How does one make a computer faster? Shrink its building blocks and pack in more of them. But there is a fundamental limit to this process: each building block cannot be smaller than a single atom. That limit is not as far-off as it may seem, and today’s chip manufacturers are quickly approaching it.
In fact, computer manufacturers routinely construct transistors, computer chip building blocks, with features assembled from only a few thousand atoms! At this scale, physicists have found that matter behaves in ways that are unpredictable from the perspective of standard mechanics. For example, electrons stop behaving like individual particles, and start behaving more like waves that can interfere and leak from one wire to another. This is one example of what physicists refer to as quantum effects, phenomena governed by the principles of quantum mechanics. Quantum behavior poses a serious challenge for chip manufacturers, who have been sidestepping certain quantum effects for decades.
Dr. Irfan Siddiqi’s Quantum Nano-Electronics Lab (QNL) at UC Berkeley aims to harness the potential benefit of the very quantum effects that plague conventional computers. They are hoping to develop the first generation of quantum computers. Recently, the team took a big step toward this goal when they directly observed the quantum behavior of a small system, called a qubit, in real time. Qubits are the potential building blocks of a future generation of powerful quantum computers.
A quantum computer employs the quantum behaviors of atoms to speedily perform complex calculations by parallel processing. Whereas each processor in a conventional computer must do computations one-by-one, or serially, quantum effects known as entanglement and superposition would allow a quantum computer to do multiple computations simultaneously, or in parallel. In fact, such a computer could rapidly find the factors of a given large integer by dividing it by all smaller integers, all at the same time—a feat that would undermine many modern encryption techniques. In a matter of seconds, a quantum computer could use this advantage to perform a factorization that might take a classical computer the entire age of the universe to compute, and use those prime factors to break a code.
How can a quantum computer do more things simultaneously than a classical one? The difference lies in how each system stores information. While a modern laptop stores information in bits, binary pieces of data that are either 1s or 0s, a quantum computer would store information in qubits, short for quantum-bits. These qubits can be 1s or 0s just like bits, but they can also be a combination of the two states. This latter combination is known as a coherent superposition, and it holds the key to a quantum computer’s potentially massive advantage.
A quarter on a table can be thought of as a bit because it can be in one of two states: heads or tails. But what if the quarter is spinning on the table? As it spins, is in neither individual state, but rather something like a superposition of both states. It is potentially both heads and tails. It is acting like a qubit. If we insist on discovering what state the quarter is in, say by touching it and knocking it down, we collapse this superposition to one of two states: either heads or tails.
Just as a quarter cannot spin forever, a qubit cannot maintain a superposition state forever. It is eventually “knocked down” due to the quantum equivalent of friction. This so-called decoherence scrambles the information stored in the superposition, and can introduce insurmountable errors in a quantum computation. As Dr. Vijay, a postdoctoral researcher at QNL, puts it, “Decoherence is our number one enemy.”
Currently, quantum computers exist only in theory, but the physicists of QNL are hoping to change all that. They began by building their own fundamental building blocks: qubits of their own design. Those made at QNL are essentially “electrical circuits made with similar techniques to [those used to make] computer chips,” according to graduate student Dan Slichter. They are unique in that they are fast, tunable, easy to manipulate, and mass producible with current technology. While the circuits’ speed is important, it comes at the cost of coherence time (how long before the superposition, containing the qubit’s information, collapses). “This is ok,” says Vijay, “so long as each decoherence event can be detected. Then, quantum error correction techniques can compensate for the information loss.” But, as Slichter points out, “individual decoherence event measurements are notoriously difficult, and scientists have been trying to do this for a long time.”
Enter the recent breakthroughs of Siddiqi, Vijay, and Slichter. Their work used qubits made from super-cooled circuits about the size of a human cell, which are too big to display quantum effects at room temperature. At just 0.03 Kelvin—barely above absolute zero—their qubit circuit becomes superconducting, meaning it offers no resistance to currents of electrons trying to flow through it. Inside the qubit, the electrons can behave in quantum-mechanical unison.
When one of these qubits undergoes a quantum jump from one state to another (like switching from “heads” to “tails”), which may be a signal of a decoherence event, it introduces a tiny shift in the electromagnetic (EM) waves in a nearby sensor. Traditionally, a chain of amplifiers would amplify this signal at the cost of adding noise. This added noise has been so large that it drowns out the quantum jump signal being amplified in the first place.
Here, QNL has made its mark. Using their unique new piece of quantum electronics called a Josephson Parametric Amplifier (JPA), the researchers have found a way to maintain the integrity of the quantum jump signal. The JPA mixes the weak jump signal with a “pump tone,” a strong EM wave at the same frequency as the weak signal. This frequency-matched carrier signal amplifies the signal from a single qubit quantum jump above the noise introduced by the later amplification.
By enabling the detection of individual decoherence events in qubits both directly and in real time, QNL cements a key step on the way to the correction of information loss due to decoherence. Their work puts a functional quantum computer, a means of making atom-sized computer building blocks, significantly closer.