Investigating the Effect of Column Geometry on Separation Efficiency

Dec 12, 2017 - A Waters Alliance 2790 HPLC system (Waters, Milford, MA, USA) with a Waters 996 photodiode array detector, controlled through Empower P...
2 downloads 6 Views 1MB Size
Subscriber access provided by READING UNIV

Article

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 Anal. Chem., Just Accepted Manuscript • Publication Date (Web): 12 Dec 2017 Downloaded from http://pubs.acs.org on December 12, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Analytical Chemistry is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Investigating the Effect of Column Geometry on Separation Efficiency using 3D Printed Liquid Chromatographic Columns Containing Polymer Monolithic Phases Vipul Guptaa, b, Stephen Beirnec, Pavel N. Nesterenkoa, and Brett Paulla, b* a

Australian Centre for Research on Separation Sciences (ACROSS), School of Physical Sciences, University of Tasmania, Sandy Bay, Hobart 7001, Tasmania, Australia b ARC Centre of Excellence for Electromaterials Science, School of Physical Sciences, University of Tasmania, Sandy Bay, Hobart 7001, Tasmania, Australia c ARC Centre of Excellence for Electromaterials Science, Intelligent Polymer Research Institute, University of Wollongong, NSW 2522, Australia ABSTRACT: Effect of column geometry on the liquid chromatographic separations using 3D printed liquid chromatographic columns with in-column polymerised 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 polymerisation of mechanically stable poly(BuMA-co-EDMA) monoliths was achieved within each design without any significant structural differences between each phase. 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 mms-1, 2.5 mms-1, 5 mms-1, and 10 mms-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.

Miniaturisation 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 so-called 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 gases2.

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 column1. 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 towards the outer 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

1 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

secondary flow can thus potentially counter the adverse racetrack effects on band broadening by reducing axial dispersion3. The effects of column geometry on unretained solutes have been studied previously both theoretically and experimentally3,4. 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 function4. 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 region4. 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 the supercritical fluid chromatography9,10. However, to the best of our knowledge, only one practical study has been reported to-date 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 functionalised cycloolefin polymer capillary column11. 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 columns12,13. The printer technology utilised, namely selective laser melting (SLM), was successfully applied to the fabrication of coiled columns of ~ 1mm I.D. and up to 60 cm in length, within a ~3 cm by 3 cm x 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 characterisation of three new 3D printed chromatographic columns, specifically a (i) 2D (planar) serpentine column design14-20, representing a predominantly straight chromatographic column with a small number of low aspect ratio turns, (ii) a 3D spiral column design12,13,21-23, 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 x 30 mm x 8 mm block. In the following paper

Page 2 of 14

we present results and discussion on their internal modification with regard to the in-column thermal polymerisation 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 computational fluid dynamic simulations, and finally their relative chromatographic performance under both isocratic and the gradient conditions.

EXPERIMENTAL SECTION Materials

Butyl methacrylate, ethylene dimethacrylate, 1-decanol, azobisisobutyronitrile (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 (αLal, from bovine milk, 14.2 kDa), myoglobin (My, from equine heart, 17 kDa), and β-lactoglobulin A & B (β-Lac, from bovine milk, ~18.4 kDa) were purchased from SigmaAldrich (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). Deionised water for use was purified through a Milli-Q water purification system (Millipore, MA, USA) with a final resistance of 18.2 MΩ. Instruments

A Waters Alliance 2790 HPLC system (Waters, Milford, MA) with a Waters 996 photodiode array detector (Waters, MA, USA), controlled through Empower Pro software (Waters, Milford, MA), and a Dionex UltiMate 3000 HPLC system (Thermo Fischer Scientific, 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, CA, USA) was used for hydroxylation, silanisation, and thermal polymerisation 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 polymerised 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

2 ACS Paragon Plus Environment

Page 3 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

All three columns were designed with the Solidworks 3D modelling and CAD software. They were designed with a planar footprint of 30 mm x 30 mm x 8 mm to precisely match a custom built Peltier thermoelectric column heater/cooler, described previously12. Detail of the heater/cooler system, dimensions and performance can be found within electronic supplementary information (ESI) (Figures S-1 to S-4). Each column was designed with two ¼ 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 & Science, WA, USA) with standard 1/16” ferrules (P-200N, IDEX Health & Science, WA, USA) 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-4V) build material, as described previously12. The columns were aligned at an angle of ca. 45° in the SLM 3D printer to minimise the cross-sectional area of each slice and hence limit the heat build-up within the structure during high-temperature 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 re-tapping 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. 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. Reynoldsaveraged 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 columns12. The iterations were manually observed for the convergence of the turbulence kinetic energy, velocity, pressure, and shear stress user points. On successful completion of each run, the results were analysed as required using the CFX-Post. In-column thermal polymerisation of the poly(BuMA-coEDMA) monolith

Each of the three printed titanium columns was first internally silanised as described previously to ensure covalent bonding of monolithic phases to the internal channel walls12,24. Poly(butyl methacrylate-co-ethylene glycol dimethacrylate) (BuMA-co-EDMA) monolithic phases were then thermally polymerised within each of the three columns. Thermal polymerisation 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 polymerisation conditions and experienced the same thermal and environmental conditions. The precise procedure, as described previously12 was followed for preparation of monoliths in all three columns. The monoliths were washed with MeOH to remove the porogen (1-decanol) and any unpolymerised material. The columns were connected to the HPLC system and were flushed with MeOH, whilst slowly increasing the flow-rate until stable pressures were observed at 30 μLmin-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 hr under atmospheric conditions, to completely remove the organic monolith and prepare the columns for re-use. The re-generated columns were then re-treated 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 4 times without any signs of physical damage. The stability and the 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 replicate readings. Printed connections were found to be leak-proof 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 (1,400 ppm), o-xylene (1,400 ppm), and naphthalene (140 ppm). The proteins mixture was composed of ribonuclease A (100 ppm), cytochrome C (150 ppm), lysozyme (70 ppm), αlactalbumin (70 ppm), myoglobin (100 ppm), βlactoglobulin A &B (100 ppm), and TFA (0.05%).

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 columns, were designed and 3D printed in titanium alloy to investigate the effects of column 3D

3 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

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 X 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 1 (a). The 3D spiral column was designed in line with the most commonly used circular spiral columns. It was designed in a bi-planar arrangement, where each plane consisted of two revolutions as shown in Figure 1 (b). The aspect ratios of the innermost and the outermost coils were 0.07 and 0.04, respectively. It was designed in a bi-planar arrangement as opposed to a mono-planar arrangement to minimise the number of revolutions in each plane, hence minimising 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 1 (c). 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 x 30 mm x 8 mm cuboid. Columns were 3D printed using the SLM technique with titanium alloy (Ti-6Al-4V) as the build material as described previously12. The SLM printed columns were blasted with glass 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 columns 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 2 (a) and (b). The observed slopes

Page 4 of 14

and intercepts for the linear fit of these pressure profiles are listed in the ESI (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 ℃ and over 30,000 column volumes of mobile phase. The 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 solvents12. As shown in Table 1, significantly different monolithic permeability 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 point towards improved reproducibility of in-column polymerisation 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 3 (a, b, and c) indicate a very similarly structured polymer monolith within all three columns. SEM micrographs of the channel cross-sections (Figures 3 (d, e, and f)) indicate conformal polymerisation of the monolith, confirming wall bonding in each column. SEM micrographs of the monolithic beds at various locations within the 2D serpentine (Figures 3 (aI, aII, and aIII)), 3D spiral (Figures 3 (bI, bII, and bIII)), and 3D serpentine (Figures 3 (cI, cII, and cIII)) columns confirmed an absence of any significant differences between intra- and inter- column polymerisation. No significant structural differences were observed in the monolithic beds of all three columns, and each monolithic bed illustrated a typical micro-globular 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 3(c) can be observed in the ESI (Figure S-5). Column efficiency

Column efficiencies for V1 and V2 monolithic columns at different linear velocities were calculated, as shown in Figure 4 (a) and Figure 4 (b), 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

4 ACS Paragon Plus Environment

Page 5 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

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 4 (a) and 4 (b) for both V1 and V2 of the various monolithic columns, respectively. For the 2D serpentine column, there was a significant and almost linear reduction in efficiency with an increase in the linear velocity, across the velocity range investigated, as shown in Figure 4 (a). The response is in general agreement with the typical behaviour of straight chromatographic columns26. Somewhat similar behaviour was observed for the 3D spiral column. However, in this case the rate of increase in plate height with flow rate observed was notably smaller than with the 2D serpentine column, as shown in Figure 4 (a) for V1 columns, and also evident in Figure 4 (b), albeit less pronounced, for V2 columns. In both cases however, the column efficiency was considerably better at all flow rates for the 3D spiral column, as compared to the 2D serpentine version. Interestingly, the 3D spiral columns produced the highest efficiency data for all column types at the lowest linear velocities tested, irrespective of their differing monolithic permeability. In notable contrast to the 2D serpentine and the 3D spiral columns, no significant increase in plate height with increasing linear velocity was observed in the case of the first (V1) 3D serpentine column (Figure 4 (a)), and only a small increase across the velocity range tested for the V2 column (Figure 4 (b)). In this case both versions of the column held monoliths of similar permeability, with the V1 column providing the slightly better efficiency across the flow rate range investigated. The relatively flat van Deemter curve sees both 3D serpentine columns produced provide the greatest efficiency of all three column designs, at the higher end of the applied flow rates (>0.6 mm/s, H ≈ 200-400 µm), demonstrating a significantly reduced C-term. However, at reduced linear velocities, the van Deemter plots reveal a reversal in relative efficiency. Here, the 3D serpentine column appears least efficient, with the V2 column exhibiting a possible Hmin at 0.2 mm/s. Under these low flow conditions, the highly convoluted 3D serpentine column would appear to be very similar in efficiency to the less convoluted 2D serpentine column. For comparison purposes, the use of similar acrylate based monoliths in conventional straight micro-bore silica and titanium columns of similar I.D, a wide range of plate heights have been reported (20 µm to 500 µm), at linear velocities of up to 1 mm/s for small molecules24,27-29. The kinetic plots for V1 monolithic columns are shown in Figure 5. These kinetic plots show a significant decrease in the chromatographic efficiencies of the 2D serpentine and 3D spiral columns with decreasing analysis time. This differs considerably to the trend shown by the 3D serpentine column. The kinetic plots also reveal the highest chromatographic efficiency for the 3D serpentine column under a rapid separation regime. As mentioned above, similar observations have been previously reported for

knitted reactors30 and band broadening of unretained solute bands3,4, although in these cases such observations have been attributed to an increase in the secondary flow with an increase in the linear velocity. To the best of our knowledge, this is the first experimental evidence for higher liquid chromatographic efficiency obtained from a three dimensionally complex (serpentine) chromatographic column for retained solutes, as compared to less convoluted (e.g. spiral) column geometries. Computational fluid dynamic simulated hydrodynamic properties

Computational fluid dynamic simulations were performed to study the flow behaviour in both open tubular 2D serpentine, 3D spiral, and 3D serpentine columns, and using simple model monoliths, in both straight and curved configurations. CFD simulations in the open tubular column were performed at a Re of 225 (as shown in the ESI), since the onset of secondary flow in coiled open tubular columns has been previously reported at similar high Re numbers31. The open tubular 2D serpentine column presented a typical laminar flow profile, as shown in the velocity contour plot (Figure S-6 (a)) and the velocity profiles mapped at the centre of the column (Figure S-6 (a1)) and near the column outlet (Figure S-6 (a2)). This is as expected for a system with a Re of 225, and predominantly straight configuration. The predicted flow profiles for the open tubular 3D spiral column demonstrated higher linear velocities near the outer wall, as shown in the velocity contour plot (Figure S-6 (b)) and the velocity profiles mapped at the innermost revolution of the bottom plane (Figure S-6 (b1)) and the outermost revolution of the top plane (Figure S-6 (b2)). These profiles are typical indicators of the onset of secondary flow, where the highvelocity streams are displaced radially towards the outer wall due to centrifugal forces. For the open tubular 3D serpentine column the model clearly predicts the generation of convective centers near the walls as shown in the velocity contour plot (Figure S-6 (c)) and the velocity profiles mapped at the anti-clockwise turn at the bottom plane (Figure S-6 (c1)) and the clockwise turn at the top plane (Figure S-6 (c2)). This indicates the generation of a significantly higher secondary flow in the higher aspect ratio turns of the open tubular 3D serpentine column, as compared to the lower aspect ratio turns of the 3D spiral column, and are in agreement with previously reported observations2,3. It is beyond the scope of this study to produce a detailed model of a monolithic network for CFD simulations. However, following the recent work of Jungreuthmayer et al.32 simple artificial monolith models were designed based on an alternating arrangement of interconnected wide and narrow channels. A 3 mm long monolithic segment was designed with four parallel flow channels, each composed of alternating wide (200 μm diameter) and narrow (50 μm diameter) channels. The parallel flow channels were interconnected through equidistant planes32. Applying these simple models, the effects of the monolithic column’s geometry on the flow behaviour was investigated within (1) a straight column channel configuration, corresponding to

5 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

the majoritively straight segments of the 2D serpentine column (Figure 6 (a)), and (2) within a curved configuration with an aspect ratio of 0.5, representing the 0.5 aspect ratio turns of the 3D serpentine column (Figure 6 (b)). As shown in Figure 6 (c), CFD simulation of 100 flow streams through the straight monolithic model resulted in only a small number of streams passing through both the wide and the narrow channels. This was found to be in agreement with previously reported observations using similar straight monolithic configurations32. For the curved monolith model, very different flow behaviour was predicted. As shown in Figure 6 (d), in this case higher velocity streams were observed near the inner wall of the curved monolith, confirming the previously discussed race track effect1. No secondary flow was observed in the high aspect ratio curved monolith model, as opposed to its open tubular counterpart due to much smaller I.D. of the flow channels (representing small domain size of the monolithic polymers). However, with the curved monolithic configuration, 43.8%, 90.0%, 100.0%, and 118.2% higher flows were predicted through the narrow channels, as compared to the straight monolithic configuration, at corresponding linear velocities of 1 mms-1, 2.5 mms-1, 5 mms-1, and 10 mms-1. This simple simulation thus predicts higher interaction between the wide and narrow channels within a curved monolithic column, as compared to its straight counterpart. In the gas phase such an increase in interaction between the flow through different sized channels has been reported to lower the theoretical plate height by reducing the mass transfer term33. Similar effects should be seen here with regard to a liquid phase, where the lower diffusion coefficients of the solutes would see increased mass transfer have a proportionally greater impact. Accordingly, the increased interactions observed between different sized channels in a curved monolithic configuration, acting to improve mass transfer, could help to explain the higher chromatographic efficiencies observed with the 3D serpentine column with its high aspect ratio turns, as compared to the 3D spiral and 2D serpentine columns. Moreover, high aspect ratio turns can also result in an increase in the effective transverse dispersion of solutes, resulting in a decrease in the effective coefficient of dispersion and hence an increase in the chromatographic efficiency2. Reversed-phase liquid chromatographic separations

Isocratic and gradient separations of various benzene derivatives and phenones were carried out using each chromatographic column to demonstrate the abovementioned effects upon standard RPLC separations. Van Deemter plots indicated that maximum efficiency using the 2D serpentine and the 3D spiral columns would be achieved at low flow rates of between 4 and 10 μLmin-1, however, at such flows run times were excessive and impractical (see ESI, Figure S-7 and S-8, respectively). Accordingly, comparative isocratic and gradient separations were performed at a flow rate of 30 μLmin-1, which resulted in a

Page 6 of 14

fair compromise between the chromatographic efficiencies and run times, as shown in Figures 7. Figure 7 (a) shows the isocratic separations (flow rate of 30 µLmin-1 and a column temperature of 58 °C) of the benzene derivatives mixture obtained on each column (V1 columns), and clearly demonstrates the improving efficiency and peak resolution when moving from the 2D serpentine column, to the 3D spiral column, and finally to the 3D serpentine column. Very similar improvements were also seen for the phenone mixture (Figure 7 (b)). For these sets of chromatograms, comparative chromatographic performance data can be found in the ESI (Table S-2). The chromatograms demonstrate similar capacity and retention, regardless of the column geometry, confirming their similar internal dimensions and monolithic integrity within each of these first batch (V1) columns. In terms of theoretical plates (N), the 3D serpentine column provided an average (from 8 solute peaks) increase of 23.4% and 244.8% in N, as compared to the 3D spiral column and the 2D serpentine column, respectively. Perhaps not surprisingly, similar results were obtained under gradient conditions for the benzene derivatives mixture and the phenone derivatives mixture, as shown in Figures 7 (c) and 7 (d), respectively. The corresponding data for these separations can also be found in the ESI (Table S3). In this case, the 3D serpentine column resulted in 15.0% and 82.4% average (8 solutes) increase in the peak capacity, as compared to the 3D spiral column and the 2D serpentine column, respectively. The above separations were also repeated using the second batch of monolithic columns (V2), with matching results, confirming the significantly higher chromatographic efficiency of the 3D serpentine column, as compared to the 3D spiral column under both isocratic and gradient conditions, as shown in the ESI (Figures S-9 and S-10). Separations of small molecules and proteins

The above results clearly demonstrate that the 3D serpentine monolithic columns can be used at higher flow rates to minimise analysis time, whilst minimizing loss in efficiency, as compared to their less convoluted counterparts. To demonstrate this practical advantage, optimised isocratic separations of a small molecules mixture, and gradient separations of mixed proteins, at higher flow rates, were carried out using the 3D serpentine column, and contrasted with the same optimized separations performed with the 3D spiral column, achieved at somewhat lower flow rates. As shown in Figure 8 (a), both the 3D serpentine and the 3D spiral columns resulted in a successful isocratic separation of a mixture of seven small molecules with a Rs ≥ 1, at a flow rate of 70 μLmin-1 and 30 μLmin-1, respectively. The 3D serpentine column could achieve the same separation in 58.0% less run time, as compared to the 3D spiral column. The peak resolution data obtained between each pair of the adjoining peaks with each column under these optimized conditions are listed in Table 2.

6 ACS Paragon Plus Environment

Page 7 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Finally, to demonstrate that the above effects were also consistent for large molecule separations, the gradient separation of a mixture of seven proteins was also carried out and optimized on both the 3D serpentine column and the 3D spiral column, again at flow rates of 70 μLmin-1 and 30 μLmin-1, respectively. As shown in Figure 8 (b), the improvement in the overall separation and peak resolution provided by the 3D serpentine column can be seen. Much poorer results were obtained for the 2D serpentine column as compared to the 3D spiral and 3D serpentine columns as shown in the ESI (Figure S-11). The 3D serpentine column resulted in an average (5 solutes) increase in the peak capacity of 74.2% as compared to the 3D spiral column. The 3D serpentine column also resulted in the higher resolutions between each pair of the adjoining peaks, as also shown in Table 2. The 3D serpentine column was able to partially separate β-lactoglobulin protein into its two components, namely β-lactoglobulin A and β-lactoglobulin B, whereas the 3D spiral column failed to do so, and again confirms the integrity and quality of the polymer monolithic bed formed within this complex column geometry with such high aspect ratio turns.

CONCLUSION Here we have presented the first experimental evidence for higher chromatographic efficiencies possible from the application of highly convoluted liquid chromatographic columns for retained solutes, as compared to their less convoluted counterparts. The 3D serpentine column with the high number of higher aspect ratio turns demonstrated a smaller rate of increase in plate height with an increase in the linear velocity and a decrease in the analysis time as compared to alternative geometries investigated. CFD simulations in simple monolith models suggested an improvement in the interaction between wide and narrow channels as a possible contributor towards the observed increase in the liquid chromatographic efficiencies. All three column geometries, namely 2D serpentine, 3D spiral, and 3D serpentine resulted in uniform and conformal polymerisation of the monoliths without any significant structural differences between their monolithic beds. The 3D serpentine column resulted in 23% and 245% average increase in the plate height for isocratic separations, and 15% and 82% average increase in the peak capacity for gradient separations, as compared to the 3D spiral column and the 2D serpentine column, respectively. It also resulted in 58% less analysis time for the isocratic RPLC separation of the small molecules mixture, and a 74% increase in the peak capacity as compared to the 3D spiral column for the separation of proteins, including the partial separation of βlactoglobulin protein into β-lactoglobulin A and βlactoglobulin B. We believe the results presented here will provide the inspiration for the future development of 3D printed high efficiency miniaturised liquid chromatographic columns.

ASSOCIATED CONTENT

Supporting Information The electronic supplementary information contains details of the heater/cooler system, velocity profiles in open tubular columns, STL files for all geometries, and chromatographic separations and observed resolutions with monolithic columns as indicated above. This material is available free of charge via the Internet at http://pubs.acs.org.

AUTHOR INFORMATION Corresponding Author *E-mail: [email protected]

ACKNOWLEDGEMENTS We would like to acknowledge the continued support of the Australian Research Council (Grant number CE140100012) and additive fabrication facilities within the Australian National Fabrication Facility (ANFF) Materials Node.

REFERENCES (1) Giddings, J. C. J. Chromatogr. A 1960, 3, 520-523. (2) Tijssen, R. Chromatographia 1970, 3, 525-531. (3) Hofmann, K.; Halász, I. J. Chromatogr. A 1979, 173, 211-228. (4) Tijssen, R. University of Technology, Delft, 1979. (5) Tijssen, R. Sep. Sci. Technol. 1978, 13, 681-722. (6) Tijssen, R.; Van den Hoed, N.; Van Kreveld, M. E. Anal. Chem. 1987, 59, 1007-1015. (7) Radadia, A.; Salehi-Khojin, A.; Masel, R.; Shannon, M. Sens. Actuator B-Chem. 2010, 150, 456-464. (8) Yuan, H.; Du, X.; Tai, H.; Li, Y.; Zhao, X.; Guo, P.; Yang, X.; Su, Y.; Xiong, Z.; Xu, M. Sens. Actuator B-Chem. 2017, 239, 304-310. (9) Springston, S. R.; Novotny, M. Anal. Chem. 1986, 58, 2699-2704. (10) Peaden, P. A.; Lee, M. J. Chromatogr. A 1983, 259, 1-16. (11) Huang, W.; Seetasang, S.; Azizi, M.; Dasgupta, P. K. Anal. Chem. 2016, 88, 12013-12020. (12) Gupta, V.; Talebi, M.; Deverell, J.; Sandron, S.; Nesterenko, P. N.; Heery, B.; Thompson, F.; Beirne, S.; Wallace, G. G.; Paull, B. Anal. Chim. Acta 2016, 910, 84-94. (13) Sandron, S.; Heery, B.; Gupta, V.; Collins, D.; Nesterenko, E.; Nesterenko, P. N.; Talebi, M.; Beirne, S.; Thompson, F.; Wallace, G. G. Analyst 2014, 139, 6343-6347. (14) Sun, J.; Cui, D.; Li, Y.; Zhang, L.; Chen, J.; Li, H.; Chen, X. Sens. Actuator B-Chem. 2009, 141, 431-435. (15) Stadermann, M.; McBrady, A. D.; Dick, B.; Reid, V. R.; Noy, A.; Synovec, R. E.; Bakajin, O. Anal. Chem. 2006, 78, 5639-5644. (16) Lee, C. Y.; Sharma, R.; Radadia, A. D.; Masel, R. I.; Strano, M. S. Angew. Chem. Int. Ed. 2008, 47, 5018-5021. (17) Radadia, A. D.; Masel, R. I.; Shannon, M. A.; Jerrell, J. P.; Cadwallader, K. R. Anal. Chem. 2008, 80, 4087-4094. (18) Radadia, A.; Salehi-Khojin, A.; Masel, R.; Shannon, M. J. Micromech. Microeng. 2009, 20, 015002. (19) Kaanta, B. C.; Chen, H.; Zhang, X. J. Micromech. Microeng. 2010, 20, 055016. (20) Bhushan, A.; Yemane, D.; Overton, E. B.; Goettert, J.; Murphy, M. C. J. Microelectromech. Syst. 2007, 16, 383-393. (21) Terry, S. C.; Jerman, J. H.; Angell, J. B. IEEE Trans. Electron Devices 1979, 26, 1880-1886. (22) Kolesar, E. S.; Reston, R. R. IEEE Trans. Compon. Packag. Manuf. Technol. B 1998, 21, 324-328. (23) Dziuban, J.; Gorecka-Drzazga, A.; Malecki, K.; Nieradko, L.; Mroz, J.; Szczygielska, M. In Optoelectronic and Electronic Sensors IV; International Society for Optics and Photonics, 2001, pp 249-257. (24) Nesterenko, E. P.; Nesterenko, P. N.; Connolly, D.; Lacroix, F.; Paull, B. J. Chromatogr. A 2010, 1217, 2138-2146.

7 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

(25) Guiochon, G. J. Chromatogr. A 2007, 1168, 101-168. (26) Siouffi, A.-M. J. Chromatogr. A 2006, 1126, 86-94. (27) Moravcová, D.; Jandera, P.; Urban, J.; Planeta, J. J. Sep. Sci. 2003, 26, 1005-1016. (28) Eeltink, S.; Herrero-Martinez, J. M.; Rozing, G. P.; Schoenmakers, P. J.; Kok, W. T. Anal. Chem. 2005, 77, 7342-7347. (29) Buszewski, B.; Szumski, M. Chromatographia 2004, 60, S261S267. (30) Engelhardt, H.; Neue, U. Chromatographia 1982, 15, 403-408.

Page 8 of 14

(31) Hofmann, K.; Halász, I. J. Chromatogr. A 1980, 199, 3-22. (32) Jungreuthmayer, C.; Steppert, P.; Sekot, G.; Zankel, A.; Reingruber, H.; Zanghellini, J.; Jungbauer, A. J. Chromatogr. A 2015, 1425, 141-149. (33) Giddings, J. C. Anal. Chem. 1962, 34, 1186-1192.

8 ACS Paragon Plus Environment

Page 9 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 1. Column renders for the (a) 2D serpentine column, (b) 3D spiral column, and (c) 3D serpentine column.

Figure 2. Column back pressures observed at different linear velocities of MeOH (□), water (○), and ACN (∆) 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.

9 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 14

Figure 3. Imaging of the poly(BuMA-co-EDMA) monolith in the 2D serpentine, 3D spiral, and 3D serpentine columns: Optical image of a transverse section of the 2D serpentine (a), 3D spiral (b), and 3D serpentine (c) columns. SEM micrograph of a channel cross-section of the 2D serpentine (d), 3D spiral (e), and 3D serpentine (f) columns. SEM micrographs of the monolithic bed in the 2D serpentine columns’ channel I (aI), channel II (aII), and channel III (aIII), 3D spiral columns’ channel I (bI), channel II (bII), and channel III (bIII), and 3D serpentine columns’ channel I (cI), channel II (cII), and channel III (cIII).

10 ACS Paragon Plus Environment

Page 11 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Figure 4. Van Deemter plots for (a) V1 2D serpentine (solid), 3D spiral (dashed), and 3D serpentine (dotted) monolithic columns for benzene (□) and acetophenone (○) using 55% ACN/45% water as the mobile phase and (b) V2 3D spiral (dashed) and 3D serpentine (dotted) monolithic columns for benzene (□), acetophenone (○), xylene (∆), and thiourea (◊) using 35% ACN/65% water as the mobile phase. The column temperature was 58 ℃ and the results were obtained at 254 nm wavelength.

Figure 5. Kinetic plots (semi-log scale) for V1 2D serpentine (solid), 3D spiral (dashed), and 3D serpentine (dotted) monolithic columns for benzene (k = 1.35) (□) and acetophenone (k = 0.62) (○) using 55% ACN/45% water as the mobile phase and a maximum operating pressure of 10 MPa. The column temperature was 58 ℃ and the results were obtained at 254 nm wavelength.

Figure 6. Computational fluid dynamic simulations in the artificial monolith: (a) artificial monolith in a straight configuration, (b) artificial monolith in a curved configuration, (c) simulated velocity streamlines in the straight monolithic configuration at an inlet velocity of 1 mms1, and (d) simulated velocity streamlines in the curved monolithic configuration at an inlet velocity of 1 mms -1.

11 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 14

Figure 7. Reversed-phase liquid chromatographic separations with the V1 3D serpentine, 3D spiral, and 2D serpentine columns as indicated: (a) isocratic separations of benzene derivatives mixture using 55% ACN/45% water, (b) isocratic separations of phenone derivatives mixture using 55% ACN/45% water. (c) gradient separations of benzene derivatives mixture using Δφ = 0.45-0.70 ACN with tG = 30 min, and (d) gradient separations of phenone derivatives mixture using Δφ = 0.45-0.85 ACN with tG = 25 min. A flow rate of 30 µLmin-1, and a column temperature of 58 °C was used. The results were obtained at 254 nm wavelength. The elution orders observed for the sub figures (a and c) were benzene (1), toluene (2), o-xylene (3), and naphthalene (4). The elution orders observed for the sub figures (b and d) were acetophenone (1), propiophenone (2), butyrophenone (3), and valerophenone (4).

Figure 8. Separations with the V1 3D serpentine and 3D spiral columns at a flow rate of 70 µLmin -1 and 30 µLmin-1, respectively: (a) isocratic separations of a small molecules mixture using 55% ACN/45% water at a column temperature of 67 °C and results were obtained at 254 nm wavelength. The elution order was thiourea (1), acetophenone (2), propiophenone (3), butyrophenone (4), toluene (5), o-xylene (6), and naphthalene (7) and (b) gradient separations of proteins mixture using Δφ = 0-0.50 ACN with tG = 60 min at a column temperature of 58 °C and results were obtained at 210 nm wavelength. The elution order was ribonuclease A (1), cytochrome C (2), lysozyme (3), αlactalbumin (4), myoglobin (5), β-lactoglobulin A (6), and β-lactoglobulin B (7).

Table 1. Linear velocity based permeability coefficients (Kp,f) of monoliths in the 2D serpentine, 3D spiral, and 3D serpentine columns using MeOH, water, and ACN mobile phases. The average and the standard deviation values of the K p,f were calculated based on the linear velocities ranging from 0.25 – 0.5 mms-1. 2D Serpentine (V1) Kp,f(m2)*10-14

3D Spiral (V1) Kp,f(m2)*10-14

3D Serpentine (V1) Kp,f(m2)*10-14

3D Spiral (V2) Kp,f(m2)*10-14

3D Serpentine (V2) Kp,f(m2)*10-14

MeOH

1.01 ± 0.27

1.11 ± 0.27

1.87 ± 0.49

5.04 ± 0.01

1.64 ± < 0.01

Water

2.08 ± 0.55

2.95 ± 0.79

3.74 ± 0.98

11.43 ± 0.06

3.39 ± < 0.01

ACN

0.98 ± 0.26

1.05 ± 0.24

1.79 ± 0.48

4.66 ± 0.00

1.58 ± < 0.01

12 ACS Paragon Plus Environment

Page 13 of 14 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Analytical Chemistry

Table 2. Resolutions obtained between each pair of the adjoining peaks with the V1 3D serpentine and 3D spiral columns for the small molecules mixture and the proteins mixture under conditions as per Figure 7. Small Molecules Mixture Peaks

Proteins Mixture

3D serpentine

3D Spiral

3D serpentine

3D Spiral

1 and 2

2.28

2.49

2.32

1.61

2 and 3

1.23

1.24

1.29

0.95

3 and 4

1.25

1.19

2.21

1.60

4 and 5

1.32

1.21

0.73

0.45

5 and 6

1.12

1.01

1.12

0.79

6 and 7

1.09

0.94

0.27

0.00

13 ACS Paragon Plus Environment

Analytical Chemistry 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 14

TOC Graphic

14 ACS Paragon Plus Environment