An electron does not have two possible positions; it has infinitely many. The ket becomes a function, sums become integrals, and position and momentum turn out to be two faces of the same object.
Normalise a wave function and compute the probability of finding the particle in an interval.
Compute \(\langle x\rangle\), \(\Delta x\) and, by Fourier transform, the momentum distribution.
Derive \([X, P] = i\hbar\) from translations and deduce \(\Delta x\,\Delta p \ge \hbar/2\).
Predict the motion and spreading of a free wave packet, and estimate a tunnelling probability.
Audio summary · 1:20Read the transcript
For a quantum object moving along a line, the state becomes a function, psi of x. The squared modulus of psi is a probability density: the probability of finding the particle between a and b is the area under this curve.
The same information can be written in terms of momentum. We go from one to the other with a Fourier transform. A packet that is narrow in position is broad in momentum, and the other way round. For a Gaussian packet, the product of the two standard deviations is exactly h-bar over two: the minimum allowed by the uncertainty relation. That relation follows directly from the commutator of position and momentum, which equals i h-bar.
The evolution is given by the Schrödinger equation. A free packet moves at the group velocity, p over m, and spreads out, because its different components do not travel at the same speed. For an electron localised within one nanometre, the width doubles in about thirty femtoseconds. For a one-gram marble, it would take far longer than the age of the Universe.
Faced with a barrier higher than its energy, a particle can still get through. This is tunnelling, and the scanning tunnelling microscope relies on it.
5.1The wave function and the presence cloud
In 1924, Louis de Broglie proposed that a material particle is accompanied by a wave. Confirmation came in 1927 at Bell Laboratories: Clinton Davisson and Lester Germer fired electrons at a nickel crystal and observed a diffraction pattern. Their sample had recrystallised by accident after being heated, which made the diffraction visible.
For a qubit, the state is expanded on two kets \(\ket0\) and \(\ket1\). For a quantum object on a line, it is expanded on infinitely many kets \(\ket x\), one per position. The rules stay the same, with three substitutions:
the sum \(\sum_i\) becomes an integral \(\int \mathrm dx\);
the Kronecker delta \(\delta_{ij}\) becomes the Dirac delta: \(\braket{x'|x} = \delta(x'-x)\);
the closure relation \(\sum_i \ket i\bra i = \mathbb 1\) becomes \(\int \mathrm dx\,\ket x\bra x = \mathbb 1\).
The kets \(\ket x\) cannot be normalised: they are computational tools, not states that can be prepared. The property of \(\delta\) to remember is \(\int_{x_1}^{x_2} f(x)\,\delta(x - x_0)\,\mathrm dx = f(x_0)\) if \(x_0 \in [x_1, x_2]\), and 0 otherwise.
Definition 5.1 — Wave function
The wave function is the component of the ket in the position basis:
It is a probability density amplitude, complex in general, and square-integrable (\(\psi\in L^2(\mathbb R)\)). The lecture notes call it the “state function”; the usual name “wave function” comes from the historical comparison with classical waves.
Principle 5.1 — Born rule in the continuum
\(|\psi(x)|^2\) is a probability density (unit: m−1). The probability of finding the quantum object in \([a, b]\) is
The probability of finding the quantum object at one exact point is zero. Only a probability over an interval is meaningful.
Mean and spread are computed as for any statistical distribution: \(\langle x\rangle = \int x\,|\psi|^2\,\mathrm dx\) and \((\Delta x)^2 = \langle x^2\rangle - \langle x\rangle^2\).
Predict, observe, explain
Before running the lab: write down your prediction, run it, compare, then explain the gap in one sentence. What is the total area under \(|\psi|^2\)? If the Gaussian is widened, what happens to the height of the peak?
The area under the presence cloud
LAB 5.1
Drag the bounds a and b
Wave function
P([a, b])
∫|ψ|² dx
⟨x⟩
Δx
|ψ(x)|² (density)ψ(x) (amplitude, dashed)
The Gaussian is \(\psi = (\pi\sigma^2)^{-1/4}e^{-x^2/2\sigma^2}\), whose standard deviation is \(\Delta x = \sigma/\sqrt2\). For the box of width \(L\), \(\psi_n = \sqrt{2/L}\,\sin(n\pi x/L)\): at level \(n = 2\), \(\psi\) changes sign at the centre, but \(|\psi|^2\) does not see this sign. The integrals are computed numerically: \(\int|\psi|^2\) shows 1.000 in every case.
Common pitfall — amplitude or probability?
\(\psi(x)\) is not a probability: it is a complex number that carries a phase. Two wave functions can have the same density \(|\psi|^2\) and different phases, hence different interference and different momentum distributions (lab 5.2).
The point \(x = -2\) lies outside the interval, so the integral is 0. This is the case of \(\int_{-1}^{1} e^{|x|+3}\,\delta(x-2)\,\mathrm dx\) in the lecture notes.
Exercise 5.2
The middle third of the box
★★☆
A quantum object is in the state \(\psi(x) = \sqrt{2/L}\,\sin(\pi x/L)\) on \([0, L]\) (zero elsewhere). Check the normalisation, then compute the probability of finding it between \(L/3\) and \(2L/3\) (three decimal places).
Solution
An antiderivative of \(\tfrac2L\sin^2(\pi x/L)\) is \(\tfrac xL - \tfrac{1}{2\pi}\sin(2\pi x/L)\). On \([0, L]\) it gives 1: the function is normalised. On \([L/3, 2L/3]\): \(\tfrac13 - \tfrac{1}{2\pi}\big(\sin\tfrac{4\pi}3 - \sin\tfrac{2\pi}3\big) = \tfrac13 + \tfrac{\sqrt3}{2\pi} \approx 0.609\). On the first quarter \([0, L/4]\), one finds \(\tfrac14 - \tfrac1{2\pi} \approx 0.091\).
Imagined dialogue: a wave of what? Born’s probabilistic reading (1926) did not convince everyone, Schrödinger first of all. The lines summarise positions; they are not quotations.
Schrödinger: “My equation describes a continuous phenomenon, a matter wave. That its square is only a probability, and that it collapses to a point on measurement, I cannot accept.”
Bohr: “The wave does not say where the electron is, but where it can be found if we look for it.”
Schrödinger: “Then the wave function is just a catalogue of our knowledge?”
Bohr: “It is the most complete tool we have. And the chance involved here is not that of our ignorance.”
5.2The other face: the momentum representation
We can just as well expand \(\ket\psi\) on the kets \(\ket p\) of definite momentum, with \(\braket{p'|p} = \delta(p'-p)\) and \(\int\mathrm dp\,\ket p\bra p = \mathbb 1\). The component \(\varphi(p) = \braket{p|\psi}\) is the wave function in momentum space. To go from one to the other, all we need is the plane wave \(\braket{x|p} = e^{ipx/\hbar}/\sqrt{2\pi\hbar}\).
Principle 5.2 — The two faces are linked by Fourier
\(|\varphi(p)|^2\) is the momentum probability density, and the norm is the same on both sides (Plancherel’s theorem): \(\int|\psi|^2\,\mathrm dx = \int|\varphi|^2\,\mathrm dp = 1\).
For the Gaussian \(\psi(x) = (\pi\sigma^2)^{-1/4}e^{-x^2/2\sigma^2}\), completing the square in the exponent gives another Gaussian:
The smaller \(\sigma\), the broader \(\varphi\): a sharply localised quantum object has a poorly defined momentum.
Clarification
The lecture notes write \(\sigma_x\sigma_p = \hbar\), taking \(\sigma\) and \(\hbar/\sigma\) as “widths”. With standard deviations, which are the ones in the Heisenberg relation, \(|\psi|^2 \propto e^{-x^2/\sigma^2}\) gives \(\Delta x = \sigma/\sqrt2\) and \(|\varphi|^2\propto e^{-p^2\sigma^2/\hbar^2}\) gives \(\Delta p = \hbar/(\sigma\sqrt2)\). The product is exactly \(\hbar/2\): the Gaussian reaches the bound. On the corresponding slide, the constant \((1/\pi\sigma^2)^{1/4}\) accompanies \(e^{-x^2/4\sigma^2}\): with that exponential, \((1/2\pi\sigma^2)^{1/4}\) is needed for the state to be normalised.
Position and momentum, side by side
LAB 5.2
State
Δx
Δp
Δx·Δp
|ψ(x)|²|φ(p)|²
Reduced units \(\hbar = 1\). For the single Gaussian, the product stays at 0.500 whatever σ. For \(\psi \propto g(x - d/2) \pm g(x + d/2)\), one has \(|\varphi(p)|^2 \propto e^{-p^2\sigma^2}(1 \pm \cos pd)\): fringes appear in momentum, and the product exceeds \(\hbar/2\). The “in phase” and “in antiphase” states have almost the same density \(|\psi(x)|^2\) when \(d\) is large, but not the same \(|\varphi(p)|^2\): the phase makes the difference.
Exercise 5.3
The width of a Gaussian
★★☆
For \(\psi(x) = (\pi\sigma^2)^{-1/4}e^{-x^2/2\sigma^2}\), what is the standard deviation \(\Delta x\)?
5.3Operators in continuous space
“Where are you?”: the position operator multiplies the wave function by \(x\), \((X\psi)(x) = x\,\psi(x)\). Its eigenstates are the \(\delta(x - x_0)\).
“Where are you going?”: momentum is defined by its geometric role. The translation operator \(T(a)\) shifts the state, \((T(a)\psi)(x) = \psi(x-a)\). For a small displacement, \(T(\mathrm da) \approx \mathbb 1 - \tfrac{i\,\mathrm da}{\hbar}P\), so \(T(a) = e^{-iaP/\hbar}\). Expanding \(\psi(x - \mathrm da) \approx \psi(x) - \mathrm da\,\psi'(x)\) gives the differential form.
The commutation relation follows from \(T^\dagger(a)\,X\,T(a) = X + a\,\mathbb 1\) to first order in \(a\). It implies, for every state, \(\Delta x\,\Delta p \ge \hbar/2\).
The expectation values are \(\langle X\rangle = \int x\,|\psi|^2\,\mathrm dx\) and \(\langle P\rangle = \int \psi^*(x)\,(-i\hbar\,\psi'(x))\,\mathrm dx\). The uncertainty is not a flaw of the instruments: no state gives narrow distributions in both \(x\) and \(p\) at the same time.
Exercise 5.4
A commutator acting on a function
★★☆
Compute \([X, P]\psi = XP\psi - PX\psi\) with \(P = -i\hbar\,\mathrm d/\mathrm dx\). What do you find?
Exercise 5.5
Mean momentum of a real function
★★☆
A normalised wave function \(\psi(x)\) is real (for example the Gaussian or a box state). What is \(\langle P\rangle\)?
Exercise 5.6
Order of magnitude of Δp
★☆☆
An electron is localised on the scale of an atom, \(\Delta x = 0.1\) nm. What is the smallest momentum spread compatible with Heisenberg? Give it in units of \(10^{-25}\ \mathrm{kg\,m/s}\).
× 10⁻²⁵ kg m/s
Imagined dialogue: is reality blurred? The relation \([X, P] = i\hbar\) fed the disagreements between Einstein and Bohr. The lines summarise their positions; they are not quotations.
Einstein: “At every instant an electron must have a definite position and momentum. If the theory does not give them, it is incomplete.”
Bohr: “Position and momentum are not properties the electron carries alone. They are defined by the measuring apparatus, and no apparatus measures both together.”
Einstein: “So the Moon exists only when we look at it?”
Bohr: “The Moon constantly interacts with its environment. An isolated electron does not: it is described by \(\psi\), not by a point.”
It is first order in time (the initial state fixes the whole future, between two measurements) and linear (any superposition of solutions is a solution).
A state of definite energy \(E_n\) evolves by a mere phase, \(\psi_n(x,t) = \phi_n(x)\,e^{-iE_nt/\hbar}\), where \(\phi_n\) satisfies the time-independent equation \(-\tfrac{\hbar^2}{2m}\phi_n'' + V\phi_n = E_n\phi_n\). Its density \(|\psi_n|^2 = |\phi_n|^2\) does not move. “Stationary” does not mean “motionless”: a plane wave \(e^{i(kx - \omega t)}\) has a uniform, constant density, yet it carries a probability current.
Tunnelling
Facing a barrier of height \(V_0\) greater than its energy \(E\), a classical ball turns back. The wave function, however, does not drop abruptly to zero: it decays as \(e^{-\kappa x}\) inside the barrier, with \(\kappa = \sqrt{2m(V_0 - E)}/\hbar\). If the barrier is thin enough, part of the wave emerges on the other side. For a rectangular barrier of width \(a\) and \(E < V_0\):
Reduced units \(\hbar = m = 1\), packet of width \(\Delta x_0 = 5\). The Schrödinger equation is solved step by step with the split-operator method (fast Fourier transform on 1,024 points). The prediction \(\bar T = \int|\varphi(p)|^2\,T(p^2/2)\,\mathrm dp\) averages \(T(E)\) over the momenta present in the packet. At the start, the whole packet is on the left (\(P(x<0) = 1\)). At the end of the run, \(P(x>a)\) is the transmission probability: it differs from the prediction by less than 0.01. Double \(a\) with \(E < V_0\): the transmission drops by a factor close to \(e^{2\kappa a}\).
Application — the scanning tunnelling microscope
A metal tip scans a conducting surface a few ångströms away. The vacuum between them acts as the barrier, with \(V_0 - E\) equal to the work function, about 4.5 eV. The current is proportional to \(e^{-2\kappa L}\): it depends so strongly on the distance \(L\) that individual atoms can be distinguished.
Exercise 5.7
Sensitivity of the tunnelling current
★★☆
With \(T \approx C e^{-2\kappa L}\), \(V_0 - E = 4.5\) eV, \(m_e = 9.11\times10^{-31}\) kg and \(\hbar = 1.05\times10^{-34}\) J s, the tip passes over an atom and the distance decreases by \(\Delta L = 0.1\) nm. What is the ratio \(I_\text{above}/I_\text{far}\)?
Clarification
The exercise in the lecture notes on the blue LED (3 nm well, transition \(n = 1 \to n = 0\)) leads to an effective mass of about \(0.045\,m_e\). The calculation is correct, but the physical interpretation is not: in an InGaN quantum-well LED, blue light comes from the recombination of a conduction-band electron with a valence-band hole. The photon energy is therefore mainly that of the band gap (about 2.7 eV), to which the confinement energies of the electron and the hole are added. A transition between two levels of the same band in a well a few nanometres wide falls in the infrared: that is the principle of quantum cascade lasers, not of LEDs.
According to an anecdote reported by Felix Bloch, Peter Debye remarked, after a seminar by Schrödinger on de Broglie’s thesis, that a wave should have a wave equation. Schrödinger would have written it a few weeks later.
Common pitfall — “stationary” does not mean “at rest”
In a stationary state it is the probability density \(|\psi(x,t)|^2\) that does not depend on time: the cloud neither deforms nor moves. The particle is not “frozen” for all that: a stationary state can describe a constant current of particles with non-zero mean momentum. The word refers to the stability of observable quantities, not to the absence of motion.
5.5A wave packet in motion
A plane wave has a perfectly defined momentum but extends over all space; a Dirac delta is perfectly localised but contains every momentum. A real quantum object lies in between: a wave packet, a superposition of plane waves with neighbouring momenta that interfere constructively in a small region. For a free quantum object, each component evolves with \(E(p) = p^2/2m\):
The group velocity, that of the envelope, is \(v_g = \mathrm dE/\mathrm dp = p_0/m\): it is the classical velocity. The phase velocity, that of the crests of the carrier wave, is \(v_\varphi = E/p = p_0/2m = v_g/2\). For a Gaussian packet of initial width \(\Delta x_0\) (standard deviation):
Exact solution of the free Schrödinger equation, in units \(\hbar = m = 1\); the curves are scaled to their maximum at \(t = 0\). The red triangle moves at \(v_g = p_0\) and stays at the centre of the packet; the blue one follows a crest of the carrier at \(v_\varphi = p_0/2\) and gets overtaken. Take \(p_0 = 0\) and \(\Delta x_0 = 0.3\): the packet does not move, but its height collapses within a few time units.
Clarification
The lecture notes attribute the spreading to the fact that the phase velocity \(p/2m\) depends on \(p\). What matters is that the components of the packet do not share the same group velocity \(p/m\): in a time \(t\), they separate by about \(\Delta p\,t/m\). This recovers the long-time behaviour, \(\Delta x(t) \approx \Delta p\,t/m\) with \(\Delta p = \hbar/(2\Delta x_0)\).
In exercise 5.6.1 of the lecture notes, if \(\Delta x_0\) and \(\Delta k\) are standard deviations (\(\Delta x_0\,\Delta k = 1/2\)), the term under the square root reads \(\hbar t/(2m\Delta x_0^2) = 2\hbar(\Delta k)^2t/m\), not \(\hbar(\Delta k)^2t/2m\). The spreading time is then \(t_0 = 2m\Delta x_0^2/\hbar = m/(2\hbar\Delta k^2)\).
Finally, the sentence “no action smaller than \(\hbar/2\) is observed” does not follow from \(\Delta x\,\Delta p \ge \hbar/2\). This inequality concerns the standard deviations of a single state, not a minimum action.
Exercise 5.8
Which is faster?
★☆☆
For a free non-relativistic electron with mean momentum \(p_0\), compare the phase velocity and the group velocity.
Exercise 5.9
An electron that spreads out
★★☆
A free electron (\(m_e = 9.11\times10^{-31}\) kg) is prepared in a Gaussian packet of width \(\Delta x_0 = 1\) nm. After how long has its width doubled? (in fs)
fs
And for a 1 g marble localised to within 1 mm?
\(t_0 = 2\times10^{-3}\times(10^{-3})^2/1.055\times10^{-34} \approx 1.9\times10^{25}\) s, so \(\tau \approx 3\times10^{25}\) s, about \(10^{18}\) years. That is nearly \(10^8\) times the age of the Universe (\(4\times10^{17}\) s). The spreading of an isolated macroscopic object is therefore invisible.
Clarification
The “Quantitative application” slide gives \(\tau \approx 30\) fs for \(\Delta x_0 = 0.1\) nm. That figure corresponds to \(\Delta x_0 = 1\) nm (exercise 5.9). For 0.1 nm, \(t_0\) is a hundred times smaller and \(\tau \approx 0.3\) fs. For the 1 g marble, the same slide says “\(10^{18}\) times the age of the Universe”: it is in fact \(10^{18}\) years, about \(10^8\) times the age of the Universe.
5.6Review exercises
Exercise 5.10
From the Gaussian to Heisenberg
★★★
Start from \(\psi(x) = (\pi\sigma^2)^{-1/4}e^{-x^2/2\sigma^2}e^{ip_0x/\hbar}\).
Show that \(\langle P\rangle = p_0\) and that \(\Delta x = \sigma/\sqrt2\).
Compute \(\varphi(p)\) by completing the square, then \(\Delta p\).
Check that \(\Delta x\,\Delta p = \hbar/2\). What happens to \(\Delta p\) over time for a free quantum object? And to \(\Delta x\)?
Hint
The factor \(e^{ip_0x/\hbar}\) simply shifts \(\varphi\): \(\varphi(p) \to \varphi(p - p_0)\). For the last question, look at how each component \(\varphi(p)\) evolves in time.
Solution
\(P\psi = -i\hbar\psi' = \big(p_0 + i\hbar x/\sigma^2\big)\psi\); the second term is odd in \(x\) and averages to zero, hence \(\langle P\rangle = p_0\). One finds \(\varphi(p) = (\sigma^2/\pi\hbar^2)^{1/4}e^{-(p-p_0)^2\sigma^2/2\hbar^2}\), so \(|\varphi|^2 \propto e^{-(p-p_0)^2\sigma^2/\hbar^2}\) and \(\Delta p = \hbar/(\sigma\sqrt2)\). The product is \(\hbar/2\).
For a free quantum object, \(\varphi(p,t) = \varphi(p,0)\,e^{-ip^2t/2m\hbar}\): only the phase changes, so \(|\varphi(p,t)|^2\) and \(\Delta p\) stay constant. On the other hand, \(\Delta x(t)\) grows (lab 5.5), and the product \(\Delta x\,\Delta p\) exceeds \(\hbar/2\) as soon as \(t > 0\).
Exercise 5.11
Assess: the cloud and \(|\psi|^2\)
★★☆
Analogy: “\(|\psi(x)|^2\) is the density of a cloud, thick where the quanton is likely to be found”. Which assessment is correct?
Evaluation grid
Mapped relations: cloud thickness ↔ probability density of presence.
Alignable difference: blur from ignorance (cloud) versus probability density of interfering amplitudes (quanton); clouds add whereas amplitudes interfere.
Status: evocative.
Correct wording: rewrite the analogy as one sentence that states its limit (“…, but …”).
Conceptual question 5.Q1
Width in position and momentum
A minimum-uncertainty Gaussian packet (no position–momentum correlation) is made narrower in position. How does its momentum width change? Discuss in pairs before answering.
Conceptual question 5.Q2
Free packet
A free minimum-uncertainty Gaussian packet evolves for \(t>0\). What happens to its width in position? Discuss in pairs before answering.
Conceptual question 5.Q3
Plane wave and localisation
A free particle is described by the plane wave \(\psi(x,t)=A\,e^{i(kx-\omega t)}\), with perfectly known momentum (\(\Delta p = 0\)). What is the uncertainty \(\Delta x\)? Discuss in pairs before answering.
File with a knowledge holder
Choose one analogy from this chapter. With the person’s consent, who may be an elder, a practitioner, a craftsperson or a musician, ask: (1) how they present this notion, in what language and with what words; (2) under what conditions this knowledge is passed on; (3) what, in their view, does not translate into physics. Write down the words in the local language, return your file to the person, then present both systems, each coherent by its own criteria, without concluding that one proves the other.
Key points of chapter 5
Main ideas.
The ket \(\ket\psi\) is projected onto the continuous basis \(\{\ket x\}\): \(\psi(x)=\braket{x|\psi}\in L^2(\mathbb R)\).
\(|\psi|^2\) is a density (m−1): only its integral over an interval is a probability.
\(\psi(x)\) and \(\varphi(p)\) are related by the Fourier transform; \(P\) generates translations.
A free wave packet moves at the group velocity and spreads; a barrier higher than the energy can still be crossed by tunnelling.
\(\Delta x\,\Delta p\ge\hbar/2\), with equality for the Gaussian.
\(v_g=\mathrm d\omega/\mathrm dk=p_0/m\); \(\sigma_x(t)=\sigma_0\sqrt{1+(t/\tau)^2}\) with \(\tau=2m\sigma_0^2/\hbar\).
Pitfalls. Taking the cloud literally (each detection is point-like); thinking the electron “grows” (it is the region of uncertainty that widens); confusing the phase velocity \(p_0/2m\) with the group velocity \(p_0/m\).
French – English glossary
Term
Terme français
Definition
Probability density
Densité de présence
\(\rho(x)=|\psi(x)|^2\), with \(\mathrm dP=\rho\,\mathrm dx\).
Quantum tunnelling
Effet tunnel
Crossing a barrier \(V_0>E\) through an evanescent wave, \(T\propto e^{-2\kappa a}\).
Wave function
Fonction d’état
\(\psi(x,t)=\braket{x|\psi(t)}\), normalised to 1.
Momentum operator
Opérateur impulsion
\(\hat p=-i\hbar\,\mathrm d/\mathrm dx\), with \([X,P]=i\hbar\).
Wave packet
Paquet d’ondes
Superposition of plane waves with nearby momenta.
Fourier transform
Transformée de Fourier
Unitary change from the position basis to the momentum basis.
What is it for in my field? — Photonics, spectroscopy
The trade-off between width in position and width in momentum governs the resolution of a spectrometer and the minimum duration of a laser pulse: the Fourier relation between duration and spectral width sets a minimum duration for a given bandwidth.
·Where do things stand in 2026?
2025 Nobel Prize in Physics: tunnelling goes macroscopic
John Clarke, Michel Devoret and John Martinis received the prize “for the discovery of macroscopic quantum mechanical tunnelling and energy quantisation in an electric circuit”. In their experiments at Berkeley in the mid-1980s, what crossed the barrier was not an electron, as in lab 5.4, but a collective variable of a superconducting circuit with a Josephson junction. These circuits are the direct ancestors of the superconducting qubits of Google and IBM.
Nobel Committee press release, 7 October 2025 — nobelprize.org
Matter waves to measure gravity
Commercial absolute gravimeters make the matter waves of laser-cooled atoms interfere. They measure \(g\) to within about \(10^{-8}\ \mathrm{m/s^2}\) and operate continuously in the field. One example is the AQG from Exail (France), which grew out of the LP2N and LNE-SYRTE laboratories. The atomic wave packets are split and then recombined, like the photons in the Mach–Zehnder interferometer of chapter 1.
Exail product sheet, accessed October 2026
Clocks: from the atom to the nucleus
NIST’s single aluminium-ion clock reaches a relative systematic uncertainty of \(5.5\times10^{-19}\) (Phys. Rev. Lett., July 2025). Very recently, two teams (TU Wien and China) operated the first nuclear clocks, locked to a transition of thorium-229 near 148 nm; for now their stability remains below that of the best atomic clocks (Nature, 7 October 2026). A redefinition of the second is not expected before the 2030 General Conference on Weights and Measures.
NIST, July 2025; Nature, October 2026; BIPM, roadmap for the redefinition of the second