The Road to the Qubit
The Road to the Qubit
From Quantum Mechanics to the Physical Architectures Behind Modern Quantum Computers
Technical Series: Building the Qubit — Part 1
Quantum computing is often introduced through a deceptively simple statement:
A classical computer uses bits. A quantum computer uses qubits.
The statement is correct, but it hides the most important engineering question in the entire field:
What is a qubit physically?
Unlike a transistor, there is no single component called “the qubit.” A qubit is an engineered quantum system capable of providing two controllable quantum states that can be used to encode information.
That system might be:
a superconducting electrical circuit,
an isolated atomic ion,
a photon,
an electron confined in a semiconductor,
a neutral atom held by laser light,
a defect inside a diamond,
a microwave cavity,
or, in more experimental approaches, a topological quantum state.
The diversity is not accidental.
Each physical implementation solves some engineering problems while creating others.
This article is the first part of Building the Qubit, a technical series about how quantum computers are physically constructed. The goal is not to start with quantum algorithms, but with the hardware itself.
We will begin with the historical path that transformed the qubit from a theoretical abstraction into an engineering object.
1. From Bits to Quantum States
A classical bit has two possible values:
[
0 \quad \text{or} \quad 1
]
The physical implementation may differ—transistors, voltage levels, magnetic states, charge states—but the computational abstraction remains binary.
A quantum bit, or qubit, is described by a quantum state such as:
[
|\psi\rangle = \alpha|0\rangle + \beta|1\rangle
]
where (\alpha) and (\beta) are complex probability amplitudes satisfying:
[
|\alpha|^2 + |\beta|^2 = 1
]
This equation immediately introduces something that does not exist in an ordinary classical bit: coherent superposition.
The qubit is not simply a probabilistic bit.
The relative amplitude and phase between its states are part of the quantum information.
When measured in the computational basis, however, the result is classical:
[
|0\rangle
]
or
[
|1\rangle
]
That creates the first fundamental engineering problem.
A quantum processor must allow us to:
prepare a quantum state,
manipulate it with sufficient precision,
preserve coherence long enough to perform computation,
entangle it with other qubits,
and finally measure it.
The physical system chosen to perform these five tasks defines the qubit architecture.
2. When Quantum Mechanics Became a Computational Resource
Quantum mechanics was originally developed as a framework for explaining the behavior of matter and light at microscopic scales.
The idea of using quantum mechanics as a computational resource emerged much later.
During the 1980s, researchers such as Richard Feynman and David Deutsch helped establish the conceptual foundations of quantum computation.
The key insight was that quantum mechanics was not merely something a classical computer could simulate.
It could become the operating principle of the computer itself.
This distinction is fundamental.
A classical computer can simulate a quantum system, but the computational resources required to represent a general quantum state can grow exponentially with the number of quantum degrees of freedom.
A quantum computer instead manipulates the quantum state directly.
But this created a problem that theory alone could not solve:
How do you physically construct a controllable quantum system?
3. The First Experimental Path: NMR Qubits
One of the earliest experimental approaches to quantum information processing was based on Nuclear Magnetic Resonance (NMR).
NMR uses the spin states of atomic nuclei.
Certain nuclei possess spin-(\frac{1}{2}) states that can be manipulated using radio-frequency pulses while the system is placed inside a strong static magnetic field.
Those states can be used as:
[
|0\rangle
]
and
[
|1\rangle
]
The source material for this series describes early NMR systems based on liquid-state molecules such as chloroform and crotonic acid, where nuclear spins were manipulated through RF pulses.
NMR provided an important experimental platform because it allowed researchers to demonstrate quantum operations and early quantum algorithms.
It also offered relatively long coherence times.
However, it introduced a fundamental scalability problem.
Instead of individually manufacturing and controlling thousands of independent physical qubits, NMR commonly operated on ensembles of molecules.
As the number of qubits increased, the useful computational signal became increasingly difficult to extract because of thermal-state averaging and signal scaling.
The lesson was important:
A physical system can be an excellent experimental platform for quantum information without being an ideal architecture for a scalable quantum computer.
NMR therefore played an important historical role without becoming the dominant architecture for large-scale gate-based quantum computing.
4. The Transition from Natural Systems to Engineered Quantum Systems
The next major step was conceptually different.
Instead of searching for a naturally occurring system that could behave as a qubit, researchers began deliberately engineering physical systems with quantum-mechanical energy levels suitable for computation.
This is where superconducting qubits became particularly important.
The idea was extraordinary from a classical engineering perspective:
Take an electrical circuit and make it behave like an atom.
5. Superconducting Qubits: The Artificial Atom
A superconducting qubit is typically based on a nonlinear electrical circuit containing a Josephson junction.
A Josephson junction consists of two superconducting regions separated by an extremely thin insulating barrier.
The junction behaves as a nonlinear, non-dissipative element.
That nonlinearity is crucial.
A simple harmonic oscillator has approximately equally spaced energy levels:
[
E_0, E_1, E_2, E_3, ...
]
If the energy spacing is perfectly uniform, selectively addressing only two levels becomes difficult because the same drive can unintentionally excite higher states.
The Josephson junction introduces anharmonicity.
This allows the lowest two levels to be isolated and used as:
[
|0\rangle,\ |1\rangle
]
The resulting circuit behaves as an engineered or synthetic atom.
The source material describes superconducting qubits as macroscopic LC circuits whose Josephson junction creates the required nonlinear energy spectrum.
This is one of the most important conceptual transitions in quantum hardware:
The qubit does not need to be a naturally occurring microscopic particle. It can be an engineered quantum system.
6. Fabricating the Qubit
Superconducting processors are manufactured using techniques derived from semiconductor fabrication.
Typical structures are fabricated on substrates such as:
silicon,
sapphire.
Thin superconducting films may use materials including:
aluminum,
niobium,
tantalum.
Josephson junctions are created by forming a very thin insulating barrier between superconducting layers.
The resulting circuits must then operate at extremely low temperatures.
The source material places superconducting qubit operation in the millikelvin regime, with dilution refrigerators providing the required thermal environment.
Why such extreme cooling?
Because quantum information is extraordinarily sensitive to unwanted energy and environmental noise.
A quantum processor must operate in a regime where:
thermal excitations are strongly suppressed,
superconductivity is maintained,
electromagnetic interference is minimized,
and unwanted coupling to the environment is controlled.
This leads to one of the defining characteristics of superconducting quantum computing:
The processor may be a lithographically fabricated chip, but the chip is only one component of a much larger cryogenic system.
7. The Other Path: Trapped-Ion Qubits
Superconducting systems represent one strategy:
engineer the quantum system.
Trapped-ion computing takes almost the opposite approach:
use the atom itself.
An ion is an electrically charged atom confined inside a vacuum system using electromagnetic fields.
A common architecture uses a Paul trap, where oscillating radio-frequency electric fields generate an effective potential that confines the ions.
The ions are then laser cooled to reduce their motional energy.
The qubit is encoded in two internal states of the ion.
These may be:
hyperfine states,
Zeeman states,
or optical transitions.
The source material identifies species such as (^{171}\mathrm{Yb}^+) and (^{40}\mathrm{Ca}^+), and describes ion confinement in ultra-high vacuum, laser cooling and optical or microwave control.
The physical system therefore looks very different from a superconducting processor.
Instead of:
chip → circuit → microwave pulse
we have:
atom → electromagnetic trap → laser cooling → optical/microwave control
8. Why Trapped Ions Are Attractive
One of the major advantages of trapped-ion systems is the quality of the physical qubit.
Atoms of the same isotope are fundamentally identical.
The quantum states are naturally defined by atomic physics rather than by lithographic fabrication.
The source material highlights several important properties:
long coherence times,
very high gate fidelities,
high measurement fidelity,
and all-to-all connectivity in sufficiently small ion crystals.
But the architecture introduces a different scaling problem.
As more ions are placed in the same system, their collective motional modes become increasingly complex.
The optical control system also becomes more difficult to scale.
Trapped-ion gates are typically slower than superconducting operations, commonly operating in the microsecond-to-millisecond regime rather than the nanosecond regime associated with superconducting circuits.
Again, the same engineering pattern appears:
Improving one dimension of the problem creates pressure somewhere else.
9. Photonic Qubits: Computing with Light
The next architecture asks a completely different question:
Does the quantum information need to be stored in matter at all?
Photonic quantum computing uses photons as the physical carriers of quantum information.
A qubit can be encoded in:
polarization,
optical path,
time-bin,
photon-number-related states,
or continuous-variable optical states.
For example:
[
|H\rangle
]
and
[
|V\rangle
]
can encode two polarization states.
Photons can be generated through several mechanisms, including:
spontaneous parametric down-conversion,
quantum dots,
integrated optical sources.
They can then be manipulated using:
beam splitters,
phase shifters,
interferometers,
optical waveguides.
These mechanisms are described in the source material's photonic-qubit overview.
10. The Photonic Advantage—and Its Fundamental Problem
Photons have exceptionally low interaction with the environment during propagation.
That gives photonic systems a major advantage:
quantum information can travel.
This makes photonic qubits particularly attractive for:
quantum communication,
quantum networking,
distributed quantum computing.
But the same weak interaction that helps photons preserve information creates a problem inside the processor.
Photons do not naturally interact strongly with each other.
A quantum computer needs entangling operations.
Implementing deterministic two-qubit interactions with photons is therefore difficult.
Another major error mechanism is photon loss.
If the photon disappears, the information carried by that photon disappears with it.
The source material identifies photon loss and the difficulty of deterministic two-qubit gates as two of the defining challenges of photonic quantum computing.
This illustrates an important principle:
A good qubit is not necessarily a good computing architecture.
The qubit must also provide a practical way to implement the operations required by a quantum computer.
11. Semiconductor Spin Qubits
The semiconductor approach takes a different route again.
Instead of using the charge of an electron, the qubit can be encoded in its spin.
Conceptually:
[
|\uparrow\rangle \rightarrow |0\rangle
]
[
|\downarrow\rangle \rightarrow |1\rangle
]
The electron can be confined inside a quantum dot.
The quantum dot is effectively a very small region in which the electron is spatially localized.
Materials used in these architectures include:
silicon,
silicon/germanium heterostructures,
silicon MOS structures.
Isotopically purified (^{28}\mathrm{Si}) can also be used to reduce magnetic noise associated with nuclear spins.
The source material emphasizes the extremely small physical footprint of these qubits and their potential compatibility with CMOS-style semiconductor manufacturing.
12. The Semiconductor Engineering Challenge
The idea of building quantum processors using semiconductor manufacturing is attractive because the semiconductor industry already has decades of experience producing structures at nanometer scales.
However, quantum devices impose requirements beyond ordinary CMOS engineering.
Spin states can be sensitive to:
charge noise,
magnetic impurities,
material disorder,
device-to-device variability.
Readout is also more difficult than simply measuring a conventional voltage.
A common strategy is spin-to-charge conversion, where the spin state is translated into a charge configuration that can be detected electrically.
The source material describes techniques involving spin-dependent tunneling and sensitive charge sensors such as SETs, QPCs and radio-frequency reflectometry.
The result is another important trade-off:
Silicon offers extraordinary manufacturing maturity, but quantum coherence imposes requirements that classical semiconductor manufacturing did not have to solve.
13. Neutral Atoms
Neutral-atom quantum computing combines some of the conceptual advantages of trapped atoms with a very different architecture.
Instead of trapping charged ions, the system manipulates neutral atoms.
Atoms are cooled and loaded into arrays formed by highly focused laser beams known as optical tweezers.
Common atomic species include:
rubidium,
cesium,
strontium,
and other suitable atoms.
The qubit is generally encoded in long-lived atomic states.
Two-qubit interactions can then be implemented through Rydberg blockade.
When an atom is excited into a highly energetic Rydberg state, it changes the energy landscape of nearby atoms.
That interaction can prevent simultaneous excitation and generate the controlled interaction needed for entangling gates.
The source material describes neutral-atom processors using optical tweezers, Rydberg interactions and dynamic two-dimensional or three-dimensional arrays.
14. A Quantum Processor That Can Rearrange Itself
This is one of the most interesting properties of neutral-atom systems.
A conventional solid-state processor typically has a fixed physical geometry.
The connections between components are determined during fabrication.
Neutral atoms are different.
Optical systems can move the atoms.
Spatial light modulators and acousto-optic systems can create and reposition optical traps.
That means the physical geometry of the processor can be dynamically changed.
The source material therefore treats neutral-atom architectures as highly reconfigurable systems capable of supporting large two-dimensional and three-dimensional arrays.
This characteristic will become especially important later in this series when we examine quantum error-correction architectures such as qLDPC codes.
15. Defect Qubits: When the Material Itself Stores the Quantum State
Another fundamentally different architecture is based on crystal defects.
One of the best-known examples is the nitrogen-vacancy, or NV center, in diamond.
An NV center consists of a nitrogen atom adjacent to a vacancy in the diamond lattice.
The defect creates quantum states that can be associated with an electron spin.
Nearby nuclear spins can also act as quantum memories.
The system can be:
initialized optically,
manipulated with microwave pulses,
and read optically.
The source material notes that NV centers can be produced through nitrogen implantation followed by annealing or during chemical vapor deposition growth.
This creates an interesting hybrid:
the crystal is both the material and the host of the quantum system.
NV centers are particularly interesting for:
quantum sensing,
quantum networking,
quantum memories,
and nanoscale quantum devices.
16. Bosonic Qubits: Beyond the Two-Level System
Up to this point, most architectures have followed an intuitive model:
one physical system → two quantum states → one physical qubit
Bosonic approaches challenge that assumption.
A quantum harmonic oscillator has many states:
[
|0\rangle, |1\rangle, |2\rangle, ..., |n\rangle
]
Instead of restricting the physical system to two levels, a bosonic architecture can encode quantum information in the larger state space of the oscillator.
One important implementation is the cat qubit.
Cat qubits use superpositions of coherent states:
[
|\alpha\rangle
]
and
[
|-\alpha\rangle
]
inside a microwave cavity.
The source material describes this as a fundamentally different approach to error management, where a single bosonic mode can provide a much larger Hilbert space than a conventional two-level physical qubit.
This architecture is particularly interesting because it introduces hardware-level error bias.
Through engineered nonlinear processes, certain error mechanisms can be strongly suppressed while others remain dominant.
That can reduce the complexity of subsequent error correction.
In other words:
Instead of correcting every type of error equally, engineer the hardware so that one class of errors becomes much less likely.
That concept will be the subject of a dedicated article later in this series.
17. Topological Qubits
Finally, we reach one of the most ambitious approaches.
Topological quantum computing attempts to store quantum information in non-local properties of a physical system.
Theoretical implementations involve concepts such as:
Majorana zero modes,
non-Abelian anyons,
topological superconductivity,
braiding operations.
The fundamental attraction is intrinsic protection.
If information is encoded non-locally, local perturbations should have greater difficulty corrupting the quantum state.
The source material identifies hybrid semiconductor-superconductor structures—including materials such as indium arsenide and aluminum—as candidate systems for realizing Majorana-based states.
However, this is where technical communication needs to become especially careful.
Topological quantum computing remains an experimental research direction.
The existence of a promising architecture is not equivalent to having demonstrated a fully functional topological qubit.
The material explicitly identifies topological qubits as an experimental, high-risk/high-reward approach whose scientific validation remains an active issue.
That distinction will be important throughout this series.
18. A Taxonomy of Physical Qubits
At this point we can build a useful engineering taxonomy.
| Qubit architecture | Physical carrier | Typical control mechanism | Major challenge |
|---|---|---|---|
| NMR | Nuclear spin | RF pulses | Scalability |
| Superconducting | Josephson-junction circuit | Microwaves | Cryogenics and decoherence |
| Trapped ion | Atomic ion | Lasers / microwaves | Optical complexity and scaling |
| Photonic | Photon | Optical components | Photon loss and interactions |
| Semiconductor spin | Electron/hole/nuclear spin | Microwave/electric control | Noise and variability |
| Neutral atom | Atomic states | Lasers / microwaves | Atom loss and optical complexity |
| NV center | Defect spin in diamond | Optical + microwave | Scaling and photon collection |
| Bosonic/cat | Cavity mode | Microwave control | Cavity engineering and error management |
| Rare-earth | Ion in crystal | Optical/spin control | Integration and individual addressing |
| Molecular | Molecular spin state | Electromagnetic control | Addressability and integration |
| Topological | Majorana/topological state | Experimental | Demonstration and materials |
This table illustrates why “qubit” is an abstraction rather than a physical component.
The information model can be the same while the physical implementation changes dramatically.
19. The Five Fundamental Engineering Questions
Every qubit architecture eventually has to answer the same five questions.
1. Initialization
How do we prepare the system in a known quantum state?
For example:
[
|0\rangle
]
2. Control
How do we apply precise operations to the quantum state?
This may involve:
microwaves,
RF pulses,
optical pulses,
electric fields,
magnetic fields,
or engineered interactions.
3. Entanglement
How do we make two qubits interact strongly enough to implement a useful two-qubit gate?
This may happen through:
electromagnetic coupling,
Coulomb interaction,
photon interference,
exchange interactions,
Rydberg blockade,
cavity-mediated interactions,
or other mechanisms.
4. Isolation
How do we prevent the environment from destroying the quantum state?
This determines the architecture's:
coherence time
and strongly influences its error rates.
5. Measurement
How do we convert the quantum state into classical information?
The physical implementation varies dramatically.
Superconducting systems can use microwave resonators.
Trapped ions and neutral atoms can use state-dependent fluorescence.
Photonic systems can use single-photon detectors.
Spin qubits can use spin-to-charge conversion.
The source material provides a broad comparison of these readout mechanisms.
20. The Real Bottleneck: Scaling
At the beginning of quantum computing, demonstrating a few controllable qubits was a major scientific achievement.
Today, the challenge is fundamentally different.
We need systems containing:
many physical qubits,
sufficiently low physical error rates,
reliable gates,
high-fidelity readout,
scalable control electronics,
and eventually quantum error correction.
A useful physical qubit is therefore only the first layer of the problem.
The system must evolve from:
[
\text{Physical qubit}
]
to:
[
\text{Many physical qubits}
]
to:
[
\text{Logical qubit}
]
and ultimately:
[
\text{Fault-tolerant quantum computer}
]
This is where quantum error correction becomes unavoidable.
The source material emphasizes that the need for error correction varies by architecture, but that large-scale fault-tolerant computing ultimately requires QEC across the major platforms.
21. Why Qubit Count Is Not Enough
One of the biggest misconceptions in quantum computing is that the number of physical qubits alone determines processor capability.
It does not.
A meaningful quantum processor needs to be evaluated through multiple parameters:
Qubit count
How many physical qubits exist?
Coherence
How long does quantum information remain usable?
Gate fidelity
How accurately can operations be performed?
Gate speed
How quickly can operations execute?
Connectivity
Which qubits can interact directly?
Measurement fidelity
How accurately can the resulting state be read?
Error model
What types of errors dominate?
Error-correction overhead
How many physical resources are required to create reliable logical qubits?
This is why two processors with similar physical-qubit counts can have completely different computational capabilities.
22. The Emerging Direction: From Physical Qubits to Logical Qubits
The industry is therefore moving toward a more meaningful unit of progress:
the logical qubit.
A physical qubit is inherently noisy.
A logical qubit is constructed by encoding quantum information across multiple physical resources and using quantum error correction to detect and suppress errors.
Different qubit architectures approach this challenge differently.
Superconducting and semiconductor-spin architectures can require substantial active error correction because their physical coherence and error characteristics create significant overhead.
Trapped-ion and neutral-atom systems benefit from high physical fidelities and long coherence times, although they still require error correction for large-scale fault tolerance.
Bosonic and topological approaches attempt to suppress certain errors directly at the hardware level.
This is the beginning of a new stage in quantum computing.
The question is no longer simply:
How many qubits can we build?
It is increasingly:
How many reliable logical qubits can we build?
23. The Quantum Computer Is Not One Technology
The history of the qubit therefore tells us something broader about the quantum-computing industry.
There is no universally superior physical implementation.
Instead, there is a set of competing architectures making different engineering trade-offs.
Superconducting systems prioritize speed and lithographic integration.
Trapped ions prioritize fidelity and coherence.
Photonic systems prioritize propagation and networking.
Neutral atoms prioritize scalability and reconfigurability.
Semiconductor spin systems prioritize density and compatibility with semiconductor manufacturing.
Defect-based systems exploit specialized materials with useful optical and spin properties.
Bosonic architectures attempt to move part of error correction into the hardware.
Topological approaches attempt to make the quantum information intrinsically resistant to local disturbances.
Rare-earth and molecular platforms explore additional combinations of long-lived quantum states, optical interfaces and material engineering.
The result is not one race toward one architecture.
It is a race between different physical strategies for solving the same fundamental problem.
24. What Comes Next?
We now have the map.
The rest of this series will examine each major architecture in detail.
For every platform, we will follow the same path:
Physical system → Materials → Fabrication → Qubit encoding → Initialization → Control → Entanglement → Readout → Error mechanisms → Error correction → Processor architectures → Advantages → Limitations
The next stop is the technology that arguably transformed quantum computing from a laboratory experiment into a rapidly industrialized hardware discipline.
The Artificial Atom
How Superconducting Circuits Become Qubits
We will begin with the Josephson junction itself.
From there we will follow the complete chain:
[
\text{Material}
\rightarrow
\text{Josephson junction}
\rightarrow
\text{anharmonic circuit}
\rightarrow
|0\rangle,|1\rangle
\rightarrow
\text{microwave control}
\rightarrow
\text{two-qubit gate}
\rightarrow
\text{readout}
\rightarrow
\text{processor}
]
Because the most useful question is no longer simply:
What is a qubit?
It is:
How do you manufacture one?
Comments
Post a Comment