Zonal rotor centrifugation revisited: new horizons in sorting nanoparticles

Abstract

Density gradient centrifugation is an effective method for the isolation and purification of small particles. Hollow rotors capable of hosting density gradients replace the need for centrifuge tubes and therefore allow separations at large scales. So far, zonal rotors have been used for biological separations ranging from the purification of whole cells down to serum proteins. We demonstrate that the high-resolution separation method opens up exciting perspectives apart from biology, namely in sorting mixtures of synthetic nanoparticles. Loading and unloading, while the rotor is spinning, avoids perturbations during acceleration and deceleration periods, and thus makes a vital contribution to sorting accuracy. Nowadays one can synthesize nanoscale particles in a wide variety of compositions and shapes. A prominent example for this are "colloidal molecules" or, generally speaking, defined assemblies of nanoparticles that can appear in varying aggregation numbers. Fractionation of such multimodal colloids plays an essential role with regard to their organization into hierarchical organized superstructures such as films, mesocrystals and metamaterials. Zonal rotor centrifugation was found to be a scalable method of getting "colloidal molecules" properly sorted. It allows access to pure fractions of particle monomers, dimers, and trimers, just as well as to fractions that are essentially rich in particle tetramers. Separations were evaluated by differential centrifugal sedimentation, which provides high-resolution size distributions of the collected nanoparticle fractions. The performance achieved in relation to resolution, zone widths, sorting efficiencies and recovery rates clearly demonstrate that zonal rotor centrifugation provides an excellent solution to the fractionation of nanoparticle mixtures.

Fig. 1. Zonal rotor centrifugation: a density gradient is introduced at the edge of a hollow rotor, while it is spinning at reduced speed. Loading starts with the lightest portion of the gradient first, followed by layers of increasing densities. Once the gradient fills the rotor completely, the sample suspension is introduced at the rotor core as the last material loaded. Separation is accomplished by sorting of the particles according to their sedimentation coefficients. At the end of the centrifuge run, the rotor speed is reduced again and its content is displaced out through the center exit by pumping a sufficiently dense solution into the edge line. Suspensions of sorted nanoparticles can be picked up using a fraction collector system.

Fig. 2. Comparison of rotors used for zone centrifugations. (A) In a swinging-bucket rotor, centrifugation tubes containing a density gradient are located in buckets, which can swing out and orient perpendicular to the axis of rotation. Different particle populations migrate as discrete zones through the gradient during centrifugation and will be separated from each other at this disposal. Although wall effects are drastically reduced as compared to fixed-angle rotors, only the particles far from the lateral wall will sediment directly to the bottom of the tube. (B) Wall effects are eliminated when using a zonal rotor, which can be regarded as a 360° extension of a swinging-bucket tube. Lateral walls are replaced by radial septa, which enables unconstrained sedimentation into sector-shaped compartments providing radial dilution. Rotor geometry has a considerable effect on the shape of a density gradient. A density gradient prepared linear with volume will assume a concave profile with radius when loaded into a zonal rotor, whereas it will be also linear with radius when placed into a centrifuge tube of a swinging-bucket rotor.

Fig. 3. Sedimentation coefficient distributions as measured by differential centrifugal sedimentation (DCS). The two samples of “colloidal molecules” contain the same type of cluster species and differ only in terms of their numerical proportions. Mixture F02 (blue line) contains higher amounts of clusters with respect to particle monomers than mixture F01 (red line). Inset: FESEM micrographs of the particles to be isolated from the mixture (particle monomers N = 1, dimers N = 2, trimers N = 3 and tetramers with tetrahedral geometries N = 4). The scale bar represents 200 nm. More information on sample composition is found in Table S1.

Fig. 4. Photographs of fractions 14 to 83 (lined up in two rows) recovered from the zonal rotor in experiment F01. Fractions 1 to 13 are not shown because they do not contain any particles. The black line reflects the absorbance of the nanoparticle fractions at a wavelength of 405 nm. The much higher amount of particles in experiment F02 did not allow for a comparable assessment of the separation by visual inspection.

Fig. 5. Fractionation of multimodal mixtures of “colloidal clusters”: gradients consisting of multiple density steps (black lines) were loaded into the zonal rotor. With time, the step gradient changed into a continuous profile (red lines and squares), as evidenced by density measurements of the fractions extracted from the rotor. The blue dots represent the absorbance of the fractions. The absorbance profiles (blue lines) thus reflect the separation of the various particle populations, which differ in the number of constituents N. Centrifugation times were optimized for the isolation of clusters with N ≤ 5. In doing so, larger species are accumulated near the rotor edge. Experiment F01: separation of 90 mg of “colloidal molecules” using a density gradient that is linear with volume (A), and concave with radius (B). Experiment F02: separation of 621.6 mg of “colloidal molecules” using a gradient that is convex both with volume and radius (C and D). See text for further explanations. Compositional analysis of each individual fraction by DCS gave the decomposition of the absorbance profiles into discrete contributions. The letters O, S, G, and C refer to the subdivision of the fluid load of the zonal rotor into overlay (O), sample zone (S), density gradient (G), and cushion (C). See experimental section for further explanations.

Fig. 6. FESEM micrographs of distinct fractions collected from the zonal rotor. (A–C) represent fractions that exclusively consist either of particle monomers, dimers, or trimers. (D) refers to a fraction that is essentially rich in particle tetramers, but also contain minor portions of trimers and pentamers. The selected fractions correspond to local maxima of the absorbance profiles shown in Fig. 5. Scale bars represent 1000 nm.

Fig. 7. Separation of mixtures of “colloidal molecules”: absorbance versus particle settling time was measured by DCS for each fraction collected in experiment F02. The individual graphs depicting particle distributions were pooled into a 3D graph, which is shown from two different perspectives (A and B). The graph shows the efficiency that is achieved in sorting clusters of up to six constituent particles (N = 1: monomers; N = 2: dimers; N = 3: trimers; N = 4: tetramers; N = 5: pentamers; and N = 6: hexamers). Comparable 3D graphs related to experiment F01 are found in the ESI (Fig. S4†).