaic mode at zero bias, which helps with noise and power consumption
- High photocurrent and responsivity for a high-speed graphene detector, with RF output around −37 dBm at 280 GHz in tests and −42 dBm used in the link demo.
Because the RF response remains flat with increasing optical power, the PD can generate different carrier frequencies on demand (e.g., 70, 140, 280 GHz) simply by adjusting the optical beat frequency. That flexibility is huge for adaptive wireless systems.
Under the hood: plasmonic Mach-Zehnder modulator (EO side)
On the receive side, the plasmonic Mach-Zehnder modulator serves as a high-speed detector that maps the 285 GHz signal back into the optical domain.
Its structure:
- Two short plasmonic phase shifters form a Mach-Zehnder interferometer.
- Each phase shifter uses a narrow metallic slot (~120 nm wide, 9 µm long) filled with an organic Pockels material.
- The sub-THz RF signal is applied to a central electrode, with surrounding electrodes grounded—this yields a push-pull configuration.
Because the slot is so tiny and the overlap between RF and optical fields is strong:
- The device behaves like a lumped capacitor with femtofarad-scale capacitance.
- It achieves an electro-optic bandwidth measured to be flat up to ~350 GHz and engineered for ~800 GHz class operation.
- Energy consumption sits around 1 fJ/bit/V² for 4-QAM formats—very low for this speed regime.
The modulator operates at its null point, so a small RF voltage swings the optical output between destructive and constructive interference, creating optical sidebands spaced by the sub-THz carrier.
The full experiment: 120 Gbit/s at 285 GHz over 5 m
In their lab demo, the authors:
- Modulate a 1550 nm carrier using a 38 GHz IQ modulator.
- Use digital Nyquist frequency-division multiplexing (NFDM) to split the data into six tributaries, each 8 GBd, for a total of 48 GBd.
- Map different QAM orders onto each tributary and perform bit-loading to match the link’s frequency response.
- Down-convert the optical data to a 285 GHz RF signal via the plasmonic PD.
- Transmit across 5 m of free space, with horn antennas and lenses compensating the ~96 dB path loss.
- Up-convert back to light with the plasmonic MZM, then process the received optical sidebands with a dual-polarization coherent receiver and linear DSP (TR, matched filtering, CMA equalization, carrier recovery, dual-sideband combining, phase recovery).
They achieve:
- Line rate: 120 Gbit/s
- Net information rate: ≈111.7 Gbit/s (from GMI and FEC assumptions)
- Error performance: BER below hard-decision FEC thresholds for all tributaries
Notably, RF amplifiers in the current setup limit the overall bandwidth to about 30 GHz. With stronger plasmonic PD output in future designs, they could reduce or remove those amplifiers and unlock much wider link bandwidths.
Why this is cool (for both geeks and non-geeks)
For non-specialists, here’s the big picture:
- This demo shows that tiny plasmonic chips can handle sub-THz wireless at fiber-like speeds.
- The devices are small, energy efficient, and manufacturable in ways that play nicely with existing silicon photonics.
- Because they work directly with light, they can seamlessly glue optical fiber networks to super-fast wireless hops.
For science and engineering folks, the excitement comes from:
- A graphene plasmonic photodetector operating at >300 GHz with high photocurrent in zero-bias PV mode.
- A plasmonic MZM with >300 GHz measured EO bandwidth, femtofarad capacitance, and realistic ~1 fJ/bit-class energy.
- An end-to-end system demo at 285 GHz, 120 Gbit/s, using mostly linear DSP because the PD and MZM behave nicely.
What could this enable next?
If this plasmonic approach scales, you can imagine:
- Fiber-wireless bridges across rivers, roads, or factory floors where pulling fiber is too expensive or impossible.
- Building-to-building or rack-to-rack links in data centers that need 100+ Gbit/s line-of-sight connections.
- Dense 6G/7G backhaul where many small cells offload data over short, high-throughput sub-THz links.
- Eventually, chip-to-chip or board-to-board optical-wireless-optical interconnects in tightly packed systems.
There’s still work to do—boosting PD output power, integrating or removing RF amplification, and ruggedizing these components—yet the direction is clear. Plasmonics is no longer just about exotic lab demos; it now underpins a credible architecture for future high-speed wireless-optical networks.
Check out the cool NewsWade YouTube video about this article!
Article derived from: Blatter, T., Koepfli, S. M., Zuerrer, A., Hess, S., Horst, Y., Destraz, M., Rieben, D., Baumann, M., Kulmer, L., Smajic, J., Fedoryshyn, Y., & Leuthold, J. (2025). All-plasmonic sub-terahertz wireless communication link. Nature Communications, 16, 9988. https://doi.org/10.1038/s41467-025-64926-6
