🔬 applied physics

Comparing optical-microwave conversion and all-microwave control schemes for a transmon qubit

This paper demonstrates that an optical control system using modulated laser light delivered via optical fiber to a photodiode at the 1K stage performs comparably to conventional coaxial microwave lines in controlling transmon qubits, showing no measurable degradation in coherence over 20-hour measurement runs and thus supporting its viability for large-scale integration.

Original authors: Volodymyr Monarkha, Massimo Borrelli, Reza Hajitashakkori Kenari, Mohammad Kobba, Eugenio Cataldo, Beer de Zoeten, Mahnaz Zarrinfar, Kamal Pandey, Abhinand Pusuluri, Filippo D. Michelacci, Eliot Jouan

Published 2026-03-20

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Volodymyr Monarkha, Massimo Borrelli, Reza Hajitashakkori Kenari, Mohammad Kobba, Eugenio Cataldo, Beer de Zoeten, Mahnaz Zarrinfar, Kamal Pandey, Abhinand Pusuluri, Filippo D. Michelacci, Eliot Jouan, Bennett Sprague, Simon Groeblacher, Thierry C. van Thiel, Robert Stockill, Russell E. Lake

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to conduct a massive orchestra of tiny, super-sensitive instruments called qubits. These instruments are the building blocks of future quantum computers. But there's a catch: they are incredibly fragile. If you get too close with a hot hand or a noisy instrument, they stop playing their tune (they lose their "coherence").

Currently, to control these qubits, scientists use thick bundles of copper wires (coaxial cables) running from the warm control room down into a giant, ultra-cold freezer (a dilution refrigerator). Think of these wires like thick, heavy fire hoses. While they deliver the necessary "water" (control signals), they also leak a lot of heat, making it hard to keep the freezer cold enough as you try to add more and more qubits.

This paper asks a simple but revolutionary question: What if we stopped using the heavy fire hoses and started using laser beams instead?

Here is the story of their experiment, explained simply:

1. The Two Competitors

The researchers set up a race between two ways of controlling a single qubit:

The Old Way (All-Microwave): Using the traditional copper wires. It's reliable, but like a heavy fire hose, it carries heat down into the cold zone.
The New Way (Optical-Microwave): This is the "laser" approach.
- Step 1: At room temperature (the warm control room), they take a laser beam and "paint" the control signals onto it, like writing a message on a beam of light.
- Step 2: They send this light down a thin, glass fiber optic cable (like a fiber internet cable) into the freezer. Glass fibers are great because they are thin and don't carry much heat.
- Step 3: At the bottom of the freezer (near the qubit), the light hits a special photodiode (a light-sensing chip). This chip acts like a translator, instantly turning the light back into the microwave signal the qubit needs to hear.

2. The Experiment: A 20-Hour Test

The team put a single qubit in the freezer and ran it for 20 hours straight. They switched back and forth between the "copper wire" method and the "laser" method.

They measured two things:

Relaxation ( $T_1$ ): How long the qubit stays excited before falling asleep.
Coherence ( $T_2$ ): How long the qubit remembers its tune before getting confused by noise.

The Result: The qubit performed exactly the same with the laser method as it did with the copper wires. The "laser" method didn't introduce any extra noise or confusion. It was just as stable as the old way.

3. The "Heat" Problem (The Catch)

There was one small concern. When the laser hits the photodiode at the bottom of the freezer, the energy has to go somewhere. It turns into a tiny bit of heat.

Think of it like this:

Copper Wires: Carry heat all the way down the line, like a long, hot pipe.
Laser + Photodiode: The light travels cold, but the "translator" (photodiode) gets slightly warm when it does its job.

The researchers calculated that if you scale this up to a massive computer with thousands of qubits, the heat from all those photodiodes would be noticeable. However, they found a clever workaround: Duty Cycling.
Instead of keeping the laser on 100% of the time, you only turn it on when you are actually sending a command. Since qubits don't need commands every millisecond, you can turn the laser off 90% of the time. This reduces the heat load significantly, making the system very efficient.

4. Why This Matters: The "Future Orchestra"

The biggest takeaway is Scalability.

Imagine trying to build a quantum computer with 1,000 qubits.

With Copper Wires: You would need 1,000 thick, heavy, heat-leaking cables. The freezer would get too hot, and the system would fail. It's like trying to fit 1,000 fire hoses into a small room.
With Lasers: You can bundle thousands of thin fiber optic cables together. They take up almost no space and carry almost no heat. The only heat comes from the tiny translators at the bottom, which can be managed with better cooling or by turning the lasers off when not in use.

The Bottom Line

This paper proves that we can control quantum computers using light instead of just wires, without hurting the performance of the qubits.

It's like discovering that you can conduct a symphony using a high-speed fiber-optic internet connection instead of shouting through a megaphone. The sound (the control signal) is just as clear, but you don't have to worry about the megaphone overheating the room. This opens the door to building the massive, room-sized quantum computers of the future.

1. Problem Statement

As quantum computers scale toward thousands of qubits, the traditional method of controlling qubits via coaxial microwave cables faces significant bottlenecks:

Thermal Load: Each coaxial cable conducts heat from room temperature to the millikelvin stage, overwhelming the cooling power of dilution refrigerators.
Physical Space: High-density coaxial arrays are difficult to route and manage within the limited volume of a cryostat.
Optical Alternative: While optical fibers offer negligible passive heat load and high bandwidth for multiplexing, converting optical signals to microwave signals at cryogenic temperatures introduces new challenges. Specifically, there are concerns regarding:
- Thermal Tradeoffs: Heat generation from the optical-to-microwave conversion process (photodiodes).
- Noise and Decoherence: High-frequency radiation or thermal noise from the optically coupled drive line could degrade qubit coherence (specifically $T_1$ and $T_2$ times) compared to standard microwave lines.

The paper aims to determine if an optical control system (using modulated laser light converted to microwaves by cryogenic photodiodes) can match the performance of conventional all-microwave control without introducing measurable decoherence.

2. Methodology

The authors conducted a comparative study using a single transmon qubit inside a Bluefors LD400 dilution refrigerator (base temperature < 10 mK). The experiment alternated between two control configurations:

Configuration A (All-Microwave): Standard control pulses were delivered via a coaxial line with attenuators and filters.
Configuration B (Optical-Microwave):
- Generation: Microwave pulses were modulated onto a 1510 nm laser beam at room temperature using an electro-optic intensity modulator.
- Transmission: The light traveled via single-mode optical fiber to the cryostat.
- Conversion: At the 1K stage, a high-speed photodiode array demodulated the light to reproduce the microwave pulses.
- Delivery: The resulting microwave signals drove the qubit.
Experimental Protocol:
- Repeated measurements of energy relaxation time ( $T_1$ ) and Ramsey coherence time ( $T_2^*$ ) were performed over 20-hour continuous runs for both configurations.
- The setup included a "hanger" readout resonator architecture.
- A typical readout chain was used without parametric amplifiers to avoid adding back-action noise.
- Statistical analysis (histograms, mean values, standard deviations) was performed to compare stability and coherence.

3. Key Contributions

Direct Performance Comparison: This is one of the first studies to directly compare qubit coherence times between a fully optical control path and a traditional coaxial path within the same cooldown cycle, eliminating variables related to qubit fabrication or environmental drift between runs.
Noise Temperature Estimation: The authors developed a method to estimate the effective noise temperature of the photodiode assembly. By comparing current data with previous cooldowns (without the photodiode), they isolated the noise contribution of the optical drive line.
Scalability Thermal Modeling: The paper provides a comprehensive thermal simulation for a hypothetical large-scale system (Bluefors XLD1000s) with ~1,000 lines, comparing three wiring scenarios:
1. All-coaxial (Normal).
2. Optical to Still + Coaxial to MXC (Normal).
3. Optical to Still + Superconducting Coaxial to MXC.

4. Key Results

A. Qubit Coherence and Stability

No Measurable Decoherence: The study found no statistically significant difference in qubit coherence between the two methods.
- Optical Control: $T_1 \approx$ (implied similar to microwave), Pure dephasing time $T_\phi = 65 \, \mu\text{s}$ .
- Microwave Control: Pure dephasing time $T_\phi = 64 \, \mu\text{s}$ .
Source Stability: The optical control system demonstrated superior stability over 20+ hours. Analysis of Ramsey fringe asymmetry showed that laser power drift was negligible compared to conventional microwave sources, ensuring consistent pulse rotation accuracy.
TLS Interaction: Periods of suppressed coherence were attributed to interactions with a slowly evolving Two-Level System (TLS) environment, affecting both configurations equally.

B. Noise and Thermal Analysis

Effective Noise Temperature: The presence of the photodiode assembly resulted in an estimated noise temperature at the qubit feedline of approximately 24 K (originating from the photodiode stage). This was higher than the ~100 mK noise floor observed in previous runs without the photodiode.
Source of Noise: The elevated noise was found to be independent of whether the laser was on or off. The authors speculate this is due to room-temperature thermal radiation leaking through the optical fiber passbands rather than shot noise or laser power dissipation.
Thermal Load:
- The photodiode dissipation (50 $\mu$ W optical power) did not measurably increase the Still stage temperature.
- However, scaling to 840 control lines would increase the Still stage thermal load by ~14 mW (due to photodiode dissipation), potentially pushing the refrigerator into a higher flow regime.

C. Scalability Simulations

Thermal Performance: Simulations for a 1,000-line system showed that the Optical + Superconducting Coax configuration offers the best thermal performance at the mixing chamber (MXC) and Cold Plate (CP) stages.
- MXC Load: Reduced from ~48 $\mu$ W (All-coax) to ~32 $\mu$ W (Optical + SC).
- CP Load: Reduced from ~2500 $\mu$ W (All-coax) to ~315 $\mu$ W (Optical + SC).
Trade-off: The optical approach shifts the thermal load to the Still stage (increasing from ~2.8 mW to ~13.9 mW). This is manageable as it is comparable to the heater power required to maintain helium flow, but it requires optimizing the refrigerator's Still stage cooling capacity.

5. Significance and Conclusion

The paper demonstrates that optical control is a viable and scalable alternative to traditional coaxial wiring for superconducting qubits.

Coherence: Optical control does not degrade qubit coherence times compared to standard microwave control.
Stability: The optical source offers excellent long-term stability, crucial for long-duration quantum algorithms.
Scalability: While the optical approach introduces a specific thermal load at the Still stage (due to photodiode heat), it drastically reduces the thermal load on the critical millikelvin stages (MXC) when combined with superconducting coaxial lines. This enables the integration of thousands of control lines in a single cryostat, a prerequisite for fault-tolerant quantum computing.

The authors conclude that future systems can utilize a hybrid infrastructure where control signals are transmitted via fiber to the 1K/4K stage, converted to microwaves, and delivered to the qubit via superconducting lines, effectively solving the "wiring bottleneck" of large-scale quantum processors.