Targeted optimization of chromatographic columns based on 3D analysis of packing microstructure
The preparation, structure, and performance of functional materials porous are strongly interrelated. Hence, a detailed analysis of the pore structure of a functional porous material in combination with a detailed characterisation of its performance can provide an understanding of the influence of i...
|PDF Full Text
No Tags, Be the first to tag this record!
|The preparation, structure, and performance of functional materials porous are strongly interrelated. Hence, a detailed analysis of the pore structure of a functional porous material in combination with a detailed characterisation of its performance can provide an understanding of the influence of individual parameters during preparation and thus identify structural limitations to an improved utilization. The obtained results can be used to tune the preparation towards a better pore structure suited for the targeted application. This work focuses on packings of silica-based particles for highly efficient chromatographic separations. The prepared packings combine an interparticle macropore space for fast flow-based transport with an intraparticle mesopore space providing high surface areas for molecule-surface interactions. Such packed columns have a wide field of application, not only in highly efficient separations, but also for catalysis, and (energy) storage However, the focus here is on separations in liquid chromatography.
In Chapter 1, the influence of the slurry concentration on separation efficiency and bed structure was investigated for capillary columns (75 µm inner diameter, 30 cm length) packed with 1.3 µm bridged-ethyl hybrid (BEH) fully porous silica particles. The slurry concentration was varied from 5 to 50 mg/mL while every other packing parameter was kept constant. Chromatographic characterisation with hydroquinone as weakly retained analyte revealed highly efficient separations (reduced plate heights as low as 1.5) at an optimal intermediate slurry concentration of 20 mg/mL for this specific set of packing parameters. Confocal laser scanning microscopy (CLSM) was utilized to conduct a three-dimensional reconstruction and to carry out a detailed morphological analysis of the column with the best performance, a column packed with a slurry concentration below the optimum, and one packed above the optimum. Two counteracting effects were revealed: Radial heterogeneities limit the separation efficiency for columns packed at low slurry concentrations. With an increase in slurry concentration, these radial effects get supressed but the number and size of large voids with a diameter similar to the mean particle diameter increase significantly. Interestingly, the reconstructions also revealed high external bed porosities between 0.47 and 0.50 which are higher than expected with respect to the random loose packing limit reported for frictional, cohesionless particles. However, no signs of bed instability could be observed demonstrating the significant impact of interparticle forces for particles as small as 1.3 µm.
In Chapter 2, the investigation of the optimal slurry concentration was expanded by analysing the effects for a different particle size to obtain a more general picture. A similar set of capillary columns (75 µm inner diameter, 45 cm length) was packed with 1.9 µm BEH particles at eleven different slurry concentrations between 5 and 200 mg/mL including additional tests for reproducibility at selected concentrations and the observation of bed formation using optical microscopy. While comparable reduced plate heights were achieved, the observed optimum of 140-160 mg/mL to pack highly efficient columns reproducibly differed significantly from the 20 mg/mL for the 1.3 µm particles identified in Chapter 1. This can be explained by the difference in the particle diameter as interparticle forces and particle aggregation become more dominant at still smaller diameters. CLSM-based reconstructions revealed similar trends in the bed structures as seen in Chapter 1. At low concentrations, pronounced ordered particle layers in the direct vicinity of the column wall, local bed densification near the column wall, and particle size-segregation limit the achieved separation efficiency. The peculiarity of the first effect is continuously decreasing with an increase in the slurry concentration even beyond the optimum while the latter two effects are already supressed at the optimal slurry concentration. On the other hand, the number and size of large voids increase with an increase in the utilized slurry concentration as already seen in Chapter 1. The videos acquired during column packing provided very helpful insights into bed formation mechanisms and thus delivered possible explanations for these structural features. At 10 mg/mL, particles arrive individually at the bed front allowing individual settlement and rearrangement on the arrival of following particles what allows a discrimination of particles according to their individual properties. The picture looks completely different for 100 mg/mL as example for higher concentrated slurries. Here, particles tend to aggregate during packing and arrive in large batches. This prevents discrimination of individual particles but significantly reduces the chances for rearrangement and is thus prone to the conservation of defects formed between the border of the arriving batches of particles and the front of the bed.
Chapter 3 is based on the results obtained during the work presented in Chapters 1 and 2. The combination of high slurry concentration and ultrasound was already proposed there as chance to keep transcolumn heterogeneities as low as possible while preventing the formation of large voids. To test this hypothesis, two sets, each consisting of three capillary columns (75 µm inner diameter, 100 cm length) were packed with 1.9 µm BEH particles at a slurry concentration of 200 mg/mL; one set under application of ultrasound during packing, the other one without. All three columns, which underwent sonication, showed significantly better performance than each of the other columns. The obtained reduced minimum plate height for a weakly retained analyte was even lower than the already impressive value of 1.5 for columns packed at a slurry concentration optimal for packing without sonication and reached values close to unity over a length of 1 m for the best-performing column. The achieved theoretical plate counts of ~500,000 demonstrate a unique potential for highly efficient separations of extremely complex samples.
In Chapter 4, the focus is shifted from capillary columns to the more common analytical format. CLSM could not be applied here as the steel columns are not transparent and extrusion of the bed is not possible without losing either stability or optical transparency. Thus, an imaging and reconstruction procedure based on focused ion beam scanning electron microscopy was developed using a commercial narrow-bore analytical column (2.1 mm inner diameter, 50 mm length) packed with 1.7 µm BEH particles. The packing was embedded with poly(divinylbenzene) prior to extrusion from the steel column in order to conserve the bed structure. Two image stacks were acquired and reconstructed at characteristic positions within the bed: one in the central section of the column along the flow direction to obtain the bulk properties of the bed and one from the column wall towards the column centre to investigate and quantify the influence of the geometrical wall effect and the second wall effect. To investigate the effect of the microstructure in the wall region on local flow through the bed, a radially resolved flow profile was obtained by lattice-Boltzmann simulations. For this column, the region affected by wall effects spanned over approximately 62 particle diameters showing a decrease in the local mean porosity by up to 10% and an increase in the local mean particle diameter by up to 3% with respect to the bulk region inducing a decrease of the local flow velocity by up to 23%. Furthermore, four more ordered layers of particles were formed directly at the hard column wall due to the geometrical wall effect leading to local velocity fluctuations by up to a factor of three. These quantified structural features are in excellent agreement with previous reports about macroscopic characterisations of the wall effects by optical or chromatographic measurements.