Multi-junction surface ion trap for quantum computing
This paper presents a multi-junction surface ion trap designed to address scaling challenges by raising the RF electrode and minimizing underlying dielectric layers to significantly reduce ohmic and dielectric power dissipation, while also characterizing heating rates across various motional frequencies and voltage sources.
Original authors:J. D. Sterk, M. G. Blain, M. Delaney, R. Haltli, E. Heller, A. L. Holterhoff, T. Jennings, N. Jimenez, A. Kozhanov, Z. Meinelt, E. Ou, J. Van Der Wall, C. Noel, D. Stick
Original authors: J. D. Sterk, M. G. Blain, M. Delaney, R. Haltli, E. Heller, A. L. Holterhoff, T. Jennings, N. Jimenez, A. Kozhanov, Z. Meinelt, E. Ou, J. Van Der Wall, C. Noel, D. Stick
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 a world where computers don't just crunch numbers but solve puzzles that would take today's supercomputers millions of years to crack. This is the promise of quantum computing, a field where information is stored not in tiny switches called bits, but in "qubits" that can exist in many states at once. One of the most promising ways to build these qubits is by trapping individual atoms (ions) in a vacuum using invisible electric fields, much like holding a marble in a bowl of air. To make a useful quantum computer, scientists need to store hundreds or thousands of these atoms and move them around to perform calculations, similar to how a conductor moves musicians around an orchestra. However, as these "traps" get bigger to hold more atoms, they face a major problem: they start to overheat. Just like a crowded room gets hot because everyone is talking and moving, these traps generate too much heat as they try to hold more ions, which can ruin the delicate quantum states needed for the computer to work.
In this paper, researchers from Sandia National Laboratories and Duke University tackle this overheating problem with a clever engineering trick. They designed a new type of ion trap called the "Enchilada trap," named for its shape, which is built to hold up to 200 ions. The main issue they solved is that as the metal electrodes in these traps get larger, they act like giant capacitors that waste energy as heat. To fix this, the team raised the main radio-frequency (RF) electrode high above the others and carved away most of the insulating material underneath it, leaving only tiny pillars for support. Think of it like lifting a heavy, hot frying pan off a stove and placing it on a few small stilts; this reduces the contact area and stops the heat from building up as much. By doing this, they calculated that the trap could dissipate significantly less power—dropping from over 100 milliwatts in a solid design to just under 40 milliwatts in their new "perforated" design. They also tested how much the trapped atoms "jiggled" (heated up) due to electrical noise, finding that while the trap works well, the biggest source of jiggling comes from tiny voltage fluctuations on the surface of the electrodes rather than the heat from the wires themselves. This work doesn't solve every problem in quantum computing, but it proves that by changing the physical shape of the trap, we can build larger, cooler, and more efficient systems for the quantum computers of the future.
Technical Summary: Multi-junction Surface Ion Trap for Quantum Computing
Problem Statement The scalability of trapped-ion quantum computers is hindered by the power dissipation challenges inherent in large-scale surface ion trap arrays. As the number of trapping sites (n) increases, the total power dissipation scales unfavorably as Ptotal∼αon3+αdn. This cubic scaling arises primarily from ohmic losses in the radio frequency (RF) electrode and dielectric losses in the insulating substrate. The large capacitance between the RF electrode and surrounding grounded electrodes, combined with the resistance of the electrode material, leads to significant current flow and heat generation. This thermal load becomes a critical bottleneck for scaling to the hundreds or thousands of ions required for practical quantum error correction and complex algorithms.
Methodology and Design To address the n3 scaling of ohmic power dissipation, the authors designed and fabricated a large multi-junction surface ion trap, termed the "Enchilada trap." The core methodology involves a geometric modification to the standard surface trap architecture:
Raised RF Electrode: An additional metal layer (M6) was added to elevate the RF electrode 4 µm above the control electrode layer.
Dielectric Removal: Extensive amounts of the underlying silicon dioxide (SiO2) dielectric were removed from beneath the RF electrode, leaving only structural support pillars. This was achieved by perforating the top metal RF layer to allow etchant access for the release of the oxide.
Capacitance Reduction: By raising the electrode and removing the dielectric, the capacitance between the RF electrode and the ground plane was significantly reduced. Simulations indicated a drop from 11.9 pF (in a standard 5-layer design) to 6.5 pF with the raised electrode, and further down to 3.7 pF (2.3 pF for the trap region only) with the perforated design.
Fabrication: The device was fabricated using a 6-metal layer CMOS-compatible process. The trap features a bowtie-shaped platform with a 1.65 mm wide central isthmus, housing six Y-junctions and five long linear sections designed to store up to 200 ions. To manage the 302 control electrodes within a 100 I/O package constraint, the design utilizes co-wiring strategies, reducing the required independent voltage signals to 75.
Key Results The paper presents theoretical calculations and experimental measurements regarding the performance of the raised-RF architecture:
Power Dissipation: Calculations for a 300 V RF amplitude at 50 MHz demonstrate that the perforated, raised-RF design reduces total power dissipation from 101.6 mW (solid dielectric) to 38.7 mW. Specifically, ohmic losses in the trap region dropped from 9.7 mW to 1.9 mW, and dielectric losses dropped from 59.1 mW to 12.9 mW.
Trapping Parameters: The geometric modification resulted in a 17% increase in radial motional frequency, allowing for reduced voltage requirements and further power savings. The ion height was adjusted to 65.7 µm above the RF electrode.
Heating Rate Characterization: The authors measured axial heating rates for a single trapped calcium ion across a range of motional frequencies (2–3 MHz) using different voltage sources and filtering configurations on a solid dielectric version of the trap with a raised RF electrode:
With a 1.3 MHz low-pass filter, the heating rate scaled as ω−10.2, indicating dominance by technical noise.
With a 206 kHz low-pass filter, the scaling was ω−3.5.
Using a battery source with the 206 kHz filter to minimize technical noise, the heating rate scaled as ω−3.8.
Noise Analysis: The authors conclude that the heating rate measured with the battery source is dominated by voltage fluctuators on the electrode surface, as the observed scaling (ω−3.8) is steeper than the Johnson noise limit (ω−1). The calculated Johnson noise contribution (15 quanta/s at 2 MHz) was found to be below the measured rates.
Significance and Claims The paper claims to demonstrate a viable architectural solution to the RF power dissipation problem that limits the scaling of surface ion trap arrays. By raising the RF electrode and removing the underlying dielectric, the authors successfully reduced both ohmic and dielectric power dissipation without compromising the trap's ability to hold large numbers of ions. The Enchilada trap is presented as a designed architecture to hold 200 ions in up to four linear chains, addressing the connectivity and storage needs for surface codes and quantum error correction. The work validates the raised-RF approach as a method to enable larger trap arrays while managing thermal loads, a prerequisite for the next generation of trapped-ion quantum computers. The authors note that future work will include comparative measurements between the perforated and solid versions of the trap, as well as demonstrations of ion transport and chains of ions.