Quantum Ligand-Binding Interrogator (QLI)

• QLI is a trapped-ion quantum sensor that detects ligand-binding-induced changes in single-molecule electrostatics using a differential gradiometer design.
• It employs a two-ion Ramsey interferometric protocol to capture state-resolved measurements with sensitivity strongly dependent on the ion-sample distance.
• The platform targets static state discrimination in vitrified samples, offering a label-free alternative to traditional binding assays.

Quantum Ligand-Binding Interrogator (QLI) denotes a proposed trapped-ion quantum sensor for label-free, single-molecule detection of ligand-binding-induced electrostatic state changes in vitrified samples. In its primary formulation, QLI is a differential sensor, or gradiometer, based on a pair of co-trapped atomic ions that measures the electric field gradient produced by a single ligand binding to its receptor, with the stated goal of distinguishing static molecular states such as bound versus unbound or apo versus holo (Huver, 17 Aug 2025). The concept is therefore not a generic binding assay, but a state-resolved electrostatic measurement platform whose central observable is the binding-induced change in local electric field at micrometer stand-off. Related quantum and hybrid quantum-classical work addresses adjacent problems—pose enumeration, docking, virtual screening, and affinity scoring—but these are methodologically distinct from the trapped-ion QLI architecture.

1. Measurement target and physical observable

QLI is built around a specific physical hypothesis: a ligand-bound receptor and an unbound receptor differ in their permanent dipole moment because of ligand charge placement, side-chain rearrangement, and bound-water reorganization. The proposed instrument attempts to read out that difference remotely through the electric field it produces, rather than through fluorescence, current, or ensemble kinetics (Huver, 17 Aug 2025).

The geometry is fixed as follows. The sample surface normal is $\hat z$, the trap axis is $\hat x$, and the two ions lie at equal height $h$ above the sample with coordinates

$\mathbf r_{S/R}=(\pm d/2,0,h).$

The measured signal is the lateral differential field

$\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$

For a binding-induced dipole change normal to the sample,

$\Delta \mathbf p = \Delta p\, \hat z,$

the free-space differential field is

$$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$

with

$$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$

The paper notes $c_{\mathrm{eff}}(0)=3$ and, for the representative geometry $d/h=0.345$, $\hat x$0 (Huver, 17 Aug 2025).

This formulation makes the central scaling law explicit. The signal is linear in the dipole change $\hat x$1, linear in the ion separation $\hat x$2, and falls as $\hat x$3. That steep distance dependence is one of the defining constraints of QLI: reducing ion-sample distance from $\hat x$4 to $\hat x$5 changes the projected signal by roughly a factor of $\hat x$6.

The proposal also includes a simple dielectric-interface correction for a vitrified biological matrix approximated as amorphous ice with relative permittivity $\hat x$7. For a dipole normal to the interface, transmission into vacuum is attenuated by

$\hat x$8

An additional isotropic root-mean-square projection factor of $\hat x$9 is then used as a conservative orientation model. This yields an effective benchmark signal

$h$0

so the experimentally relevant observable is already a reduced version of the free-space maximum (Huver, 17 Aug 2025).

2. Two-ion Ramsey gradiometer and phase readout

The sensing protocol is a Ramsey-style interferometric measurement on a two-ion entangled state coupled to the axial stretch mode. Differential electric forces couple to the stretch mode, whereas spatially uniform fields mainly drive the center-of-mass mode. This gives the device a native preference for field gradients rather than absolute fields (Huver, 17 Aug 2025).

The sequence begins in the spin and motional ground state,

$h$1

A Mølmer–Sørensen gate prepares the Bell-like state

$h$2

The paper emphasizes two reasons for this choice. First, the state is sensitive to differential perturbations acting oppositely on the two ions. Second, it lies in a decoherence-free subspace against common-mode magnetic-field noise, because uniform magnetic fluctuations shift both components similarly (Huver, 17 Aug 2025).

During interrogation, near-resonant spin-dependent optical dipole forces are applied around the stretch-mode frequency. The relevant Hamiltonians are

$h$3

Here $h$4 is the spin-dependent-force coupling, $h$5 is the detuning from the stretch mode, $h$6 are stretch-mode ladder operators, $h$7 is the oscillator length, and the external field enters through the differential force

$h$8

Using a second-order Magnus expansion, the relative phase between $h$9 and $\mathbf r_{S/R}=(\pm d/2,0,h).$0 is

$\mathbf r_{S/R}=(\pm d/2,0,h).$1

with transduction gain

$\mathbf r_{S/R}=(\pm d/2,0,h).$2

For a closed-loop sequence satisfying

$\mathbf r_{S/R}=(\pm d/2,0,h).$3

this simplifies to

$\mathbf r_{S/R}=(\pm d/2,0,h).$4

The field-dependent phase therefore grows linearly with the number of phase-space loops $\mathbf r_{S/R}=(\pm d/2,0,h).$5 (Huver, 17 Aug 2025).

After interrogation, the entangling operation is reversed:

$\mathbf r_{S/R}=(\pm d/2,0,h).$6

The measured bright-state population is then

$\mathbf r_{S/R}=(\pm d/2,0,h).$7

After $\mathbf r_{S/R}=(\pm d/2,0,h).$8 repetitions, the phase uncertainty scales as

$\mathbf r_{S/R}=(\pm d/2,0,h).$9

The proposal does not claim a beyond-standard-quantum-limit protocol; its emphasis is instead on common-mode rejection, gradient selectivity, and compatibility with long averaging through repeated shots (Huver, 17 Aug 2025).

3. Experimental architecture and operating regime

The proposed hardware is a cryogenic linear Paul trap operating near $\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$0 K in ultra-high vacuum, with $\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$1 as the sensing ion and a possible coolant ion such as $\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$2 for sympathetic recooling between shots. The biological sample is mounted on the apex of an AFM-style quartz stylus, vitrified by plunge-freezing, and transferred into the cryogenic vacuum chamber. This stylus architecture is the bridge between trapped-ion metrology and biological specimens (Huver, 17 Aug 2025).

Vitrification is not incidental. The proposal is explicitly aimed at static-state discrimination in a vitrified sample, not live real-time kinetics. The envisioned assay compares frozen preparations of different states—such as apo versus holo or ligand-free versus ligand-bound—rather than watching a binding trajectory unfold in one molecule. The paper is explicit that QLI is intended to distinguish static molecular states in a vitrified preparation (Huver, 17 Aug 2025).

The representative target stand-off is

$\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$3

with representative ion spacing

$\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$4

for an axial trap frequency

$\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$5

Operation near $\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$6 stylus-trap separation has precedent, but the authors identify reduction toward $\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$7 as a key engineering milestone because of the $\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$8 signal law (Huver, 17 Aug 2025).

Long averaging does not require a single coherent superposition to persist for minutes. Instead, the proposal uses repeated shots of coherent duration $\Delta E_x \equiv E_x(+d/2,h)-E_x(-d/2,h).$9 followed by cooling, preparation, and readout overhead $\Delta \mathbf p = \Delta p\, \hat z,$0. For total averaging time

$\Delta \mathbf p = \Delta p\, \hat z,$1

and duty cycle

$\Delta \mathbf p = \Delta p\, \hat z,$2

the signal-to-noise ratio obeys

$\Delta \mathbf p = \Delta p\, \hat z,$3

The paper cites realistic coherent live times of $\Delta \mathbf p = \Delta p\, \hat z,$4–$\Delta \mathbf p = \Delta p\, \hat z,$5 ms under dynamical decoupling and lock-in-style protocols, with a practical reference frequency around $\Delta \mathbf p = \Delta p\, \hat z,$6 Hz and

$\Delta \mathbf p = \Delta p\, \hat z,$7

The proposed development path is staged. It begins with calibration against a metallic nano-tip carrying a known voltage, proceeds through ion-surface proximity studies and measurement of bare-vitrified-sample backgrounds, and only then advances to biomolecular proof-of-principle targets such as DNA hybridization and, later, protein-ligand apo/holo discrimination (Huver, 17 Aug 2025).

4. Sensitivity estimates, noise model, and feasibility gate

The paper anchors feasibility to a benchmark dipole change

$\Delta \mathbf p = \Delta p\, \hat z,$8

motivated by a few-angstrom relocation of fractional or integer charge. For the representative geometry $\Delta \mathbf p = \Delta p\, \hat z,$9, $$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$0, and $$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$1, the geometry factor is $$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$2. The vacuum maximum differential field is

$$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$3

Including dielectric attenuation gives

$$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$4

and after the isotropic RMS orientation factor the projected signal becomes

$$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$5

These values are then compared to experimentally reported single-ion low-frequency sensitivities

$$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$6

from Bonus et al. (Huver, 17 Aug 2025).

At $$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$7, the projected sensor-limited figures are:

Mode	Benchmark at SNR $$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$8	Benchmark at SNR $$\Delta E_x(h,d;\Delta p) = \frac{\Delta p}{4\pi\epsilon_0}\,\frac{3 h d}{\big(h^2+(d/2)^2\big)^{5/2}} = \frac{\Delta p}{4\pi\epsilon_0}\; d\,h^{-4}\; c_{\mathrm{eff}}\!\left(\frac{d}{h}\right),$$9
AC / lock-in	$$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$0	$$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$1
DC-style	$$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$2	$$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$3

The distance sensitivity is severe. At

$$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$4

the paper computes

$$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$5

leading to

$$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$6

By contrast, increasing the two-ion baseline to $$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$7 yields projected times of roughly

$$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$8

This identifies stand-off reduction and baseline extension as the dominant geometry levers (Huver, 17 Aug 2025).

The decisive unknown is not the trapped-ion sensitivity itself but the electrostatic stability of vitrified samples. The paper models the sample contribution through a differential field-noise power spectral density

$$c_{\mathrm{eff}}(u) \equiv \frac{3}{\big(1+(u/2)^2\big)^{5/2}}.$$9

with amplitude spectral density

$c_{\mathrm{eff}}(0)=3$0

Instrument and sample noise combine as

$c_{\mathrm{eff}}(0)=3$1

Because averaging time scales as

$c_{\mathrm{eff}}(0)=3$2

the slowdown relative to the sensor-limited case is

$c_{\mathrm{eff}}(0)=3$3

To keep the integration-time penalty below $c_{\mathrm{eff}}(0)=3$4, the sample noise should satisfy

$c_{\mathrm{eff}}(0)=3$5

Numerically, the feasibility targets are approximately

$c_{\mathrm{eff}}(0)=3$6

for the AC protocol and

$c_{\mathrm{eff}}(0)=3$7

for DC-style operation. The paper explicitly identifies these as quantitative feasibility gates (Huver, 17 Aug 2025).

5. Related computational programs and broader QLI interpretations

A broader QLI ecosystem has emerged around computational interrogation of ligand binding, even though these works do not implement the trapped-ion sensor. In structure-based virtual screening, one line of work uses quantum or hybrid quantum-classical models as binding-affinity estimators on 3D protein-ligand complexes. A quantum convolutional neural network trained on PDBbind v2020 reached a test-set Pearson correlation coefficient of $c_{\mathrm{eff}}(0)=3$8 and RMSD $c_{\mathrm{eff}}(0)=3$9 kcal/mol, with the authors emphasizing that under simulated noise the correlation remained largely stable while RMSD worsened, which suggests stronger support for relative ranking than for calibrated thermodynamic prediction (Yang, 13 Jul 2025). A related parameterized-quantum-circuit regressor reported RMSD $d/h=0.345$0 kcal/mol and Pearson correlation $d/h=0.345$1 with six quantum circuit units, and stated that predictions remained consistent at $d/h=0.345$2 shots (Yang, 24 Jul 2025). A multimodal hybrid quantum neural network, HQDeepDTAF-NN-Angle with 9 qubits, reported MAE $d/h=0.345$3, RMSE $d/h=0.345$4, and $d/h=0.345$5, with about $d/h=0.345$6 fewer trainable parameters than DeepDTAF (Jeong et al., 14 Sep 2025). These results suggest a computational QLI role centered on rescoring, reranking, or prioritization.

Ligand-based screening has also been cast in quantum-embedding terms. In optimized quantum data embeddings for ligand-based virtual screening, Neural Quantum Embedding and projected quantum kernels were evaluated on LIT-PCBA and COVID-19 tasks. The clearest low-data result reported is a balanced accuracy of $d/h=0.345$7 for PQK with ZZ at 4 qubits on the COVID-19 set, with the authors arguing that advantages are strongest in limited-data and class-imbalanced conditions (Choi et al., 18 Dec 2025). This suggests a complementary QLI mode in which ligand interrogation is supervised by known actives rather than inferred from structure.

Docking and pose search form a second adjacent cluster. A neutral-atom maximum weighted independent set formulation solved a 540-node docking interaction graph for the TACE-AS complex and recovered the exact optimum MWIS weight $d/h=0.345$8 on that instance, but the reconstructed ligand pose had RMSD $d/h=0.345$9 to the crystallographic pose, so the result is better interpreted as contact-set selection than as chemically complete docking (Garrigues et al., 25 Aug 2025). Quantum encoding of rigid 3D ligand poses has been proposed as a pose-generation front end that coherently represents $\hat x$00 translated configurations, but it does not provide a scoring oracle or an end-to-end docking protocol (Yang, 14 Dec 2025). On quantum annealers, weighted subgraph isomorphism has been used to formulate geometric docking search, and a later physically informed extension added Coulomb, van der Waals, hydrogen-bond, and hydrophobic corrective terms to the QUBO (Triuzzi et al., 2024, Micucci et al., 10 Apr 2026). These docking papers support a broader interpretation of QLI as a pose-interrogation engine, albeit one distinct from the trapped-ion sensing proposal.

A third cluster is quantum-informed physics-based scoring. Qenergy-VM2 combines Mining Minima sampling, QM-refined ligand charges, QM/MM interaction evaluation, and a VQE-based electronic correction; across 23 protein targets and 543 ligands it reported mean absolute error about $\hat x$01 kcal/mol, Pearson $\hat x$02, Spearman $\hat x$03, and Kendall $\hat x$04 (Molani et al., 5 Dec 2025). Full-complex DFT rescoring of 22 MCL1 ligands reported $\hat x$05, Spearman $\hat x$06, and PI $\hat x$07 with about 40 minutes per calculation on AWS (Mardirossian et al., 2020). In a more specialized electronic-structure setting, DFT+DMFT showed that many-body effects at the Fe center are necessary to bring myoglobin $\hat x$08 and CO binding energetics into near balance, which suggests that ligand interrogation in correlated metalloproteins may require beyond-standard-DFT electronic structure (Weber et al., 2014). A plausible implication is that experimental QLI-style electrostatic measurements and computational quantum scoring could become mutually constraining rather than competing paradigms.

6. Limitations, misconceptions, and prospective role

QLI is not presently an integrated demonstrated instrument. The trapped-ion proposal is theoretical and feasibility is dominated by an unmeasured materials question: whether vitrified biological samples mounted on a nearby stylus are electrostatically quiet enough in the relevant low-frequency band. The paper is explicit that the decisive unknown is the electrostatic stability of vitrified samples at $\hat x$09 stand-off and 4 K (Huver, 17 Aug 2025).

It is also not a live kinetic assay in its current form. The observable is a static electrostatic contrast between molecular states in vitrified preparations. The strongest directly supported use case is therefore bound-versus-unbound or apo-versus-holo discrimination across frozen snapshots, not continuous tracking of association and dissociation in one molecule (Huver, 17 Aug 2025).

A second common misconception is to conflate the trapped-ion QLI with computational quantum docking or screening papers. Those studies are relevant to a broader ligand-interrogation agenda, but most remain proof-of-concept, simulator-based, benchmark-specific, or ranking-oriented. Several explicitly support reranking or prioritization more strongly than absolute free-energy reporting, and several do not establish superiority over strong classical baselines (Yang, 13 Jul 2025, Yang, 24 Jul 2025, Jeong et al., 14 Sep 2025).

The narrow but distinctive niche of QLI is therefore clear. Compared with fluorescence-based single-molecule methods, it aims to avoid perturbative labels. Compared with ensemble electrostatic probes, it targets single-molecule state resolution. Compared with classical docking and scoring, it offers a possible experimental route to directly benchmark electrostatic consequences of ligand binding at the level of one vitrified molecule. If realized, QLI would not replace pharmacology assays, docking, or free-energy simulation; it would supply a new kind of state-specific electrostatic datum that those computational and biochemical methods currently lack (Huver, 17 Aug 2025).