Fullscreen HTML5Med Res (??MB)HTML4

QUANTUM COMPUTERS “No, you’re not going to be able to understand it... You see, my physics students don’t understand it either. That is because I don’t understand it. Nobody does... The theory of quantum electrodynamics describes Nature as absurd from the point of view of common sense. And it agrees fully with experiment. So I hope you can accept Nature as She is – absurd.” (Nobel laureat Richard Feynman, 1985) Quantum computers are based on a different type of architecture to conventional computers and can solve problems using the strange properties of quantum mechanics such as superposition and entanglement. By Dr David Maddison I n recent times computers have become enormously powerful and can be used to solve extremely complicated problems such as in fluid dynamics. But the architecture of conventional computers is unsuited for certain classes of problems. Solving those problems would take impractically long periods of time or be altogether impossible. Examples of problems that quantum computers could or should be good at solving include simulation of quantum mechanical systems. For example, it may be possible to accurately simulate interactions that occur in a particle accelerator such as the large Hadron Collider. Chemical re14  Silicon Chip actions could also be simulated including extremely complex ones like photosynthesis. New drugs could also be developed more quickly because large numbers of configurations of drug molecules could be evaluated quickly and the ones most likely to work selected for further testing. They could be used for breaking of certain types of encryption codes (with Shor’s algorithm, for example) or searching very large unsorted databases (with Grover’s algorithm). A classic example given is searching a phone book for a certain phone number when the name is not known. If a book had one million numbers it could be searched in siliconchip.com.au one thousand steps with a quantum computer instead of on average one half million steps as with a conventional computer and a naive search method! Weather forecasting and climate models are other possible uses. There are huge numbers of ways that different parts of a weather or climate system can interact. A quantum computer could analyse all possibilities at once and find the optimal answer. Travel routes and schedules could be quickly optimised. For example, for the classic “travelling salesman problem” the order of which cities to visit in order to minimise the distance travelled and not to visit any city twice could be quickly evaluated. A commercial quantum computer, the D-Wave Vesuvius solved such a problem in less than half a second compared with a conventional computer which took 30 minutes. Other possible uses include machine learning, pattern recognition, image classification and analysis, optimisation problems, quantum communication for guaranteed privacy and quantum teleportation where information is transferred from one quantum system to another with no wires, radio or light transmission. Note that quantum computers are unlikely to replace the computers we use now but will coexist with them and be used only for the types of problem they are best at solving. Quantum mechanics Before discussing how a quantum computer works it is first necessary to discuss some basic principles of quantum mechanics. Quantum mechanics is a branch of physics that describes the behaviour of the very small objects such as atoms, sub-atomic particles and photons and is distinct from traditional classical mechanics that describes the behaviour of larger scale objects. In classical mechanics, objects exist in a specific and definite place and time, something we are all used to. But in quantum mechanics, particles exist in a “cloud of probability” so that the location of a particle is described by a probability distribution. In addition, in quantum mechanics, particles are restricted to certain particular values of properties such as how much energy they have or a property known as spin The probability distribution of an electron in a particular orbital of an atom. The darker the “cloud”, the higher the probability of finding the electron. It does not have an orbit analogous to a planet orbiting the sun as in the traditional simplistic view that many people are familiar with although electrons will have the highest probability of being at certain energy levels. For a further explanation of energy levels in atoms see SILICON CHIP, November 2015, page 17. (Image credit: RJ Hall) siliconchip.com.au How safe is encryption against attack by a Quantum Computer? It has been claimed that quantum computers will be able to break certain types of cryptography by their ability to factor large numbers which are the basis of certain types of encryption schemes. Schemes claimed to be at risk include those based on symmetric key algorithms (block cyphers) and asymmetric public key algorithms (RSA, DSA, ECC). Acknowledging the risk, the US National Security Agency has already announced it will be moving toward using encryption schemes which are resistant to attack by quantum computers. It should be noted that for the foreseeable future, there are no conceivable realistic quantum computers that are able to come close to factoring the numbers required to break the above schemes (when the key length is long enough) so they should be safe for a long time. For example, using Grover’s algorithm to factor a large number would enable the calculation to be done in the square root of the time taken by a classical computer (say 10 days instead of 100 days) but the security of the encryption could be maintained if the key length were doubled which is relatively easy to do. where a particle must be either “up” or “down”, much like the north or south on a compass. The values of the properties of particles are regulated like the clicks on a dial and are said to be quantised. Another main property of quantum mechanics is that elementary particles sometimes act like waves and at other times like particles, so-called “wave-particle” duality. There is also the uncertainty principle which states that for a given particle we cannot measure with precision its properties of both position and its momentum. The more accurately one value is known, the less accurately the other is known. In fact, accurately knowing these two properties together is a meaningless concept in nature. Any attempt at measurement of one property will alter the other property of the particle so it is impossible to ever know both values. Incidentally, this uncertainty also applies to macroscopic objects but is so small as to be of no consequence, eg, the uncertainty of position of a thrown cricket ball would be around 10-30mm. Superposition is the condition whereby a particle can be in a combination of two or more quantum states simultaneously. For example, rather than having a spin of “up” or “down”, an electron can be (3/5) up or (4/5) down. In this case, the RMS sum of the coefficients must remain as one. If up and down corresponded to the binary numbers zero and one we would say (simplistically) that it partially had the values of zero and one at the same time. Any attempt to read or measure the value of the particle, however, causes its quantum state to collapse or de-cohere into one of the values it possesses such as a “one” and superposition is lost. This phenomenon is known as quantum decoherence. While superposition is a characteristic of one-particle systems, a property that pairs or groups of particles can have is entanglement. In this case the quantum state of the pair or group is described collectively as it is shared and it is not possible to describe the state of an individual particle independently. Consider a pair of entangled particles which are known to have a total spin of zero. These entangled particles are March 2016  15 SOME SIGNIFICANT DEVELOPMENTS IN THE HISTORY OF QUANTUM COMPUTING There are far too many developments relevant to quantum computers to list them all here, so only a selection is given. 1975 RP Poplavskii showed the impossibility of simulating quantum systems on classic computers due to superposition. 1976 Roman Stanisław Ingarden published work on quantum information theory. 1980 Yuri Manin proposed the idea of a quantum computer in his work “The computable and the non-computable” (in Russian). 1981 Richard Feynman said in a talk that it seemed impossible to simulate quantum systems on classical computers and proposed a basic theoretical model for a quantum computer. For those interested they can read a transcript of this talk at www.cs.berkeley.edu/~christos/classics/Feynman.pdf 1982 Paul Benioff proposed a comprehensive theoretical model for a quantum computer. 1985 David Deutsch described a theoretical model of universal quantum computer that can be used to model other quantum computers and the algorithms they use. 1991 Artur Ekert invents secure communication based on quantum entanglement. 1993 Dan Simon invents a problem that would be exponentially faster for a quantum computer to solve than a classical one. 1994 Peter Shor, incorporating Dan Simon’s ideas from above, discovers a method to factor large integers quickly. The factoring of large integers is the basis of many modern cryptography systems such as credit card transactions and this algorithm could theoretically break many such systems. This lead to a tremendous interest in quantum computation. 1995 Peter Shor and Andrew Steane propose a method for quantum error correction. Also Christopher Monroe and David Wineland experimentally produce the first quantum logic gate based on a trapped atom. 1996 Lov Grover invents a quantum algorithm to search databases that is much faster than would be achievable on classic computers. David P. DiVincenzo published a list of the physical requirements for a quantum computer. 1998 First demonstration of a quantum algorithm run on a two qubit quantum computer. First three qubit quantum computer invented. Grover’s algorithm (1996) run on quantum computer. 2000 First five qubit and then seven qubit quantum computer and also partial execution of Shor’s algorithm (1995). 2001 Full execution of Shor’s algorithm (1995) to factor the number 15. 2002 Quantum Computation Roadmap developed to facilitate the future development of quantum computation. The document is constantly updated. See http://qist.lanl.gov/qcomp_map.shtml 2003 The US Defense Advanced Research Projects Agency (DARPA) implements a quantum network using optical fibres to transmit information securely using entangled photons. Any attempt to improperly intercept the data will result in a loss of entanglement of the photons and an inability to read the data. Also the University of Queensland demonstrate quantum NOT gates. 2005 First quantum byte created, known as a qubyte. 2006 First 12 qubit quantum computer. 2007 Commercial company D-Wave Systems announce working 28 qubit quantum computer. 2008 Qubits based on graphene quantum dots. D-Wave Systems announce working 128 qubit quantum computer chip. 2009 Qubits with lifetimes of hundreds of milliseconds. Google and D-Wave Systems collaborate in the area of using quantum computation for image searches. 2010 Single electron qubit demonstrated. 2011 D-Wave produces first commercial quantum computer. Error correction in quantum processor developed. Decoherence minimised using high magnetic fields. Record low error rates are achieved for a quantum computer. An error rate of one in 10,000 logic operations was considered a benchmark target but a rate of one in 50,000 was achieved. A group at the University of New South Wales and the University of Tokyo achieve a breakthough in quantum teleportation, successfully transmitting quantum information without error or affecting the superpositions of qubits. 2012 D-Wave produced quantum computer with 84 qubits. Single atom transistor developed. 1QB Information Technologies founded, the world’s first company to write quantum computer software. See www.1qbit.com/ Decoherence was kept suppressed for 2 seconds. A group at the University of New South Wales develop the first qubit based on a single atom of silicon which would enable quantum computers to be built in silicon like conventional computers with similar fabrication technology. 2013 Three billion qubits were held in a state of superposition for 39 minutes, exceeding the previous record of 2 seconds (2012). 2014 Leaked documents show that the US National Security Agency is interested in quantum computing for cryptography purposes. Quantum teleportation demonstrated over 3 metres. This is necessary for a quantum-based Internet to make it secure and fast. The largest number ever factored on a quantum computer was achieved, 56,153 exceeding the previous record of 143. University of New South Wales researchers embedded qubits in silicon to protect them and give them longer decoherence times. 2015 D-Wave Systems announce a 1,000 qubit system. University of New South Wales researchers build the world’s first quantum logic gate in silicon. 16  Silicon Chip siliconchip.com.au The Titan supercomputer at the Oak Ridge National Laboratory, Tennessee, USA is the most powerful classical computer in the Western world and the most powerful supercomputer that is freely accessible. It uses 18,688 AMD Opteron 6274 16-core CPUs and the same number of Nvidia Tesla K20X GPUs or graphics processing units. It has a benchmark of 17.59 petaFLOPs (where peta is 1015 or 1,000,000,000,000,000 and a FLOP is one floating point operation per second). The computer runs the Cray Linux Environment and it consumes 8.2MW. There is a more powerful Tianhe-2 supercomputer in China; however it uses US-made CPUs, is not freely accessible and has been criticised for its difficulty of use. Quantum computers will not replace computers such as these but will supplement them. in a state of superposition. If a measurement is made (thus destroying superposition) on one particle and it is found to have an up spin, for example, the other particle will automatically acquire a down spin as the total spin of the pair must be zero (an up spin plus a down spin). The particle that is not measured changes its quantum state as if to “know” a measurement has been made on its partner. This happens no matter by what distance the particles are separated and would happen even if the particles were at opposite ends of the universe. Furthermore, the change is instantaneous, not propagated at the speed of light as might be expected. The information travels at an infinite speed, although it cannot be used for faster-than-light communication. Einstein called this phenomenon “spooky action at a distance” and felt it meant that the description of reality by quantum mechanics was incomplete. Bits and Qubits The basic unit of information in a conventional computer is the bit which can have a value of either zero or one. It is typically physically implemented by the use of a transistor which is in either an “off” or an “on” state representing either zero or one or a capacitor which is either charged or discharged. For 2015 the commercial CPU with the largest number of transistors, 5.5 billion, was Intel’s 18-core Xeon Haswell-EP. A qubit is the quantum equivalent of a bit which when read (measured) will result in an answer equivalent to 0 Bloch sphere diagram representing a qubit. x, y and z represent the axes of the sphere, the north and south poles represent the basis states and the  represents the superposition of |0> and |1>.  and  represent angles. Image credit: Glosser.ca [CC BY-SA 3.0] siliconchip.com.au or 1. Due to the principle of quantum superposition as explained above, the qubit can have a combination of these values at the same time whereas a conventional bit must be either zero or one but not both at any given time. A qubit can be physically represented by the states of various quantum particles such as the spin of electrons (which are either up or down) or other quantum-dominated systems (see below). A qubit is regarded as the superposition of two basis states which are denoted mathematically as |0> and |1> (spoken as ket 0 or ket 1) and are equivalent to 0 or 1 in classical computing. While an ordinary bit in classical computing can be represented in a diagram by either a simple 0 or 1 a qubit is a bit more complicated and is represented by a Bloch Sphere as shown. On the Bloch spere, the “north” and “south” poles represent the basis states of |0> and |1> which physically might Simulating a Quantum Computer without yet having one! There are a lot of problems to solve with quantum computers but algorithms and computer code still need to be developed to solve these problems. Microsoft have developed a software simulation tool called LIQUi|> or Language-Integrated Quantum Operations (the symbols at the end a notation used in quantum computing) that transforms a higher level computer language such as F# that is coded to represent a quantum operation into one specific to low level operations in quantum computers. It allows researchers to write and develop quantum code on conventional computers in the absence of access to full scale quantum computers that Microsoft judges to be 10-20 years away, notwithstanding the developments described here. If you are interested in looking at this it can be downloaded free from https://github.com/msr-quarc/liquid That version allows for the simulation of up to 23 qubits. Among specific algorithms that can be simulated and which are included as examples are: simple quantum teleportation, Shor’s factoring algorithm, quantum chemistry, computing the ground state energy of a molecule, quantum error correction, quantum associative memory and quantum linear algebra. March 2016  17 represent spin up or spin down states. The superposition of these states – the qubit - is represented by some point anywhere on the sphere. When the state of a qubit is measured there is a loss of superposition and thus the system can no longer be in two states simultaneously due to quantum decoherence. The result is |0> or |1>, equivalent to 0 or 1 in classical computing. When multiple qubits exists in a system they can possess the property of entanglement, mentioned above. This means that, for example, a pair of entangled qubits will maintain a relationship with each other so if one is measured (thus causing quantum decoherence) and found to have a spin up state, the other will automatically have a down spin. Entanglement is one method by which multiple qubits can be made to “work together” and thus solve more complex problems. Information representation in bits and qubits Consider the information that can be represented in a 2-bit system. Two bits can be represented as either 00, 10, 01 or 11. Two bits can therefore represent only one of four different values and to use all four values in some given computation the computer would have to execute at least four cycles so that each value could be loaded and then used in a calculation. On the other hand, a 2-qubit quantum computer can contain and utilise for a calculation all those four values (above) simultaneously so only one computer cycle is necessary to operate on all four items of data. In other words, two bits contain information about only one value and two qubits contain information about four values. In fact, quantum computers scale the information that can be contained in the qubits exponentially according to 2n where n is the number of qubits. A 4-qubit computer could, for example, simultaneously hold sixteen values (24), ie, 0000, 0001, 0010, 0011, 0100, 0101, 0110, 0111, 1000, 1001, 1010, 1011, 1100, 1101, 1110 and 1111. In contrast, a conventional 4-bit computer could store only one of those sixteen values and would have to repeat an operation 16 times to do the same computation SQUID as used for the qubit in the D-Wave quantum computer. The horizontal arrows represent the possible directions of current and the vertical arrows represent the two possible spin states, up or down corresponding to zero or one. as the quantum computer could do just once. A multiprocessor classical computer increases its power directly in proportion to the number of processors it possesses. The ability for data representation to scale exponentially in a quantum computer compared to a classical computer and subsequent processing of that data is a key to its theoretical power, providing that can be implemented in a practical manner. In other words a quantum computer is not simply the same as a classical parallel processing computer. What can be used as a qubit? Almost any system that displays quantum mechanical phenomena can be used as the basis of a qubit as long it is capable of possessing two different quantum mechanical states, such as spin up or spin down. Any real quantum computer might have a combination of different two state systems just as a classical computer uses the state of a transistor in a CPU, capacitors in RAM, the pit or absence thereof in optical media such as a DVD or the state of a magnetic domain on a hard disk. Systems proposed include but are not limited to: electrons (spin up or down), light (amplitude or phase “squeezed”), Josephson junction and SQUIDs (direction of current), photon (vertically or horizontally polarised), atomic nucleus (spin up or down), optical lattice (spin up or down), quantum dot (spin up or down), graphene quantum dot (spin up or down), trapped ion (state of ions), nuclear magnetic resonance of liquid molecules (nuclear spin state) and diamonds (nuclear spin of atomic vacancies). Note: What would program code for a Quantum Computer look like? Anyone who has learned to program has probably started with a simple program such as the classic one that prints “Hello world”. What would a very simple program on a quantum computer look like? No one yet knows how quantum computers and their programming languages will evolve but it might look like the following. Consider a quantum computer language with just four instructions N (create qubit), E (entangle qubit), M (measure qubit) and X (execute operation). This program creates an ancilla, a special bit used for quantum error correction, entangles it with the input qubit, measures the input qubit and conditionally performs an operation on the ancilla. After the operation, qubit 2 contains the state of qubit 1 after a Hadamard transformation has been performed. A Hadarmard transformation is a one qubit rotation whereby two qubit states are mapped onto two superposition states with the same computational state as the original qubits (more generally it is a class of Fourier transforms). Note that this is very low level programming, equivalent to as18  Silicon Chip sembly language in a conventional computer and coding would not normally be done at such a low level – much higher level programming languages would be used. N2 # create a new quantum bit and identify it as ‘2’ E 1 2 # entangle qubits ‘1’ and ‘2’, qubit 1 already exists and is considered input M 1 0 # measure qubit ‘1’ with an angle of zero (angle can be anything in [0,2pi] # qubit ‘1’ is destroyed and the result is either True or False # operations beyond this point can be dependent on the signal of ‘1’ X 2 1 # if the signal of qubit ‘1’ is True, execute the Pauli-X operation on qubit ‘2’ Reference: http://cstheory.stackexchange.com/questions/9381/ what-would-a-very-simple-quantum-program-look-like siliconchip.com.au This graph shows “Rose’s Law” demonstrating the steady increase in the number of qubits in the D-Wave quantum computer which is analogous to Moore’s Law with the number of transistors in a classical computer, SQUID is a superconducting quantum interference device. Note that a qubit does not have to be physically small, although that is desirable so many qubits can be placed on one chip. Basic elements of a quantum computer A practical quantum computer must have certain basic requirements (DiVincenzo’s criteria) some of which also differ from a conventional computer as explained below. 1) It must be scalable to enable a reasonable number of qubits just as a conventional computer must have a reasonable number of bits for efficient operation. 2) The qubits must be able to be set to a common initial state such as all zeros, just as are the bits in a conventional computer. 3) The state of the computer must be controllable using universal gates such as quantum logic gates. They are analogous to the logic gates in conventional digital computer circuits (but unlike in a conventional computer they are reversible). 4) To enable logic operations to be performed by the logic gates the decoherence times of the qubits must be long enough for the gate operation to complete. Decoherence can be suppressed by error correction techniques and fault tolerant computation. The logic state of a conventional digital circuit will remain indefinitely but qubits are inherently unstable and will eventually revert to an alternative state. A stability time of somewhere between nanoseconds and seconds is required. 5) There has to be a means to read the quantum state of the processor. In quantum mechanics, the very act of taking a reading or measurement will alter the state of the system. Conventional digital circuits can be read without altering the state of the system. siliconchip.com.au Quantum decoherence As mentioned above, quantum decoherence can happen due to making a measurement or reading but it can also happen for unwanted reasons and this represents one of the greatest challenges of quantum computing. A quantum system can decohere due to thermal vibrations in the atomic lattice (if a crystal-based system is used) or other subatomic or macro scale phenomena. One partial solution is to cool the quantum processor to extremely low temperatures in order to reduce thermal vibrations. INTO MODEL RAILWAYS IN A BIG WAY? With lots of points, multiple tracks, reversing loops, multiple locos/trains, – in other words, your model trains are more a passion than just a hobby? Then you might be interested in these specialised model train projects from March 2013 Automatic Points Controller (Supplied with two infrared sensor boards) (PCB 09103131/2)........................$13.50 Frog Relay Board (09103133)............$4.50 Capacitor Discharge for Twin-Coil Points Motors (PCB 09203131)..................$9.00 See article previews at www.siliconchip.com.au ORDER NOW AT www.siliconchip.com.au/shop March 2016  19 (Above): closeup of the D-Wave 1000 qubit quantum processor. (Right): D-Wave processor package mounted on dilution refrigerator to keep it at a temperature close to absolute zero. Temperatures as low as 20mK or twenty thousandths of a degree above absolute zero are required. This corresponds to -273.15° Celsius and is much colder than anywhere in the universe, which doesn’t get much colder than about 3° above absolute zero. Cooling won’t necessarily remove all instances of decoherence and it is necessary to use quantum error correction to detect and reduce errors however this comes at the cost of the requirement for many more qubits in the system. Conventional computers, it should be noted, also use extensive error correction to ensure they operate correctly and in very early digital computers it was necessary to run a program several times to ensure the same result was obtained each time and if it was, confidence could be had in the result! Operation of a Quantum Computer To operate a quantum computer the qubits are first set to an initial state representing the problem and then those qubits are manipulated using quantum logic gates which are operated in a sequence according to a quantum algorithm. Quantum logic gates are like logic gates in classical computers (although their operation is reversible). A quantum algorithm consists of the step-bystep instructions for solving the problem but is specifically designed to utilise features of the quantum computer such as superposition and entangle20  Silicon Chip ment. Algorithms from classical computers can also be implemented on a quantum computer. Two widely known quantum algorithms are Shor’s algorithm for factoring and Grover’s algorithm for searching unstructured databases. Once a quantum computer has finished running an algorithm, a measurement of the qubits is made which collapses the qubits into their basis states, representing a zero or one to yield the result. Some quantum algorithms give the correct answer only with a certain probability and may give a different result each time the algorithm is run! This is the case with some algorithms run on the D-Wave computer discussed next. When these algorithms are run multiple times the most common result is likely to be the correct one. The commercial D-Wave Quantum Computer The only company making quantum computers on a commercial basis is DWave Systems (www.dwavesys.com), a Canadian company founded in 1999. D-Wave’s computers run a very specialised type of process called quantum annealing which is used for solving problems involving optimisation where a huge number of options are reduced to the best choice. One way to think of these problems is to think of a metaphor involving a vast landscape with many hills and valleys. The object is to find the low- est valley (the best choice) and the way to do it is either to 1) survey the whole landscape by walking up and down the hills looking for the lowest valley as a conventional computer would do or 2) use the quantum computer to effectively tunnel through the hills to quickly find the lowest point. The basis of the qubit in the D-Wave computers is a SQUID or Superconducting QUantum Interference Device. The device is made of a ring of superconducting niobium and a junction. Current within the ring can flow in one direction or the other, resulting in magnetic spin states which are either up or down although before measurement the device is in a superposition of both states, effectively meaning that the current flows in both directions at once. The D-Wave computer quantum processor must be kept at a temperature close to absolute zero to minimise quantum decoherence and also to ensure that the SQUID devices can operate in their superconducting state. The large size of the computer is primarily due to the cooling equipment. In the quantum annealing process, the algorithm used to run calculations tries to predict what states the qubits will be in when the temperature of the SQUIDs is increased, thus finding the solution or set of solutions for the lowest point in the valley in the landscape metaphor described above. As mentioned previously, this computer does not necessarily give the siliconchip.com.au same answer to a problem if run a second time however the more answers it repeatedly gives which are the same, the greater the confidence one has in the result. D-Wave sees this as an advantage as it assists in determining the confidence the computer has in the result of complex computer-based decisions in machine learning applications. The D-Wave computer is in use by Google, NASA, Lockheed Martin and others. Google hopes to use the computer for image and news classification, spoken word recognition, machine learning and understanding natural language and is doing research into other possible uses. The D-Wave computer has been criticised because it is not a “universal quantum computer” meaning that it cannot run any type of calculation but is limited to just “combinatorial optimisation problems” and it thus cannot run Shor’s algorithm, for example. Another criticism relates to whether it truly is a quantum computer andwhether it uses entangled states. The reality is that no one fully understands how it works in all aspects, not even the designers, although it is now generally agreed that it is indeed a real quantum computer. Other issues relate to questions of how to benchmark the speed of such a computer and compare it to classical computers. Making single atom qubits, atomic wires Australia is a world leader in aspects of quantum computing. The Centre of Quantum Computation and Communication Technology (www.cqc2t.org/) is a collaboration between The University of NSW, The University of Melbourne, Australian National University, Griffith University, The University of Queensland and The University of Sydney. It is undertaking work involving a diverse area of quantum communication and quantum computing. One (1) To make an image of an atomic structure the probe of a scanning tunnelling microscope (STM) is moved along the surface of a silicon crystal and an image of the surface is obtained by measuring a current flowing between the crystal and the tip which varies according to the position on the crystal surface. An STM can also be used to manipulate single atoms on the crystal surface. It is important to map the crystal surface so the exact location of the qubit is known. 22  Silicon Chip (4) Phosphine gas, consisting of phosphorus (red) and hydrogen, is introduced and the molecule of gas settles in the place where the two hydrogen atoms were removed. (5) The phosphorus atom of the gas molecule now lies on the surface of the silicon crystal. Conclusions The dream of quantum computing has been around for a while and now there is one type of specialised quantum computer in commercial production with major research in other areas of quantum computing, with Australia being a key player. Quantum computers will not replace classical computers but will supplement them by solving specialised types of problems for which they are suited. It is also important to distinguish hyperbole from reality. Most likely quantum computers will be introduced slowly, at first solving a limited number of problems and then, perhaps, the market will expand as they solve problems with widespread demand, such as understanding and interpreting spoken language, recognising objects or even artificial intelligence. SC particular project is the Precision Qubit Program. This program involves making qubits using single atoms and aims to “position, control and read out the electron spin on a single (phosphorus) atom in silicon which acts as a quantum bit or qubit”. Single electron transistors and microwave strip lines are used to both read and manipulate the electron spin on a single phosphorus atom embedded in a crystal of silicon. The ability to create a single atom (2) A layer of hydrogen atoms (light colour) is laid down on the silicon surface to create the desired types of surface chemical bonds. A pulse of current is then applied to the STM probe which removes one hydrogen atom. (3) A second pulse of current is then applied to the STM tip to remove a second hydrogen atom. (6) The hydrogen atoms are removed. (7) More silicon atoms are added to the surface, embedding the phosphorus atom deep in the atomic structure where it is not affected by undesired interference from the crystal surface. (Diagrams captured from https://youtu. be/0dXNmbiGPS4) siliconchip.com.au and the single atom transistor qubit and support structures such as nano-wires to access the qubit is a remarkable achievement and only possible due to the recent development of techniques to reproducibly manipulate single atoms and also to know exactly where those single atoms are located within the crystal lattice. The illustrations in the numbered images in the box show how a single atom of phosphorus is embedded into a specific location within a silicon crystal. Actual STM image of a phosphorus atom (centre) located on the surface of a silicon crystal at step 6. The scale bar represents one nanometre, one millionth of a millimetre. The ability to accurately place a single atom at a precise location plus the ability to make an atomic scale wire allow the fabrication of a single atom transistor. Such a transistor can be used as a qubit or as a component of a classical computer. While making such a device is a fantastic start, practical computers need large numbers of devices on the one chip. Also, according to Moore’s Law for classical computing which says that the number of transistors on a chip doubles every 12 to 18 months, the size will need to reach the atomic scale by 2020 if that rate of advancement is to be maintained. Obviously beyond the point of a single atom transistor, no further size reduction is possible. STM image of single atom transistor. The single phosphorous atom is at the centre and the atomic scale wires are shown in pink. siliconchip.com.au Dr Matthew House with Honours student Kirsti Date studying deterministic placement of single donors in silicon at the Atomic Fabrication Facility at the University of New South Wales. An atomic scale wire just one atom tall and four atoms wide. This is the type of wire that may be used to connect to single atom qubits. It was made by using an STM to create a channel in the silicon and then exposing the area to phosphine gas to make a line of phosphorus atoms and then depositing silicon atoms on top of the phosphorus atoms (similar to with the numbered images). The phosphorus atoms, which were placed at a spacing of less than one nanometer, doped the region around their vicinity causing it to become conductive and act as a wire. A similar conductivity and current carrying capability as copper was achieved. This particular work also proved that Ohm’s law operates at the atomic scale which was not an expected result as quantum effects were though to dominate at this size scale. On the other hand, a concern that has been raised from the knowledge that Ohm’s law still works at this scale is that nonquantum affects may dominate making a qubit difficult to implement. Another important outcome of this work relates to conventional silicon chip fabrication. Companies such as Intel have become increasingly worried that the feature size on microprocessors is becoming so small that quantum effects will soon start to dominate and no further miniaturisation can occur. Already transistor gate sizes are at 22nm which is about 100 times the spacing of silicon atoms. This work suggests that miniaturisation can continue for some time and down to much smaller feature sizes. Image Courtesy of the Centre for Quantum Computation & Communication Technology. March 2016  23