Results and discussion
In principle, the entire dataset can be evaluated globally, and efforts are underway to incorporate such routine into the Ultrascan evaluation software package (Demeler 2005). However, even then we are confronted with a confounded polydispersity of both optical and colloidal/hydrodynamic properties. In the left side of Fig. 2, the assumed core-shell structure of a β-carotene microparticle is shown (Auweter et al. 1999). Such a complex hybrid particle exhibits several levels of polydispersity, which impact the distribution of sedimentation coefficients observed in an AUC. They are for example:diameter of the inner core;
chemical composition, esp. oil content, of the inner core;
concentration of the adsorbed protection colloid (gelatin);
degree of swelling of the gelatin.
Parameters 1 and 2 determine the optical properties and bioavailability that are decisive for the commercial application profile. For smallest particle sizes, β-carotene is an H-aggregate while for the biggest particle sizes β-carotene forms J-aggregates. Intermediate particle sizes are assumed to integrate H- and J-aggregates in differing ratio (Auweter et al. 1999). Parameters 3 and 4 determine the thickness of the protection colloid layer, which is typically 40 nm in pure water. The buoyant density of gelatin is rather high (above 1.3 g/cm3), and cannot be matched with a non-interfering solvent such as heavy water.
All parameters 1–4 enter into the calculation of the effective density and the hydrodynamic diameter of the hybrid particle. The frictional force under sedimentation depends on the ion concentration and pH because the gelatin may collapse or swell thus changing the effective frictional forces (and thus changing the observable sedimentation constant) although the chemical composition and buoyant density, which in principle could be measured in a Krattky gauge or density gradient, stay the same. The swelling of the gelatin corona alone impedes an exact conversion from measured sedimentation constants to hydrodynamic diameters. Considering that also parameters 1 and 2 contribute to the polydispersity in the observable distribution of sedimentation coefficients, we decided to limit ourselves to a conservative evaluation on the level of sedimentation fractions, not sizes.
We now discuss the optical properties that result from the specific colloidal microstructures as discussed above. Due to different preparation conditions, the morphology of the β-carotene core changes. H- and J-aggregates have different UV/vis spectra, shown as visual impression on the right side of Fig. 2. Figure 1 dot line curve shows the UV/vis spectrum of 0.05 g/l product without any ultracentrifugation. Four peaks at 288, 449, 478 and 518 nm can be seen. The three peaks in the visible can be attributed to the 1Ag− (S0)–1Bu+ (S2) transition with the vibrational progression 2–0, 1–0, 0–0 of the C–C stretch vibration along the alternatingly double bonded electronically conjugated backbone of the carotenoid (Polivka and Sundstrom 2004).The UV peak partially can be attributed also to the carotenoid transition 1Ag−–1Ag+, which is forbidden by symmetry, but becomes allowed in the crystalline assembly. The spectrum of the composite particle indicates the β-carotene J-aggregate (Auweter et al. 1999). Figure 1 solid curve shows the pure gelatin spectrum for 1 g/l. It can be seen that gelatin only contributes to the UV region of the spectra below 280 nm. However, the contribution of gelatin is vanishing compared to the three times stronger absorption of the composite sample at 20 times lower overall concentration. Another component that presumably contributes to the UV absorption is the ascorbylpalmitat dispersant that is added during the co-precipitation.
Figure 3 shows four of the 40 experimental scans. If we put all these 40 scans in sequence, we can form a 3D movie of the sedimentation process. This movie is available as supplementary information. Figure 3.1 shows scan 1 where particles just have been transferred from the reservoir to the sample column. The baseline offset is 0.05 (purple) due to the absorption calculation with an empty cell as reference. We see two main peaks. The peak in the visible region is assigned to β-carotene. In the UV range, there is an overlay of two peaks, one is the UV peak of β-carotene (see Fig. 2) and the other is the UV signal of gelatin. After 15 min of sedimentation, fractionation of the sample is obvious and the first sedimentation fraction proceeds to the bottom of the cell. Scan 10 (15 min) is the last scan where the entire particle range can be seen before the first particles reach the bottom. If we compare the height of the peak in the UV and visible region at different radial positions, the ratio changes. For the faster sedimenting particles, the ratio of β-carotene to gelatin is higher. This is the first important result, demonstrating that the sample is not homogenous. Instead, the particles change their colloidal properties in correlation with the optical properties. The observed effect can be explained by a higher content of stabilizing agent that induces smaller particle sizes. Note that the shape of the peak at 288 nm (Fig. 1) does not exactly match the gelatin absorption and that the expected contribution of gelatin is weak at the applied concentrations, hinting at a combined action of both gelatin and the ascorbylpalmitat added during the co-precipitation in particle synthesis. The third part of Fig. 3 shows the scan after 27 min. Here, the fastest particles have sedimented already. In the fourth part of Fig. 3, we see the last fraction that remained after 60 min of sedimentation, which is mainly composed of gelatin. However, some β-carotene absorption is still visible, which seems to be solubilized in small amount by the excess gelatin or excess ascorbylpalmitat. We do not detect free gelatin in the analysis. In an independent experiment we measured the characteristic sedimentation behavior of gelatin with the interference optics of the Beckman XLI AUC at 44,000 rpm. We found a sedimentation constant distribution from 2 to 10 Sved. This confirms our assignment that the last fraction cannot be pure gelatin.
To summarize the global evaluation, Fig. 3 demonstrates the power of MWL-AUC. We can differentiate particles, observe the full UV/vis spectra of the particles and can already draw conclusions about the different components in the complex sample mixture without any further evaluation, as the Y-axis shows the full UV/vis wavelength range. This is a key feature of MWL-AUC. Such analysis was impossible in analytical ultracentrifugation experiments before.
We can also use projections of the data onto individual axes and proceed thus to a more quantitative evaluation. In order to calculate the full s-distribution of all particles, we have selected scan 10 for all further evaluation as this scan shows fractionation of the mixture while no particles are yet lost due to complete sedimentation. More information is potentially available with a global evaluation of the entire dataset. In Fig. 4, the s-distribution of the particles is shown for 5 different representative wavelengths out of 330 (250–750 nm with a wavelength resolution of 1.5 nm). We have selected the wavelengths according to the peaks of the β-carotene microparticles: 260, 280, 450, 480 and 520 nm in Fig. 1.
Fig. 4 Sedimentation coefficient distributions at different wavelengths
The s-distribution is obviously very broad. Due to the chemical heterogeneity of the particles and the resulting density distribution, it is not possible to convert the sedimentation coefficient to the particle size. However, all important sample characteristics can be discussed for the s-distributions. From Fig. 4, we conclude that there are at least three fractions in the sample: small hybrid particles with s < 25 S, a main fraction around 100 S and larger particles around 200 S. Their absorption spectra (and chemical composition) are clearly different as can be seen in Fig. 5.
Fig. 5  Top Normalized UV/vis spectra of particles with different sedimentation coefficients, bottom zoom the range around 450 nm and peak positions of 10.6 S (448 nm) up to 232 S (439 nm)
In Fig. 5, seven representative UV/vis spectra are shown. The spectra agree well with those of pure H-aggregates (Auweter et al. 1999). However, the original sample contained J-aggregates too (Fig. 1, dashed line). We believe that the J-aggregates were already precipitated before the first scan in the AUC cell was taken. Indeed precipitation of particulate material inside the reservoir of the vinograd cell was observed visually after cell disassembly following the experiment. However, irrespective of the actual nature of the particulate material that remained in the Vinograd cell reservoir, the result stands for itself that the coloristic polydispersity (Fig. 1) is not due to an intra-particle but inter-particle distribution of morphologies (Fig. 6). This result is contrary to the previous assumption that is sketched in Fig. 2, where H/J-aggregates would coexist in the particles.
Fig. 6 Ad hoc structural model of the β-carotene microparticle system on basis of the presented AUC results. The different color of the samples does not originate from the intraparticular coexistence of H- and J-aggregates as assumed before (Fig. 2) (Auweter et al. 1999) but instead separate particles contain pure H- or J-aggregates and the concentration ratio between these particles determines the colour of the final sample. Note the difference to Fig. 2
The peak around 520 nm slightly shifts to lower wavelength with increasing sedimentation coefficient and the peak height also decreases. The same is true for the peak at 480 nm. Therefore, this excitation of β-carotene microparticles decreases with increasing sedimentation coefficient. For the peak at 450 nm, only the spectral shift to lower wavelength is observed with increasing sedimentation coefficient. The effects of this inhomogeneity were not known before in such detail. We suspect that the displacement of the electronic potential energy surfaces changes, such that the Frank–Condon—factors change for the vibrational progression, due to the changing incorporation of the chromophore into the partially crystalline assembly.
As discussed above, an overlay of the signal from β-carotene, gelatin and the ascorbylpalmitat is observed in the UV region. In this range, a shift of the peak maximum to higher wavelength with increasing sedimentation coefficient is detected. In addition, a drastic decrease of the peak height relative to the 450 nm peak is observed with increasing sedimentation coefficient. The particles that sediment slower show stronger UV absorption, which we attribute to a higher content of ascorbylpalmitat, and hence smaller particle diameters due to the formation process in co-precipitation Auweter et al. 1999). To our knowledge, it is the first time that relatively small steps of spectral shift among an H-aggregate are shown for composite particles. All of the 200 different detected spectra follow the same trend. Although the spectral changes appear to be continuous, that does not exclude defined and different spectra for different particle populations as the detected raw signals are those of a sample band, which is broadened by polydispersity in size, composition and diffusional broadening.