Zero-Point Energy

Cool anything far enough, classical physics says, and it stops. Every vibration dies, every atom settles to the bottom of its potential well, and the energy books read zero. Helium never got the memo: at atmospheric pressure it refuses to freeze no matter how close to absolute zero you take it — the only element with that distinction — and you must squeeze it to about 25 atmospheres before a solid forms. The culprit is the subject of this article: zero-point energy, the irreducible minimum energy a confined quantum system keeps in its ground state, strictly above the bottom of its potential well. You have already met the two cleanest specimens. The Infinite Square Well's lowest level sits at $E_1 = \pi^2\hbar^2/2mL^2$ , not zero; the harmonic oscillator's ladder starts at $E_0 = \tfrac12\hbar\omega$ , the famous half-quantum that survives at absolute zero. This article is where those two facts become one principle. The engine is the uncertainty relation: pinning a particle down ( $\sigma_x$ small) forces its momentum to spread ( $\sigma_p \gtrsim \hbar/2\sigma_x$ ), and momentum spread is kinetic energy. The ground state is not "at rest at the bottom" — it is the best compromise the particle can negotiate between a potential that wants it narrow and a kinetic term that punishes narrowness. Here we derive that compromise honestly — a lower bound $\langle H\rangle \ge \tfrac12\hbar\omega$ with the Gaussian as the unique state that achieves it — then let the number out into the world: why hydrogen and deuterium have measurably different chemistry, why helium stays liquid, and what zero-point motion does to the mirrors of a gravitational-wave detector.

Key takeaways

Zero-point energy (ZPE) is the irreducible ground-state energy of any confined quantum system — always strictly above the potential minimum.
For the harmonic oscillator, $E_0 = \tfrac12\hbar\omega$ , split exactly half-and-half: $\langle T\rangle = \langle V\rangle = \tfrac14\hbar\omega$ .
It follows from the uncertainty principle alone: $\langle H\rangle = \sigma_p^2/2m + \tfrac12 m\omega^2\sigma_x^2 \ge \tfrac12\hbar\omega$ , with equality only for the Gaussian ground state.
Confinement sets the scale: for a box, $E_1 = \pi^2\hbar^2/2mL^2$ — smaller box or lighter particle means more ZPE.
ZPE is real and measured: H₂ vs D₂ bond energies differ by the predicted 0.08 eV, and helium's ZPE outweighs its ~11 K well depth, keeping it liquid.
Key idea: you cannot make both the position spread and the momentum spread vanish, so a bound particle can never sit still at the bottom of its well — some energy always remains.

The core insight

Confinement plus the uncertainty principle forces a positive minimum energy: the ground state of any bound system lies strictly above the potential minimum, exemplified by $E_0 = \tfrac12\hbar\omega$ for the oscillator.

Where this lives today

Zero-point motion is a daily engineering line-item wherever light atoms or precise mirrors matter. Neutron and X-ray diffraction analyses correct structure factors for the Debye–Waller factor, whose zero-temperature limit reflects the atoms' zero-point motion; hydrogen-storage and battery chemists see ZPE shift reaction free energies whenever H is swapped for D. The Casimir force — vacuum ZPE differing between plate geometries — is a real stiction budget in MEMS devices at sub-micron gaps, measured to percent accuracy since Lamoreaux (1997). And at the grandest scale, LIGO's mirrors: a 40 kg test mass on a 1 Hz suspension has a zero-point RMS motion of only $4.6\times10^{-19}\,\mathrm{m}$ — yet detector sensitivity now brushes this "standard quantum limit" (in certain situations), and squeezed-light injection redistributes the vacuum's zero-point fluctuations between quadratures to beat it. The half-quantum in equation (1) of 1912 is today a noise term in a billion-dollar instrument's error budget.

The physical picture

Try to pin a quantum particle to the bottom of a bowl. Squeeze its wavefunction narrower and the potential energy happily drops — but the momentum spread balloons ( $\sigma_p \ge \hbar/2\sigma_x$ ), and with it the kinetic energy, $\sigma_p^2/2m \propto 1/\sigma_x^2$ . Let the wavefunction relax wider and kinetic energy falls, but now the particle samples the rising walls of the bowl and potential energy climbs as $\sigma_x^2$ . One cost falls as $1/\sigma_x^2$ , the other rises as $\sigma_x^2$ : the total has a valley, and the valley floor is not at zero.

Figure

The tug-of-war — kinetic cost of narrowness vs potential cost of width, and the ħω/2 valley between them

That picture already contains every qualitative fact about ZPE. Stiffer confinement (larger $\omega$ , smaller $L$ ) raises the valley floor. Heavier particles feel it less — the kinetic cost carries $1/m$ , so ZPE fades as mass grows, which is why the effect owns the light end of the periodic table (H, He) and vanishes for billiard balls. And nothing thermal is involved anywhere: the argument uses only $\hbar$ , never temperature. Zero-point energy is what remains after every thermal quantum has been extracted. The ladder picture makes the same point in energy space: the oscillator's levels are $E_n = (n+\tfrac12)\hbar\omega$ , and cooling removes quanta $\hbar\omega$ one at a time until the last rung — but the last rung hangs $\tfrac12\hbar\omega$ above the bottom of the parabola, and no process that leaves the particle bound can take that away.

Figure

Two confinements, one law — the oscillator's half-quantum offset and the box's confinement floor

The bound, derived honestly

The claim to prove: no normalized state of the harmonic oscillator has mean energy below $\tfrac12\hbar\omega$ . Not just the stationary states — any state whatsoever. The mean energy of a state with spreads $\sigma_x, \sigma_p$ splits into two pieces:

\langle H\rangle = \frac{\langle p^2\rangle}{2m} + \frac{1}{2}m\omega^2\langle x^2\rangle \qquad (1)

where: $\langle p^2\rangle = \sigma_p^2 + \langle p\rangle^2$ and $\langle x^2\rangle = \sigma_x^2 + \langle x\rangle^2$ — second moments, which exceed the variances whenever the state is displaced or moving. Since displacement and drift only add energy, the minimum lives among centered states, and there $\langle H\rangle = \sigma_p^2/2m + \tfrac12 m\omega^2\sigma_x^2$ . Now the uncertainty principle enters as the only quantum input.

Derivation AThe zero-point bound from the uncertainty principle alone

We derive $\langle H\rangle \ge \tfrac12\hbar\omega$ for every normalized state, and identify the unique state that holds the equality. The strategy: reduce to centered states, use $\sigma_x\sigma_p \ge \hbar/2$ (the Heisenberg relation, taken as given from Standard Deviation and Quantum Uncertainty) to eliminate $\sigma_p$ , and minimize the resulting one-variable function of $\sigma_x$ . The subtle point worth watching is the logic of the two inequalities: we need both to be equalities simultaneously for the bound to be reached, and both point at the same state — the Gaussian.

Step 1. Center the state. For fixed spreads, $\langle H\rangle$ is smallest when $\langle x\rangle = \langle p\rangle = 0$ , since equation (1) contains $\langle x\rangle^2$ and $\langle p\rangle^2$ with positive coefficients. So it suffices to bound

\langle H\rangle = \frac{\sigma_p^2}{2m} + \frac{1}{2}m\omega^2\sigma_x^2

Step 2. Apply Heisenberg: $\sigma_p \ge \hbar/2\sigma_x$ , so

\langle H\rangle \;\ge\; \frac{\hbar^2}{8m\sigma_x^2} + \frac{1}{2}m\omega^2\sigma_x^2 \;\equiv\; f(\sigma_x)

Why this is legitimate: the kinetic term is monotonically increasing in $\sigma_p$ , so replacing $\sigma_p$ by its smallest allowed value can only lower the right side — the inequality survives.

Step 3. Minimize $f$ . Setting $f'(\sigma_x) = -\hbar^2/4m\sigma_x^3 + m\omega^2\sigma_x = 0$ gives

\sigma_x^2 = \frac{\hbar}{2m\omega}, \qquad f_{\min} = \frac{\hbar\omega}{4} + \frac{\hbar\omega}{4} = \frac{\hbar\omega}{2}

$f''(\sigma_x)$ at this point is positive, which confirms this stationary point is a minimum. *Why the split is half-and-half: at the minimum the two costs are equal — the general feature of $a/\sigma^2 + b\sigma^2$ optimizations (the AM–GM inequality gives $f \ge 2\sqrt{ab} = \hbar\omega/2$ directly, same answer). Equal sharing of kinetic and potential energy is the uncertainty compromise made visible.*

Step 4. Identify the state when the equality holds. Equality in Step 2 requires a minimum-uncertainty state ( $\sigma_x\sigma_p = \hbar/2$ exactly), which is a Gaussian; equality in Step 3 fixes its width to $\sigma_x^2 = \hbar/2m\omega$ . That state is precisely the oscillator ground state $\psi_0 = (m\omega/\pi\hbar)^{1/4}e^{-m\omega x^2/2\hbar}$ — so the bound is not merely a bound but the exact ground-state energy:

E_0 = \frac{\hbar\omega}{2}

As a check: the minimization can be redone symbolically, confirming $\sigma_x^2 = \hbar/2m\omega$ , $f_{\min} = \hbar\omega/2$ , and second derivative $4m\omega^2 > 0$ (a minimum, not a maximum); the AM–GM route gives the identical bound.

The same logic with rougher arithmetic explains the box. In the Infinite Square Well (which owns the exact solve — recall $E_1 = \pi^2\hbar^2/2mL^2$ ), the state is confined to $\sigma_x \lesssim L$ , so the kinetic floor is of order $\hbar^2/2mL^2$ ; the exact prefactor $\pi^2$ is what the boundary conditions add to the uncertainty estimate. Same principle, different geometry:

E_1 = \frac{\pi^2\hbar^2}{2mL^2} \qquad (2)

where: $L$ — the box width; note the pure confinement character: there is no $\omega$ , no force at all inside the box, yet the ground state carries energy simply because it is caged. Halve the box, quadruple the floor.

What the ground state actually does

"The particle has energy but sits still" is a contradiction the ground state resolves in a specific, measurable way: it spreads. The two variances of $\psi_0$ , computed as Gaussian integrals, are

\langle x^2\rangle = \frac{\hbar}{2m\omega}, \qquad \langle p^2\rangle = \frac{m\hbar\omega}{2} \qquad (3)

where: the product gives $\sigma_x\sigma_p = \hbar/2$ exactly — the ground state is a minimum-uncertainty state, sitting precisely on the Heisenberg boundary. Its energy budget splits evenly, $\langle T\rangle = \langle V\rangle = \tfrac14\hbar\omega$ , the equal sharing that Step 3 predicted. The RMS excursion $\sqrt{\langle x^2\rangle} = \sqrt{\hbar/2m\omega}$ is the zero-point motion — the "breathing" of a bond at absolute zero that neutron and X-ray scattering measure directly as the Debye–Waller factor, and whose polynomial bookkeeping lives in Hermite Polynomials and Oscillator Eigenfunctions.

Go deeper · Derivation B · 1 minGround-state spreads and the exact half-quantum budget

Derivation BGround-state spreads and the exact half-quantum budget

We compute $\langle x^2\rangle$ and $\langle p^2\rangle$ in the ground state directly, confirm the minimum-uncertainty product, and verify that kinetic and potential energies each carry exactly $\tfrac14\hbar\omega$ . The route is honest Gaussian integration — no ladder shortcuts — so that the equal split appears from the wavefunction itself. (The algebraic route via $\hat a, \hat a^\dagger$ , which gets these in two lines, belongs to Ladder Operators and the Algebraic Oscillator.)

Step 1. The normalized ground state is the Gaussian $\psi_0(x) = (m\omega/\pi\hbar)^{1/4}\,e^{-m\omega x^2/2\hbar}$ ; its normalization $\int\psi_0^2\,\mathrm{d}x = 1$ is the standard Gaussian integral.

Step 2. Second moment of position:

\langle x^2\rangle = \int_{-\infty}^{\infty} x^2\,\psi_0^2\,\mathrm{d}x = \frac{\hbar}{2m\omega}

Why it comes out this way: $\psi_0^2$ is a Gaussian of variance $\hbar/2m\omega$ ; the integral is just that variance.

Step 3. Second moment of momentum, via $\langle p^2\rangle = \int\psi_0\,(-\hbar^2\psi_0'')\,\mathrm{d}x$ :

\langle p^2\rangle = \frac{m\hbar\omega}{2}

Step 4. Assemble the budget:

\langle T\rangle = \frac{\langle p^2\rangle}{2m} = \frac{\hbar\omega}{4}, \qquad \langle V\rangle = \frac{1}{2}m\omega^2\langle x^2\rangle = \frac{\hbar\omega}{4}, \qquad \sigma_x\sigma_p = \frac{\hbar}{2}

Why the equal split matters: it is the quantum virial theorem for the oscillator, and it says the zero-point energy is genuinely half motion, half stretch — not a bookkeeping fiction parked in one column. *As a check: all three integrals evaluate exactly to $\langle x^2\rangle = \hbar/2m\omega$ , $\langle p^2\rangle = m\hbar\omega/2$ , uncertainty product $\hbar/2$ , and each energy budget line $\hbar\omega/4$ .*

The isotope dial — ZPE you can measure on a bench

Here is the cleanest experimental handle. Two isotopologues — H₂ and D₂ — share the same electronic potential curve to excellent approximation (electrons do not care about neutron count), so the spring constant $k$ is common and only the mass differs: $\omega = \sqrt{k/\mu}$ . Doubling the mass divides the frequency, and with it the zero-point energy, by $\sqrt{2}$ :

\frac{E_0^{\mathrm{D_2}}}{E_0^{\mathrm{H_2}}} = \sqrt{\frac{\mu_{\mathrm{H_2}}}{\mu_{\mathrm{D_2}}}} = 0.707 \qquad (4)

where: $\mu$ — the reduced mass ( $m/2$ for a homonuclear molecule). The numbers land on target: H₂ vibrates at $4401\,\mathrm{cm}^{-1}$ , predicting $4401 \times 0.7072 = 3113\,\mathrm{cm}^{-1}$ for D₂ against a measured $3115.5$ — agreement at the 0.1% level. The ZPEs are then $2201\,\mathrm{cm}^{-1}$ (0.273 eV) and $1556\,\mathrm{cm}^{-1}$ , so D₂ sits $644\,\mathrm{cm}^{-1} \approx 0.080\,\mathrm{eV}$ deeper in the shared well: it takes measurably more energy to pull D₂ apart. This ZPE difference is the root of the kinetic isotope effect — C–D bonds react up to several times slower than C–H at room temperature, a workhorse tool for tracing reaction mechanisms — and of isotope fractionation in atmospheric and geological chemistry. Historically it ran the other way: in 1924, before quantum mechanics was even complete, Mulliken found band spectra of boron monoxide isotopes inexplicable without a half-quantum offset — the first experimental proof that $\tfrac12\hbar\omega$ is real.

Limiting cases

case	result	meaning
harmonic oscillator	$E_0 = \tfrac12\hbar\omega$	equal kinetic/potential shares $\tfrac14\hbar\omega$ ; vanishes only as $\omega \to 0$ (no confinement)
infinite square well	$E_1 = \dfrac{\pi^2\hbar^2}{2mL^2}$	pure confinement energy; $L \to \infty$ releases it
stiff confinement, $\omega$ or $1/L$ large	ZPE grows	squeezing costs energy — the quantum-dot and nuclear scales
heavy mass, fixed spring	$E_0 = \tfrac12\hbar\sqrt{k/m} \propto m^{-1/2}$	ZPE fades for heavy particles: H and He feel it, xenon barely
classical limit $\hbar \to 0$	$E_0 \to 0$	the particle may rest at the well bottom — classical intuition restored
free particle	no ZPE	no confinement, no floor: ZPE is a property of bound systems

Common trapZero-point energy is a reservoir you could tap if you were clever enough

It is the opposite: ZPE is defined as the energy of the lowest state, so there is no lower rung to drop to and nothing to extract — any device "harvesting" it would need somewhere lower to put the system, which by definition does not exist. Perpetual-motion schemes built on "vacuum energy extraction" fail at this first line. Keep two other confusions out of your notes as well: ZPE is not thermal (it survives at exactly $T = 0$ ; temperature never enters the derivation), and the offset $\tfrac12\hbar\omega$ has no observable meaning by itself for a single fixed system — what experiments see are ZPE differences: between isotopes (Mulliken's bands), between geometries (Casimir), between reactant and transition state (kinetic isotope effect). Energy differences, always.

Worked examples

Example 1 — the frozen vibration of HCl (confidence builder). The HCl stretch has wavenumber $\tilde{\nu} = 2990\,\mathrm{cm}^{-1}$ . Find the zero-point energy and compare it with thermal energy at room temperature. Is the vibration "on" at 300 K?

Step 1. Frequency: $\omega = 2\pi c\tilde{\nu} = 2\pi(2.998\times10^{10})(2990) = 5.63\times10^{14}\,\mathrm{rad/s}$ .

Step 2. Zero-point energy: $E_0 = \tfrac12\hbar\omega = \tfrac12(1.055\times10^{-34})(5.63\times10^{14}) = 2.97\times10^{-20}\,\mathrm{J} = \boxed{0.185\,\mathrm{eV}}$ .

Step 3. Thermal scale: $k_BT = 25.9\,\mathrm{meV}$ at 300 K. The ZPE alone is 7.2 times the thermal energy, and a full quantum $\hbar\omega = 0.371\,\mathrm{eV}$ is 14.3 times it.

Step 4. Boltzmann factor for the first excited state: $e^{-\hbar\omega/k_BT} = e^{-14.3} \approx 6\times10^{-7}$ — fewer than one molecule in a million is vibrationally excited. At room temperature the vibration is thermally frozen, yet every single molecule still carries 0.185 eV of vibrational energy it can never shed. Zero-point energy is the part of the energy that does not obey the thermometer.

Example 2 — why helium will not freeze (exam level). Model a helium atom in a would-be solid as a particle in a 3D box of side $L = 3\,\text{Å}$ (a typical lattice cell). Compare its confinement ZPE with the He–He attractive well depth, $\varepsilon/k_B \approx 11\,\mathrm{K}$ , and pronounce a verdict.

Step 1. One-dimensional floor: $E_1 = \pi^2\hbar^2/2m_{\mathrm{He}}L^2$ with $m_{\mathrm{He}} = 6.65\times10^{-27}\,\mathrm{kg}$ and $L = 3\times10^{-10}\,\mathrm{m}$ gives $E_1 = 9.2\times10^{-23}\,\mathrm{J} = 0.57\,\mathrm{meV}$ .

Step 2. In kelvin: $E_1/k_B = 6.7\,\mathrm{K}$ per dimension.

Step 3. Three dimensions triple it (the 3D box ground energy is $3E_1$ ): confinement ZPE $\approx \boxed{20\,\mathrm{K}}$ .

Step 4. Verdict: the caging energy (~20 K) exceeds the ~11 K well depth that would hold the lattice together — the crystal is energetically a losing proposition, and helium stays liquid at any temperature under its own vapour pressure. Pressure (~25 atm) is needed to force the atoms close enough that the well deepens and wins. Hydrogen, though lighter, does freeze — its intermolecular well is several times deeper — so helium's liquidity is a genuine two-parameter conspiracy: light mass and feeble attraction. This estimate is crude (a real solid is not a hard box), but the physics — ZPE versus binding — is exactly the accepted explanation.

The man who found energy at absolute zero

Origins · 1 minWalther Nernst and how this was discovered

Walther Nernst

1864–1941 · Berlin

Nernst owned the physics of absolute zero. His heat theorem of 1906 — the third law of thermodynamics — showed that entropies flatten as $T \to 0$ , and his laboratory's heroic low-temperature calorimetry made the approach to zero an experimental science. So it carried weight when, in 1916, Nernst argued that emptiness itself should not be empty: taking Planck's "second theory" of 1912, in which each oscillator retains $\tfrac12 h\nu$ , he proposed that the electromagnetic field's modes keep their half-quanta too — a zero-point energy of the vacuum, decades before quantum field theory could make the idea precise. He even speculated about its cosmological weight, anticipating a problem that still torments physics. The molecular half-quantum he championed was confirmed by Mulliken's 1924 isotope band spectra; the vacuum half-quantum, by Casimir's force between metal plates. Nernst received the 1920 Nobel Prize in Chemistry for the heat theorem.

History

1900

— Planck quantizes oscillator energy exchange; his ladder has no half-quantum yet

1912

— Planck's "second theory" introduces a residual ½hν per oscillator — the first appearance of zero-point energy

1913

— Einstein and Stern argue (from hydrogen's specific heat) for zero-point motion in molecules

1916

— Nernst extends zero-point energy to the electromagnetic vacuum itself

1924

— Mulliken's boron monoxide isotope bands give the first compelling spectroscopic evidence of the half-quantum vibrational offset

1926

— Schrödinger's oscillator solution derives E₀ = ħω/2 rather than postulating it

1948

— Casimir predicts an attractive force between conducting plates from vacuum zero-point energy

1997

— Lamoreaux measures the Casimir force to ~5%, closing the loop Nernst opened

Going further

References. Griffiths & Schroeter, Introduction to Quantum Mechanics, 3rd ed., Ch. 2 · Sakurai & Napolitano, Modern Quantum Mechanics, 2nd ed., Ch. 2 · Shankar, Principles of Quantum Mechanics, 2nd ed., Ch. 7 · Cohen-Tannoudji, Diu & Laloë, Quantum Mechanics, Vol. I, Ch. V · Townsend, A Modern Approach to Quantum Mechanics, 2nd ed., Ch. 7 Key results used throughout this article: oscillator spectrum $E_n=(n+\tfrac12)\hbar\omega$ with ground state $E_0=\tfrac12\hbar\omega$ ; infinite-well spectrum $E_n=n^2\pi^2\hbar^2/2mL^2$ ; ground-state variances $\langle x^2\rangle=\hbar/2m\omega$ and $\langle p^2\rangle=m\hbar\omega/2$ , verifiable directly from $\psi_0$ ; Casimir pressure coefficient $\pi^2\hbar c/240$ (Casimir 1948); helium solidification pressure ~25 atm; hydrogen ground state $-13.6$ eV.

Builds on: Time-Independent Schrödinger Equation in One Dimension · Quantum Harmonic Oscillator: Analytic Solution · Ladder Operators and the Algebraic Oscillator · Bound States versus Scattering States

Leads to: Infinite Square Well · Hermite Polynomials and Oscillator Eigenfunctions · Delta-Function Potential Well and Bound State · Coherent States of the Harmonic Oscillator

If the ground state has energy, why can't I extract it?

Extraction means the system ends in a lower state, and the ground state is the lowest — that is its definition. The half-quantum is not stored fuel; it is the floor of the building. What you can do is change the floor's height: alter the confinement (move Casimir plates, break a bond, swap an isotope) and the ZPE difference between configurations does real thermodynamic work. That is why observable ZPE physics is always difference physics.

Is zero-point motion actual motion? Is the particle moving or not?

It is not classical motion — the ground state is stationary, its probability density frozen in time, and $\langle p\rangle = 0$ . But it is not rest either: $\langle p^2\rangle = m\hbar\omega/2 > 0$ , so momentum measurements return nonzero values with an RMS spread, and the position density has width $\sqrt{\hbar/2m\omega}$ rather than a point. The honest statement: the particle has no trajectory at all, and its state permanently carries both position spread and momentum spread. "Zero-point motion" is shorthand for those irreducible spreads — which scattering experiments genuinely see.

Why doesn't the infinite square well's ground energy look like ħω/2?

Because $\omega$ is an oscillator parameter — the well has no spring. Each confinement geometry converts the same uncertainty logic into its own formula: the box gives $\pi^2\hbar^2/2mL^2$ (scale set by $L$ ), the oscillator gives $\hbar\omega/2$ (scale set by stiffness), hydrogen's Coulomb well gives 13.6 eV binding with the Bohr radius as the compromise width. Universal principle, geometry-dependent number: ZPE $\sim \hbar^2/2m\ell^2$ with $\ell$ the confinement length, prefactors from the details.

Does the thermal energy kT just add on top of the zero-point energy?

Essentially yes, and the two parts behave differently. The mean oscillator energy at temperature $T$ is $\langle E\rangle = \hbar\omega/2 + \hbar\omega/(e^{\hbar\omega/k_BT} - 1)$ : a constant quantum floor plus a thermal part that dies exponentially once $k_BT \ll \hbar\omega$ (Planck's function). At room temperature a molecular vibration like HCl's has essentially only its zero-point part — the thermal term is one part in a million — while a soft lattice mode can be fully classical. The crossover $k_BT \sim \hbar\omega$ is the boundary between "frozen" and "thermal" degrees of freedom, and it is why specific heats fall at low temperature.

Does zero-point energy gravitate? I've heard it predicts a catastrophic cosmological constant.

This is a genuine open problem, and this article can only point at it. Summing $\tfrac12\hbar\omega$ over the electromagnetic field's modes up to any plausible cutoff gives a vacuum energy density wildly larger — famously by ~120 orders of magnitude — than the observed cosmological constant. Whether and how vacuum ZPE gravitates is unresolved; Nernst himself worried about it in 1916. What is settled is that ZPE differences have ordinary energy's properties: they shift masses (binding energies), exert forces (Casimir), and drive chemistry (isotope effects). The absolute vacuum offset is the part still in dispute.

Is the Casimir effect really proof of vacuum zero-point energy?

It is the standard interpretation and the cleanest mental picture: between conducting plates the allowed field modes are restricted, so the vacuum ZPE depends on plate separation, and its gradient is a measurable attractive pressure ( $\pi^2\hbar c/240a^4$ for ideal plates). The measured force (Lamoreaux 1997 and successors) matches. A caveat belongs in your notes: the same force can be derived from fluctuating charges in the plates (van der Waals language) without invoking vacuum energy explicitly — so Casimir demonstrates ZPE-difference physics rather than settling the absolute-vacuum-energy question of the previous FAQ.

Do nuclei and quantum fields have zero-point energy too, or is this a molecular phenomenon?

Everything bound and quantum has it. Protons and neutrons carry large zero-point kinetic energies (tens of MeV) inside the ~femtometre nuclear well — the uncertainty estimate $\hbar^2/2m\ell^2$ at $\ell \sim 2\,\mathrm{fm}$ says so immediately. Electrons in atoms: the 13.6 eV scale of hydrogen is confinement energy negotiated against Coulomb attraction. And in quantum field theory every mode of every field is an oscillator with its half-quantum — the "second quantization" story previewed at this chapter's edge and taken up properly in the many-body and relativistic tracks.