Characterisation of silica nanoparticles prior to in vitro studies: from primary particles to agglomerates

Abstract

The size, surface charge and agglomeration state of nanoparticles under physiological conditions are fundamental parameters to be determined prior to their application in toxicological studies. Although silica-based materials are among the most promising candidates for biomedical applications, more systematic studies concerning the characterisation before performing toxicological studies are necessary. This interest is based on the necessity to elucidate the mechanisms affecting its toxicity. We present here TEM, SAXS and SMPS as a combination of methods allowing an accurate determination of single nanoparticle sizes. For the commercial material, Ludox TM50 single particle sizes around 30 nm were found in solution. DLS measurements of single particles are rather affected by polydispersity and particles concentration but this technique is useful to monitor their agglomeration state. Here, the influence of nanoparticle concentration, ionic strength (IS), pH and bath sonication on the agglomeration behaviour of silica particles in solution has been systematically investigated. Moreover, the colloidal stability of silica particles in the presence of BSA has been investigated showing a correlation between silica and protein concentrations and the formation of agglomerates. Finally, the colloidal stability of silica particles in standard cell culture medium has been tested, concluding the necessity of surface modification in order to preserve silica as primary particles in the presence of serum. The results presented here have major implications on toxicity investigations because silica agglomeration will change the probability and uptake mechanisms and thereby may affect toxicity.

Supplementary section: comparison of methods for particle size determination

In Fig. 8, the impact of concentration on the diffusion coefficient and apparent particle diameter from DLS as calculated by the cumulants method is shown. In all cases, the average particle size decreases with the silica concentration, i.e. the diffusion coefficient increases. This indicates a repulsive interaction between the particles in solution (Evans and Wennerström 1994). For concentrated samples (c ≥ 5 wt%) the PDI increases up to 22% and the analysis by NNLS shows a bimodal distribution indicating the formation of agglomerates at high concentrations. Normally, one can expect erroneous particle size values at high concentrations in DLS measurements due to multi-scattering phenomena (Urban et al. 2000). However, the measurement of solutions of Ludox TM50 at different concentrations by the use of Nanophox^©, which uses PCCS technology and is capable of eliminating the multi-scattering effects, led to similar results. This proves that the observed effect with the concentration is only due to the interparticle interactions and not due to multi-scattering effects.

Fig. 8

Apparent average particle diameter(filled squares) and diffusion coefficient (empty triangles) calculated by the cumulants method for successive dilutions of Ludox TM50 in MilliQ water by use of PCS. The empty squares show the average particle size obtained by PCCS. The apparent particle size decreases by increasing the silica concentration due to the repulsive interparticle interactions

In Fig. 9, the SAXS scattering curves for samples at concentrations 1 and 0.1 wt% and the corresponding fittings and particle sizes are shown. The analytical fitting shows significant deviations with the experimental data in the low q region for the sample at silica concentration 1 wt%. This behaviour is also indicative of interactions between the particles, and can be evaluated by the structure factor computed using the square well potential model (see Graph B, Fig. 9).

Fig. 9

A Experimental SAXS curves (empty squares) and analytical fittings (white lines) for aqueous solutions of Ludox TM50 at concentration (a) 1 wt% and (b) 0.1 wt%. B structure factor showing the increase of interparticle interactions with the concentration

We show here that the DLS measurements provide for dilutions of Ludox TM50 significant larger values for the particle sizes than measured by SAXS. This is because the scattering intensity is dominated in DLS by larger particles (Rübe et al. 2005). On the other hand, the particle sizes found by TEM and SAXS are in a good agreement proving that for Ludox TM50 there is only a slight shrinking effect in the silica matrix during the drying process. However, this cannot be generalised to any silica system since depending on their chemical composition silica nanoparticles can suffer a drastic shrinking effect due to the drying process and sizes found by TEM can be much smaller than the sizes in solution (Costa et al. 2003). Another disadvantage of TEM compared to scattering methods in solution is the fact that the particle size distributions obtained by TEM are computed over a relative low number of particles.

Since a solvated surface layer can only exist in a wet environment and shrinking only occurs when particles are dried for a longer period, we analysed the particles with SMPS, which measures particles with dried surface layers and not completely dried cores (Bresch et al. 2008). The peaks at larger sizes correspond to agglomerates containing up to 10 particles. Agglomeration takes place while the particles in the electrosprayed droplets are drying or due to particles' collisions (Cho et al. 2007). The small peak at 20 nm belongs to doubly charged single particles. As explained above, predefined relative sizes were used for the agglomerates fits. Since this is only an extrapolation to the nanoparticle regime, the values might be not valid. In fact, better fits were obtained if bigger agglomerate mobility diameters were assumed. Nevertheless, the shift on the main peak maximum would be only 0.2 nm to bigger sizes proving the accuracy of the analysis method.