An Overview Of Commercial Qubit Development Technologies
If you take a brisk walk through the device you are reading this on, you’ll find processors native to the language of 0s and 1s.
Our system of digital electronics finds its origin in classical mechanics, according to which, the flow of electrons determines whether a bit contains a 0 or a 1. With the advent of transistors, computation has made a leap from the ENIAC to Fugaku — the fastest supercomputer as of the writing of this sentence. In fact, humanity itself has developed extensively from the age of the abacus.
The rise of quantum mechanics, as well as its evidence and exploitable properties, is now taking us towards a new wave — one that does not take off from bits, but qubits.
A qubit is to quantum computer engineering what a bit is to traditional computer engineering, with a much more far-reaching potential. This is a result of the exploitation of three properties of quantum mechanics: superposition, coherence, and entanglement.
Theory of information has been metamorphosed by the discovery that quantum systems can run exponentially faster than their classical versions, and only need to be supplemented by the invention of quantum error-correction and coherence-maintenance protocols. Most such strategies have been based on atomic degrees: dipoles (two-state systems), the spin of electrons or nuclei, or even ions suspended in a vacuum.
For technology behemoths and startups alike, the development of qubits is extremely important. A mastery over individual quantum degrees of freedom and their interactions will make coherent quantum computation accessible, in turn allowing them to tap into markets much faster. The competition is forcing hundreds of companies on the front to allocate more and more resources for research in quantum science and engineering.
Now, developing a qubit is essentially a hardware process. We know that qubits are environment-sensitive and are flippant with data. Thus, not only must the quantum systems be isolated from their environment, but the preservation of data must also be tackled (implying that decoherence times need to be long) and the observer (or readout) should be able to measure the state of the qubit (the lack of which will make computation impossible).
Additionally, being able to individually address those qubits, developing two-qubit interactions, and being able to initialize the system to a known state also pose as requirements. The main challenge of these implementations is reducing decoherence from parasitic environmental nodes while enhancing qubit coupling and entanglement.
Ion trap
The trapped ion method, one way to realize a quantum computer, involves the use of electromagnetic forces to trap charged particles (ions).
Ions held in electromagnetic traps make headway as productive qubit registers with long coherence times, a result of extremely weak ion-atmosphere interaction. Although the small distance between the localized ions makes it tough to address one singularly, lasers focussed at the particular qubit of interest have proven to manipulate the qubit’s orientation without disturbing the others.
For building a trapped-ion computer, the arrangement of ions can be maneuvered in the form of strings or using photons. In the string method, a string of ions is suspended between electrodes. Lasers can flip the ions’ magnetic orientations (each targeted ion is positioned in a way to also influence its like-charge or unlike-charge neighbor) which encodes the string data — a 1 can correspond to an up orientation, while a 0 to a down orientation.
The scalability of this system is debatable, so an alternative approach is to link the ions using their emitted photons, the frequencies of which depend on the magnetic orientation of the ions. A photon emitted by an ion in a superpositioned state would also be in superposition.
A light beam splitter is used to direct the photons in the same photodetector if they are in the same state, and to separate photodetectors if not. The case follows that experimenters cannot tell which photon originated from which ion, thus confirming the entanglement of the ions.
Superconducting qubits
The ability of certain materials to conduct electricity with practically zero resistance — superconductivity — is currently one of the leading ways to realize a quantum computing system.
Similar to other kinds of qubits, the creation of a solid-state (semiconductors and superconductors) qubit too requires an isolated two-level quantum system. Superconductors possess coherence superiority because the electrons encapsulate into certain pairs that form a solitary superfluid.
The amplitude and phase of the wavefunction that governs superfluid electrons are conjugate variables, essentially related to the main types of superconducting qubits: charge qubits (associated with amplitude) and flux qubits (associated with phase). A third type of qubit, called a phase qubit, is also associated with phase and like the other types, employs a Josephson energy junction.
The Josephson junction is an extremely thin (in the order of nanometers) layer of aluminum oxide sandwiched between two superconducting layers of aluminum, and partners with electronic components like capacitors to make up a superconducting qubit.
Though the strong coupling provided by this technology is benign, it is also detrimental to coherence. This is an implication to maintain a balance and prioritize according to the researcher’s need. It is yet to be seen if superconducting qubits can be scaled up successfully, but if at all, it is practically possible only after a couple of decades.
Semiconductors
Semiconducting qubits are as versatile in terms of quantum applications as they are in the current digital age. This particular class of materials has been the forebearer of transistor and chip advancements, and its strength can be utilized to make arrays of qubits that make up a quantum processor.
There are three main ways in which semiconductors are manufactured to construct quantum systems: heterostructures, SRT-embedded heterostructures, and quantum dot arrays.
Made of manifold semiconductor layers that have the same elemental anatomy and vary only in fractions, heterostructures have a barrier layer that caps it on the top, a series of doped gates, and qubits within the heterostructure layers. The qubits in these configurations are solitary but indistinct, and thus, adaptations built on this configuration are also aimed at furnishing discernibility.
Some semiconducting heterostructures can be trumped up with spin resonance transistors (SRT) lodged within the heterostructure to enable quantum entanglement between the qubits. It allows for the rooting of a logic gate that provides more precise control of inter-qubit linkage. The position of the qubit is in such a manner that we can determine the electronic state by the flow of current alone.
Semiconducting quantum dot arrays are placed on top of a semiconductor heterostructure. A two-by-two grid provides control to couple the qubits along divergently. In practice, a qubit architecture requires the qubits to be linked along two dimensions to allow for substantial, effective and larger scaling.
Extensive studies of gallium arsenide, silicon, and germanium have demonstrated two-qubit logic, but interconnecting larger numbers of qubits in semiconductor devices has remained tedious. Semiconductor quantum dots however possess the most benefits–lengthy coherence times, scaling potential, and compatibility with current semiconductor research.
Magnetic Resonance
Many principal results are coming from one of the primeval areas of quantum physics: NMR (nuclear magnetic resonance). We can use the nuclear spin of atoms within a molecule as a quantum mechanical system (especially a spin-1/2 nuclei). When such spin nuclei are subjected to a strong static magnetic field, they can orient in a way similar to a classic bar magnet. We could label those two states as a logical 0 or 1.
Every different nuclei’s spin-1/2 can be manipulated individually by applying a pulse tuned to its resonance frequency. We can tip the spin into any direction using this pulse, sanctioning control over the implementation of single gates. Two-qubit gates in this method require natural scalar coupling to encourage entanglement between the qubits.
Photonic
In the case of photonic quantum circuits, the qubits are based on photons and used as information carriers. Although photon-based quantum processors have not yet taken off compared to the other methods in this list, they are reliable candidates for quantum error-correction protocols and channeling quantum data over distances due to weak photon-atmosphere interaction. In fact, a single-qubit gate can be constructed using a regular beam-splitter.
However, the weak interaction makes the implementation of two-qubit gates difficult, because as previously established–coupling, nonlinearity and grids are integral for it. According to Appendix D: Other Approaches to Building Qubits of 2019 report Quantum Computing: Progress and Prospects, there are two methods to overcome this challenge:
- single-photon actions and assessments to implement a two-qubit gate, guided by linear optics quantum computing, and
- utilizing optical defects and quantum dots to induce strong inter-photon communication.
Typically, each qubit will be an individual photon, with its states corresponding to the photon’s polarization. The leads of this technique are numerous. With a significant scaling potential, such a system will also be relatively easy to operate at room temperature, apart from painless interfacing with quantum communication technologies.
What are some of the companies at the frontline of this battle?
D-wave, a Canadian quantum computing company, recently made headlines for the benchmark of a 5000-qubit quantum computer. D-wave’s strategies rely on the method of quantum annealing, a technique which is best used for operation’s research or optimization problems. Quantum annealing utilizes the process of quantum fluctuations to find the most logical solution to a problem that a user is solving.
While IBM’s Eagle–a quantum processor of 127 qubits–has also made its appearance, Google is yet to catch up after the release of its 2D array 54-qubit Sycamore processor last year.
Most companies like Intel, IBM, and Honeywell are on the frontline of this battle, working on quantum gating computers–yet another technique that expresses the interaction between qubits as gates (fun fact: quantum logic gates are reversible, unlike many classical gates).
Substantial research has been funded by these companies in an effort to scale their quantum systems, many being specific towards the listed methods and more. From trapped ion, (IonQ, Quantinuum), superconducting qubits (IBM, Google), quantum dot arrays (Fujitsu, Toshiba) to even nitrogen-vacancy qubits contained in a diamond, the new wave is filled with the buzz of everything quantum.
Bonus fact: studies have also attempted to implement quantum circuits using classical simulation with the help of tensors, thus opening up a new field of quantum machine learning. It may be too early to decide which technology is the most suitable, and the tiniest step ahead in this revolution may be decades away, but in the spirit of academia, it is essentially something to keep looking forward to.
Bibliography
https://www.cs.virginia.edu/~robins/Quantum_Computing_with_Ions.pf
https://physicsworld.com/a/superconducting-quantum-bits/
https://www.tudelft.nl/en/2021/tu-delft/semiconductor-qubits-scale-in-two-dimensions
https://www.nature.com/articles/s41586-021-03332-6
https://qutech.nl/2022/01/19/semiconductor-spin-qubits/
https://www.azom.com/article.aspx?ArticleID=17173
https://web.physics.ucsb.edu/~msteffen/nmrqc.htm
http://citeseerx.ist.psu.edu/viewdoc/downloaddoi=10.1.1.129.7114&rep=rep1&type=pdf
https://www.nap.edu/read/25196/chapter/1
