A team of researchers has devised perhaps the world's most intricate coin toss, a device utilizing vacuum chambers, magnetic fields, lasers and microwave pulses to produce a random string of 0s and 1s—each representing heads or tails, essentially. The complexity is necessary to move the generation of random numbers beyond the hard-to-predict but fundamentally deterministic world of classical physics and into the realm of quantum mechanics, where uncertainty takes hold.

Antonio Acín, a physicist at the Institute of Photonic Sciences in Spain and an author of a paper describing the approach in the April 15 issue of Nature, says that true randomness is elusive. "If you go to a casino and play roulette, or you flip a coin, if you had access to the initial position and speed of the ball or coin, you could predict the result with certainty," he says. "The randomness that we have in our world is because of lack of knowledge." (Scientific American is part of Nature Publishing Group.)

For many purposes that is enough—there is a reason that casino owners tend to fare better than their patrons—but Acín notes that genuinely random numbers are desirable for advanced cryptography and other such applications. "You need some randomness that is not based on lack of knowledge" for those realms, Acín says, because "maybe someone has better knowledge than you."

To that end the researchers utilized a pair of ytterbium ions as quantum bits, or qubits, each confined to a private vacuum chamber about a meter apart in an experimental system at the Joint Quantum Institute of the University of Maryland and the National Institute of Standards and Technology. Depending on the state of the ions, a resonant laser pulse will either cause them to emit a photon, representing a binary 1, or remain dark, representing a zero. Each atom's state cannot be known with certainty until it is measured with the laser pulse—that is, it is probabilistic rather than deterministic—so the measurements can be used to generate an intrinsically random string of binary digits.

Acín's group used statistical tests to show that the output from the new device indeed stems from quantum uncertainty rather than from residual deterministic—and hence predictable—effects. Using so-called Bell inequalities, the researchers demonstrated that the two qubits shared a quantum-mechanical link known as entanglement, meaning that the measurement of one qubit's quantum-mechanical state instantaneously affects that of the other qubit. Bell inequalities, named for Irish physicist John Bell, mark how much correlation a purely deterministic, non-entangled system should have. (In other words, they dictate how the qubits should behave if measurement of one has no effect on the other.) If those inequalities are violated, some unseen and instantaneous link must be in play that allows distant systems to influence each other. Entanglement is not possible in classical physics, so the nature of the system must be governed by quantum randomness.

The group's apparatus is not the first to harness quantum effects to produce random bits—some are already commercially available. But to ensure that the outcome of most such devices is truly random, a skeptical user would have to crack it open and parse its inner workings, says Valerio Scarani, a physicist at the National University of Singapore who wrote a commentary accompanying on the research in Nature. In contrast, notes study co-author Dzmitry Matsukevich, a Joint Quantum Institute postdoctoral researcher, the new setup functions as a kind of black box—in principle, one need not understand its workings to see that its output is indeed truly random. "Even if I don't trust my co-workers who build our ion traps and suspect that they put some device inside our vacuum chamber to simulate trapped ions, I still can certify using the Bell inequality violation that the system obeys quantum mechanics," Matsukevich says.

Powerful though it may be, do not expect a random-number generator based on entangled qubits to find its way into digital slot machines anytime soon. Even in an extraordinarily well-controlled system such as that at the Joint Quantum Institute, in which microwave bursts can rotate the atoms around an axis set by a magnetic field, entanglement is a finicky state that can be spoiled by any number of outside influences. Despite attempting to verify entanglement between the qubits 95,000 times per second, the researchers could only establish an entangled pair every eight minutes or so, ultimately yielding only a few dozen random bits over the course of the experiment. Because the Bell violation is a statistical test requiring a substantial sample of trials, notes Joint Quantum Institute physicist and study co-author Christopher Monroe, "we have to waste time building up statistics just to show a violation."

Scarani notes that the group's present approach is far from practical. "With some assumptions and the hard work of several experienced post-docs and PhD students, they managed to extract 40 bits from two high-vacuum chambers cooled down to very low temperatures," he says, noting that such random-number generators will move toward practicality only if the market deems them necessary. Otherwise, Scarani says, "the idea will remain a beautiful demonstration of the power of quantum physics."