Cette thèse a pour but de démontrer la faisabilité de la technique de Nucléation non-Photochimique Induite par Laser (NPLIN) appliquée aux composés pharmaceutiques organiques. Avec nos résultats expérimentaux, ceux obtenus dans la littérature et calculs théoriques ab initio, nous avons discuté du mécanisme de la méthode NPLIN.Cette thèse a décrit en premier lieu, le nouveau montage expérimental semi-automatisé adapté aux exigences des études NPLIN développé à CentraleSupelec. Des expériences NPLIN sur le sulfathiazole (STZ), l-acid glutamique (LGA) et la glycine (GLY) sont menées pour étudier l’impact des paramètres du laser et de la sursaturation des solutions sur leur cristallisation. Les résultats expérimentaux montrent que la technique NPLIN permet d’obtenir des cristaux de STZ, LGA and GLY. L’efficacité de la nucléation croit avec l’augmentation de la densité d’énergie du laser et de la sursaturation. Un indice nouveau Ind50 a été défini correspondant au couple densité d’énergie/sursaturation où 50% nucléation est atteinte. Son comportement est discuté. Il est trouvé que le nombre cristaux STZ induits par laser dépend linéairement avec le temps d’irradiation. Une dépendance du polymorphe des cristaux induit par le laser avec la polarisation du faisceau laser est également découvert pour STZ et GLY. Un autre indice Det(A) est utilisé pour caractériser l’impact de la polarisation sur le polymorphisme. Des calculs théoriques ab initio par le logiciel Gaussian09 permettent de donner une estimation des énergies d’interaction des différents dimères pour des polymorphes de STZ, LGA, GLY, l-histidine (LH) et urée. L’empilement des molécules dans les clusters pré-existant est prédit en accord avec la détermination des énergies d’interaction. L’analyse de corrélation entre la symétrie d’empilement et des résultats expérimentaux souligne l’hypothèse de l’effet Kerr pour expliquer cet impact de la polarisation du faisceau laser sur le polymorphisme. / This thesis concerns the demonstration of the feasibility of Non-Photochemical Laser Induced Nucleation (NPLIN) of some organic pharmaceutical compounds. Using our experimental results and those obtained in literature together with ab initio theoretical calculations we have been able to discuss the mechanism of the NPLIN method.This thesis presents a new experimental set-up developed at CentraleSupelec and dedicated to perform NPLIN experiments. NPLIN experiments on sulfathiazole (STZ), l-glutamic acid (LGA) and glycine (GLY) have been carried out to examine the impact of laser parameters and solution supersaturation on their crystallization. Experimental results show that crystals of STZ, LGA and GLY have been obtained by means of NPLIN. For these compounds, nucleation efficiency increases with laser power density and solution supersaturation. A new index Ind50 corresponding to the couple (energy density/supersaturation) where 50% of nucleation is teached, has been defined. Its behavior has been discussed. It was found that laser induced STZ crystal number depends almost linearly on exposure duration. Moreover, for STZ and GLY, a dependance of laser induced crystal polymorph on laser polarization has been found. Another new index Det(A), has been used for characterization of the impact of the polymorphism. Ab initio quantum computations using Gaussian09, provided an interaction energy estimate for different dimers in different polymorphs of STZ, LGA, GLY, l-histidine (LH) and urea. Packing mode in pre-existing clusters is predicted in agreement with interaction energy determinations. Correlation analysis between packing symmetry and experimental results, shed new light on the Kerr effect hypothesis relative to the impact of laser polarization on polymorphism.

NPLIN was studied in supersaturated solutions (S = 1.06) of potassium chloride (KCl) and bromide (KBr). The fraction of samples nucleated (f) follows a nonlinear dependence on peak power density that approaches f = 1 at higher incident powers. It is shown that a lower threshold power is required for nucleation using 532 nm laser pulses than at 1064 nm, and that a higher fraction of samples nucleate when exposed to 532 nm pulses at a given laser power. Comparison with KCl shows higher fractions of KBr samples nucleate with lower threshold values at both wavelengths. Samples of KCl of equal supersaturation at two different temperatures (23 and 33 °C) exposed to 1064 nm pulses show that those at 33 °C are significantly more labile to nucleation. The ratio of samples nucleated at 33 °C compared to those at 23 °C was 2.11 ± 0.47. A classical nucleation model based on activation of subcritical solute clusters accounts remarkably well for the experimental data and provides phenomological values of the crystal–solution interfacial tension (γ) at 23 °C for KCl and KBr of 5.283 and 4.817 mJ m-2. At 33 °C, the model yields a best-fit value of γ = 5.429 mJ m-2 for KCl. As an extension of this work the use of an evanescent wave (ew) generated by total internal reflection was investigated as a method to cause nucleation in supersaturated KCl solution. Evanescent wave NPLIN (ew-NPLIN) was shown to cause nucleation. The results showed a higher laser-power threshold required for nucleation and sample lability greater than that of bulk NPLIN. In a second approach to understanding NPLIN, the structures of concentrated solutions were probed by a series of laser scattering experiments. Evidence showing populations of particles in solution was provided by Rayleigh laser scattering (RLS) experiments. Scattering in solutions prepared to be nearly saturated (S = 0.95) was observed using a low magnification (×10) microscope objective; almost all solutions showed the presence of scattering objects moving freely in solution. For those that showed no particles, it was noted that the solutions were typically of higher solute concentration (> 11 mol% solute). Ammonium nitrate solution showed no particles using ×10 magnification, however particles were identified when higher magnification was used (×50 and ×100). Video footage of the Rayleigh scattering observed in aqueous solutions of glycine, urea and ammonium nitrate obtained using ×50 magnification were analysed using a custom nanoparticle tracking software. The results showed a population of particles in aqueous urea and glycine solutions with particle concentrations of the order 108 particles cm-3 and mean hydrodynamic diameter of approximately 267 ± 1 and 173 ± 2 nm respectively. Not enough particles were identified in ammonium nitrate solution to complete the tracking analysis; however a fluctuating background scatter suggested a population of particles with sizes below the limit of resolution of the optical system. Using aqueous urea solution as a model system the structure of the particles identified in solution was investigated using scanning microscopy. The second-harmonic scattering (SHS) signal measured in concentrated aqueous urea solution was measured as a function of solution concentration (C) over a range of saturation conditions from undersaturated (S = 0.15) to supersaturated (S = 1.86). The results show a non-linear increase in SHS signal with local maxima near S = 0.95 and 1.75 suggesting a change in solution structure near these points. Rayleigh scattering images indicate the presence of particles in nearly saturated (S = 0.95) urea solutions. Time-dependent SHS measurements indicate that signals originate from individual events encountered during scanning of the sample through the focal volume of the probe laser, consistent with second harmonic generation (SHG) from particles. SHG from aqueous dispersions of barium titanate (BaTiO3) nanoparticles with diameters < 200 nm, showed signals ~20 times larger than urea solutions. The results suggest the existence of a population of semi-ordered clusters of urea that changes with solution concentration.

The aim of this thesis was to discover the mechanism for non-photochemical laser-induced nucleation (NPLIN), which is a technique for inducing nucleation of crystals with laser light without absorption. The mechanism of the optical Kerr effect (OKE) was suggested by Garetz et al. [Physical Review Letters 77, 3475–3476 (1996)] to give an explanation for NPLIN. Since the feasibility of the OKE mechanism for NPLIN has been questioned, a series of experiments on NPLIN of aqueous supersaturated urea were carried out to quantify the relationship between crystal alignment and laser polarization. Digital imaging of crystal growth during laser irradiation showed that nascent needle-shaped crystals of urea were not aligned with the direction of the electric field of the laser. Additionally, work on glycine was aimed at verifying the possibility to control the polymorph of the obtained crystal via the laser polarization. However, our finding shows that the probability of γ-glycine is more likely to increase with increasing supersaturation; and the influence of laser polarization on the resulting morphologies is not strong as reported by Sun et al. [Crystal Growth & Design 6, 684–689 (2006)]. Furthermore, in another work on NPLIN of L-histidine, based on Sun et al. [Crystal Growth & Design 8, 1720–1722 (2008)], we were unable to reproduce the results as stated in Sun’s published paper. We find their results exhibit a large uncertainty when recalculated through the Wilson score interval for binomial distributions. On account of these revised uncertainties, it is unlikely that laser polarization gives polymorphism control. Comparison with the nucleation probability of unfiltered samples in the work of urea and glycine shows that the number of filtered samples nucleated as a result of NPLIN was largely decreased. Moreover, experiments on NPLIN of NaCl and KCl also exhibited that the number of filtered samples nucleated was significantly lower than that of unfiltered samples. This downward tendency in nucleation probability after filtration cannot be explained by Garetz’s OKE mechanism. On account of this, an alternative mechanism named particle-heating mechanism was proposed, and supported by experiments on NPLIN of sodium acetate. Sodium acetate experiments showed that the crystallization of sodium acetate can be induced by a single pulse of a nanosecond laser (1064 nm) with minimum power of 0.1 J cm−2. As discovered by Oliver et al. [D. Oliver, PhD Thesis, Edinburgh University, 2014], anhydrous or trihydrate sodium acetate can be formed under the effects of different organic and inorganic additives, such as poly- (methacrylic acid) and disodium hydrogen phosphate. We demonstrate that crystalline growth velocities and crystal morphology can be influenced by these additives. We find that high levels of additive cause only nucleation of bubbles. By counting the number of crystals, which is approximately consistent to the number of bubbles observed, video microscopy of laser-induced crystallization of sodium acetate has revealed that the general mechanism of NPLIN is most likely to be caused by a particle-heating mechanism. Chapter 8 of the thesis describes a number of solute molecules that were tested using NPLIN, but failed. In terms of improvements for future work or a perspective on NPLIN, detailed suggestions have been described in Chapter 9, which also gives a summary of all work presented.

No / We demonstrate that nanofiltration of aqueous glycine solutions has a pronounced effect on laser-induced nucleation. Two nucleation regimes were observed in nonfiltered, irradiated solutions under isothermal conditions: a rapid initial regime associated with laser-induced nucleation and a second much slower spontaneous nucleation regime. Filtration of the solutions prior to irradiation greatly suppressed the rapid regime, while the slow regime was similar regardless of filtration or irradiation, for all supersaturations studied. A clear effect of filtration on crystal polymorphism was also observed. Nonfiltered irradiated solutions at a lower supersaturation almost exclusively yielded the α-polymorph, while at higher supersaturations there was significant presence (∼40%) of the γ-polymorph. On the other hand, filtered solutions almost exclusively yielded the α-polymorph of glycine at all supersaturations studied. These surprising results challenge some established ideas about laser-induced nucleation, showing that previously reported laser-induced nucleation phenomena in glycine aqueous solutions can be effectively suppressed by filtration, so that the underlying mechanism is unlikely to be based on molecular scale interactions involving just the solute and the solvent alone. Instead, laser-induced nucleation in this system appears to be related to either colloidal scale solution clusters or foreign solid or molecular impurities that can be removed by nanofiltration.

Advances in molecular-simulation methods allow for ever larger systems of particles to be studied and on longer timescales. Calculations are reaching such a scale that they can be used to address a vast range of key questions across chemistry, physics, and engineering. In this work, molecular dynamics and Monte Carlo simulations are employed to address two key areas: the structure and dynamics of simple aqueous ionic salt solutions at high concentrations; and the structure, dynamics, and phase behaviour of dipolar fluids (such as colloidal ferrofluids). The first part of the work begins with a study of the structure and dynamics in metastable, supersaturated, aqueous solutions of potassium chloride, and the possible relevance of these to the phenomenon of non-photochemical laser-induced nucleation (NPLIN). It is thought that the potassium and chloride ions form long-lived, amorphous clusters that may, under the influence of nanosecond laser pulses, undergo structural reorganisation to form post-critical crystal nuclei. It is found that spontaneous nucleation does not occur on the simulation timescale, but that amorphous clusters do form with cluster lifetimes comparable to those of the shortest laser pulses that can be used in NPLIN ( 100 picoseconds). Next, an alternative scenario for NPLIN involving rapid laser heating of impurity particles is examined by simulating heated carbon nanoparticles in saturated aqueous solutions of sodium chloride. The concentration at which an aqueous sodium chloride solution first crystallises on the simulation timescale is determined. A spherical carbon impurity is then added to a system with concentration close to, but lower than, the concentration at which crystallisation occurs on the simulation timescale. The effects that adding, and heating, this impurity has on the structure of this near-crystallising system are then observed. The second part of the work discusses model dipolar fluids, of direct relevance to colloidal ferrofluids (suspensions of magnetised nanoparticles in simple carrier liquids). The two-body, dipole-dipole interaction is long-ranged and anisotropic, and it is computationally expensive to handle in molecular simulations. Here a new method is proposed that relies on a formal mapping between the partition function of a dipolar fluid and that of a hypothetical fluid with many-body, short-ranged, isotropic interactions. Only the leading-order two-body interactions (akin to the van der Waals attraction) and three-body interactions (corresponding to the Axilrod-Teller potential) are retained. It is shown that this simple model is sufficient to reproduce the characteristic particle chaining and the associated disappearance of the vapour-liquid phase transition of dipolar fluids. Finally, the dynamical response of ferrofluids to oscillating magnetic fields (the dynamic magnetic susceptibility [DMS]) is studied. The DMS of ferrofluids, predicted by a new theory that takes into account the leading-order effects of dipole-dipole interactions, are critically compared to those found using Brownian-dynamics simulations of monodisperse systems of dipolar particles. This new theory is found to provide more accurate predictions of the DMS than previous theories, with the DMS predicted to a high degree of accuracy for systems with dipolar coupling strength in the experimentally achievable region.

The practical application of non-photochemical laser-induced nucleation (NPLIN) to continuous flow was investigated. Supersaturated aqueous solutions were screened with a 5 ns pulsed laser (532 nm 44 MW cm-2) for NPLIN activity. Upon irradiation succinic acid nucleated at S20 = 4.3 and adipic acid at S20 = 2.0 - 3.0. NPLIN activity is reported for the first time in nicotinic acid (S20 = 2.6 - 3.0). No overall pattern was observed of chemical structure on NPLIN activity. From inorganic compounds similarly screened, ammonium chloride (S20 = 1.04 - 1.20) was identified as most suitable for further tests. It was shown to have an increase of NPLIN crystals with higher supersaturation from 13 at S = 1.038 to 252 at S = 1.135. A quadratic increase in number of crystals with increased laser power. The effects of NPLIN upon ammonium chloride are diminished upon filtration through a 0.2 μm poly (ether sulfone) filter, reducing the number of crystals from 350 to 10 per 70 mJ pulse (25 MW cm-2). The use of NPLIN in continuous flow was demonstrated from the first time. A S23 = 1.1 solution of aqueous ammonium chloride in flow produced crystals when irradiated by 10 pulses s-1 of a 1064 nm 6 ns laser. When the laser was stopped, crystals were no longer produced and the system returned to flowing supersaturated solution. Lab scale apparatus for continuous NPLIN experiments was developed. A design involving a re-dissolution step and loop flow was constructed for both laminar and slug-flow regimes. Nucleation of ammonium chloride (S = 1.1) was demonstrated in both systems. Repeatable NPLIN experiments were hindered by spontaneous nucleation. Spontaneous nucleation in flow was observed around areas where supersaturated solution passed from one component to another. Spontaneous nucleation was also observed upon cooling (25 to 10 °C). Filtration was observed to both suppress NPLIN and spontaneous nucleation in flow.