Measurement-Enhanced Quantum Battery

• Measurement-enhanced quantum batteries are systems where measurement techniques define battery capacity, alter charging dynamics via postselection, and enable work extraction from charger correlations.
• Protocols employ diverse approaches such as projective filtering, continuous monitoring, and state stabilization to improve ergotropy and ensure robustness against noise.
• Experimental implementations, including room-temperature collective-spin batteries, validate these methods by demonstrating minimal deviation from tomography-based capacity measurements and controlled dephasing effects.

Measurement-enhanced quantum battery denotes a class of quantum-battery protocols in which measurement does more than provide terminal readout. In the current literature, measurement is used to operationally define battery capacity, to alter charging dynamics through postselection or continuous monitoring, to stabilize open batteries, to amplify capacity by local projective filtering, and to extract information-assisted work from battery–charger correlations (Li et al., 19 Apr 2026, Yan et al., 2022, Manickam et al., 29 May 2026, Gherardini et al., 2019, Zhang et al., 2024, Wang et al., 18 Jun 2025, Satriani et al., 2024). The term therefore covers several technically distinct mechanisms: measurement as metrology, as dynamical back-action, as a source of conditional control, and as an information-theoretic thermodynamic resource.

1. Conceptual scope and relation to standard quantum-battery theory

Quantum-battery theory predates explicitly measurement-based protocols. A central benchmark is collective unitary charging, where charging power can exceed that of local charging under fair Hamiltonian constraints; this framework establishes that quantum advantage can arise from many-body coherent dynamics alone and does not require measurement assistance (Campaioli et al., 2016). In that benchmark literature, the relevant observables are work, average power, and the scaling of the quantum advantage $\Gamma$ under constraints on operator norm, interaction order $k$, and connectivity $m$ (Campaioli et al., 2016).

Recent work separates naturally into three categories. First, some experiments are measurement-characterized rather than measurement-enhanced in the strict dynamical sense. In superconducting transmon arrays, quantum charging advantage was demonstrated for $N=2$ to $12$ cells using measurements of populations, $g^{(2)}$, coherent and incoherent ergotropy, and second-order Rényi entropy; the enhancement itself came from nearest-neighbor pairwise double-excitation charging rather than from measurement back-action (Hu et al., 9 Feb 2026). Second, some platforms use measurement to make an abstract battery quantity operationally accessible, as in direct capacity metrology of a thermal-vapor spin battery (Li et al., 19 Apr 2026). Third, genuinely measurement-assisted protocols use projective, weak, or indirect measurements to change the accessible state, steady-state current, or extractable work (Yan et al., 2022, Manickam et al., 29 May 2026, Satriani et al., 2024).

A further terminological distinction concerns the figure of merit. Several measurement-assisted papers are formulated in terms of ergotropy or charging rate (Yan et al., 2022, Manickam et al., 29 May 2026, Satriani et al., 2024), whereas the local-projective-measurement literature adopts a spectral notion of battery capacity that is invariant under unitary transformations and Schur-convex in the state spectrum (Zhang et al., 2024, Wang et al., 18 Jun 2025). This distinction is substantive: a protocol can enhance charging power or ergotropy without addressing spectral capacity, and conversely can enhance spectral capacity by measurement-induced spectral filtering.

2. Operational capacity metrology in a room-temperature collective-spin battery

A particularly explicit realization of a measurement-enhanced quantum battery is the room-temperature thermal-vapor battery based on approximately $10^{12}$ $^{87}\mathrm{Rb}$ atoms in a paraffin-coated cell, with coherence times exceeding $110$ ms (Li et al., 19 Apr 2026). The battery is a macroscopic collective atomic-spin ensemble polarized along a bias magnetic field and modeled in an effective two-level Zeeman subspace. Its state is described by

$$\hat{\varrho}=\frac{1}{2N}\sum_{n=1}^N\!\left(\mathbb{I}+\sum_{k=x,y,z}\langle j_k^n\rangle \sigma_k\right),$$

or equivalently by the normalized collective spin $k$0. In the fully polarized state prepared by optical pumping, $k$1, while $k$2 (Li et al., 19 Apr 2026).

The distinctive feature of this experiment is the tomography-free operational definition of capacity. For a closed system with Zeeman Hamiltonian $k$3, the capacity is defined as

$k$4

and for the collective-spin platform this becomes

$k$5

Experimentally, the extremal energies are found by hierarchical scans over rotations about the $k$6, $k$7, and $k$8 axes, with a coarse scan followed by a finer scan. The measured capacities differ from tomography-based values by less than $k$9, validating the operational protocol (Li et al., 19 Apr 2026).

The same experiment verifies a decomposition of capacity into coherent and incoherent parts. In the energy eigenbasis of $m$0, the off-diagonal sector contributes a coherent capacity $m$1, the population sector contributes an incoherent capacity $m$2, and the data confirm the quadratic relation

$m$3

This establishes experimentally that coherence contributes to storage capability independently of level populations (Li et al., 19 Apr 2026).

The platform also connects capacity to state mixedness. The measured states satisfy an entropy–capacity inequality for von Neumann entropy,

$m$4

an analogous Tsallis-entropy relation for $m$5, and the linear-entropy identity

$m$6

These constraints were verified for coherent and incoherent states with error levels below a few percent (Li et al., 19 Apr 2026).

Controlled dephasing provides the clearest demonstration of measurement-enhanced thermodynamic characterization. A magnetic-field gradient from anti-Helmholtz coils destroys transverse coherence while leaving $m$7 essentially unchanged. For a maximally coherent state, coherence decays from $m$8 to $m$9, while capacity falls from $N=2$0 eV to $N=2$1 eV (Li et al., 19 Apr 2026). The experiment thus turns capacity, coherence loss, and entropy production into jointly measurable quantities in a macroscopic, non-cryogenic battery.

3. Projective and postselected charging protocols

A different branch of the field uses measurement to drive the charging process itself. In the repeated-collision protocol “Charging by quantum measurement,” an $N=2$2-level battery interacts sequentially with disposable qubit chargers, and each interaction is followed by a projective measurement on the charger (Yan et al., 2022). In the principal “power-on” setting, chargers are prepared in $N=2$3 and postselected in $N=2$4. The inter-measurement time is optimized adaptively,

$N=2$5

so that, under near-resonant conditions, the battery population approximately obeys

$N=2$6

The protocol can therefore transfer roughly one excitation per successful round, generate population inversion, and drive the ratio of ergotropy to energy close to unity from a thermal initial state in less than $N=2$7 measurements, without requiring any initial coherence in either battery or chargers (Yan et al., 2022).

The same paper analyzes the converse “power-off” setting, where chargers start in $N=2$8 and are postselected in $N=2$9. There the battery is charged by useful work extracted by measurement alone rather than by excitation transfer. The mechanism is less efficient, and the success probability is much smaller than in the power-on scheme (Yan et al., 2022). This contrast clarifies that postselection is not merely a readout step: it reshapes the battery distribution nonunitarily and thereby changes the achievable energetic trajectory.

Measurement can also synthesize a coherent battery from many small coherent units. In “Measurement Induced Synthesis of Coherent Quantum Batteries,” a global measurement diagonal in the energy basis filters out the all-ground component of independent partially coherent two-level systems (Gumberidze et al., 2022). For the two-TLS projector pair $12$0 with $12$1, the success probability is

$12$2

The successful branch has increased energy, reduced energy variance, and, for sufficiently small $12$3, increased relative entropy of coherence; numerically, $12$4 approximately up to $12$5 (Gumberidze et al., 2022). Because the failure branch can be recycled, the protocol admits a repeat-until-success form with failure probability $12$6 after $12$7 repetitions (Gumberidze et al., 2022).

A third projective route exploits non-Markovian memory. In a charger-mediated two-qubit battery with time-dependent dephasing of the charger, the dephasing rate can become temporarily negative in the early stage, signaling memory effects that increase the maximal ergotropy relative to a Markovian approximation (Wang et al., 15 Feb 2026). The paper proposes a discrete-time measurement-enhanced implementation consisting of a short global unitary

$12$8

followed by a probabilistic local flip $12$9 on the charger. The timing relative to the negative-rate window is essential; in one example, a larger discrete time step produces about $g^{(2)}$0 ergotropy at $g^{(2)}$1, whereas the more faithful small-step simulation remains much lower (Wang et al., 15 Feb 2026). Here measurement-inspired interruption is used to steer the system into a more favorable later-time charging state.

4. Continuous monitoring, daemonic extraction, and measurement-resource accounting

Weak continuous measurement provides a distinct mechanism from postselection. In a two-qubit charger–battery system coupled to thermal reservoirs, quantum point contact (QPC) detectors continuously monitor the qubits and modify the nonequilibrium steady-state charging rate (Manickam et al., 29 May 2026). The charging-rate enhancement is defined relative to the unmeasured baseline $g^{(2)}$2, and the enhancement is non-monotonic in the measurement temperature and potential gradients. The paper identifies a plateau of near-optimal enhancement rather than a single sharply tuned operating point (Manickam et al., 29 May 2026).

Configuration	Optimal point $g^{(2)}$3	$g^{(2)}$4
Single QPC	$g^{(2)}$5	$g^{(2)}$6
Two independent QPCs	$g^{(2)}$7	$g^{(2)}$8
Coherent two-QPCs	$g^{(2)}$9	$10^{12}$0

The hierarchy is explicit: coherent two-QPC $10^{12}$1 two independent QPCs $10^{12}$2 single QPC, both in the magnitude of enhancement and in the reduction of measurement resources needed to reach it (Manickam et al., 29 May 2026). The coherent scheme introduces an additional correlated operator $10^{12}$3, which generates a zero-frequency dissipator and an additional dissipation-assisted transport pathway (Manickam et al., 29 May 2026).

Information-assisted work extraction appears even more explicitly in the daemonic protocol based on strong battery–charger coupling (Satriani et al., 2024). The battery is charged by thermalization with a charger, then an auxiliary memory measures the charger while leaving the battery intact. Conditional on outcome $10^{12}$4, one extracts the battery ergotropy from $10^{12}$5; averaging gives the daemonic ergotropy

$10^{12}$6

The memory can itself retain ergotropy, so the total extracted work is $10^{12}$7, while the measured efficiency obeys

$10^{12}$8

The protocol can surpass the unmeasured efficiency for favorable temperature, coupling, and measurement-basis choices (Satriani et al., 2024).

The same work makes the informational cost explicit. For a fully degenerate memory, reset obeys the Landauer bound $10^{12}$9, and the dissipated work satisfies

$^{87}\mathrm{Rb}$0

Two measurement schemes related by permutation of the measurement operators can yield the same daemonic ergotropy and the same dissipated work while leaving the memory in active and passive final states, respectively (Satriani et al., 2024). The resulting distinction between battery ergotropy and memory ergotropy is specific to information-assisted cycles and has no counterpart in purely unitary charging models.

5. Local projective measurements as capacity amplifiers

A separate line of work studies measurement-enhanced capacity in the spectral sense. For a $^{87}\mathrm{Rb}$1-dimensional state $^{87}\mathrm{Rb}$2 with Hamiltonian $^{87}\mathrm{Rb}$3 and ordered eigenvalues $^{87}\mathrm{Rb}$4, the battery capacity is defined as

$^{87}\mathrm{Rb}$5

This quantity is unitarily invariant and Schur-convex, so measurement can enhance capacity if it produces a post-measurement state with a more favorable spectrum (Zhang et al., 2024, Wang et al., 18 Jun 2025).

In the bipartite setting, a local rank-1 projective measurement on subsystem $^{87}\mathrm{Rb}$6 can increase the capacity of subsystem $^{87}\mathrm{Rb}$7, and in some parameter regimes also that of the full system $^{87}\mathrm{Rb}$8 (Zhang et al., 2024). For Bell-diagonal two-qubit states, equal averaging of the two post-measurement branches reduces the whole-system capacity and leaves the reduced capacity unchanged, but random weighting of the branches can always increase the subsystem capacity when $^{87}\mathrm{Rb}$9 and $110$0 (Zhang et al., 2024). For Werner states, this subsystem enhancement occurs even in the separable regime $110$1 (Zhang et al., 2024). For a representative two-qubit X state, the equal-average prescription always improves the subsystem capacity for $110$2, and weighted mixtures enhance the full-system capacity in an explicit parameter region (Zhang et al., 2024).

The tripartite extension introduces two schemes and the notion of an optimal local projective operator (Wang et al., 18 Jun 2025). Scheme 1 measures one subsystem and seeks to enhance the capacity of a two-body subsystem or the total system. Scheme 2 measures two subsystems and seeks to enhance the capacity of the remaining subsystem. For general three-qubit X states, explicit analytical expressions show that the post-measurement conditional states have at most four nonzero eigenvalues in Scheme 1 and at most two in Scheme 2 (Wang et al., 18 Jun 2025). Because the capacity is Schur-convex, the optimal operator is the one whose conditional reduced state majorizes the alternatives (Wang et al., 18 Jun 2025).

Noise robustness is part of the same spectral picture. Under white noise, capacity decreases monotonically, but optimal local projective operators improve the robustness of subsystem and total-system capacity against white noise for general tripartite X states (Wang et al., 18 Jun 2025). Under dephasing,

$110$3

capacity also decreases in general, but the post-measurement X states in the proposed schemes are diagonal, so dephasing has no further effect on them (Wang et al., 18 Jun 2025). The paper therefore reports complete robustness against dephasing noise for the relevant post-measurement capacities (Wang et al., 18 Jun 2025). These results show that projective measurement need not be capacity-destructive; under majorization-favorable spectral filtering it can be capacity-enhancing.

6. Stabilization, thermodynamic interpretation, and experimental outlook

Measurement can also be used to preserve stored energy rather than to inject it. In “Stabilizing Open Quantum Batteries by Sequential Measurements,” an open quantum battery with Hamiltonian $110$4 is protected against free-energy leakage by fast unitary initialization followed by repeated projective measurements in the energy eigenbasis (Gherardini et al., 2019). Once the battery is projected into the maximally charged state $110$5, repeated measurements at interval $110$6 produce a Zeno stabilization regime. The measurement is energy-neutral on average,

$110$7

but its information record carries thermodynamic cost through Landauer erasure (Gherardini et al., 2019).

The same protocol derives a second-law-like lower bound on the stabilization power and makes the measurement-frequency trade-off explicit. As $110$8, Zeno freezing improves, but the Landauer cost diverges because the number of recorded outcomes grows without bound (Gherardini et al., 2019). This establishes a general point that recurs across measurement-enhanced battery models: informational control cannot be assessed independently of memory and reset costs.

Experimentally, the field now contains both measurement-enhanced and measurement-characterized platforms. The thermal-vapor collective-spin battery is a direct room-temperature implementation in which measurement operationally defines capacity and tracks coherence-driven performance degradation (Li et al., 19 Apr 2026). By contrast, the scalable solid-state transmon battery realizes quantum charging advantage using nearest-neighbor pairwise interactions, with measurement revealing nonzero coherent ergotropy, incoherent ergotropy, entanglement, and $110$9 but not acting as the charging resource itself (Hu et al., 9 Feb 2026). Likewise, the room-temperature organic-microcavity battery demonstrates superextensive charging, metastabilization of stored energy, and superextensive electrical power, but the reported measurements are diagnostic—transient reflectivity, pump–supercontinuum spectroscopy, and electrical readout—rather than back-action-based enhancement (Hymas et al., 27 Jan 2025).

This suggests two converging research directions. One direction treats measurement as an active thermodynamic resource that modifies ergotropy, capacity, or charging rate through postselection, continuous monitoring, or information-to-work conversion (Yan et al., 2022, Manickam et al., 29 May 2026, Satriani et al., 2024). The other treats measurement as the means by which battery performance becomes operationally accessible in scalable hardware, as in the tomography-free collective-spin capacity protocol (Li et al., 19 Apr 2026). The combination of these directions is likely to determine whether measurement-enhanced quantum batteries remain primarily a conceptual tool of quantum thermodynamics or become a standard control primitive in experimentally scalable energy-storage devices.