Quantum computing

Putting it into practice

Article metrics

Will quantum information theory ever lead to practical quantum information technologies? At a conference reviewing the advances of the past two years, delegates looked to the future with cautious optimism.

Quantum information processing1 (QIP) combines the counterintuitive theory of quantum mechanics with the practical field of information technology. Instead of worrying about the strange behaviour of quantum systems, researchers have harnessed this 'quantum weirdness' to perform apparently impossible types of information processing. Although the first ideas were formulated more than 30 years ago2, it is only in the past decade that serious proposals for quantum technologies have been made. These involve individual quantum systems, such as atoms, ions, nuclei or photons, which can be manipulated to store and process information — but which must be carefully isolated from their surroundings, as any uncontrolled interactions, known as decoherence3, will destroy the fragile quantum effects.

One of the first methods used to investigate QIP involves trapping small numbers of ions with electromagnetic fields and manipulating them with laser pulses. As reported at a recent conferenceFootnote 1, and on page 48 of this issue4, ion-trap methods have now been used to implement the simplest possible quantum computation, the Deutsch–Jozsa algorithm (R. Blatt, Univ. Innsbruck). Although the basic elements involved have been demonstrated before5,6, this is the first unambiguous demonstration of quantum computation. The Deutsch–Jozsa algorithm is most accurately described in terms of function evaluation, where it provides a method for determining the parity of a binary function using only one function evaluation. An easier analogy is to imagine tossing a coin. A normal coin has two sides, heads and tails, but a trick coin may show the same pattern on both sides. To recognize a trick coin, it would be necessary to look at first one, then the other side of the coin. But, by the Deutsch–Jozsa algorithm, in a quantum device this check could be made in effectively a single glance.

The original ion-trap approach to quantum computing, in which all the ions are held in a single trap, unfortunately cannot be extended to build large-scale devices. An alternative is to use a network of memory traps, in which ions are stored when not in use, and interaction traps, where computations occur. It has now been shown that ions can be moved between two traps while maintaining their quantum states (D. Wineland, NIST Boulder).

Ions interact strongly with electromagnetic fields and with one another, making them easy to trap and manipulate, but also rendering them vulnerable to decoherence. Atoms interact less strongly, so decoherence should be less of a problem. A lattice of traps can be built using intersecting laser beams to produce a standing light wave. Inside the trap, atoms seek the regions of lowest light intensity, which occur at the nodes of the standing wave. These traps, however, are very shallow, making it easy for atoms to escape.

The solution is to start with ultracold atoms, which have so little thermal energy that they remain trapped in this optical lattice (W. Phillips, NIST Gaithersburg). Ideally there should be only one atom at each lattice site7. The atoms can be made to interact by altering the properties of the laser beams, causing them to move. Although it is not yet possible to control the atoms individually, detailed control of the whole lattice is possible (I. Bloch, MPI für Quantenoptik, Garching), allowing complex entangled states of many atoms to be created.

Entanglement, in which the properties of different atoms are inextricably intertwined, is a key quantum phenomenon that underlies quantum information processing. Entangled atoms could be used to implement certain quantum algorithms (I. Cirac, MPI für Quantenoptik, Garching). In particular, it should be possible to use this system to simulate the properties of other, less well-controlled, quantum systems.

Trapped atoms and ions can be studied today, but many researchers believe that the future of QIP lies with solid-state devices (A. Briggs, Univ. Oxford). One early proposal8 that generated a great deal of excitement uses the spin of phosphorus nuclei implanted in a silicon matrix to store the quantum information. The nuclei are controlled and observed through the behaviour of their surrounding electrons.

Building such a device would be extremely difficult, involving placing individual atoms at precisely known locations inside a silicon chip and then building tiny electrodes and transistors around them. These ideas have been demonstrated (R. Clark, Univ. New South Wales) and a device with two phosphorus atoms has been constructed. Early results suggest that these two atoms can be controlled and observed. Another proposal based on superconducting quantum interference devices (SQUIDs) has also made substantial progress (G. Wendin, Chalmers Univ., Göteborg). Several working single-SQUID devices have been constructed, and attempts to couple two SQUIDs are under way.

In addition to the new results announced at the meeting, several themes for future work emerged. Although many simple demonstrations of QIP have been achieved, only one of these, quantum cryptography (J. Rarity, QinetiQ), could really be described as practical9. But before we can make existing toy systems do something useful, we will have to greatly improve the precision of quantum logic gates (J. Jones, Univ. Oxford) and of sources and detectors (I. Walmsley, Univ. Oxford).

It has long been noted that there seems to be a close connection between the controllable interactions needed to implement quantum logic, and the uncontrollable interactions that cause decoherence. Is this simply an artefact of the systems studied so far, or does it reflect some underlying principle? In fact, whenever two quantum objects interact with one another they must also interact with their environment (A. Fisher, Univ. College London). The strengths of these desirable and undesirable interactions are inevitably related, and real quantum computers will need this ratio to be as large as possible. To achieve this will require an extremely careful choice of environment.

Will these ideas lead to a practical quantum computer? The progress so far is remarkable, but the difficulties can hardly be overstated. Whether or not we eventually get there, we will certainly see some interesting scenery on the way.


  1. 1.

    *Practical Realizations of Quantum Information Processing, The Royal Society, London, UK, 13–14 November 2002.


  1. 1

    Bennett, C. H. & DiVincenzo, D. P. Nature 404, 247–255 (2000).

  2. 2

    Wiesner, S. SIGACT News 15, 78–88 (1983).

  3. 3

    DiVincenzo, D. & Terhal, B. Phys. World 11 (3), 53–57 (1998).

  4. 4

    Gulde, S. et al. Nature 421, 48–50 (2003).

  5. 5

    Monroe, C., Meekhof, D. M., King, B. E., Itano, W. M. & Wineland, D. J. Phys. Rev. Lett. 72, 4714–4717 (1995).

  6. 6

    Jones, J. A. & Mosca, M. J. Chem. Phys. 109, 1648–1653 (1998).

  7. 7

    Greiner, M., Mandel, O., Esslinger, T., Hänsch, T. W. & Bloch, I. Nature 415, 39–44 (2002).

  8. 8

    Kane, B. E. Nature 393, 133–137 (1998).

  9. 9

    Kurtsiefer, C. et al. Nature 419, 450 (2002).

Download references

Author information

Correspondence to Jonathan Jones.

Rights and permissions

Reprints and Permissions

About this article

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.