Making Photonic Qubits

A couple of earlier posts here covered superconducting and trapped ion qubits. Every approach hits the same wall: scaling. Photonic qubits encode information on individual photons, which largely ignore their surroundings and pick up less noise as a result. Their specific scaling bottleneck is producing single photons reliably.

Think of quantum dots like Dots candy. Each one is a small,
self-contained unit that delivers exactly one thing when you bite into it.

A photonic qubit stores quantum information on a single particle of light. Simple to say. Hard to build.

Choosing Your Encoding

We've discussed encoding in earlier posts. Pick a property of a photon to represent 0 and 1. Polarization is the most common choice: horizontal = 0, vertical = 1. Because in quantum mechanics, a single photon can be in both states simultaneously.

Other options include arrival time (time-bin), path through a chip (path encoding), or photon presence or absence. Each trades off differently depending on what you need downstream.

Creating Single Photons

The first real problem is producing single photons on demand.

• Spontaneous Parametric Down-Conversion (SPDC): Shine a laser into a special crystal. Occasionally one laser photon splits into two. Detect one, and you know the other exists. The success rate is roughly 1 to 10 out of every 100 laser pulses. Most pulses produce nothing.
• Quantum Dots: These are tiny semiconductor structures that emit exactly one photon per excitation. The best versions now exceed 99% reliability. The cost is that they only work near absolute zero, around -452°F. The photons travel fine at room temperature, but the sources need dilution refrigerators.

Manipulation and Interaction

Single-qubit operations on polarization qubits are straightforward. Wave plates rotate polarization by precise amounts. This is well understood and cheap.

Two-qubit operations are the hard part. Photons normally ignore each other completely. In 2001, three physicists showed you could use measurements and fast switching to simulate an interaction between photons. It works, but the basic two-qubit gate historically succeeded about 1 time in 16.

Scaling with Fusion

A newer approach skips reliable two-photon interaction entirely. You create small entangled photon groups, then merge them with simple optical components. When a merge fails, error correction absorbs the loss. PsiQuantum and others are pursuing this fusion-based architecture to reach millions of qubits.

Manufacturing

Photonic chips use the same fabrication processes and factories as conventional silicon chips. No custom facilities are required. One catch is that while the chips are silicon, reading results often requires superconducting detectors. Photons are stable at room temperature, but a full-scale photonic quantum computer will still likely sit inside a cryogenic system to keep the detectors and quantum dot sources cold enough to function.

Photonic qubits have a manufacturing path that superconducting and trapped ion systems do not: existing silicon fabs. The open question is whether photonic qubits can be mass-produced at the volumes a million-qubit machine requires.

From the Classroom to the Job Site: The Rise of the Applied Engineer

For years, the gold standard for a Bachelor of Science in Engineering was "theoretical mastery." We taught our students to live in the world of complex differentials and perfect simulations. But as I look at the 2026 labor market and the "Great Trade Shift" recently detailed by the Washington Post it is clear that the "Desk Engineer" is a disappearing species.

In its place, we are seeing the rise of the Applied Engineer.

The Pedagogy of the Physical

In my conversations with industry partners and alumni, the feedback is consistent: they don’t just need people who can solve equations; they need people who can solve problems where the digital meets the physical. This is the "Applied" mindset. It’s the realization that an engineering degree is not a pass to avoid the "dirty" work, but a license to master it.

The Applied Engineer is a hybrid professional. They possess the rigorous academic foundation of a BS degree, but they have paired it with the tactical fluency of a master tradesperson. For example:

• Mechanical Engineers are now expected to understand the "feel" of metallurgy and the constraints of a 5-axis CNC machine.

• Civil Engineers are bridging the gap between CAD models and the real-world variables of a construction site.

• Electrical Engineers are increasingly required to move from the circuit simulation to the physical integration of high-voltage hardware and PLC systems.

Why "Applied" is the New "Essential"

This shift is a direct response to two forces: the AI Revolution and the Infrastructure Renaissance.

While generative AI is becoming exceptionally good at creating optimized designs and writing code, it cannot go into a 1 gigawatt (GW)  data center and troubleshoot a hardware failure. It cannot oversee the physical reshoring of a semiconductor fab. The market is placing a historic premium on "un-automatable" skills - the kind that require a human being to bridge the gap between a theoretical model and a functioning machine.

A Call to My Peers and My Students

As educators, we have to ask: are we graduating engineers who are afraid to pick up a tool? The most successful graduates of 2026 are those who respect the craft of the technician as much as the math of the scientist.

We are moving toward an economy that rewards "Technical Literacy" over "General Management." Whether you are holding a wrench or a scientific calculator, your value is defined by your ability to make things work in the physical world. For the modern engineer, the path to career resilience isn't found in more abstraction - it’s found in the application.

Making Trapped Ion Qubits

In an earlier post I discussed how qubits are made using the Superconducting Qubit method. In this post I discuss the Trapped Ion Qubit method.

What Is a Trapped Ion Qubit?

A quantum computer needs a basic unit of information called a qubit (short for "quantum bit"). A classical bit is either 0 or 1. A qubit can also be in a combination of 0 and 1 at the same time, a property called superposition. When multiple qubits interact, they can become entangled, meaning the state of one depends on the state of another, no matter how far apart they are. These two properties, superposition and entanglement, give quantum computers their potential power.

A trapped ion qubit uses a single atom that has been stripped of one electron (making it a positively charged ion) as the qubit. The ion is held in place by electric fields inside a vacuum chamber and controlled with laser beams. Two specific energy levels inside the atom serve as the 0 and 1 states. Companies building quantum computers this way include IonQ, Quantinuum (Honeywell), Alpine Quantum Technologies and Oxford Ionics.

Why Use Atoms?

Every atom of the same type is identical. Every ytterbium-171 ion in the universe has exactly the same internal structure. There is nothing to manufacture; nature provides the qubit. The competing approach, superconducting circuits (used by IBM and Google), builds qubits out of tiny electrical components on a chip. These fabricated qubits always have small manufacturing differences that require individual tuning.

Atoms hold quantum information for a long time. A qubit is useful only as long as it maintains its quantum state, a property measured by "coherence time." Trapped ion qubits hold their state for 10 seconds to several minutes. Superconducting qubits typically last about 100 to 500 microseconds (millionths of a second), roughly 10,000 times shorter. Longer coherence time means more operations can be performed before errors accumulate.

Common ion species. The atoms most often used are ytterbium-171 (¹⁷¹Yb⁺, used by IonQ), barium-137 (¹³⁷Ba⁺, used in newer IonQ systems), and calcium-40 (⁴⁰Ca⁺, used by AQT and Oxford Ionics). Each species has different laser wavelength requirements, which affects the engineering complexity of the system.

How the Trap Works

A basic law of physics (Earnshaw's theorem) says you cannot hold a charged particle in place using only constant electric fields. Trapped ion systems get around this with a device called a Paul trap, which uses rapidly oscillating electric fields.

The Paul trap. Electrodes surrounding the ion produce a radiofrequency (RF) electric field that flips direction millions of times per second (typically 10 to 100 million times). The ion cannot keep up with the rapid switching, so it experiences an average force that pushes it toward the center of the trap. Additional constant-voltage electrodes keep the ion from drifting along the length of the trap. The result: the ion floats in empty space, about 50 to 100 micrometers (roughly the width of a human hair) above the electrode surface.

Modern trap chips. Early traps used hand-assembled metal rods. Today's systems use microfabricated chips, similar to computer chips, with all electrodes printed on a flat surface using standard semiconductor manufacturing. This allows many trapping zones on one chip, so ions can be moved around for different computations.

The vacuum. The ion must not collide with air molecules, which would knock it out of its quantum state or eject it from the trap. The chamber is pumped down to a pressure of about 10^-11 torr, roughly one hundred-billionth of atmospheric pressure. At this pressure, a stray gas molecule would hit the ion only once every few hours. Achieving this requires baking the chamber at high temperature for days and using specialized pumps.

Controlling the Ion with Lasers

Lasers do all the work: cooling the ion down, setting its initial state, performing computations, and reading the result.

Cooling. After the ion is loaded into the trap, it vibrates too much for precise control. A laser tuned to a specific frequency repeatedly hits the ion with photons (particles of light) in a way that slows it down, similar to how throwing tennis balls at a moving shopping cart from the front would slow it. This "laser cooling" brings the ion nearly to a standstill, at a temperature below one-thousandth of a degree above absolute zero.

Setting the starting state. Before a computation begins, the qubit must be set to a known state (0). A technique called optical pumping uses a laser to drive the ion into the 0 state with over 99.9% reliability.

Performing operations (gates). To change a single qubit's state (a single-qubit gate), a pair of laser beams is aimed at the ion. By adjusting the laser frequency, intensity, and duration, the ion can be rotated between its 0 and 1 states in any desired proportion. These operations succeed better than 99.99% of the time.

To entangle two qubits (a two-qubit gate), laser beams are aimed at both ions simultaneously. The beams couple the ions' internal states through their shared vibration in the trap, temporarily exchanging energy through the motion and then returning the vibrational energy to its starting point while leaving the qubits entangled. This is called the Mølmer-Sørensen gate. It works between any pair of ions in the chain, giving trapped ions "all-to-all" connectivity. Superconducting qubits can only interact directly with their immediate neighbors on the chip. Two-qubit gate success rates are 99.5% to 99.9%.

Reading the result. A detection laser is aimed at the ion. If the ion is in state 0, it glows brightly (scattering thousands of photons). If it is in state 1, it stays dark. A camera or photon detector records which ions are bright and which are dark. This measurement takes 100 to 500 microseconds and is correct over 99.5% of the time.

The laser challenge. Each laser must be held at precisely the right frequency: a drift of just 10 kHz (ten thousand cycles per second out of trillions) can cause a computation to fail. A system with N qubits needs roughly 3 to 5 laser beams per qubit, each individually controlled. A 32-qubit system requires over 100 beams. Scaling to hundreds of qubits will require miniaturized optics built directly onto the trap chip.

How Trapped Ions Compare to Superconducting Qubits

Feature	Trapped Ion	Superconducting
What is the qubit?	A single atom	A fabricated electrical circuit
How long does it hold state?	10 seconds to minutes	100 to 500 microseconds
Gate accuracy (1 qubit)	> 99.99%	> 99.9%
Gate accuracy (2 qubit)	99.5 to 99.9%	99.0 to 99.5%
How fast are gates?	1 to 200 microseconds	10 to 100 nanoseconds (1,000x faster)
Which qubits can talk?	Any pair (all-to-all)	Only neighbors on the chip
Biggest challenge	Laser complexity per qubit	Manufacturing consistency

 

Scaling Up

Today's trapped ion processors have 20 to 56 qubits in a single chain. Adding more ions to the chain makes the system harder to control because the ions share an increasingly complex set of vibrational patterns. Beyond about 40 ions, gate operations slow down and lose accuracy.

Splitting into zones. The leading solution is the quantum charge-coupled device (QCCD) architecture. Instead of one long chain, the trap chip has multiple small zones (5 to 10 ions each) connected by transport channels. Voltage adjustments on the electrodes shuttle ions from one zone to another in 10 to 100 microseconds, with very high reliability (over 99.99% per shuttle). Quantinuum's H-series processors use this approach.

Connecting multiple chips. To reach thousands of qubits, separate trap chips must be linked together. The current method uses photons (light particles). An ion on one chip emits a photon that carries its quantum state through a fiber optic cable to another chip. When the photons from two chips are compared at a detector, the two distant ions become entangled. The success rate per attempt is low (about 0.01% to 1%), so this process must be repeated many times. Current linking speeds are 1 to 100 connections per second. For practical large-scale computing, this needs to reach thousands per second, a target for 2027 to 2030.

Where Errors Come From

Vibration noise from the trap surface. The electrode surfaces emit small, random electric fields that cause the ion to vibrate. This "motional heating" is the largest error source for two-qubit gates, because those gates work by coupling through the ion's motion. Heating rates depend strongly on how close the ion is to the surface. Cleaning or polishing the electrodes, or cooling the entire trap to very low temperatures (4 to 77 Kelvin), reduces this noise by 10 to 100 times.

Laser light hitting the wrong ion. Ions in a chain are only about 5 micrometers apart. When a laser targets one ion, a small amount of light spills onto its neighbors (called crosstalk). Careful beam focusing and pulse design reduce this to acceptable levels.

Toward error correction. Every physical qubit makes occasional errors. Quantum error correction uses groups of physical qubits to form a single "logical qubit" that can detect and fix its own errors, similar in concept to how classical computers use checksums. This only works if the physical error rate is low enough, typically below about 0.1%. Trapped ions are at or near this threshold. In 2023, Quantinuum demonstrated the first instance where a logical qubit made from trapped ions actually had a lower error rate than any of its physical components, a milestone for the field.

Current Systems (2024-2025)

•       Quantinuum H2: 56 qubits, QCCD architecture, 99.8% two-qubit gate accuracy. First real-time error correction demonstrations.

•       IonQ Forte Enterprise: 36 qubits, available through cloud services (AWS, Azure, GCP), 99.5% two-qubit gate accuracy.

•       Alpine Quantum Technologies: 24 qubits in a compact, rack-mounted system designed for European deployment.

•       Oxford Ionics: Developing microwave-based control (instead of lasers) to reduce optical complexity.

What Comes Next

Trapped ions have the highest gate accuracy and longest coherence times of any qubit technology. The tradeoff is slower gate speeds and the difficulty of managing many precision laser beams. The path to larger systems depends on three developments: miniaturized optics built into the trap chip to replace bulky laboratory lasers, faster photonic links to connect multiple trap modules, and better understanding of the surface noise that limits gate accuracy.

If these engineering problems are solved, systems of 1,000 to 10,000 qubits could be built within the next five years. Because trapped ion gates are already so accurate, fewer physical qubits are needed per error-corrected logical qubit compared to lower-fidelity platforms. That efficiency advantage becomes increasingly important as systems grow.

Physics Works, Engineering Makes It Work

My post Saturday on Making Quantum Supercomputing Qubits received an interesting (and excellent) comment from a reader on LinkedIn. 

Summarizing, the commenter argued quantum computing metrics (coherence times, yields, precision) are measured on idle qubits, not under real computational loads with circuits, crosstalk, and error correction. Once you add gates and routing, stability requirements exceed current improvements by orders of magnitude. And, after decades, there's still no error-corrected logical qubit running useful work. 

My comment back “Agree. The physics works. Scaling remains unsolved."This got me thinking this morning. STEM students often ask about major differences. One of the most common: "What separates physics from engineering?" Let's try to answer that using quantum as an example.

Physics discovers principles. Engineering builds systems that exploit those principles at scale. The gap between the two defines most hard technology problems.

Take quantum computing. Physicists proved you can trap ions, manipulate superconducting circuits, or use topological states to create qubits. The math works. Lab demonstrations show quantum advantage for specific problems. Physics is satisfied.

Engineering asks different questions. How do you manufacture 1,000 identical qubits when each one requires nanometer precision? How do you cool them to 15 millikelvin and hold that temperature while running computations? How do you shield them from electromagnetic interference in a data center? How do you get signals in and out without destroying coherence? How do you do all this reliably, repeatedly, and affordably?

Physicists build one qubit that works beautifully under perfect conditions. Engineers must build systems where hundreds of qubits work together under real conditions. Every quantum computing company today is struggling to bridge this gap.

The physicist optimizes for understanding. The engineer optimizes for constraints: cost, yield, thermal management, signal integrity, maintenance, supply chains. A physics experiment might use custom components that cost $500,000 and require manual calibration. An engineering solution needs off-the-shelf parts and automated processes.

You see this everywhere in technology. Physicists demonstrated photovoltaic effects in 1839. Engineers spent 150 years making solar panels cheap enough to matter. Physicists proved nuclear fusion in the 1930s. Engineers still cannot build a reactor that produces more energy than it consumes at useful scale.

The difference is not just scale though. It is thinking about failure modes, manufacturing tolerances, quality control, serviceability, and integration with existing infrastructure. Physics assumes ideal conditions. Engineering assumes Murphy's Law.

This creates tension. Physicists get frustrated when engineers say "that will never work in production." Engineers get frustrated when physicists dismiss practical constraints as details. Both are wrong. You need physics to know what is possible. You need engineering to make it real.

Quantum computing sits in this gap right now. The physics is spectacular. The engineering is brutal. Whoever solves the engineering problem first wins the market. That is always how it works. 

Making Quantum Superconducting Qubits

I’ve written about qubits in the past - the basic unit of quantum information that can exist in a
superposition of both 0 and 1 states simultaneously until measured, unlike a classical bit which is always either 0 or 1. Let’s take a closer look on how a qubit can be made - there are three ways:

·      Superconducting Qubits

·      Trapped Ion Qubits

·      Photonic Qubits

Superconducting qubits are what IBM and Google use and I’ll cover in this post. They're tiny electrical circuits that only work at temperatures colder than deep space.

What Makes Them Quantum

At room temperature, these are just pieces of metal. Cool them to 15 millikelvin and they become superconductors. Electricity flows without resistance. Electrons move in perfect sync, acting like one quantum wave instead of individual particles.

The circuit has specific energy levels, like rungs on a ladder. Ground state is 0, first excited state is 1. The quantum effect lets the circuit be in both states at once until you measure it.

The Josephson Junction

The heart of a superconducting qubit is the Josephson junction. Two pieces of aluminum separated by an insulating barrier 1 to 2 nanometers thick. About 10 atoms wide.

At cryogenic temperatures, electrons quantum tunnel through that barrier even though classical physics says they can't. This creates a nonlinear inductance. Combined with a capacitor, you get unequally spaced energy levels.

Why does that matter? A regular circuit has evenly spaced levels. You can't use it as a qubit because when you try to flip between 0 and 1, you accidentally excite higher levels too. The Josephson junction creates anharmonicity. The gap between 0 and 1 differs from the gap between 1 and 2. This lets you address just the first two levels with microwave pulses.

The Fabrication Process

Start with silicon: High purity silicon wafer, extremely flat and clean. Any contamination creates
defects.

Deposit aluminum: Vacuum chamber, molecular beam epitaxy. Deposit 100 to 200 nanometers of aluminum in ultra-high vacuum to prevent oxidation.

Pattern the circuit: Photolithography for the basic shapes. Electron beam lithography for the junction because you need nanometer precision. An electron beam writes the pattern point by point.

Create the junction: The Dolan bridge technique works well. Evaporate aluminum at an angle, deposit the oxide barrier, evaporate more aluminum at a different angle. The two layers overlap slightly with oxide between them. The overlap area is your junction.

Getting the oxide thickness right is critical. Too thick and electrons can't tunnel. Too thin and you get leakage. You're aiming for 1 to 2 nanometers with sub-nanometer precision.

Add control circuitry: Microwave transmission lines, coupling capacitors, and resonators. The readout resonator is a microwave cavity whose frequency shifts depending on qubit state. Send in a microwave pulse, measure the reflected signal. The tiny frequency shift tells you if the qubit is 0 or 1.

Why It's Hard

Junction uniformity: A 5% variation in junction area changes qubit frequency by hundreds of megahertz. Hitting the right target across a whole chip is brutal.

Material defects: Impurities create two-level systems that absorb energy and cause decoherence. You need ultra-pure materials and ultra-clean fabrication. Cosmic rays passing through can disrupt qubits.

Yield: When IBM makes a chip with 50 qubits, maybe 30 to 40 work well. The rest have junction defects, frequency problems, or excessive noise. No way to repair a bad qubit.

Coherence times: Superconducting qubits lose their quantum state in 100 to 500 microseconds. Some designs reach milliseconds but that took years of material science improvements.

Different Designs

Transmon: Most common. Large capacitor reduces sensitivity to charge noise. Coherence around 100 microseconds. Relatively easy to make.

Flux qubit: Superconducting loop with Josephson junctions. Sensitive to magnetic flux. Harder to isolate from noise.

Fluxonium: Long chain of Josephson junctions as a superinductor. Can hit 1 millisecond coherence but harder to fabricate and control.

The Support System

Dilution refrigerator: Pumps helium-3 and helium-4 to reach millikelvin temperatures. Takes 24 hours to cool down. Costs approx. $2 million.

Microwave control: Room temperature electronics generate 4 to 8 gigahertz pulses. Signals travel down coaxial cables into the fridge. A 50 qubit chip needs 100+ cables.

Magnetic shielding: Mu-metal around the refrigerator, sometimes superconducting shields at the coldest stage. Stray fields from power lines or passing cars disrupt qubits.

Signal processing: Low-noise amplifiers, high electron mobility transistor amplifiers, fast analog-to-digital converters. Software extracts qubit states from noisy signals.

Current Performance

Best superconducting qubits today:

• 100 to 500 microsecond coherence
• 20 nanosecond gate operations
• 99%+ two-qubit gate fidelity
• 99.9%+ single-qubit gate fidelity

You can do roughly 1,000 to 10,000 operations before decoherence. Not enough for most useful algorithms yet, which need millions of operations. That's why quantum error correction is critical.

Scaling Challenges

Wiring: Can't run a million coax cables into a fridge. Need multiplexing and cryogenic control electronics inside the refrigerator.

Crosstalk: Packed qubits interfere with each other. Control signals leak to neighbors.

Uniformity: Making 1,000 nearly identical qubits pushes fabrication limits.

Materials: Better materials with fewer defects would improve coherence directly.

Five years ago, 50 microsecond coherence was state of the art. Now it's 500 microseconds. Ten years ago, chips had 5 qubits. Now they have hundreds.

The physics works. Scaling remains unsolved. I'll describe trapped ion and photonic qubits here in future posts.