Bo Zhang’s team has developed a new strategy for the precise manufacturing of liquid chromatographic microspheres – novel fully porous particles – that allows for independent control over five key properties: particle size, particle size distribution, morphology (shape), internal mesopore structure, and material chemistry (see:A Revolution in Chromatographic Performance? Let’s Discuss.) This is a significant contribution to the chromatography field – and could be the foundation of future performance breakthroughs, which I’ll come to. But does their new manufacturing process – used to make particles for slurry-packed HPLC and UHPLC columns – represent a “revolution in chromatographic performance” as we stand today?
We first have to consider what state-of-the-art column performance looks like. Enhanced column performance includes both its ability to run faster analyses (in this case, scientists are looking at the permeability of the packed column) and to better resolve close by peaks (here, scientists are interested in its efficiency or packing uniformity).
Regarding speed of analysis, the permeability of packed bed is scaled to the third power of the interparticle void fraction (the empty space between them) and is inversely proportional to the square of the surface area per unit column volume (Kozeny-Carman). However, for the very same interparticle void fraction and average particle size, commercially available polydisperse particles are always providing a smaller specific surface area than monodisperse particles (Vasseur) and are then inherently advantageous in HPLC and UHPLC columns. In the end, a subtle tradeoff has yet to be found between the external particle morphology and its polydispersity for optimizing column permeability and stability.
As far as column efficiency is concerned, there is a long history of records in column performance available in the literature: in the 1990s, Jorgenson and colleagues observed a minimum RPH (hmin) of 0.9 with standard 5 um particles packed in 20 um i.d capillary columns. In the late 2000s, 4.6 mm i.d. columns packed with sub-3 core-shell particles showed hmin around 1.5 (Kirkland). In the 2010s, Jorgenson and Godhino packed standard polydisperse 2 um particles in a 100 um x 1 m long capillary column with hmin around 1.1. In the 2020s, unconsolidated 4.6 mm x 150 mm beds of standard 4.0 um HILIC particles, packed under so-called stress-free packing conditions, generated minimum RPH of only 1.0 (Deloffi and Gritti). Also, we should not forget that packing an LC column is always both an art and a science, as the late Jack Kirkland wrote. The nature of the particle alone cannot ensure the production of highly efficient columns. For the very same particles and column format, talented column packers can achieve RPH of 1.7-1.9 while others will barely reach values below 3.0 (and I can tell you I belong to the second category of column packers!).
Given historical achievements in LC column performance (i.e. a reduced plate height of h = 0.8–0.9 for fully porous randomly jammed sphere packings with a 40 percent interparticle void fraction), the observation of a minimum reduced plate height in the range of 1.7–1.9, as reported by Bo’s team, cannot be regarded as “significant” in the sense of representing a breakthrough in column efficiency or LC resolution. This simply denotes a very well-packed column with spherical particles using an optimized packing procedure as has been done with conventional fully porous particles.
Nevertheless, this new particle manufacturing process paves the way for more fundamental investigations into column performance. Among the five controllable particle properties, particle size clearly dominates in determining column efficiency – but that’s well established. For small, fast-diffusing analytes, neither internal mesopore size and structure (mesopore size > 50 Angstrom) nor surface chemistry substantially influence efficiency, and size monodispersity is not a fundamentally determining factor for randomly packed columns. The remaining frontier lies now in demonstrating the scale up of this novel manufacturing process to be compatible with narrow-bore and 4.6 mm i.d. columns, expanding the chemical and mechanical suitability and durability spectrum of current chromatographic materials. This could enable high-performance separations under unconventional and extreme conditions, a set of variables still largely underexplored in the LC community. Advancing this area could significantly enhance permeability, bed stability, and resolution performance. We’d be very interested to hear more about the potential of these new silica materials in that context.
Exciting avenues to explore
However, irrespective of whether Bo’s team made a significant breakthrough in separation performance, we still need to understand why they observed a decrease in RPH from the standard value of 2.0 for slurry-packed columns with fully porous particles down to 1.7–1.9. In my view, this can be fully explained by a decrease in trans-column, or long-range, eddy dispersion.
Long-range eddy dispersion accounts for the small but impactful velocity biases occurring from the center bulk region to the thin wall region of the column. No other explanation is plausible when working with small analytes. This relative performance improvement of -0.1-0.3 RPH cannot be due to a decrease of the so-called short-range or bulk eddy dispersion. This is clear when we consider Tallarek’s elegant work comparing the axial bulk dispersion of fully porous particles (wide particle size distribution, RSD ~25 percent) and superficially porous particles (tight particle size distribution, RSD ~3.5 percent).
But what, then, caused this reduction in the long-range eddy dispersion? Only two possible explanations are left: the particle morphology and/or the optimized packing recipe employed in Bo’s article. Size distribution does not matter in random structures. It obviously does in ordered structures.
To test their hypothesis (e.g., that particle monodispersity matters), Bo’s team could first investigate the impact of slightly modifying the packing protocol (slurry and pushing solvents, packing pressure, packing flow rate program, etc.) and assess the robustness of the packing process based on the observed minimum reduced plate heights for the same batch of monodisperse particles. The goal would be to exclude “rare events” and those packed columns that fall outside the ±3σ range of the efficiency distribution curve, which is inherent to a random slurry-packing process.
Secondly, because Bo’s team can accurately control particle size, they could blend approximately ten batches of such new monodisperse silica particles, with sizes ranging from 1.5 to 2.5 µm, to construct a polydisperse batch with an average particle diameter of 2.0 um and an RSD of around 20 percent comparable to the RSD of currently available 2.0 um polydisperse particles. Note that the inverse approach was investigated in our group (starting from the polydisperse particles, the RSD of the size distribution was reduced from 20% to around 9%) and no change in column performance was observed under optimally packed conditions (Shiner, Izzo, and Gritti). Would such a polydisperse mixed batch be more difficult to pack as efficiently as the monodisperse ones? I’d like to find out!
A foundation for future breakthroughs
Overall, this work is highly valuable in the field of separation science and column performance, as it will enable scientists to design the most appropriate experiments to determine the optimal combination of particle shape, external morphology, particle size distribution, and packing protocols for achieving maximum permeability, packing quality, and minimizing trans-column eddy dispersion. The ultimate goal: a minimum RPH of 0.8-0.9 (1.7 from Bo’s team so far).
To summarize, up to now our “physician” knows everything about the diagnostics and symptoms of the current “disease” of random sphere packings in cylindrical tubes: the average void fraction in the thin wall region does not exactly match that in the bulk region. Unfortunately, the physician still has no remedy to offer to column manufacturers. Perhaps – and we all hope – Bo’s team’s future research will finally provide one based on their microfluidic precision manufacturing.
The Discussion Continues
Frantisek Svec: Monodisperse Microspheres: No “Revolution” – But Excitingly Scalable
Newsletters
Receive the latest analytical science news, personalities, education, and career development – weekly to your inbox.
