Helicity-Flip Bookkeeping for Quantum Dot Ligand Exchange

Quantum dots are semiconductor crystals a few nanometers across, and the molecules bound to their surfaces largely set how they behave. Replacing the long, insulating molecules they are grown with by short or strongly polar ones, while the dots are still suspended in liquid, can shift their energy levels by up to 2.0 eV (Kroupa et al.). The same replacement can also flip abruptly from almost no exchange to almost complete exchange, because neighboring ligands attract or repel one another (Bronstein et al.). What the field lacks is a direct count of how many molecules leave and enter the surface at each step of the reaction. Without that count, the strength of the neighbor interactions that drive these abrupt transitions cannot be measured. It can only be inferred from optical shifts, which Kroupa et al. showed are not simply predicted by the ligand's own dipole.

We propose to count those molecules with gas chromatography-mass spectrometry (GC-MS). This instrument separates a mixture in time by boiling point and affinity for a column coating, then identifies each component by how it fragments. The difficulty is precision. Near an abrupt transition, the useful signal is a small change in released ligand between adjacent titration points, sitting on a large and drifting background. For the measurement design we borrow the central trick of parity-violating electron scattering. That field extracts parts-per-million asymmetries by rapidly flipping the beam's spin between two states, so that slow drifts cancel in the difference. It then hunts down the "helicity-correlated" errors that do not cancel. Here, the physics of parity violation is not the mechanism. The experimental discipline that makes tiny differential signals trustworthy is.

Core Claim

A paired, pseudorandomly interleaved GC-MS measurement can resolve the incremental release of oleate during solution-phase exchange precisely enough to fit a lattice model with nearest-neighbor interactions to exchange isotherms. The two arms are dot-containing and dot-free, run with the same ligand addition, and the assignment of isotope-labeled internal standards between arms is swapped periodically. The fit yields a lateral interaction energy for each incoming ligand. This number does not currently exist for these systems. It can be tested against the cooperative step locations reported qualitatively by Bronstein et al. and against the packing-dependent band shifts of Kroupa et al.

The core mechanism is GC-MS mass balance. The application regime is solution-phase exchange on lead sulfide (PbS) dots with substituted cinnamic acids. Parity-violating scattering contributes the measurement design. We deliberately do not propose polarized-electron probes of chiral molecules. Weak-force asymmetries measured at GeV energies are about 24 ppm even for a proton target (Ahmed et al.), and parity-violating effects in molecules at chemical energies are smaller by many orders of magnitude. That direction is not a viable experiment.

Research Program

The chemistry follows Bronstein et al. We start with oleate-capped PbS dots and titrate in cinnamic acids carrying para substituents that span a range of dipole moments. The released oleic acid and the unconsumed incoming acid are recovered from aliquots, converted to volatile methyl or trimethylsilyl derivatives, and quantified by isotope dilution against deuterated spikes.

Two chromatographic issues need care. First, cinnamates come in cis/trans and positional isomers, and trace isomeric impurities could change how the ligands pack on the surface. We use the retention-time model of Lin et al. to choose a temperature program that separates these isomers before running experiments. That model combines a 3D molecular graph network with a recurrent encoder of the oven temperature program, and it reports errors under 0.7 minutes on cis/trans and positional isomer groups. We use it as a design tool, not as a claimed contribution. Second, derivatization converts oleic acid and lead oleate into the same ester. The measurement therefore reports total oleate leaving the surface. It cannot distinguish proton-driven exchange from displacement of lead-oleate complexes, and we state this limit rather than hide it.

The borrowed design has three parts, each mapped from its scattering-experiment counterpart. First, rapid state reversal becomes interleaving: dot and blank injections alternate in a pseudorandom order, so instrument drift enters both arms equally and cancels in the normalized difference $$A = \frac{N_{\text{dot}} - N_{\text{blank}}}{N_{\text{dot}} + N_{\text{blank}}},$$ where $N$ is the isotope-dilution-corrected oleate amount in each arm.

Second, regression against beam monitors becomes regression of $A$ against recorded nuisance variables. These include internal-standard absolute areas, inlet temperature, derivatization time, and injection order.

Third, the half-wave plate reversal becomes label swapping. In scattering experiments, inserting a half-wave plate flips the sign of the true asymmetry but not of certain instrumental false asymmetries. Here we periodically swap which deuterated standard goes to which arm. A label-correlated bias then changes sign while the chemistry does not, and averaging over both configurations removes it. Null runs with blanks in both arms play the role of null-asymmetry measurements and must give zero within error. Analysis is blinded with a hidden offset until the systematic budget is frozen.

The resulting coverage-versus-concentration isotherms are fit with a lattice-gas model in which each surface site feels an exchange free energy plus a lateral term proportional to the number of like neighbors. The fit returns one interaction parameter per ligand, which is then correlated with substituent dipole.

Baselines

The naive reference assumes that every added cinnamic acid binds, with one-to-one release of oleate. It predicts a linear isotherm with no step, so any measured step and any failure of mass balance directly refutes it.

The standard in the field is solution proton NMR of free versus bound ligand, together with optical absorbance shifts, as used by Kroupa et al. and Bronstein et al. We run NMR on the same aliquots with the same total instrument time. The research file notes that broadened peaks from bound ligands hinder integration.

The strongest competitor is conventional isotope-dilution GC-MS: the same derivatization and the same deuterated standards, but samples run in blocks without interleaving, label swaps, or monitor regression. It uses the same aliquots, the same injection count, and the same analyst time. This baseline doubles as the ablation that isolates the proposal's key ingredient.

The proposal beats these baselines if its increment uncertainty near the cooperative step is at least twice smaller, its null runs show no bias outside error while the block design's null runs do, and its fitted interaction parameters are stable across replicate batches. No reported result for ligand-exchange isotherm precision, or for any quantitative cooperativity parameter, was found on Lacuna. The GC figures in the file, such as R² = 0.931 on unseen compounds for Lin et al., describe retention-time prediction, not quantitation, and are not a baseline number for this question.

Why not just…?

Why not just use conventional isotope-dilution GC-MS without the interleaving machinery? This is the real threat. Internal standards already cancel much of the instrument-response drift. What they do not cancel are biases tied to one arm, such as oleate sticking to glassware in the presence of dot residue or the order of derivatization. The block-versus-interleaved comparison with label swaps measures exactly this. If the two designs give equal precision and equal null bias, the scattering-inspired contribution vanishes, and the project reduces to GC-MS exchange isotherms. That would still be the first direct measurement of cooperativity, but we would report it as such.

Why not just use NMR, including diffusion-resolved NMR? NMR measures bound and free ligand at once without derivatization, and that is a real advantage. However, peak broadening and overlapping signals from closely related cinnamate isomers limit integration near the step, which is where the resolution matters. Running NMR on the same aliquots shows directly whether it resolves the increments.

Why not just infer coverage from optical absorbance shifts? Kroupa et al. found that band shifts depend on how the ligands pack, not only on how many are bound. Using absorbance to infer coverage, and then using coverage to explain absorbance, is circular. Counting molecules independently is what breaks that loop.

Proposed Experiments

For each of four to six para-substituted cinnamic acids, we record a twelve-point titration isotherm in triplicate batches and fit the lattice model. There are four decisive checks. Mass balance must close, meaning oleate released equals cinnamate consumed within error. Both-blank null runs must be consistent with zero. The interaction parameter must increase in magnitude with substituent dipole and predict where the step sits for a held-out ligand. Finally, the pairing of coverage with absorbance shift across ligands should show that ligands with the same coverage can give different band shifts, which would quantify the packing effect Kroupa et al. described only qualitatively.