Salts induce the agglomeration and assembly of gold nanocrystals that are stabilized by charged ligand shells. This destabilization is known to partially deviate from the predictions of classical DLVO theory for larger colloids, but existing studies focus on limited concentration ranges or ion types. Here, we use a high-throughput approach to test the agglomeration efficacy of 17 different salts at concentrations ranging from 0.16 mM to 2 M on negatively charged nanocrystals with shells of 11-mercaptoundecanoic acid and/or triethylene glycol mono-11-mercaptoundecyl ether. Automated pipetting is used to create a large dataset of close to 10000 UV–vis absorbance spectra. We analyze the spectral shifts to find the onset of agglomeration, identify critical salt concentrations, and characterize the nature of the agglomeration transition. The results are compared to classical DLVO theory using conventional analysis, and the effects of ion concentration and anion valency are shown to be consistent with DLVO predictions. Cation valencies only partially follow the predictions, suggesting local ion-specific interactions that dominate when screening lengths reach molecular length scales. A Random Forest Regression model is used as additional “black box” analysis of the results to correlate the ionic strength, ligand shell composition, and intrinsic ion properties, and to rank their relative importance to the colloidal stability of gold nanocrystals. The ranking combines classical DLVO effects and specific ion interactions with subtle effects on the agglomerate structure that affect plasmon resonance shifts, providing a complementary interpretation of the data.
Nexha et al. (2026) studied this question.