What Is Quantum Sensing? How Quantum Sensors Detect Tiny Signals

Quantum sensing uses a quantum system, property or phenomenon to measure a physical quantity. A field, force, frequency shift or acceleration changes a controlled system, and a readout converts that change into an estimate.

Different sensors use different resources: discrete energy levels, spin evolution, superconducting interference or specially prepared states. Entanglement can improve some protocols, but is not a requirement for every quantum sensor. ^[1]

Understanding a result means following the full measurement: preparation, interaction, readout and uncertainty. Detecting a tiny change, identifying an absolute reference and demonstrating a useful instrument are distinct achievements.

What makes a quantum sensor quantum?

All physical devices ultimately obey quantum mechanics. The label becomes informative when an identifiable quantum property is central to how the measurement works.

An atomic clock references a quantum transition. A spin sensor measures how states respond to a field. A superconducting quantum interference device converts magnetic flux into an electrical response through phase-sensitive superconducting behaviour.

NIST’s overview describes this family of sensing approaches. ^[2] It does not follow that a quantum instrument is automatically preferable for every task.

A practical comparison must include size, operating temperature, measurement speed, dynamic range and calibration, alongside the smallest signal the instrument can distinguish. A more sensitive laboratory device may still be unsuitable for a particular field application.

Follow a spin measurement from preparation to readout

Consider a simplified magnetic-field sensor. A control sequence prepares a known spin state and then creates a superposition of two levels.

During a controlled waiting time, the relative phase evolves at a rate affected by the field. A later pulse converts that phase into differences in outcome probabilities. Repeated readout allows the field to be estimated.

The preparation establishes the reference. The interaction encodes the signal. The final pulse and detector make that information accessible.

A longer interaction time can increase the accumulated phase, but environmental noise can reduce coherence or introduce unwanted phase shifts. The optimal time is therefore not always the longest possible time.

This sequence illustrates why the sensor is the complete controlled experiment, rather than simply the atom or defect at its centre.

Sensitivity, accuracy and range answer different questions

Sensitivity concerns the size of a change that can be distinguished under stated conditions. Accuracy requires understanding how the reported value relates to the quantity being measured, including systematic corrections and uncertainty.

Dynamic range describes the usable range of inputs. Bandwidth concerns how rapidly signals can vary while still being measured reliably.

Imagine a magnetometer that detects tiny variations but has an unknown offset. It can be highly useful for tracking changes while being less useful for identifying the absolute field.

Likewise, a device excellent at measuring a slowly varying signal may miss a brief pulse. A sensitivity figure without bandwidth, averaging time and operating conditions is incomplete.

These are engineering distinctions with direct consequences for how a research claim should be interpreted.

How a SQUID measures magnetic flux

A superconducting quantum interference device, or SQUID, combines a superconducting loop with one or more Josephson junctions. A junction is a weak link that allows a superconducting current with a phase-dependent response.

In a common two-junction design, magnetic flux changes the phase relationship around the circuit and alters its electrical behaviour. Suitable current biasing produces a voltage response that can be read out.

The input is flux, which depends on the field through an effective area. Estimating a magnetic field therefore also requires knowledge of geometry, coupling and calibration. NIST describes SQUIDs as established magnetometers and as readout components for other sensitive detectors. ^[3]

The term quantum does not remove that calibration step. The device must still be connected to the desired external quantity through a justified response model.

Why identifying zero field is a separate problem

A conventional SQUID response repeats as flux changes. Repetition is useful for tracking variation, but creates ambiguity when a single output value is used to infer an absolute flux.

An unlabelled clock face provides an analogy: the hand positions identify a location within a repeating cycle, not which day it is.

Combining loops with different effective responses can produce a distinctive central feature useful for identifying a reference field. The usable range and response must still be characterised; a reference feature is not permission to assume an unlimited, ambiguity-free measurement range.

Absolute-reference identification and small-signal sensitivity therefore solve related but different problems. A sensor may improve at one without demonstrating an improvement at the other.

What the October SQUID-array experiment changed

Monaghan and colleagues introduced selected superconducting loops without Josephson junctions into two-dimensional SQUID arrays. Their model relates the arrangement of these bare loops to an effective spread of loop areas, called synthetic area spread.

Fabricated arrays showed responses consistent with this approach, allowing a zero-field reference feature without imposing the corresponding spread of physical SQUID-loop areas. The journal paper appeared on 26 October 2026; the first preprint was posted in November 2025. ^[4]

The demonstrated result is the engineered response, not a measured attainment of quantum-limited sensitivity or a clinical improvement. The manuscript discloses a patent application concerning the concepts. ^[5]

Why the best reference point may not be the best operating point

A smooth symmetric response minimum can identify zero. At the exact bottom, however, the first derivative is zero: a very small input change produces only a second-order output change.

The steepest part of the response lies elsewhere. Biasing, modulation or feedback can help exploit a useful slope while preserving knowledge of the reference.

This distinction is not a criticism of zero-field sensing. It separates calibration from signal conversion.

The same issue appears in spectroscopy. Locating a resonance centre, generating an error signal and measuring small changes around a chosen operating point can require different processing of the same underlying response.

A complete instrument must connect those operations and quantify the noise added at each stage.

Is amplifying the signal better than reducing noise?

Increasing the signal helps only in relation to the relevant noise. An amplifier that multiplies the signal and its existing noise equally leaves that signal-to-noise ratio unchanged.

Amplification can nevertheless help overcome noise introduced later by a detector or electronic readout. Its value depends on where the dominant uncertainty enters.

Quantum squeezing takes another approach. It redistributes fluctuations between suitable conjugate observables, reducing noise in a selected measurement direction while increasing it in another. It does not make every observable arbitrarily precise.

Gravitational-wave detectors provide an established application of squeezed light. Their improvements depend on matching the squeezing to the frequency-dependent measurement noise. ^[6]

An amplification claim and a noise-reduction claim should therefore be compared through the final estimation uncertainty, not simply the size of the output.

What the oscillator proposal does and does not say

A theoretical paper by Jesús Fuentes Aguilar, published on 26 October 2026, designs a driven oscillator cycle with transient amplification. In the ideal closed, signal-free cycle, the state returns to vacuum. With a signal present, the final state carries a displacement that encodes information; it is not simply unchanged vacuum.

The paper also analyses damping and the noise associated with the transient dynamics. ^[7]

This is a proposed sensing protocol, not a demonstrated device. Its comparison depends on the drive, signal model, duration, loss and other resource assumptions. Returning an ideal baseline state to its initial form does not, by itself, establish that a practical repeated-measurement instrument needs no resetting or recalibration.

Where quantum sensing is already useful

SQUIDs measure weak magnetic signals and read out superconducting detectors. Atomic magnetometers infer fields from atomic behaviour. Diamond defects can provide magnetic information on small spatial scales. ^[3]

These platforms offer different combinations of sensitivity, spatial resolution, bandwidth and operating conditions. None is best on every axis.

A local magnetic image, for example, answers a different question from a very precise measurement of a nearly uniform field. Portability may also matter more than the best laboratory noise floor for a particular use.

The same principle applies to dark-matter detectors: choose a defined signal model and then assess whether the complete instrument can distinguish it from background.

What repeated measurements improve

For independent measurements with the same finite variance, the statistical uncertainty of the mean generally falls with the square root of the sample count. One hundred times as many measurements then gives about ten times smaller statistical uncertainty.

A shared calibration offset does not average away by that rule. Correlated noise and drift can also prevent the expected improvement.

Quantum protocols can change particular measurement limits, but a claimed advantage needs a fair resource comparison. The number of probes, total measurement time, control power, losses and readout conditions can all matter.

The foundations connect to quantum mechanics through coherent evolution and measurement. The practical gain appears only when those resources improve the final estimate under the stated conditions.

How to read the next quantum-sensing headline

First identify the measured quantity. Then identify the demonstrated change: a steeper response, less noise, improved calibration, greater range, faster measurement or a more practical package.

Next separate a theoretical proposal, a component experiment and a complete instrument tested in its intended environment. Each can advance the field, but they support different conclusions.

The SQUID-array work changes a magnetic response. The oscillator paper analyses an alternative control protocol. Neither should acquire an unmeasured clinical benefit or universal sensitivity advantage through the language used to describe it.

A useful claim makes the signal pathway and uncertainty budget explicit enough that another experiment can test the same performance.