Structural characterization of the phospholipid stabilizer layer at the solid-liquid interface of dispersed triglyceride nanocrystals with small-angle x-ray and neutron scattering

Dispersions of crystalline nanoparticles with at least one sufficiently large unit cell dimension can give rise to Bragg reflections in the small-angle scattering range. If the nanocrystals possess only a small number of unit cells along these particular crystallographic directions, the corresponding Bragg reflections will be broadened. In a previous study of phospholipid stabilized dispersions of $\beta$-tripalmitin platelets [Unruh, J. Appl. Crystallogr. 40, 1008 (2007)], the x-ray powder pattern simulation analysis (XPPSA) was developed. The XPPSA method facilitates the interpretation of the rather complicated small-angle x-ray scattering (SAXS) curves of such dispersions of nanocrystals. The XPPSA method yields the distribution function of the platelet thicknesses and facilitates a structural characterization of the phospholipid stabilizer layer at the solid-liquid interface between the nanocrystals and the dispersion medium from the shape of the broadened 001 Bragg reflection. In this contribution an improved and extended version of the XPPSA method is presented. The SAXS and small-angle neutron scattering patterns of dilute phospholipid stabilized tripalmitin dispersions can be reproduced on the basis of a consistent simulation model for the particles and their phospholipid stabilizer layer on an absolute scale. The results indicate a surprisingly flat arrangement of the phospholipid molecules in the stabilizer layer with a total thickness of only 12 Å. The stabilizer layer can be modeled by an inner shell for the fatty acid chains and an outer shell including the head groups and additional water. The experiments support a dense packing of the phospholipid molecules on the nanocrystal surfaces rather than isolated phospholipid domains.

Figure 1

Freeze fracture electron micrograph taken of a tripalmitin suspension with S100 ($10\%$ tripalmitin, $2.4\%$ S100, $0.6\%$ NaGC, in H${}_2$O) showing the characteristic thin platelets of the $\beta$-triglyceride nanoparticles.

Figure 2

SAXS and SANS patterns for concentrated (10wt%) tripalmitin suspensions stabilized by S100 (black $\diamond$ and $\square$) or DLPC (blue $▿$ and $▵$). Dashed lines mark the positions of the 001 and 002 Bragg reflections for bulk $\beta$-tripalmitin. For the suspension with DLPC, the first five (two) orders of the inter-particle interference maxima in the SAXS (SANS) data are labeled by numbers. The red ($\circ$) SANS curve stems from a dilute suspension with deuterated tripalmitin-d98 stabilized with chain-deuterated DMPC-d54. For better clarity, the black ($\square$ and $\diamond$) and blue ($▿$) curves were divided by 1.5, 10 and 5, respectively.

Figure 3

Cryo-TEM exposure taken from an aqueous tripalmitin suspension stabilized with S100 ($10\%$ tripalmitin, $2.4\%$ S100, $0.6\%$ NaGC, in H${}_2$O). At tripalmitin concentrations above $4\%$ a fraction of the particles starts to self-assemble in stacks as shown here [7]. The image reveals the internal structure of the individual platelets that are composed of multiple molecular layers of tripalmitin [5].

Figure 4

Chemical structure of phospholipids with different degrees of saturation of the C${}_{18}$ acyl chains in S100. One should keep in mind that S100 does not consist of solely monoacid phospholipids but also of PC with mixed fatty acids. (a) 18:0 (DSPC), (b) 18:1 (DOPC), (c) 18:2, and (d) 18:3. All phospholipids possess the same head group. For the unsaturated phospholipids the first double bond in the acyl chains is always on the ninth carbon atom as measured from the carboxyl group facilitating a kinked or bent shape of the chains in their natural cis configuration. (e) depicts the structure of the bile salt NaGC used as a costabilizer.

Figure 5

Visualization of the simulation model for an assembly of nanocrystals. For efficient computations the current version of the program allows only the simulation of stacks in which the crystals are parallel to each other. For concentrated triglyceride suspensions this is a good approximation. The vector ${}_n{\mathbf{R}}_k$ points to the $k$th crystal in the stack. The stabilizer layer consists of two shells dedicated to the phospholipid chains and head groups, respectively. In the lower crystal for some unit cells the tuning fork shape of the tripalmitin molecules in their stable $\beta$ modification (cf. Fig. 10 in Appendix  2) is indicated.

Figure 6

$\mu$DSC heating and cooling runs for dilute ($3\%$) tripalmitin suspensions stabilized with the phospholipid DOPC (18:1) and purified soybean lecithin (S100).

Figure 7

Best simultaneous fits using model A for the SAXS and SANS patterns of dilute ($3\%$) tripalmitin suspensions stabilized with either DOPC or S100 (for clarity the measured dark green ($\diamond$), blue ($▵$) and black ($\square$) curves and their fits are divided by factors 2, $5,$ and 10, respectively). Panels (a), (b), (c) (for DOPC) and (d), (e), (f) (for S100) display the platelet thickness distribution and the electron and neutron SLDs for the inner and outer stabilizer layers (isl, osl) and the dispersion medium (dm).

Figure 8

Models for the molecular arrangement of the phospholipids in the stabilizer layer. (a) Model B: The phospholipid chains of DOPC and S100 reside in the inner layer, while the head groups stick out in the outer layer. H${}_2$O/D${}_2$O penetration is allowed for each layer. (b) Model C: The phospholipid molecules are situated only in the inner shell at the interface; the outer shell is virtually not existent since it is filled with the dispersion medium. (c) Model D: A fraction of the phospholipids resides completely in the inner layer and the remaining fraction occupies both the inner layer with its chains and the outer with the head groups. Water penetration is allowed for the outer layer.

Figure 9

Best simultaneous fits for the dilute ($3\%$) tripalmitin suspensions with DOPC or S100 obtained with model B for the stabilizer layer shown in Fig. 8a. As in Fig. 7, the fitted platelet thickness distributions of the $c_i$ and the fitted electron densities and neutron SLDs for the stabilizer shells and the dispersion medium are shown.

Figure 10

Visualization of the transformed triclinic unit cell of tripalmitin's stable $\beta$ phase used in the simulations (view to the $b$-$c$ plane). The structural data of $\beta$ tripalmitin from Ref. [65] are summarized in the table as well as the data for the transformed unit cell. The unit cell contains $Z=2$ tripalmitin molecules that adopt a characteristic tuning fork structure in the $\beta$-phase [66].

Figure 11

Simulated SAXS powder patterns for 100 tripalmitin platelets with statistically varying diameters and thicknesses for different spherical codes and point densities (i.e., number of orientations). For all tested spherical codes the particles are exactly the same. Patterns obtained with the Monte Carlo (MC) method spreading the orientations randomly on a sphere are taken as reference. Only if the $\sin \vartheta$ weighting factor is included with the grid algorithm is a reasonable agreement with the MC curves obtained.

Figure 12

Sketch for a dilute dispersion (D) with concentration ${\phi }_{\mathrm{cry}}$ comprising of individual platelets with thicknesses of 1, $2,$ and 3 unit cells. The other three spheres represent the simulated ensembles $i$ of platelets with thicknesses of $i$ unit cells. As in a typical simulation all ensembles possess the same concentration ${\phi }_{\mathrm{cry}}$ and the same number of members $M^i$. The sphere volumes represent the associated dispersion volumes corresponding to ${\phi }_{\mathrm{cry}}$. For the sake of simplicity the lateral platelet diameters are assumed in the drawing to be the same for all platelets; in a real suspension and in the simulation runs the diameters are subject to a statistical distribution. For this simplified exemplary suspension the volume fractions of the three ensembles are $c_1=2/9$, $c_2=4/9$, and $c_3=1/3$, respectively. For the simulated ensembles this corresponds to volume fractions of ${\tilde{c}}_1=1/3$, ${\tilde{c}}_2=1/3$, and ${\tilde{c}}_3=1/6$, respectively.

Figure 13

X-ray (brown and black curve) and neutron (orange and red) diffractograms for tripalmitin crystals in vacuum with different numbers of unit cells, $N_3$, perpendicular to the $\left(001\right)$ planes. The integrated peak intensity for neutron scattering is nearly independent of the number of unit cells due to its slowly varying structure factor in the Bragg reflection range [cf. Fig. 14a] providing rather symmetric peaks. In contrast, for SAXS the integrated peak intensity converges to the theoretical value only for larger crystal sizes perpendicular to the $\left(001\right)$ planes. The inset magnifies the $s$ range around the 001 Bragg reflection. Vertical brown and black dashed lines indicate the integration range for the 001 Bragg reflection for $N_3=10$ and 100, respectively, for both SAXS and SANS.

Figure 14

(a) X-ray and neutron structure factor ${\left|F\left(s\right)\right|}^2$ calculated in a 001 Bragg geometry for $\beta$-tripalmitin (PPP) and deuterated $\beta$-tripalmitin (PPP-d98) in vacuum in the small-angle range. The positions of the $00l$ Bragg reflections are also marked (magenta dashed lines). While for neutron scattering the structure factor for PPP is a slowly varying function in the range of the Bragg reflections (red $\circ$), the sharp minima of the structure factor for X-ray scattering are close to the Bragg reflections (black line). (b) Structure factors $|F_{00l}{|}^2={\left|F\left(s_{00l}\right)\right|}^2$ ($s_{00l}=l·0.2487$ nm${}^{-1}$) in the small-angle range ($l=1,2,3$) extracted from (a). For SAXS the 001 and 003 structure factors dominate, while the 002 peak is small and often disappears in the background. For SANS the 001 and 003 peaks are considerably weakened while the 002 reflection is enhanced. The latter effect is much more pronounced for PPP-d98, where experimentally only the 002 reflection can be observed with SANS (cf. Fig. 2).

Figure 15

Estimated total stabilizer layer thickness (values on the color scale are given in Å) for platelets with diameter $D$ and height $h$ according to Eq. (I10). The plot shows that for diameters in the range of $50–200$ nm and thicknesses of 2–3 unit cells ($8–12$ nm) the expected total stabilizer layer thickness is typically in the range of $8–14$ Å.