A dual growth mode unique for organic crystals relies on mesoscopic liquid precursors

The study investigates how organic solvents, where van der Waals interactions dominate, influence crystal growth mechanisms compared to aqueous systems governed by hydrogen bonding and polar interactions. It addresses whether organic crystals grow by classical monomer-by-monomer addition or via nonclassical pathways involving precursors. Focusing on etioporphyrin I crystallized from neat 1-octanol, the authors note that prior nonclassical growth observations largely involve aqueous media, leaving a gap for organic systems. They hypothesize that weaker, less directional interactions in organic solute–solvent pairs may favor liquid precursors and nonclassical growth, and seek to identify molecular-level parameters steering the transition between classical and nonclassical regimes, as well as clarify the role of liquid droplets/clusters versus amorphous particles in crystal growth.

Classical nucleation and growth theories posit sequential monomer attachment to nuclei and steps, supported by experiments at moderate driving forces. However, numerous systems (proteins, biominerals, organic compounds) exhibit nonclassical behaviors: nucleation within dense liquids, growth via preformed molecular assemblies, and oriented attachment of nanocrystallites. Prior suggestions link nonclassical transitions to specific conditions (solvent choice, additives, acidity, supersaturation) and to strong nonspecific attractions distinct from crystallization-driving forces. There is also a disconnect between nonclassical nucleation (often in liquids/clusters) and growth (often framed as nanocrystallite or amorphous particle attachment). The present work situates etioporphyrin I within this context to test nonclassical growth in an organic medium.

- Bulk crystallization: Supersaturated etioporphyrin I solutions (0.5 mM) prepared in 1-octanol at 70 °C, filtered (0.22 µm PTFE), concentrations measured by UV–Vis. Solutions cooled slowly to room temperature over glass disks to grow crystals. - UV–Vis spectroscopy: DU 800 spectrometer (200–800 nm); calibration via Soret/Q-bands with OD at 492 nm for concentration; solubility (Ce) determined at 29 °C by repeated equilibration and measurement. - Scanning electron microscopy: Crystals grown by slow cooling, rinsed, Au-coated (10–20 nm), imaged at 2×10⁻⁵ mbar, 3 kV. - Microfluidics growth-rate measurements: Crystals on glass disk placed in PDMS channel; growth solutions (0.25–0.5 mM) flowed at 0.1 mL/min; imaging every 5 min (Leica DMi8); growth rate R of axial {101} faces determined by linear regression across 20 measurements from 5–10 crystals; constant concentration maintained by continuous flow. - In situ AFM: Bruker Multimode (tapping mode, SNL-10C probes, 32 kHz). Temperature in liquid cell 29 ± 0.1 °C. Monitored {010} faces with spiral steps (h = 0.99 ± 0.10 nm). Time-resolved imaging to track step displacements and velocities v versus concentration C, and to visualize precursors landing, evolving into multi-layer stacks, merging, and dislocation emergence. Tested step–step interactions and supply pathways by comparing single/double steps and varying separations; assessed whether solute reaches steps via surface diffusion or direct incorporation. - Oblique illumination microscopy (OIM/Nanosight LM10-HS): Tracked mesoscopic aggregates in undersaturated solutions at 29 °C using 532 nm laser, 30 fps. Extracted particle trajectories, diffusion coefficients, hydrodynamic radii Re, and number densities N in a known imaging volume (120 × 80 × 5 µm³). Evaluated dependence of Re and N on initial concentration C0; assessed equilibrium concentrations Cr after centrifugation and removal of clusters. - Transmission Mueller matrix imaging polarimetry: Custom dual rotating compensator microscope on Zeiss Z1 Observer. Measured linear retardance (LR) at 500 nm for crystals grown at C = 0.3 mM (classical-dominant) and 0.5 mM (nonclassical-dominant) under mother liquor to evaluate optical anisotropy and strain-related quality. - Additional analyses: All-atom molecular simulations and absorption spectroscopy (DFT-supported) to probe etioporphyrin I dimer formation energies/geometries (details referenced to SI). Theoretical estimates of translational/rotational diffusivities to assess feasibility of oriented attachment for ~100 nm vs ~1 nm precursors.

- Growth kinetics: For axial {101} faces, R increases quasi-exponentially for C just above Ce = 0.25 mM, reaching ~20-fold higher at C = 0.35 mM (C/Ce − 1 = 0.4), then linearly for 0.35 < C < 0.47 mM (C/Ce − 1 up to 0.88), with some slowdown above 0.47 mM due to solute supply limits. No dead zone observed. - AFM step dynamics: {010} faces show spiral growth from screw dislocations. Step velocity v is steady and linear in C up to C/Ce − 1 ≈ 0.6. Step edges are kink-rich near the thermodynamic limit at Ce, arguing against kink scarcity as a cause of superlinear R(C). Step velocities for single vs double steps and for closely spaced steps are similar (~20 nm/s), indicating lack of step–step supply competition. - Exclusion of alternative mechanisms: Data are inconsistent with two-dimensional nucleation dominance, kink scarcity, or step pinning by inhibitors as causes of quasi-exponential R(C). - Liquid precursor mediation: In situ AFM at supersaturation (C/Ce − 1 = 0.56) reveals ~400 nm diameter, 6–12 nm height dome-like precursors landing on {010} surfaces, growing, flattening within ~5 min into stacks of ~25–50 layers that spread laterally aligned with the substrate lattice. Stacks from neighboring precursors merge seamlessly; dislocations (spiral steps) emerge on flat tops. - Direct solute supply: Lack of v dependence on step spacing implies solute incorporation occurs directly from solution to kinks, bypassing adsorption and surface diffusion. This enables rapid lateral spreading of densely spaced layer stacks without supply-field overlap stalling. - Strain assessment: Steps on precursor-generated flat tops grow 20–40% slower than on underlying crystal; layers within a stack grow about half as fast. Thicker stacks (later times) show faster steps. Strain likely due to slight lattice misfit and higher condensate viscosity but not sufficient to suppress growth. - Mesoscopic clusters in solution (OIM): Undersaturated solutions (C up to 0.06 mM ~ 4× below Ce) contain dynamic aggregates with narrow size distribution, Re = 75 ± 12 nm, stable for ≥6 h. Number density N decreases strongly with C (from ~1.2×10⁹ to ~0.02×10⁹ cm⁻³ as C decreases from 0.06 to 0.02 mM), while Re remains ~constant, consistent with mesoscopic solute-rich clusters in dynamic equilibrium. Clusters occupy a minute volume fraction (φ ~ 10⁻⁷ at 0.06 mM). Equilibrium concentration after cluster removal, Cr, increases with initial C0 (Cr ≈ C0 but not constant), inconsistent with standard phase equilibria and consistent with cluster models involving transient oligomers. - Origin of clusters: Simulations and spectroscopy support weakly bound, mostly parallel etioporphyrin dimers with binding energies of several kBT, sufficient to drive mesoscopic cluster formation. Weak, short-lived oligomeric interactions (π-stacking, quadrupole, dispersion) promote cluster formation in organic media. - Oriented attachment unlikely for ~100 nm precursors due to insufficient rotational diffusion within the short-range interaction zone; feasible only for ~1 nm scale units. - Amorphous particles: Some particles landing on crystal surfaces shrink (dissolve) over 10–15 min despite supersaturation and nearby growing steps, indicating a higher free energy amorphous solid phase that does not integrate. Mesoscopic clusters likely age into amorphous particles in solution absent a crystal template; such particles are poor growth precursors and potential defect sources if entrapped. - Crystal quality: Mueller matrix polarimetry at 500 nm shows similar linear retardance for crystals grown predominantly classically (0.3 mM) and nonclassically (0.5 mM): LR ~0.019 vs 0.023, indistinguishable within pathlength accuracy, indicating that fast, nonclassical growth does not degrade optical quality.

The work demonstrates that in an organic solute–solvent system (etioporphyrin I in octanol), crystal growth proceeds via a dual mode: classical dislocation-driven monomer addition and nonclassical liquid-precursor-mediated layer stack formation. The quasi-exponential rise of R with C near Ce and linear regime at higher C are explained by concentration-dependent numbers of mesoscopic liquid precursors that land and transform into crystallographically aligned multilayer stacks. AFM and step-kinetics analyses show direct incorporation from solution without surface diffusion, enabling rapid lateral spreading of closely spaced steps—behavior facilitated by weak solute–solvent interactions typical of organic media. The identification of mesoscopic solute-rich clusters as precursors connects nonclassical nucleation (in clusters/dense liquids) to growth. Oriented attachment is unlikely for the observed precursor sizes, supporting a liquid-to-crystal ordering upon surface contact. While stack tops show modestly reduced v indicating some strain, polarimetric measurements reveal no detrimental effects on optical quality, suggesting fluidity of the precursors enables seamless lattice integration. Amorphous particles resulting from cluster aging do not integrate and may seed defects, highlighting the importance of controlling precursor aging and delivery. Overall, the findings clarify molecular conditions favoring nonclassical growth in organic systems and suggest that weak, short-lived oligomeric interactions and direct solute supply pathways are key determinants.

The study reveals a rare dual classical–nonclassical growth mode for organic crystals: mesoscopic, liquid etioporphyrin-I-rich clusters land on crystal surfaces, order in registry, and generate stacks of up to ~50 layers that then spread laterally while incorporating solute directly from solution. Weak intermolecular interactions in organic media favor both the formation of such clusters (via transient oligomers) and direct step supply, making nonclassical growth likely prevalent in organic crystallization. Rapid growth at moderate supersaturation is achievable without compromising optical quality. Amorphous particles, likely formed by cluster aging in solution, do not integrate and may introduce defects, indicating that not all particulate precursors are beneficial. Future directions include mapping solvent/solute conditions and supersaturation to control the balance between classical and nonclassical modes, strategies to promote fresh liquid-cluster delivery while limiting aging to amorphous solids, and extension to other organic systems to generalize these mechanisms.

- Cluster characterization at higher concentrations was limited by strong absorption of etioporphyrin I at the OIM wavelength, restricting quantitative N and Re measurements above ~0.06 mM (well below Ce). - Some thermodynamic interpretations assume activity coefficients near unity at C < 1 mM; deviations could modify quantitative estimates of driving forces. - AFM and microfluidic measurements focus on specific faces ({010}, {101}) and a single solute–solvent pair at ~29 °C; generality to other systems and temperatures requires further validation. - Simulation and spectroscopy evidence for dimers inform the mechanism but detailed oligomer distributions and lifetimes are not fully resolved within the main text. - Oriented attachment feasibility analysis is theoretical; direct in situ orientation dynamics of sub-10 nm species were not observed in this system.