Promote the miniaturization of quantum computers! Quantum RAM achieves important breakthrough

A new quantum random-access memory (RAM) device uses chirped electromagnetic pulses, superconducting resonators to read and write information, making hardware much more efficient than previous devices. On November 7, the related results were published in Physical Review X under the title "Random Storage Quantum Memory Using Chirped Pulse Phase Encoding"[1].

01Quantum RAM: Speeding up algorithm execution and increasing information storage density

Random access memory (or RAM) is an integral part of a computer that acts as a short-term memory bank that can quickly recall information. Apps on our phones or computers use RAM so that tasks can be switched in the blink of an eye. Researchers working on building future quantum computers hope that such systems could one day operate using similar components of quantum RAM: This would speed up the execution of quantum algorithms, or increase the density of information that can be stored in quantum processors.

Now, James O'Sullivan and colleagues at the London Centre for Nanotechnology have taken an important step towards realizing quantum RAM: they have demonstrated a hardware-efficient method of using chirped microwave pulses to store and retrieve atoms in spinning atoms quantum information.

Just like quantum computers, experimental demonstrations of quantum memory devices are in their early stages. A leading chip-based quantum computing platform uses circuits made of superconducting metals; in this system, central processing is done by superconducting qubits, which send and receive information via microwave photons. However, there are no quantum storage devices that can reliably store these photons for long periods of time.

Fortunately, scientists have some ideas.

One idea is to exploit the spins of impurity atoms in chips embedded in superconducting circuits. Spin, one of the fundamental quantum properties of atoms, acts like an internal compass needle, aligned with or against an applied magnetic field. The two arrangements are similar to the 0s and 1s of classical bits and can be used to store quantum information. If the chip contains many impurity atoms, the atoms' spins can act as a "multimode" memory device: the information contained in many photons can be stored simultaneously.

For atomic spins, information storage times can be orders of magnitude longer than for superconducting qubits. For example, researchers have shown that bismuth atoms placed inside a silicon chip can store quantum information for longer than a second [2]. One might ask: why not use spin qubits instead of superconducting qubits? It is true that there are research groups working on atom-based quantum computers, but the control and measurement of atomic spin has its own unique challenges. A hybrid approach would be to use superconducting qubits for processing and atomic spins for storage, but the challenge here is how to use microwave photons to transfer information between the two systems. While researchers have demonstrated the absorption and retrieval of information from microwave photons by ensembles of atomic spins [3], these demonstrations require the use of strong magnetic field gradients or specialized superconducting circuits, both of which increase the complexity of quantum memory hardware .

02Quantum RAM: Achieving Hardware Efficiency, but Insufficient Information Retention

O'Sullivan and his colleagues provide a solution for microwave photonic information storage and retrieval that employs a hardware-efficient approach.

The researchers developed a RAM device from a superconducting circuit resonator and a silicon chip embedded with bismuth atoms; chirped microwave pulses pass quantum information back and forth between the resonator and the bismuth atoms, and the information is stored in the atoms' self-contained cells. in the spin state.

The team's device consists of a superconducting circuit resonator on a silicon chip embedded with bismuth atoms. The team fed weak microwave excitations containing about 1,000 photons into the resonator, which were absorbed by the spin of the bismuth atoms. They then hit the resonators with electromagnetic microwave pulses that increased in frequency over time, an effect known as the "chirp effect." Because of this, the quantum information contained in the photons imprints a unique "phase" identifier on the spins, which captures the relative orientation of neighboring spins. The team then retrieved this information to transfer the photons back into the superconducting circuit by hitting the spin ensemble with an identical pulse, which they found reversed the phase of this imprint.

O'Sullivan and colleagues show that their memory device is capable of simultaneously storing multiple photons of information in the form of four weak microwave pulses. Importantly, they also demonstrated that this information can be read back in any order, making their device a true RAM.

Device schematic. (a) Schematic illustration of the near-surface-implanted Bi layer on the silicon wafer and the Nb resonator on the surface. (b) Simulated Bi implant profile versus depth below the substrate surface. (c) Finite element simulation of the magnetic field generated near the inductance of the resonator (d) Copper sample box with an asymmetric antenna.

In the first demonstration, the team reported an efficiency of 3 percent, indicating that most of the information was lost by memory. Therefore, their device is still some distance from the fidelity storage and retrieval required for future quantum computers. However, analysis of the potential source of this inefficiency suggests that it does not come from the transfer process, but from a potentially solvable limitation of the device.

The team believes that by increasing the number of spins, they can drastically improve the efficiency of the device.

03Application areas: reducing the size of quantum computers, increasing the density of qubits

In addition to storing information, quantum RAM components could also help increase the density of qubits in quantum processors.

In September 2022, IBM launched the Goldeneye [9], the world's largest dilution refrigerator. The ultracold behemoth, larger than three household refrigerators, will host IBM's next-generation superconducting quantum computer. Current superconducting quantum computers have a qubit density of less than 100 per square millimeter, while conventional computer chips contain 100 million transistors per square millimeter: it is therefore understandable why IBM needed such a large refrigerator.

O'Sullivan's team's spin-based quantum memory device could, in principle, store multiple qubit states in the space currently occupied by only one qubit, which could one day help alleviate this size problem.

Reference link:

[1]https://journals.aps.org/prx/abstract/10.1103/PhysRevX.12.041014

[2]https://www.nature.com/articles/nnano.2013.117

[3]https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.105.140503