Article Cite This: Anal. Chem. 2018, 90, 1186−1194
pubs.acs.org/ac
Investigating the Effect of Column Geometry on Separation Efficiency using 3D Printed Liquid Chromatographic Columns Containing Polymer Monolithic Phases Vipul Gupta,†,‡ Stephen Beirne,§ Pavel N. Nesterenko,† and Brett Paull*,†,‡ Australian Centre for Research on Separation Sciences (ACROSS) and ‡ARC Centre of Excellence for Electromaterials Science, School of Physical Sciences, University of Tasmania, Sandy Bay, Hobart 7001, Tasmania, Australia § ARC Centre of Excellence for Electromaterials Science, Intelligent Polymer Research Institute, University of Wollongong, Wollongong, NSW 2522, Australia Anal. Chem. 2018.90:1186-1194. Downloaded from pubs.acs.org by UNIV OF NEW ENGLAND on 10/23/18. For personal use only.
†
S Supporting Information *
ABSTRACT: Effect of column geometry on the liquid chromatographic separations using 3D printed liquid chromatographic columns with in-column polymerized monoliths has been studied. Three different liquid chromatographic columns were designed and 3D printed in titanium as 2D serpentine, 3D spiral, and 3D serpentine columns, of equal length and i.d. Successful in-column thermal polymerization of mechanically stable poly(BuMA-co-EDMA) monoliths was achieved within each design without any significant structural differences between phases. Van Deemter plots indicated higher efficiencies for the 3D serpentine chromatographic columns with higher aspect ratio turns at higher linear velocities and smaller analysis times as compared to their counterpart columns with lower aspect ratio turns. Computational fluid dynamic simulations of a basic monolithic structure indicated 44%, 90%, 100%, and 118% higher flow through narrow channels in the curved monolithic configuration as compared to the straight monolithic configuration at linear velocities of 1, 2.5, 5, and 10 mm s−1, respectively. Isocratic RPLC separations with the 3D serpentine column resulted in an average 23% and 245% (8 solutes) increase in the number of theoretical plates as compared to the 3D spiral and 2D serpentine columns, respectively. Gradient RPLC separations with the 3D serpentine column resulted in an average 15% and 82% (8 solutes) increase in the peak capacity as compared to the 3D spiral and 2D serpentine columns, respectively. Use of the 3D serpentine column at a higher flow rate, as compared to the 3D spiral column, provided a 58% reduction in the analysis time and 74% increase in the peak capacity for the isocratic separations of the small molecules and the gradient separations of proteins, respectively.
M
Giddings suggested that in a coiled column molecules near the inner walls travel at a greater velocity as compared to the molecules near the outer wall, resulting in a nonequilibrium state laterally across the column.1 This phenomenon was termed the racetrack effect. Within a chromatographic column, or similar open tubular fluidic system, such an effect typically results in an increase in the axial dispersion of a solute band (band broadening), and a consequent decrease in the chromatographic efficiency. Alternatively, secondary flow can be developed due to strong centrifugal forces acting on the fluid flowing through a coiled system. In the open tubular format, a straight column generates a parabolic flow profile under a laminar flow regime, whereas a coiled column exerts a higher centrifugal force on the central flow streams (due to higher velocities) as compared to the boundary flow streams, displacing the central flow streams radially toward the outer
iniaturization and portability of a chromatographic system can be facilitated by coiling long capillary columns into small footprints. However, the effect of column coiling, and related high aspect ratio (column radius/turn radius) column geometries, on the chromatographic efficiency remains somewhat unclear, particularly in the case of packed and monolithic columns, and more particularly in the case of liquid phase separations. As far back as 1960, Giddings1 discussed the racetrack effect in coiled columns and its adverse effects upon chromatographic efficiency with respect to coiled open tubular gas chromatographic columns. However, a decade later in 1970, Tijssen2 demonstrated that column coiling does not always result in lower chromatographic efficiency, demonstrating how the combinations of increased mobile phase velocity and column coiling could achieve higher chromatographic efficiency, due to the development of socalled secondary flow. Tijssen also concluded that such effects are likely to be more pronounced in liquid phase separations due to the higher densities of the liquids as compared to gases.2 © 2017 American Chemical Society
Received: September 14, 2017 Accepted: December 11, 2017 Published: December 12, 2017 1186
DOI: 10.1021/acs.analchem.7b03778 Anal. Chem. 2018, 90, 1186−1194
Article
Analytical Chemistry
computational fluid dynamic simulations, and finally their relative chromatographic performance under both isocratic and gradient conditions.
wall. This phenomenon results in the development of a secondary flow in the radial direction, in addition to the primary flow in the axial direction. The generation of said secondary flow can thus potentially counter the adverse racetrack effects on band broadening by reducing axial dispersion.3 The effects of column geometry on unretained solutes have been studied previously both theoretically and experimentally.3,4 However, the effects of column geometry on retained solute bands are much more complex, as the flow profile in a chromatographic column including a stationary phase, with a convoluted 3D geometry, is difficult to predict, resulting in unknown values for the velocity profile factor and retention function.4 Such predictions also require consideration of additional terms beyond diffusion coefficients, namely the resistance to mass transfer in the stationary phase5 and the resistance to mass transfer in the interphase region.4 Further to this, changes in the column geometry can also result in imperfections and/or changes in the properties of the stationary phase. These combined complications effectively render such purely theoretical investigations rather irreproducible and complex. Therefore, interest remains for experimental investigation of such effects, albeit with limitations on what can be practically achieved using traditional column formats. Experimental observations have been discussed previously in relation to both gas chromatography6−8 and supercritical fluid chromatography.9,10 However, to the best of our knowledge, only one practical study has been reported todate on the effects of column 3D geometry on retained solutes using liquid chromatography, which was a recent brief study investigating an open-tubular coiled anion exchange functionalized cycloolefin polymer capillary column.11 Although relatively minor, the coiled column exhibited reduced b and c terms in relation to a straight capillary column of similar functionality and dimensions. Recently, we reported on the feasibility of 3D printing metal liquid chromatographic columns.12,13 The printer technology utilized, namely selective laser melting (SLM), was successfully applied to the fabrication of coiled columns of ∼1 mm i.d. and up to 60 cm long, within a ∼3 cm × 3 cm × 5 mm titanium or stainless steel block. This printer’s capabilities can be used to design and fabricate geometrically complex columns, allowing for the first time a practical investigation into the effects of column 3D geometry on liquid chromatographic efficiency. Accordingly, herein we report upon the production and characterization of three new 3D printed chromatographic columns, specifically (i) a 2D (planar) serpentine column design,14−20 representing a predominantly straight chromatographic column with a small number of low aspect ratio turns, (ii) a 3D spiral column design,12,13,21−23 representing a chromatographic column with medium aspect ratio turns, and (iii) a novel 3D serpentine column design, representing a chromatographic column with a high number of repeating high aspect ratio turns. These three differing columns were designed with identical column dimensions of i.d. and length, prepared using the same build material, and printed within a similarly sized 30 mm × 30 mm × 8 mm block. Herein we present results and discussion on their internal modification with regard to the in-column thermal polymerization of poly(BuMA-co-EDMA) monolithic stationary phases, their comparative chromatographic performance for small and large molecules in relation to mobile phase velocity, insights into the effects of column 3D geometry using
■
EXPERIMENTAL SECTION Materials. Butyl methacrylate, ethylene dimethacrylate, 1decanol, azobis(isobutyronitrile) (AIBN), 3-trimethoxysilylpropyl methacrylate (TMSPM), organic solvents, ACN, MeOH, trifluoroacetic acid, acetone, sodium hydroxide pellets, ribonuclease A (RNase A, from bovine pancreas, 13.7 kDa), cytochrome c (Cyc, from equine heart, 11.7 kDa), lysozyme (Lys, from chicken egg white, 14.3 kDa), α-lactalbumin (α-Lal, from bovine milk, 14.2 kDa), myoglobin (My, from equine heart, 17 kDa), and β-lactoglobulin A and B (β-Lac, from bovine milk, ∼18.4 kDa) were purchased from Sigma-Aldrich (St. Louis, MO, USA). All the reagents were analytical grade and the mobile phases were HPLC grade. Hydrochloric acid (37%) was purchased from Merck (Darmstadt, Germany). Deionized water for use was purified through a Milli-Q water purification system (Millipore, Billerica, MA, USA) with a final resistance of 18.2 MΩ. Instruments. A Waters Alliance 2790 HPLC system (Waters, Milford, MA, USA) with a Waters 996 photodiode array detector, controlled through Empower Pro software (Waters), and a Dionex UltiMate 3000 HPLC system (Thermo Fischer Scientific, Waltham, MA, USA), with variable wavelength UV−vis detector, controlled through Chromeleon software, were used for all liquid chromatography work. A small heating furnace (Woodrow Kilns, Sydney, Australia) was used for thermal oxidation. A gas chromatography oven (Hewlett Packard 5890, Palo Alto, CA, USA) was used for hydroxylation, silanization, and thermal polymerization of the monolith. A Realizer SLM 50 system (Realizer GmbH, Borchen, Germany) was used for selective laser melting (SLM) 3D printing. A field emission scanning electron microscope (Hitachi GmbH SU-70, Europe) was used to obtain SEM micrographs. A BalTec SCD 050 sputter coater (Leica Micro-systems, North Ryde, Australia) was used to sputter coat platinum. SEM micrographs of the polymerized monolith in all three columns were obtained from transverse sections of the column bodies. The transverse sections were sputter coated with Pt for 15 s. The SEM micrographs were obtained using a 1.5-kV electron beam. Design and 3D Printing of Liquid Chromatographic Columns. All three columns were designed with the Solidworks 3D modeling and CAD software. They were designed with a planar footprint of 30 mm × 30 mm × 8 mm to precisely match a custom built Peltier thermoelectric column heater/cooler, described previously.12 Detail of the heater/cooler system, dimensions, and performance can be found in the Supporting Information (SI) Figures S-1 to S-4. Each column was designed with two 1/4 female unified fine pitch threads (UNF) to allow direct connections using conventional HPLC fittings and tubing. Short 1/4−28 flangeless fittings (XP-208, IDEX Health and Science, Oak Harbor, WA, USA) with standard 1/16 in. ferrules (P-200N, IDEX Health and Science) were used to connect HPLC tubing to the columns through the UNF ports. All three columns were 3D printed using SLM and the titanium alloy (Ti-6Al-4 V) build material, as described previously.12 The columns were aligned at an angle of ca. 45° in the SLM 3D printer to minimize the cross-sectional area of each slice and hence limit the heat buildup within the structure during high-temperature 1187
DOI: 10.1021/acs.analchem.7b03778 Anal. Chem. 2018, 90, 1186−1194
Article
Analytical Chemistry
replicate readings. Printed connections were found to be leakproof up to a pressure of ca. 12 MPa, hence the maximum operating pressures were limited to 10 MPa. Chromatographic Conditions. Van Deemter plots, isocratic LC separations, and gradient LC separations for the initial batch of monolith columns (V1, as described below) were obtained using the Waters HPLC system. For the second batch of columns (V2), the Dionex HPLC system was used. The gradient separation of proteins was performed using 0.05% (v/v) TFA in water and 0.04% (v/v) TFA in ACN as the mobile phases. The small molecules mixture was composed of thiourea (40 ppm), acetophenone (25 ppm), propiophenone (30 ppm), butyrophenone (30 ppm), toluene (1400 ppm), o-xylene (1400 ppm), and naphthalene (140 ppm). The proteins mixture was composed of ribonuclease A (100 ppm), cytochrome C (150 ppm), lysozyme (70 ppm), α-lactalbumin (70 ppm), myoglobin (100 ppm), β-lactoglobulin A and B (100 ppm), and TFA (0.05%).
laser melting. A supporting scaffold structure was used between the column and the build platform to ensure proper adhesion and support. All three columns were processed similarly in terms of fabrication, cleaning of the remaining powder particles, and retapping of the threads. Computational Fluid Dynamic Simulations. Computational fluid dynamic simulations were performed using ANSYS 17.0 software with CFX solver R17.0 Academic (Ansys, Inc., Canonsburg, PA, USA). The open tubular 2D serpentine, 3D spiral, and 3D serpentine column designs were meshed similarly and resulted in 4.8 million, 4.8 million, and 3.1 million nodes, respectively. The straight and the curved monolithic model configurations were meshed similarly and resulted in 1.6 million nodes each. Reynolds-averaged Navier− Stokes (RANS) simulations were performed using the shear stress transport (SST) turbulence model with water as the fluid material. A no-slip wall condition with a roughness of 20 μm was prescribed for the open tubular columns in accordance with the internal wall surface roughness of the titanium alloy SLM printed columns.12 The iterations were manually observed for the convergence of the turbulence kinetic energy, velocity, pressure, and shear stress user points. Upon successful completion of each run, the results were analyzed as required using the CFX-Post. In-Column Thermal Polymerization of the poly(BuMA-co-EDMA) Monolith. Each of the three printed titanium columns was first internally silanized as described previously to ensure covalent bonding of monolithic phases to the internal channel walls.12,24 Poly(butyl methacrylate-coethylene glycol dimethacrylate) (BuMA-co-EDMA) monolithic phases were then thermally polymerized within each of the three columns. Thermal polymerization of the monoliths was achieved using 24 wt % butyl methacrylate (0.48 g), 16 wt % ethylene dimethacrylate (0.32 g), and 60 wt % 1-decanol (1.2 g) with 1 wt % AIBN (0.008 g), with respect to monomers, at 45 °C for 20 h. All three column monoliths were produced in batch mode to ensure each was produced under the same polymerization conditions and experienced the same thermal and environmental conditions. The precise procedure, as described previously,12 was followed for preparation of monoliths in all three columns. The monoliths were washed with MeOH to remove the porogen (1-decanol) and any unpolymerized material. The columns were connected to the HPLC system and were flushed with MeOH, while slowly increasing the flow rate until stable pressures were observed at 30 μL min−1. As the 3D titanium alloy printed columns were both extremely robust and thermally stable, they could be reused multiple times. Accordingly, all three columns from the initial batch produced, following full chromatographic evaluation, were heated to 500 °C for 6 h under atmospheric conditions, to completely remove the organic monolith and prepare the columns for reuse. The regenerated columns were then retreated as mentioned above, and used to generate the second batch of monolithic columns for evaluation. In total, these 3D printed titanium alloy columns were reused in this way at least four times without any signs of physical damage. The stability and permeability of all three monolithic columns were assessed using the Waters Alliance 2790 HPLC system with ACN, MeOH, and water mobile phases. The flow rates for all three mobile phases were increased and the corresponding back pressure was monitored for 5 min. The average pressure during this time was recorded for three
■
RESULTS AND DISCUSSION 3D Printed Liquid Chromatographic Columns. Three different chromatographic columns with different column geometries, namely a 2D serpentine, a 3D spiral, and a 3D serpentine column, were designed and 3D printed in titanium alloy to investigate the effects of column 3D geometry on their liquid chromatographic performance. The 2D serpentine column was precisely designed in line with conventional serpentine columns. Within the 3D build area limits of the SLM printer used (∼70 mm × 74 mm), this predominately straight design was included to provide a reference point against which the more complex 3D designs could be evaluated. The 2D serpentine column was composed of ca. 75% straight channels with 5 intermittent turns with an aspect ratio of 0.5 each as shown in Figure 1a. The 3D spiral column
Figure 1. Column renders for the (a) 2D serpentine column, (b) 3D spiral column, and (c) 3D serpentine column.
was designed in line with the most commonly used circular spiral columns. It was designed in a biplanar arrangement, where each plane consisted of two revolutions as shown in Figure 1b. The aspect ratios of the innermost and the outermost coils were 0.07 and 0.04, respectively. It was designed in a biplanar arrangement as opposed to a monoplanar arrangement to minimize the number of revolutions in each plane, hence minimizing the difference between the aspect ratios of the innermost and the outermost coils. The 3D serpentine column was designed with ca. 100 repeating turns of aspect ratio 0.5 each as shown in Figure 1c. All three columns were designed with an i.d. of 1 mm and a total column length of 300 mm. They were integrated with 1/4 UNF threaded ports at the inlet and the outlet and were enclosed within a 30 mm × 30 mm × 8 mm cuboid. Columns were 3D printed using the SLM technique with titanium alloy (Ti-6Al-4 V) as the build material as described previously.12 The SLM printed columns were blasted with glass 1188
DOI: 10.1021/acs.analchem.7b03778 Anal. Chem. 2018, 90, 1186−1194
Article
Analytical Chemistry
permeability of each monolithic column for each mobile phase tested was calculated as per Darcy’s law25 and is listed in Table 1. The permeability for different mobile phases followed the order water > MeOH ∼ ACN for all three columns. Higher monolithic permeability for the aqueous mobile phase as compared to the organic mobile phase has also been previously observed for similar monoliths and is presumably due to a minor degree of swelling of the polymer monolith within organic solvents as compared to the aqueous solvents.12 As shown in Table 1, significantly different monolithic permeabilities were observed for V1 and V2 of the 3D spiral columns, whereas similar monolithic permeability were observed for V1 and V2 of the 3D serpentine columns. The reasons for the differences are not clear, although they point toward improved reproducibility of in-column polymerization within the 3D serpentine column as compared to the 3D spiral column, possibly due to the better support structure provided to the monolith by the repeating high aspect ratio turns in the former design. The polymer monolith in the 2D serpentine, 3D spiral, and 3D serpentine columns were imaged using SEM to study the monolithic structure within each column at various points. Figures 3a, b, and c indicate a very similarly structured polymer monolith within all three columns. SEM micrographs of the channel cross sections (Figures 3d, e, and f) indicate conformal polymerization of the monolith, confirming wall bonding in each column. SEM micrographs of the monolithic beds at various locations within the 2D serpentine (Figures 3aI, aII, and aIII), 3D spiral (Figures 3bI, bII, and bIII), and 3D serpentine (Figures 3cI, cII, and cIII) columns confirmed an absence of any significant differences between intra- and intercolumn polymerization. No significant structural differences were observed in the monolithic beds of all three columns, and each monolithic bed illustrated a typical microglobular polymeric structure with interspersed flow-through pores. Additional SEM micrographs of the monolithic bed in the 3D serpentine column at locations marked by IV, V, and VI in Figure 3c can be found in SI Figure S-5. Column Efficiency. Column efficiencies for V1 and V2 monolithic columns at different linear velocities were calculated, as shown in Figure 4a and Figure 4b, respectively. With regard to small molecule separations, four solutes, namely benzene, acetophenone, xylene, and thiourea, were used to determine the column efficiencies. The retention factors calculated using thiourea as the void marker for 45% and 65% water in ACN mobile phases were 1.35 and 1.39 (benzene), 0.62 and 0.56 (acetophenone), and 2.36 and 2.58 (xylene). All four solutes exhibited similar trends for plate height (based on the peak width at half height) with respect to the linear velocity, and typical responses obtained are shown in Figures 4a and 4b for both V1 and V2 of the various monolithic columns, respectively. For the 2D serpentine
beads to remove any loosely bound material from the exterior, and they were tapped and sonicated in IPA to remove unfused powder particles from the channels. The averaged i.d. of each column was calculated using column volume measurements, assuming accurate column length ( 0.99) with an increase in the flow velocity for MeOH, ACN, and water, as shown in Figures 2a and b. The observed slopes and intercepts
Figure 2. Column back pressures observed at different linear velocities of MeOH (□), water (○), and ACN (Δ) mobile phases at a column temperature of 44 °C for (a) V1 2D serpentine (solid line), 3D spiral (dashed line), and 3D serpentine (dotted line) monolithic columns and (b) V2 3D spiral (dashed line) and 3D serpentine (dotted line) monolithic columns.
for the linear fit of these pressure profiles are listed in SI Table S-1. The pressure profiles obtained confirm the mechanical stability of the wall bound monolithic phases in all columns. In terms of stability, no signs of change to the monolithic structure over time were observed for any of the columns, which were tested to a maximum operating temperature of 70 °C and over 30 000 column volumes of mobile phase. The
Table 1. Linear Velocity Based Permeability Coefficients (Kp,f)a of Monoliths in the 2D Serpentine, 3D Spiral, and 3D Serpentine Columns Using MeOH, Water, and ACN Mobile Phases Kp,f (m2) × 10−14 MeOH water ACN
2D Serpentine (V1)
3D Spiral (V1)
3D Serpentine (V1)
3D Spiral (V2)
3D Serpentine (V2)
1.01 ± 0.27 2.08 ± 0.55 0.98 ± 0.26
1.11 ± 0.27 2.95 ± 0.79 1.05 ± 0.24
1.87 ± 0.49 3.74 ± 0.98 1.79 ± 0.48
5.04 ± 0.01 11.43 ± 0.06 4.66 ± 0.00
1.64 ±
