Improved packing of preparative biochromatography columns by mechanical vibration

The bioprocessing industry relies on packed‐bed column chromatography as its primary separation process to attain the required high product purities and fulfill the strict requirements from regulatory bodies. Conventional column packing methods rely on flow packing and/or mechanical compression. In this work, the application of ultrasound and mechanical vibration during packing was studied with respect to packing density and homogeneity. We investigated two widely used biochromatography media, incompressible ceramic hydroxyapatite, and compressible polymethacrylate‐based particles, packed in a laboratory‐scale column with an inner diameter of 50 mm. It was shown that ultrasonic irradiation led to reduced particle segregation during sedimentation of a homogenized slurry of polymethacrylate particles. However, the application of ultrasound did not lead to an improved microstructure of already packed columns due to the low volumetric energy input (~152 W/L) caused by high acoustic reflection losses. In contrast, the application of pneumatic mechanical vibration led to considerable improvements. Flow‐decoupled axial linear vibration was most suitable at a volumetric force output of ~1,190 N/L. In the case of the ceramic hydroxyapatite particles, a 13% further decrease of the packing height was achieved and the reduced height equivalent to a theoretical plate (rHETP) was decreased by 44%. For the polymethacrylate particles, a 18% further packing consolidation was achieved and the rHETP was reduced by 25%. Hence, it was shown that applying mechanical vibration resulted in more efficiently packed columns. The application of vibration furthermore is potentially suitable for in situ elimination of flow channels near the column wall.

ultrasound and mechanical vibration during packing was studied with respect to packing density and homogeneity. We investigated two widely used biochromatography media, incompressible ceramic hydroxyapatite, and compressible polymethacrylatebased particles, packed in a laboratory-scale column with an inner diameter of 50 mm.
It was shown that ultrasonic irradiation led to reduced particle segregation during sedimentation of a homogenized slurry of polymethacrylate particles. However, the application of ultrasound did not lead to an improved microstructure of already packed columns due to the low volumetric energy input (~152 W/L) caused by high acoustic reflection losses. In contrast, the application of pneumatic mechanical vibration led to considerable improvements. Flow-decoupled axial linear vibration was most suitable at a volumetric force output of~1,190 N/L. In the case of the ceramic hydroxyapatite particles, a 13% further decrease of the packing height was achieved and the reduced height equivalent to a theoretical plate (rHETP) was decreased by 44%. For the polymethacrylate particles, a 18% further packing consolidation was achieved and the rHETP was reduced by 25%. Hence, it was shown that applying mechanical vibration resulted in more efficiently packed columns. The application of vibration furthermore is potentially suitable for in situ elimination of flow channels near the column wall.

K E Y W O R D S
bed compaction, column packing method, mechanical vibration, preparative biochromatography, ultrasound-assisted sedimentation

| INTRODUCTION
Packed-bed column liquid chromatography constitutes the main separation step for the purification of products in the bioprocessing industry. 1 Its basic principle relies on the flow of a mobile phase containing product and impurities through a packed bed composed of porous particles (the stationary phase). The interaction between the solute components and chromatographic media results in a selective separation of the target product from the impurities. Such interactions can be based on size exclusion, ion exchange, hydrophobic interaction, affinity, or mixed mode, among others. 1 While alternative approaches to packed beds such as monolith-based columns have gained a level of acceptance in the last decades, packed-bed chromatography still represents the main method. 2 For applications in biochromatography, compressible, spherical, porous particles based on cross-linked organic polymers such as agarose or polymethacrylate are frequently employed. Additionally, rigid, near-spherical, sintered ceramic hydroxyapatite particles, 3,4 and irregularly shaped, porous glass particles 5 are used. Regardless of particle type, an efficient column requires a porous bed packing with a homogeneous structure to avoid uneven mobile phase distribution and trans-column dispersion, which result in solute band broadening. 6,7 Furthermore, for an efficient process, the packed column should also be stable and not deteriorate over time, evident as particle rearrangement or additional consolidation during operation. 8,9 Unwanted consolidation and solute band broadening arise from the presence of void spaces formed during packing. Other packing defects can take place during operation, such as cracks in the bed, particle detachment from the column wall leading to flow channeling and formation of irregular aggregates. [10][11][12] The degree of packing defects in a bed can be assessed through the packing density, which is used as a criterion to evaluate column packing protocols. 13 A bed with a high porosity usually denotes a large proportion of void spaces in its structure and often implies a low packing quality. However, it has been observed that bed homogeneity plays a larger role with regard to column efficiency than low overall bed porosity. A heterogeneous void distribution causes higher levels of hydrodynamic dispersion, which negatively affects column quality. 13 Thus, a column with a larger void volume fraction, which is more homogeneously distributed, can outperform a packing with fewer voids, which are more heterogeneously located. The higher importance of packing homogeneity compared to global bed porosity has been studied in 3D computer simulations 10 and verified experimentally. 14 Bed porosity and homogeneity are determined by the packing method used, which mainly rely on the use of hydrodynamic flow to force a particle slurry to form a packed bed. 15 During packing of compressible resins, this is often complemented by the application of a mechanical axial force on the formed bed to achieve an additional level of axial compression. 16 The correlation between packing procedure, bed homogeneity, and solute band dispersion has been the subject of many studies, 17 and involves complex interactions of packing pressure, flow ramps, slurry concentration, and the applied bed densification steps. 18 Packing procedures are often kept a secret and are described as an art rather than a science in the open literature. 7,10,19 Advances in particle measurement techniques and simulation capacity have allowed recent studies to focus on the examination of beds at the scale of individual particles and how such structures are affected by different packing methods. The 3D reconstruction of real packings enables the geometric analysis of their structure 20,21 or their use in posterior computer simulations. 22 Additionally, granular simulations of in silico generated packings permit the study of events difficult to access in real packings, 23 thus enabling the study of timedependent processes such as structure evolution under operation or particle migration. Among several works following this approach, 6,13 the effect of the slurry concentration during packing on the bed structure and efficiency was studied. 18 The authors found that at higher slurry concentrations, the packing had an increased number of voids and a lessened degree of particle size segregation with the ideal concentration corresponding to a compromise between the degree of column heterogeneity and bed density.
An additional aspect of column packing is the behavior of chromatographic particles as microscopic frictional granular matter. It is well-known that such materials interact dissipatedly through inelastic collisions and a network of frictional force-chains which influence bed formation and long-term bed stability. 23,24 This was observed during the study of the influence of particle friction during the packing of two particle types with distinct and different levels of surface roughness. It was observed that smooth particles with lower friction slipped more readily during packing, resulting in a denser, albeit more inhomogeneous, microstructure. In comparison, media with rougher surfaces resisted movement under consolidation and delivered less dense and more homogeneous structures. 14 Furthermore, the granular properties of the packed bed also determine stability during operation. It has been previously established that granular packings come to rest in a jammed state which may be far from the most stable and dense configuration. 25 In such a metastable state, the structure is stabilized through cooperative frictional structures of several particles, forming bridges or arches which enable the packing to withstand external loads and behave locally as a solid. 26 These structures are also responsible for the formation and persistence of void spaces. 27 Unless perturbed by an external force, the system remains static 24 and is unable to reach its most homogeneous structure. 15,23 To escape the meta-stable state, energy must be supplied to the system. External perturbations can unlock the packing and allow the structure to explore the phase space 24 and densify the packing as particles fill void spaces made accessible by bridge collapses. 28 This principle also applies to chromatographic beds and it has been argued to be the only means to minimize packing defects. 10 Granular densification is a matter of interest for various industries with different applied approaches such as thermal cycling, 29  The application of ultrasound for the packing of chromatographic media has been studied in thin capillaries from micrometer range 40,41 up to 4.6 mm in diameter, 42 and microchips at the micrometer scale. 43 In most cases, the capillaries were submerged in an ultrasonic bath.
While some studies carried out a simultaneous ultrasound irradiation of the capillaries with flow packing, 7,41 others decoupled both processes. 40 Though some authors reported the packing of more efficient and stable columns measured by improved values of the height equivalent to a theoretical plate (HETP), it was also found that ultrasonic vibrations can cause the bed to become more heterogeneous. 42 The use of mechanical vibration to achieve further compaction in chromatographic beds has also been studied previously. However, this has only been performed using rigid, nonspherical, irregularly shaped, porous glass-based media. 15,44 Furthermore, these studies were limited to the application of a rotational type of pneumatic vibrators.
In this work, we investigate the use of ultrasound during sedimen- 2 | EXPERIMENTAL METHODS

| Chemicals and materials
Acetone and sodium di-hydrogen phosphate monohydrate were purchased from Carl Roth GmbH & Co. (Karlsruhe, Germany). Sodium hydroxide was acquired from NeoFroxx GmbH (Einhausen, Germany).
CHT ceramic hydroxyapatite is an incompressible mixed-mode chromatography media made of a near-spherical, macroporous form of hydroxyapatite. In this study, CHT particles type I with a mean particle diameter of 40 μm were used. For CHT particles, a working buffer of 500 mM NaOH and 50 mM NaH 2 PO 4 was used. Toyopearl SP-650M is a compressible anion exchange resin with a porous polymethacrylate base matrix with a mean particle diameter of 65 μm.
For Toyopearl particles, a solution of 10 mM NaOH was employed.

| Flow equipment
Flow for the packing procedures was provided by a computer-

| Columns
The columns were made from a poly(methyl methylacrylate) (PMMA) cylinder with an inner and outer diameter of 50 and 60 mm, respectively, with a static and a movable flow distributor which could be placed either at the top or the bottom of the column. Flow distributors were machined out of stainless steel and fitted with a stainless steel frit with a mean pore size of 8 μm (Tridelta Siperm GmbH, Dortmund, Germany) to retain the particles. The movable distributor was attached to a trapezoidal screw allowing the column height to be changed, and the distributor to be driven without rotating.

| Pneumatic vibration
The vibrators were attached to the column in two possible configurations, lateral or axial. In the lateral case, the vibrators were fixed near the bottom of the column outer wall using a PMMA clamp ( Figure 2).
In the axial case, the vibrators were attached to a V2A steel bracket directly connected to the flow distributor through two apertures in the bottom flange of the column (Figure 3).

| Preliminary pneumatic vibration experiments
The compression-relaxation dynamics at the single particle level play a role during the packing of chromatographic beds composed of compressible media. 51

| Axial vibration
Two vibrator types were explored in the axial study. Either a single K-

| Pneumatic vibration packing and column quality testing
Packing procedures for CHT and Toyopearl particles were derived from the respective manufacturer's notes and guidelines. 46

| CHT ceramic hydroxyapatite vibration packing
The standard CHT-packing procedure was a single flow-packing step with no mechanical compression due to the rigid nature of the sintered particles. 47 The column fill height was set to obtain a slurry concentration of 33% vol/vol following the manufacturer's packing advice. 47 The packing procedure was started by linearly increasing the downward flow during 30 s until a target flow rate of 300 cm/h was reached which was then held for 2 min. 3

| Toyopearl SP-650M vibration packing
The column fill height was set to obtain a slurry concentration of 50% vol/vol after particle resuspension following the manufacturer's packing advice. 46

| Column quality testing
Once packed, the beds were equilibrated at a constant flow of 100 cm/h until ultraviolet absorbance at 265 nm of the effluent liquid became stable. Elution chromatograms of the aqueous solution of 2% acetone were measured in triplicate and the resulting peaks were employed to calculate the quality of the packing. HETP was defined as: where L is the bed height, t r the retention time of the acetone peak, and W 50% the peak width at half peak height. HETP was divided by the respective mean particle diameter to obtain the reduced HETP (rHETP). Asymmetry was defined as: where a is the peak width (left half) at 10% of peak height and b the peak width (right half) at 10% peak height.

| Sedimentation of a chromatographic slurry in an ultrasonic field
The difference in settling behavior of polymethacrylate particles between ultrasound-assisted sedimentation and the standard case without ultrasound can be seen in Figure 4, in which the change in time of the slurry height is shown for both cases.
The time course during ultrasound-assisted settling was sigmoidal. Sedimentation was slow in the beginning but increased with time.
A relatively high sedimentation rate was observed before the final packing height was reached. In contrast to this, the time course of the gravity settling was linear at a lower sedimentation rate until the final packing height was reached. However, the obtained ultrasoundassisted packing consolidation was only~1% higher compared to gravity settling.
The ultrasound-assisted settling behavior can be explained by the acoustic effects. In general, the application of an ultrasonic field in a fluid can generate two distinct phenomena: acoustic streaming and acoustic radiation forces. 54 While several types of acoustic streaming can be defined, 55  In our system, the particles were attracted to and collected in the pressure nodes. As the particles group along the nodal planes, they form clusters with a decreased surface to volume ratio equivalent to "super-particles" with an increased settling velocity. 57 This effect explains the observed accelerated settling. As particles of different sizes agglomerate, they form a denser structure in contrast to Stokes sedimentation. In the latter case, particles settle according to their size with the larger particles at the bottom and the smaller ones at the top resulting in an unwanted particle segregation.
Once the bed had consolidated, the application of ultrasound caused no further change in the packing. Further experiments also showed no or very little effect of ultrasonic vibrations in already consolidated beds. A possible explanation is that the magnitude of the generated ultrasonic waves was not high enough to overcome the frictional resistance of the particles in the consolidated packings in order to accomplish particle rearrangement. Consequently, the application of mechanical vibration using pneumatic vibrators was explored. These not only generate larger amplitudes than ultrasonic waves but also experience lower acoustic attenuation, which is proportional to the wave frequency. While ultrasonic waves propagate at frequencies starting at 20 kHz, pneumatic vibrators commonly oscillate in the three-digit Hz range. Hence, mechanical vibration can deliver higher energy and the mechanical waves are less attenuated as they propagate.  The observed time course agrees with theoretical work on vibration of granular materials describing the densification of packings as an inverse exponential process. 28 As the bed becomes denser, additional compression requires the simultaneous rearrangement of an increasingly larger number of particles, thus becoming less frequent. 59 The homogeneity of the achieved bed was tested by tracer experiments and the calculation of asymmetry factor and rHETP. Figure 7a shows the elution chromatograms for standard and vibration-assisted CHT packings. Figure 7b shows compared to CHT particles resulting in less resistance to particle rearrangement due to lower friction between particles. Figure 9a shows the elution chromatograms for standard and vibration-assisted polymethacrylate packings. Figure 9b shows the rHETP of the standard and vibration-assisted packing. The asymmetry factor of the standard packing was 4.4 ± 0.9 while the asymmetry factor of the vibration packed columns was reduced to 3.7 ± 0. 5

| CONCLUSIONS
Biochromatography columns packed by conventional methods frequently do not reach their most dense state and their packing structure is not as homogeneous as possible. The generation of an ultrasonic field inside the column can be used to delay the onset of particle sedimentation. This is achieved by acoustic streaming and acoustic radiation forces, which keep the particle slurry longer in suspension, and thus may reduce the level of particle segregation due to University of Munich, Germany, is gratefully acknowledged.