Synthesis of Superparamagnetic Particles with Tunable Morphologies

Feb 1, 2013 - James J. O'Mahony, Mark Platt,. †. Devrim Kilinc, and Gil Lee*. School of Chemistry and Chemical Biology, University College Dublin, B...
3 downloads 0 Views 356KB Size
Download PDF
Article pubs.acs.org/Langmuir

Synthesis of Superparamagnetic Particles with Tunable Morphologies: The Role of Nanoparticle−Nanoparticle Interactions James J. O’Mahony, Mark Platt,† Devrim Kilinc, and Gil Lee* School of Chemistry and Chemical Biology, University College Dublin, Belfield, Dublin 4, Ireland S Supporting Information *

ABSTRACT: Superparamagnetic microparticles are extensively used in the purification of biomolecules due to the speed and ease of magnetic separation. It is desirable that the microparticles used in biological affinity separations have both high surface area and high magnetic mobility to facilitate a high binding capacity of target biomolecules and their rapid removal from solution, respectively. Scaling laws for conventional spherical superparamagnetic microparticles are such that increasing the microparticle specific surface area results in a significant decrease in the magnetic mobility. More favorable combinations of these key parameters can be found if alternative microparticle morphologies are developed for use in affinity separations. Emulsion-templated self-assembly of iron oxide nanoparticles into microparticles using oil-in-water emulsions was carried out using a modified Couette shear mixer with separate inlet ports for the oil and aqueous phases, enabling high throughput microparticle synthesis. By controlling the dissolved nanoparticle concentration and nanoparticle surface activity at the droplet interfaces, the resulting microparticles were tuned to spherical, dimpled, or crumpled morphologies. The specific binding capacity and magnetic mobility of each type of microparticle were measured by a peroxidase-based colorimetric assay and by their magnetic field-induced motion in a viscous fluid, respectively. Superparamagnetic microparticles with dimpled and crumpled morphologies were found to have higher specific binding capacities compared to spherical microparticles, while maintaining high magnetic field velocities due to their high iron oxide content. Superparamagnetic microparticles with these novel morphologies would make excellent tools for affinity-based bioseparations where binding capacity and magnetic mobility are key factors.

■

INTRODUCTION Superparamagnetic microparticles (SMPs) have applications in targeted drug delivery1,2 and diagnosis3,4 and are widely used in affinity separations due to the speed, ease, efficiency, and inexpensive nature of magnetic separation.5,6 The microparticles are coated with a polymer that stabilizes them in aqueous solution and provides chemical groups to which biomolecules such as DNA and proteins can be conjugated.7−9 Commercially available SMPs, defined as microparticles over 100 nm in diameter, are typically spherical and composed of an assembly of iron oxide (Fe3O4 or Fe2O3) superparamagnetic nanoparticles with diameters of 3−50 nm. The nanoparticles can either be distributed in a polymer microparticle matrix10−13 or self-assembled to form a tightly packed spherical SMP without the addition of a polymer component.8,14,15 It is desirable that SMPs used in biological affinity separations have both high surface area and high magnetic mobility. The large surface areas provide, when functionalized with the appropriate ligands, a higher binding capacity, and therefore a lower mass of SMPs is required to achieve high analyte recovery. High magnetic mobilities enable fast separation of SMPs and reduce the time required to separate the beads from the supernatant after binding occurs. Although reducing the diameter of © 2013 American Chemical Society

spherical SMPs increases the surface area per mass, the magnetic mobility of the SMPs is reduced significantly, as it is proportional to the square of the microparticle radius.16 A solution to this limitation is provided by changing the SMP morphology from spherical to either hollow, porous, or buckled morphologies. A number of efforts have been made in this respect, with the primary focus on the use of the development of hollow SMPs for drug encapsulation for targeted delivery.13,17,18 Hollow SMPs consisting of iron oxide nanoparticles embedded in polystyrene have been fabricated using an inverse microemulsion polymerization process. The surfaces of the nanoparticles have been made amphiphilic through functionalization with both active hydrophilic hydroxyl groups and hydrophobic oleic ester groups. The amphiphilic nanoparticles could then be used as colloidal stabilizers for inverse water-inoil emulsions and also mediated the polymerization of styrene monomers in the oil phase at the droplet interface to form a hollow polystyrene iron oxide composite SMP up to a couple of Received: November 30, 2012 Revised: January 28, 2013 Published: February 1, 2013 2546

dx.doi.org/10.1021/la3047565 | Langmuir 2013, 29, 2546−2553

Langmuir

Article

micrometers in diameter. 13 Hollow and porous Gd 2 O 3 microspheres have been synthesized through the deposition of superparamagnetic material together with a polymer on the surface of or within a spherical gelatin template, which was later removed using calcination leaving behind a hollow or porous SMP roughly 100 or 200 nm in diameter, respectively.18 Polymer−iron oxide composite SMPs have been synthesized using water−oil−water double emulsions in which the middle oil phase consists of a mixture of both a liquid prepolymer and superparamagnetic nanoparticles. After UV curing and drying of the internal water phase, polymer magnetite iron oxide SMPs with a hollow hemispherical morphology ∼12 μm in diameter were formed.17 These methods for the production of SMPs either involve a large number of steps such as the removal of a sacrificial template on which the superparamagnetic material has been deposited or incorporate a polymeric material, which reduces the magnetic moment of the resulting SMPs by lowering the fraction of iron oxide in each sample. A more facile and straightforward synthesis route for high surface area SMPs involving fewer steps is emulsion-templated self-assembly of iron oxide nanoparticles using single emulsions.14,15 In the most basic form, involving single oil-in-water emulsions, emulsion-based self-assembly of microspheres consists of an emulsification step followed by a drying step. The former involves the emulsification of the oil soluble polymer or nanoparticle in an immiscible aqueous phase, with the use of a surfactant to stabilize the droplets. Emulsification techniques commonly employed include membrane emulsification,8,19 Couette type shear devices,20−22 and microfluidic devices.23−25 Here we use a Couette type shear device, which provides a higher throughput and is easier to use when compared to either membrane or microfluidic emulsification methods. The device consists of a rotor and stator arrangement, with a narrow annular gap between the concentric cylinders, where one of the cylinders rotates at a fixed speed, i.e., the rotor. In previous works, a crude emulsion was first prepared by manually stirring the two phases together and then pumped into the Couette device through a single inlet.20,22 Fragmentation of the crude emulsions took place within the device under the action of the shearing force, and the homogenized emulsion emanated through an outlet port.20,22 This emulsion synthesis method can be problematic due to the inconsistencies in the premixing step potentially leading to the formation of very small droplets and limited throughput time due to the phase separation of the metastable premixed emulsions.20 Even if the phase separation has not progressed to the point where the homogenization needs to be halted, the change in the dispersed phase volume fraction, which occurs as the phases in the premixed emulsion begin to separate, results in the variation of droplet size and uniformity over the course of the synthesis.22,26 To overcome these issues and facilitate higher throughputs over longer time periods, we modified the design of the Couette mixer to eliminate the premixing step. After the emulsion synthesis is complete, the drying step involves the evaporation of the solvent and the crystallization/aggregation of the dissolved material in the droplets to form individual microparticles.8,24 The size and size distribution of the droplet, and therefore of the microparticle, are determined in the first step, whereas the morphology of the microparticle is determined in the drying step. The situation is slightly different when discussing double emulsions, i.e., water−oil−water emulsions, where a polymeric material or nanoparticle is dissolved in the oil phase and the innermost water phase droplets dry to leave a

hollow imprint within the microsphere.27−29 The single oil-inwater emulsion method was used to synthesize polymer−iron oxide composite SMPs with a crumpled morphology by adjustment of the hydrophobicity of the iron oxide nanoparticles and therefore their affinity for the droplet interface.30 In this study we demonstrate the synthesis of dimpled and crumpled SMPs using single oil-in-water emulsion-templated assembly, without the use of polymer in the dispersed oil phase. We show that, by controlling the dissolved nanoparticle concentration and nanoparticle surface activity at the droplet interfaces, we can manipulate the morphology of the resulting SMPs. A particular emphasis is given here on the characterization of the SMPs toward their application in magnetic affinity separations. Our results demonstrate the role of nanoparticle interactions at the droplet interface in determining the drying routes in emulsion-templated microparticle synthesis and the potential benefits of tailoring microparticle morphologies for different applications.

■

EXPERIMENTAL SECTION

List of Chemicals. All chemicals were obtained from SigmaAldrich (Wicklow, Ireland) and used without further purification, unless otherwise stated. Iron chloride tetrahydrate (FeCl2·4H2O), iron chloride hexahydrate (FeCl3·6H2O), ammonium hydroxide, ethanol, perchloric acid, hexane, sodium dodecyl sulfate (SDS), Triton X-100, sodium chloride, potassium sulfate (K2SO4), sulfuric acid (H2SO4), potassium hydroxide (KOH), polyethylenimine (PEI; 750 000 g/mol), poly(acrylic acid-co-maleic acid) (PAAMA; 3000 g/mol), 2-(Nm o r p h o li n o ) e t h a n e s u l f o n i c a c i d ( M E S ) , 1 - e t h y l - 3 - ( 3 (dimethylamino)propyl)carbodiimide (EDC), acetone, sulfo-N-hydroxysuccinimide (Sulfo-NHS), phosphate buffered saline (PBS), Tween 20, bovine serum albumin (BSA), sodium azide, sucrose, oleic acid (Fisher Scientific, Dublin, Ireland), avidin (Fisher), biotin-labeled horseradish peroxidase (Biotin-HRP; Fisher), 2,2′-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS; Fisher), dextran (60 000−90 000 g/mol; Affymetrix, High Wycombe, UK), and epoxy resin (TAABEpon; Agar Scientific, Essex, UK). Deionized (DI) water was obtained from a Milli-Q system (Millipore, Cork, Ireland). Synthesis of Hydrophobic Nanoparticles (Ferrofluid). FeCl2·4H2O (48 g) and FeCl3·6H2O (98 g) were dissolved in deoxygenated water (250 mL) in a 1 L three-neck flask under a N2 atmosphere. The flask was placed into an ice bath, and ammonium hydroxide (200 mL) was added rapidly with vigorous stirring. The solution was kept at 0 °C for 45 min and heated to and kept at 85 °C for 1 h. Oleic acid (30 mL) was added, and the solution was continued to be heated for another hour. The flask was allowed to cool to room temperature (RT) before being transferred to a 600 mL beaker. A magnet was placed next to the beaker to collect the black precipitate, which was washed with ethanol (3 × 200 mL). After each wash a magnet was placed to the side of the beaker, and the ethanol solution was poured to waste. This process was repeated 3× with DI water, 3× with 20% perchloric acid, 3× with DI water, and finally 3× with ethanol (each wash with a volume of 200 mL). After the last ethanol wash the solution was poured away, and hexane 400 mL was added to the beaker. The black precipitate of hydrophobic iron oxide nanoparticles was dispersed in hexane. This suspension was called the ferrofluid. When additional washing steps were required, stock ferrofluid (5 mL, 90 mg/mL) was mixed with ethanol (35 mL) in a 50 mL centrifuge tube. The nanoparticles were dispersed in the mixture by sonication and vigorous mixing. The mixture was then centrifuged at 4000 rpm in a Rotina 420R centrifuge (Hettich, Tuttlingen, Germany). The supernatant was decanted, and a fraction of the sediment (ca. 100 mg) was removed and dissolved in hexanethis ferrofluid was washed once. The remainder of the sediment was resuspended in ethanol (40 mL) and centrifuged as before. This procedure was repeated 3×, after which the iron oxide nanoparticles could not be fully redissolved in hexane. 2547

dx.doi.org/10.1021/la3047565 | Langmuir 2013, 29, 2546−2553

Langmuir

Article

Calculation of Surface Ligand Coverage on Nanoparticles Based on Elemental Analysis. Elemental analysis was carried out to determine the iron and carbon contents of the dried nanoparticles, after solvent evaporation. The radius of the nanoparticles was measured using Axiovision optical sizing software (Zeiss) and TEM images (see Supporting Information) to be 5.39 ± 1.86 nm. A CE-440 elemental analyzer (Exeter Analytical Inc., North Chelmsford, MA) was used to measure carbon content, and a SpectrAA 55B atomic absorption spectrometer (Varian Inc. (Aligent Technologies), Victoria, Australia) was used to measure iron content. As the elemental analysis returned the composition in percentage form, to simplify the calculations, the total mass was assumed to be 1 g. The mass of iron oxide, MFe3O4, was calculated using the experimentally determined mass of iron. The molecular weight of iron oxide is 231.5 g/mol. Iron makes up a mole fraction of 0.724 of magnetite, Fe3O4. The mass of iron oxide present in the sample, MFe3O4, was calculated by dividing the experimentally determined mass of iron by 0.724. The number of nanoparticles present, Nn, in a given mass of iron oxide, MFe3O4, was calculated as Nn MFe3O4/ρnVn, where the nanoparticle density, ρn, was assumed to be that of pure magnetitie, i.e., 5.17 g/cm3, and Vn is the nanoparticle volume, calculated assuming spherical nanoparticles with the above radius. The total surface area of the nanoparticles, Snt, was then calculated as Snt = NnSn, where Sn is the nanoparticle surface area, calculated assuming spherical nanoparticles with the above radius. The quantity of carbon measured was used to calculate the mass of oleic acid present. The molecular weight of oleic acid is 282.5 g/mol, of which carbon accounts for a fractional mass of 0.766. The mass of oleic acid, Mol, present in the sample was calculated by dividing the experimentally determined mass of carbon present by 0.766. The number of oleic acid ligands in the sample, Nol, was calculated as Nol = (Mol/Mwol)NA, where Mwol is the molecular weight of oleic acid and NA is Avogadro’s contsant. The density of oleic acid ligands, Dol, on the iron oxide nanoparticles was then calculated as Dol = Nol/Snt and expressed in chains/nm2. Emulsification Using a Couette Flow Mixer. A Couette mixer was constructed using a HT-2 viscometer sample chamber (Brookfield Viscometers Ltd., Essex, UK) as the stator. Two inlet ports for the oil and water phases were welded onto the chamber perpendicular to each other at the base and an outlet port for the homogenized emulsion at the top of the chamber. A cylindrical rotor was constructed to fit into the chamber with a gap of ∼100 μm between the rotor and the stator. An electric motor (Arrow Engineering, NJ; Model 1750) was coupled to the rotor, and the rotor speed was measured using a photo tachometer (Extech Instruments, Waltham, MA; Model 461825). A solution containing 25% (w/w) dextran and 2% (w/w) SDS (60 mL total) and ferrofluid (20 mL, unwashed) with nanoparticle concentration varying between 0.1 and 5 mg/mL were pumped into the Couette flow mixer through separate inlet ports (Figure 1) at 0.75 and 0.25 mL/min flow rates, respectively. The emulsion droplet size was adjusted by varying the shear rate, which for low dispersed phase volume fractions can be approximated as γ̇ = Riω/(Ro − Ri), where Ro is the outer radius of the rotor, Ri is the inner radius of the stator, and ω is the angular velocity of the rotor.22 In the geometry used here, Ri is 18.81 mm, Ro − Ri is ∼100 μm, the maximum rotor angular velocity, ω, used in this setup was 63 rad s−1 (600 rpm), and therefore the maximum applied shear rate, γ̇, was ∼6300 s−1. The sheared emulsion was collected in 0.5% (w/w) Triton X-100 (850 mL) under gentle agitation. Hexane was allowed to evaporate for 24 h to form micrometer-sized aggregates of iron oxide nanoparticles. The microparticles were collected using a magnet for 2 h and washed with 0.5% Triton X-100 solution 2× to remove residual dextran. Microparticle Surface Modification and Centrifugation. Dried microparticles were resuspended in a solution of 0.05% (w/ w) Triton X-100, 250 mM NaCl, and 4.5% (w/w) PEI (50 mL total). The microparticle suspension was gently mixed, sonicated for 2 min and incubated at RT for 16 h. The microparticles were washed 3× with 250 mM NaCl 25 mL with 30 s sonication steps between each wash. The microparticles were then resuspended in a solution composed of 250 mM NaCl and 5% (w/w) PAAMA (50 mL total; pH = 4.0). The

Figure 1. Couette shear mixer showing individual inlets for aqueous and dispersed phases, with emulsion outlet on top right. Droplet size was adjusted through variation of the shear rate, γ′, which for low dispersed phase volume fractions can be approximated as γ′ = Riω/(Ro − Ri), where Ri is the inner rotor radius, Ro is the outer radius of the stator, and ω is the angular velocity of the rotor. Teflon O-rings were used to prevent leakage. suspension was mixed, sonicated for 2 min, and incubated at RT for 6 h. The microparticles were washed 3× with 250 mM NaCl (25 mL), resuspended in MES buffer 40 mL (pH = 6.3), and sonicated for 2 min. EDC (10 mg) in MES buffer (1 mL) (pH = 6.3) was added to the microparticles, and the suspension was mixed and incubated for 30 min. The carboxylated microparticles were then washed 3× with and resuspended in DI water. For isolation of the largest size fractions for magnetic mobility measurements, 0.5 mg/mL microparticle suspension (20 mL) in 50 mL centrifuge tubes was centrifuged at 1000 rpm in a Marathon 10K centrifuge (Fisher) for 3 min. The sediment microparticles were resuspended in DI water and sonicated for 1 min, and the process was then repeated 10×. Characterization of Emulsion Droplets and SMPs. Images of the emulsion droplets were obtained using an inverted microscope (Zeiss, Herefordshire, UK). 5 μL samples of the emulsion were dropped onto a glass slide, and a coverslip was placed over the drop. A 100× oil immersion objective or 63× objective in differential interference contrast mode was used to obtain images. For SEM imaging, small quantities (50 μg) of microparticles were dried on a copper strip and mounted on stubs using double-sided carbon tape. Samples were then imaged using either a Hitachi TM-1000 (Hitachi High Technologies Ltd.) or a FEI Quanta 3D FEG DualBeam (FEI Ltd., Hillsboro, OR) when higher resolution images were desired. Axiovision software (Zeiss) was used to measure droplet and microparticle size distributions from the obtained images. More than 80 data points were used when measuring size distributions of superparamagnetic particles and emulsion droplets. Zeta-potential measurements were performed using a Zetasizer Nano ZS (Malvern Instruments Ltd., Worcestershire, UK) using DTS-1060 cells. 1 mM K2SO4 solutions were made up to pH 4, 6, 8, and 10 using concentrated H2SO4 or KOH. Solution conductivity was maintained between 0.3 and 0.35 mS/cm for each pH. 100 μg of microparticles was washed three times and resuspended in K2SO4 (1 mL) at each pH. Samples for transmission electron microscopy (TEM) were dehydrated in a graded ethanol series (30%, 50%, 70%, 90%, and 100%). Following the dehydration samples were transferred from 100% ethanol to acetone, from acetone to a 1:1 mixture of acetone and epoxy resin for 1 h. To complete the resin infiltration the samples were placed in 100% resin at 37 °C for 2 h. Finally, samples were embedded in resin were placed at 60 °C for 24 h until polymerization was complete. For orientation purposes sections from each sample were cut at 1 μm and examined by light microscopy (Leica DMLB, Leica Microsystems, Wetzlar, Germany). From this survey, sections areas of interest were identified and ultrathin (50−80 nm) sections of these areas were cut using a Leica EM UC6 ultramicrotome (Leica Microsystems, Wetzlar, Germany). These sections were collected on 200 mesh thin bar copper grids for imaging. Avidin Conjugation and Measurement of SMP Binding Capacity. Carboxylated microparticles 1 mg were suspended in 1 mL 2548

dx.doi.org/10.1021/la3047565 | Langmuir 2013, 29, 2546−2553

Langmuir

Article

MES buffer (pH = 6.3) containing EDC (10 mg) and Sulfo-NHS (10 mg) and incubated at RT for 30 min. Microparticles were washed and incubated in MES buffer 1 mL (pH = 6.3) containing avidin (1 mg) at RT for 2 h on a rotating wheel. Avidin-functionalized microparticles were washed and incubated in PBS buffer (1 mL) containing 0.1% (w/ w) Tween 20 and 0.2% (w/w) BSA (pH = 7.4) at RT for 30 min. Microparticles were then washed and incubated in PBS (1 mL) containing 0.1% (w/w) Tween 20 and Biotin-HRP 200 μg (pH = 7.4) at RT for 30 min on a rotation wheel. Microparticles were thoroughly washed and resuspended in PBS (500 μL) containing 0.1% (w/w) Tween 20 (pH = 7.4). 5, 10, or 20 μL of this solution was diluted with PBS (200 μL) containing 0.1% (w/w) Tween 20 (pH 7.4), followed by the addition of 150 μL of ABTS indicator solution. The reaction was stopped after 3 min by adding 0.02% (w/w) sodium azide (50 μL), and the absorbance at 410 nm was measured using a Nanodrop 2000c spectrophotometer (Thermo Scientific, Dublin, Ireland). Magnetic Mobility Measurement. The magnetic mobility of microparticles was measured by applying magnetic forces on particles immersed in a viscous fluid, as reported in our previous work.16 A magnetic field was imposed by two pairs of NdFeB magnets (12.7 × 12.7 × 6.35 mm per pair; Apex Magnets, Petersburg, WV) with the same poles facing each other over a 1 mm air gap. An aluminum magnet holder attached to a micromanipulator (Eppendorf, Hamburg, Germany) was used to position the magnets at specified distances from the imaging field of an inverted microscope (Zeiss). 20 μL of magnetic microparticle suspension, containing ca. 50 000 particles in 55% (w/w) sucrose solution (viscosity = 28 mPa·s), was added to a polystyrene microwell (Nunc, Rochester, NY) and covered with a glass coverslip (Menzel, Braunschweig, Germany) to minimize evaporation. Fluororesbrite Plan YG 1 μm fluoroscent particles (Polysciences Inc., Eppelheim, Germany) were also present in the well as positional references. The force acting on a SMP in high magnetic field gradient is calculated as F = mMs∇H, where m is the mass of the microparticle, Ms is the saturation mass magnetization of the microparticle, and H is the external magnetic field.31 The acting force is equal and opposite to the drag force, which is calculated using the Stokes−Einstein relation: F = 6πμrv, where μ is the dynamic viscosity of the fluid and r and v are the radius and the drag velocity of the microparticle, respectively.32 Hence, the drag velocity is linearly related to the saturation magnetization of microparticles with equal diameter, submerged in the same fluid and subjected to the same magnetic field, and is described by v = mMs∇H/6πμr.

Figure 2. SEM images of SMPs produced with ferrofluid nanoparticle concentrations of (A) 1 g/L, (B) 0.5 g/L, and (C) 0.1 g/L, with insets in (B) and (C) showing magnified images. All other factors were kept constant. A transition from spherical to dimpled and crumpled microparticles occurs with decreasing nanoparticle concentration. Continuous phase dextran 25% w/w, SDS 2% w/w, flow rates of continuous and dispersed phases 0.75 and 0.25 mL/min, respectively. TEM images of cross sections of microparticles from (A), (B), and (C) and shown in (D), (E), and (F), respectively. Hollow areas inside the dimpled and crumpled particles suggest that during drying a nanoparticle shell first develops at the surface of the droplet and then collapses inward as drying continues. Scale bars = 11 μm in (A−C) (500 nm in insets in (B) and (C)). Bars = 250 nm in (D−F).

microparticles synthesized with varying concentrations of iron oxide nanoparticles in hexane. At ferrofluid nanoparticle concentrations equal to or above 1 g/L the synthesized microparticles are smooth and spherical. However, as the concentration decreases to 0.5 g/L and to 0.1 g/L, dimpled and crumpled morphologies are observed, respectively. Crosssectional TEM images (Figures 2D−F) show that the SMPs produced at higher nanoparticle concentrations are homogeneous with no evidence of internal void spaces. However, the dimpled and crumpled microparticles have hollow cores and anisotropic shell thicknesses. The cross sections suggest that crumpled and dimpled particles are formed by a mechanism involving adherence of the nanoparticles to the droplet interface during drying and the formation of a shell of densely packed nanoparticles. At some point the stresses driving droplet shrinkage overcome the electrostatic forces stabilizing the nanoparticle shell and buckling occurs. The formation of a shell during droplet drying may be explained by an increased affinity of the iron oxide nanoparticles for the oil−water interface. Oleic acid chains bound to the nanoparticles make them oilsoluble and also act as a stabilizer to prevent undesired aggregation.34 Incomplete coverage of the nanoparticles with oleic acid may result in a reduction of their hydrophobicity and, correspondingly, an increased affinity for the oil−water interface, leading to their accumulation and locking at the surface of the droplet. In order to investigate this hypothesis, the density of oleic acid side chains on the iron oxide nanoparticles was measured for the stock ferrofluid (the resulting microparticles are shown in Figure 2) and after washing with ethanol. The oleic acid chain density on the nanoparticles in the stock ferrofluid, and after a number of ethanol washes, is shown in Figure 3. A close-packed monolayer of oleic acid on a flat surface corresponds to ∼4.17 chains/ nm2.35 The stock nanoparticles (5.48 chains/nm2) therefore have a monolayer of adsorbed oleic acid, with the density higher than the close-packed value attributable to the curvature of the nanoparticles. The chain density is depleted to 2.31 chains/nm2 after two ethanol washing steps, which corresponds

■

RESULTS AND DISCUSSION SMPs with controlled morphologies were synthesized by creating microemulsions via a modified Couette flow-based shear mixer. The device used here is a modified Couette mixer (Figure 1) that has separate inlets for the oil and aqueous phases, such that the emulsification does not occur until the phases enter the device, and therefore no premixing step is required. This novel design circumvents the aforementioned issues associated with premixing, such as accidental production of very small droplets and phase separation during homogenization. Larger scale synthesis is also made possible, since the throughput time is limited only by the syringe capacity. Microemulsions were produced with varying concentrations of iron oxide nanoparticles in the dispersed phase, while keeping other factors, such as flow rates, surfactant concentrations, and drying bath volumes constant. As the solvent within each emulsion droplet evaporates, SMPs with different sizes are formed depending upon its initial size and ferrofluid concentration of the droplet. These parameters were adjusted empirically (see Supporting Information) in order to produce SMPs with similar diameter (average: 0.5 μm; maximum: 1.2 μm) and a coefficient of variation (CV) of ∼50%, while noting that microparticles with a CV above 25% are considered polydisperse.33 Figures 2A−C show SEM images of the 2549

dx.doi.org/10.1021/la3047565 | Langmuir 2013, 29, 2546−2553

Langmuir

Article

to buckling, and therefore likely to collapse inward at only one location, unlike the thinner shells that form at low nanoparticle concentrations and buckle at multiple locations as drying proceeds. In this way, dimpled and crumpled microparticles form at higher and lower nanoparticle concentrations, respectively. In addition, buckling of the shell may also be hindered due to spatial constraints within the droplet at higher nanoparticle concentrations.37 In the case of the microparticles shown in Figure 2B,C, diluting the stock to the desired nanoparticle concentration with hexane also decreases the free oleic acid concentration, which in turn resulted in the detachment of oleic acid from the nanoparticles to maintain the equilibrium between bound and free oleic acid moieties (the solubility of oleic acid in hexane is assumed to be similar to that of stearic acid, i.e., ∼16 g/L38). Hence, the density of oleic acid chains on the nanoparticle surface decreases when the stock ferrofluid is diluted. Diluting the stock therefore had the combined effect of reducing both the nanoparticle oleic acid density and the nanoparticle concentration in the dispersed phase, both of which lead to shell formation and subsequent buckling during drying to form dimpled and crumpled SMPs. Implication of Surface Morphology in Bioaffinity Separations. In order to functionalize the synthesized microparticles, their surfaces were coated with polymer layers. The outer carboxyl layer provides a suitable base for further functionalization and has a high absolute value of zeta-potential, minimizing particle aggregation during handling and storage.8 Figure 4 shows the zeta-potential of the particles over a range of

Figure 3. Oleic acid chain density on iron oxide nanoparticles before and after washing steps with ethanol. Removal of oleic acid side chains on nanoparticles decreases their hydrophobicity and therefore increases their affinity for the oil−water interface.

to a submonolayer coverage of ∼42% on the nanoparticles. Hence, the contact angle of the nanoparticles changes after washing so that they have a greater affinity for the oil−water interface and also a greater tendency to aggregate with each other.34 After a third ethanol wash, nanoparticles could no longer be fully redissolved in hexane due to the decrease in their hydrophobicity. In order to assess the effect of oleic acid density on microparticle morphology, SMPs were synthesized with nanoparticles that were washed twice with ethanol. Microparticles produced with 1 g/L ethanol-washed nanoparticles are mostly crumpled (Figure 5A). SMPs produced using the unwashed nanoparticles with a monolayer of oleic acid and at the same nanoparticle concentration (1 g/L) are spherical (Figure 2A). Therefore, when the nanoparticle concentration is kept constant, the nanoparticle chain density influences the drying route to that of either spherical or crumpled microparticles. This indicates that the nanoparticle chain density determines their affinity for the interface and aggregation to each other and thus the likelihood of the formation of a shell of nanoparticles at the interface. To investigate if the nanoparticle concentration influences the resulting microparticles, SMPs were synthesized with 5 g/L ethanol-washed nanoparticles (i.e., submonolayer oleic acid coverage). The SMPs were found to be dimpled (Figure 5B), suggesting that nanoparticle concentration, independently of nanoparticle oleic acid coverage, influences the drying route and resulting microparticle morphology. From these results it can be inferred that the microparticle morphology depends on two factors: (i) the density of oleic acid chains attached to the nanoparticles and (ii) the concentration of iron oxide nanoparticles in the ferrofluid. The oleic acid chain density on the nanoparticles affects the rate of nanoparticle aggregation events and therefore the number of seed nuclei. The nuclei are drawn to the energy well at the oil−water interface due to amphiphilic nature of the oleic acid-deficient nanoparticles. The concentration of nanoparticles then determines the growth rate of the seed nuclei and, by mass balance, the thickness of the shell of densely packed nanoparticles that forms at the droplet interface. The thickness of the formed nanoparticle shell in turn affects its structural stability, as the critical buckling stress of a thin elastic spherical shell is proportional to the square of the thickness, Pc = 2Et2/(r2(3(1 − υ2))1/2), where E is the Young’s modulus, r is the shell radius, υ is the Poisson ratio, and t is the shell thickness.36 The thicker shells that form at higher nanoparticle concentrations are stronger and more resistant

Figure 4. Zeta-potential values of spherical SMPs at different steps during the polymer coating procedure, from uncoated particles (squares) to amine- (circles) and carboxyl-functionalized (triangles) microparticles. The change in the value of zeta-potential indicates that each coating step successfully modified SMP surfaces. Error bars show standard deviation.

pH values for the precoated particles and after the PEI and PAAMA coating steps. The change in the value of zetapotential indicates that each coating step successfully modified SMP surfaces. An outer carboxyl later is preferred over the amine layer due to its higher zeta-potential (Figure 4) and lower tendency to bind nonspecifically. From the observed differences in morphology it would be expected that the spherical, dimpled, and crumpled particles have different surface area per mass. Therefore, when modified with a biological capture ligand, their binding capacity should be higher than that of spherical SMPs of equivalent size. In order to accurately compare the binding capacity of microparticles with different morphologies, the shear rate applied in the Couette mixer was tuned to obtain equivalent size distributions 2550

dx.doi.org/10.1021/la3047565 | Langmuir 2013, 29, 2546−2553

Langmuir

Article

Figure 5. SEM images of SMPs produced with ferrofluid nanoparticle concentrations of (A) 1 g/L and (B) 5 g/L, both washed twice with ethanol, with insets showing magnified images. Continuous phase dextran 25% w/w, SDS 2% w/w, flow rates of continuous and dispersed phases 0.75 and 0.25 mL/min, respectively. Microparticles produced with 1 g/L nanoparticle concentration show signs of crumpling, whereas the microparticles produced with 5 g/L nanoparticle concentration are dimpled. Scale bar = 5 μm (1 μm in insets).

in each sample (see Supporting Information). The SMP concentration was also kept constant for each sample. To measure the binding capacity, each type of microparticle was functionalized with avidin, followed by incubation with an excess of biotin-HRP. ABTS indicator was used to colorimetrically quantify the amount of biotin-HRP bound per milligram of microparticles (Figure 6A). The dimpled SMPs were shown to have a 1.2 times higher binding capacities than spherical SMPs, but by far the biggest increase was observed with the crumpled microparticles which have a binding capacity approximately 6−7 times that of the spherical SMPs. Another possible implication of the different morphologies is the way the beads move through solution in a presence of magnetic field. To test this, SMPs were synthesized having maximum diameter 1.2 ± 0.1 μm, which were then isolated through centrifugation. These fractions were suspended in a viscous solution in a sealed well, and the particles were pulled with permanent magnets positioned at varying distances. Since the dimpled and crumpled particles have approximately the same cross-sectional area as spherical particles, i.e., each has one dimension ∼1.2 μm in length, it is possible to compare acting magnetic force by comparing their drag velocities. As expected, drag velocities of each type of SMP decrease with increasing distances from the magnet, since the magnetic field decays exponentially.32 The velocities of dimpled and crumpled SMPs are 1.3 times and 6.2 times lower than the spherical SMPs, respectively. This can be attributed to the lower nanoparticle content per SMP due to the hollow cores of the dimpled and crumpled particles and the higher drag coefficients of nonspherical microparticles. In order to obtain values for the drag coefficients of the crumpled and dimpled microparticles, the change in velocity due to both the volume and the drag was isolated for the dimpled and crumpled SMPs. The Stokes− Einstein relation can be modified for nonspherical microparticles in low Reynolds number flows, by multiplying the drag force by a correction factor, k, which accounts for the change in resistance of microparticles due to their morphology.39 The velocity of dimpled microparticles can therefore be described

Figure 6. (A) Biotin-HRP binding capacity of avidin-functionalized SMPs with different morphologies. Carboxyl-functionalized spherical SMPs were used as a blank. The dimpled SMPs have slightly higher binding capacities (1.2×) than spherical SMPs, the crumpled microparticles have binding capacities 6−7 times higher than the spherical SMPs. Error bars represent standard deviation. (B) Velocity of SMPs of diameter 1.2 ± 0.1 μm (largest cross-sectional diameter measured for dimpled and crumpled microparticles) with distance from a fixed magnetic field source. The velocities of dimpled and crumpled SMPs are 1.3 times and 6.2 times lower than the spherical SMPs, respectively. This can be attributed to the lower nanoparticle content per SMP due to the hollow cores of the dimpled and crumpled particles. Error bars represent standard deviation.

by vd = (md/kd)(Ms∇H/6πμr), where md and kd are the mass and resistance coefficients of the dimpled microparticles, respectively. A similar approach was used for the crumpled microparticles, with mc and kc representing the mass and resistance coefficients. The second set of terms in the above equation was experimentally derived using the spherical SMPs. The mass of the dimpled and crumpled microparticles was found using the cross-sectional TEM images in Figures 2E and 2F to be approximately 0.70 ± 0.02 and 0.27 ± 0.02 times that of spherical microparticles, respectively. The resistance coefficients kd and kc could then be estimated by fitting to the measured velocity curves (Figure 6B) and were found to be 1.03 ± 0.03 and 1.54 ± 0.11 for dimpled and crumpled microparticles, respectively (see Supporting Information). These resistance coefficients are in line with published values for ellipsoidal and irregularly shaped colloidal particles.39,40 It is desirable for magnetic affinity separations that the particles are uniform in size and have high binding capacity and high magnetic mobility.8 Scaling laws make the combination of high binding capacity and high magnetic mobility difficult to achieve with conventional spherical microparticles. Halving the 2551

dx.doi.org/10.1021/la3047565 | Langmuir 2013, 29, 2546−2553

Langmuir

■

diameter of a spherical microparticle doubles the surface area to volume ratio (A/V) but will result in a 4-fold decrease in the magnetic mobility, as it scales with the square of the microparticle radius. The advantages of the crumpled SMPs over spherical ones are highlighted in considering the A/V and velocity of the particles; crumpled microparticles have a A/V ratio 7 times higher than spherical microparticles of equivalent diameter and a magnetic mobility 6−7 times lower, indicating the A/V ratio is inversely proportional to the magnetic mobility for crumpled microparticles. Therefore, if one decreases the radius of spherical and crumpled SMPs by a factor of 2, the A/V ratio increases by a factor of 2 for both types of microparticles, but the velocity will decrease by a factor of 2 and 4 for the crumpled and spherical particles, respectively. The spherical SMPs discussed here have 6.1 times greater magnetic mobility compared to commercial spherical microparticles (Dynal beads, Invitrogen) of equivalent diameter (unpublished data). The crumpled microparticles presented here have a magnetic mobility comparable with the commercial beads but have an A/V ratio 6−7 times higher. The cost-effectiveness of crumpled SMPs over spherical ones becomes especially significant when large-scale separations, such as the downstream processing of biopharmaceuticals, are considered.6

■

Article

ASSOCIATED CONTENT

S Supporting Information *

Characterization of twin entry port Couette shear device. This material is available free of charge via the Internet at http:// pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Present Address †

Department of Chemistry, Loughborough University, Loughborough LE11 3TU, United Kingdom. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS This material is based upon works supported by the Science Foundation Ireland under Grants 08/RP1/B1376 and 08/IN1/ B2072. The authors acknowledge the Nano Imaging and Material Analysis Centre (NIMAC), University College Dublin, for assistance in preparing and imaging specimens for this research.

■

CONCLUSIONS

REFERENCES

(1) Al-Deen, F. N.; Ho, J.; Selomulya, C.; Ma, C.; Coppel, R. Superparamagnetic Nanoparticles for Effective Delivery of Malaria DNA Vaccine. Langmuir 2011, 27, 3703−3712. (2) Polyak, B.; Friedman, G. Magnetic Targeting for Site-Specific Drug Delivery: Applications and Clinical Potential. Expert Opin. Drug Delivery 2009, 6, 53−70. (3) Chang, W.-S.; Shang, H.; Perera, R. M.; Lok, S.-m.; Sedlak, D.; Kuhn, R. J.; Lee, G. U. Rapid Detection of Dengue Virus in Serum Using Magnetic Separation and Fluorescence Detection. Analyst 2008, 133, 233−240. (4) Schutt, W.; Gruttner, C.; Hafeli, U.; Zborowski, M.; Teller, J.; Putzar, H.; Schumichen, C. Applications of Magnetic Targeting in Diagnosis and Therapy–Possibilities and Limitations: a Mini-Review. Hybridoma 1997, 16, 109−17. (5) Balmayor, E. R.; Pashkuleva, I.; Frias, A. M.; Azevedo, H. S.; Reis, R. L. Synthesis and Functionalization of Superparamagnetic Poly-εCaprolactone Microparticles for the Selective Isolation of Subpopulations of Human Adipose-Derived Stem Cells. J. R. Soc. Interface 2011, 8, 896−908. (6) Muzard, J.; Platt, M.; Lee, G. U. Affinity Separation: M13 Bacteriophage-Activated Superparamagnetic Beads for Affinity Separation. Small 2012, 8, 2289−2289. (7) Uhlen, M. Magnetic Separation of DNA. Nature 1989, 340 (6236), 733−734. (8) Shang, H.; Chang, W.-S.; Kan; Majetich, S. A.; Lee, G. U. Synthesis and Characterization of Paramagnetic Microparticles through Emulsion-Templated Free Radical Polymerization. Langmuir 2006, 22, 2516−2522. (9) Muzard, J.; Fields, C.; O’ Mahony, J. J.; Lee, G. U. Probing the Soybean Bowman-Birk Inhibitor Using Recombinant Antibody Fragments. J. Agric. Food Chem. 2012, 60, 6164−6172. (10) Liu, X.; Liu, H.; Xing, J.; Guan, Y.; Ma, Z.; Shan, G.; Yang, C. Preparation and Characterization of Superparamagnetic Functional Polymeric Microparticles. China Particuol. 2003, 1, 76−79. (11) Ugelstad, J.; Soderberg, L.; Berge, A.; Bergstrom, J. Monodisperse Polymer Particles - A Step Forward for Chromatography. Nature 1983, 303, 95−96. (12) Levison, P. R.; Badger, S. E.; Dennis, J.; Hathi, P.; Davies, M. J.; Bruce, I. J.; Schimkat, D. Recent Developments of Magnetic Beads for Use in Nucleic Acid Purification. J. Chromatogr., A 1998, 816, 107− 111.

The morphology of superparamagnetic microparticles produced using a single emulsion-based self-assembly technique can be controlled through variation the nanoparticle concentration in the dispersed phase and ligand density on the nanoparticle surfaces. The collapsed shell and hollowness of the dimpled and crumpled particles suggest that they are formed via the following mechanism: Dissolved nanoparticles build up and aggregate at droplet interfaces during drying, resulting in the formation of a shell of densely packed nanoparticles. This shell then collapses as drying proceeds. The formation of the dimpled and crumpled SMPs is the result of two factors: (i) decreased oleic acid density on nanoparticle surfaces reduces their hydrophobicity and increases their affinity for the droplet interface, and (ii) the variation in the nanoparticle concentration changes the shell thickness and results in different microparticle buckling modes and therefore different microparticle morphologies. The results confirm the previously reported finding that nanoparticle surface activity is critical in determining the morphology of resulting microparticles in emulsion-templated self-assembly.30 SMPs with dimpled and crumpled morphologies have higher binding capacity per mass than spherical SMPs, while maintaining high magnetic field velocities due to their large size and high iron oxide content. The crumpled microparticles in particular have a magnetic mobility similar to that of a commercial alternative of equivalent size (Dynal beads, Invitrogen) but a surface area to volume ratio 6−7 times higher. These SMPs would be suitable for use in magnetic-based affinity separations where high microparticle surface area and magnetic mobilities are desirable. The novel Couette mixer technique used here for microemulsion synthesis, with separate inlet ports for the continuous and dispersed phases, avoids the need to prepare a premixed emulsion and is useful when high throughput synthesis of emulsions is required. 2552

dx.doi.org/10.1021/la3047565 | Langmuir 2013, 29, 2546−2553

Langmuir

Article

(13) Yang, S.; Liu, H.; Zhang, Z. A Facile Route to Hollow Superparamagnetic Magnetite/Polystyrene Nanocomposite Microspheres via Inverse Miniemulsion Polymerization. J. Polym. Sci., Part A: Polym. Chem. 2008, 46, 3900−3910. (14) Lee, M.; Cho, Y.; Park, J.; Chung, H.; Jeong, S.; Choi, K.; Moon, D.; Kim, S.; Kim, I.-S.; Kwon, I. Size Control of Self-Assembled Nanoparticles by an Emulsion/Solvent Evaporation Method. Colloid Polym. Sci. 2006, 284, 506−512. (15) Qiu, P.; Jensen, C.; Charity, N.; Towner, R.; Mao, C. Oil Phase Evaporation-Induced Self-Assembly of Hydrophobic Nanoparticles into Spherical Clusters with Controlled Surface Chemistry in an Oilin-Water Dispersion and Comparison of Behaviors of Individual and Clustered Iron Oxide Nanoparticles. J. Am. Chem. Soc. 2010, 132, 17724−17732. (16) Kilinc, D.; Blasiak, A.; O’Mahony, J. J.; Suter, D.; Lee, G. U. Magnetic Tweezers-Based Force Clamp Reveals Mechanically Distinct apCAM Domain Interactions. Biophys. J. 2012, 103, 1120−1129. (17) Koo, H. Y.; Chang, S. T.; Choi, W. S.; Park, J.-H.; Kim, D.-Y.; Velev, O. D. Emulsion-Based Synthesis of Reversibly Swellable, Magnetic Nanoparticle-Embedded Polymer Microcapsules. Chem. Mater. 2006, 18, 3308−3313. (18) Huang, C.-C.; Liu, T.-Y.; Su, C.-H.; Lo, Y.-W.; Chen, J.-H.; Yeh, C.-S. Superparamagnetic Hollow and Paramagnetic Porous Gd2O3 Particles. Chem. Mater. 2008, 20, 3840−3848. (19) Yanagishita, T.; Fujimura, R.; Nishio, K.; Masuda, H. Fabrication of Monodisperse Polymer Nanoparticles by Membrane Emulsification Using Ordered Anodic Porous Alumina. Langmuir 2009, 26, 1516− 1519. (20) Mason, T. G.; Bibette, J. Shear Rupturing of Droplets in Complex Fluids. Langmuir 1997, 13, 4600−4613. (21) Goubault, C.; Pays, K.; Olea, D.; Gorria, P.; Bibette, J.; Schmitt, V.; Leal-Calderon, F. Shear Rupturing of Complex Fluids: Application to the Preparation of Quasi-Monodisperse Water-in-Oil-in-Water Double Emulsions. Langmuir 2001, 17, 5184−5188. (22) Mabille, C.; Schmitt, V.; Gorria, P.; Leal Calderon, F.; Faye, V.; Deminiare, B.; Bibette, J. Rheological and Shearing Conditions for the Preparation of Monodisperse Emulsions. Langmuir 1999, 16, 422− 429. (23) Umbanhowar, P. B.; Prasad, V.; Weitz, D. A. Monodisperse Emulsion Generation via Drop Break Off in a Coflowing Stream. Langmuir 1999, 16, 347−351. (24) Martín-Banderas, L.; Rodríguez-Gil, A.; Cebolla, Á .; Chávez, S.; Berdún-Á lvarez, T.; Fernandez Garcia, J. M.; Flores-Mosquera, M.; Gañ ań -Calvo, A. M. Towards High-Throughput Production of Uniformly Encoded Microparticles. Adv. Mater. 2006, 18, 559−564. (25) Sugiura, S.; Nakajima, M.; Iwamoto, S.; Seki, M. Interfacial Tension Driven Monodispersed Droplet Formation from Microfabricated Channel Array. Langmuir 2001, 17, 5562−5566. (26) Schmitt, V.; Leal-Calderon, F.; Bibette, J.; Antonietti, M. Preparation of Monodisperse Particles and Emulsions by Controlled Shear. Colloid Chem. II 2003, 227, 195−215. (27) Rosca, I. D.; Watari, F.; Uo, M. Microparticle Formation and Its Mechanism in Single and Double Emulsion Solvent Evaporation. J. Controlled Release 2004, 99, 271−280. (28) Rosa, G. D.; Iommelli, R.; La Rotonda, M. I.; Miro, A.; Quaglia, F. Influence of the Co-encapsulation of Different Non-Ionic Surfactants on the Properties of PLGA Insulin-Loaded Microspheres. J. Controlled Release 2000, 69, 283−295. (29) Yang, Y.-Y.; Chia, H.-H.; Chung, T.-S. Effect of Preparation Temperature on the Characteristics and Release Profiles of PLGA Microspheres Containing Protein Fabricated by Double-Emulsion Solvent Extraction/Evaporation Method. J. Controlled Release 2000, 69, 81−96. (30) Liu, B.; de Folter, J. W. J.; Mohwald, H. Magnetic Nanoparticles-Induced Anisotropic Shrinkage of Polymer Emulsion Droplets. Soft Matter 2011, 7, 3744−3749. (31) Shang, H.; Lee, G. U. Magnetic Tweezers Measurement of the Bond Lifetime-Force Behavior of the IgG-Protein A Specific Molecular Interaction. J. Am. Chem. Soc. 2007, 129, 6640−6646.

(32) Tanase, M.; Biais, N.; Sheetz, M. Magnetic Tweezers in Cell Biology. In Methods in Cell Biology; Academic Press: New York, 2007; Vol. 83, pp 473−493. (33) Mabille, C.; et al. Monodisperse Fragmentation in Emulsions: Mechanisms and Kinetics. Europhys. Lett. 2003, 61, 708. (34) Wu, W.; He, Q.; Jiang, C. Magnetic Iron Oxide Nanoparticles: Synthesis and Surface Functionalization Strategies. Nanoscale Res. Lett. 2008, 3, 397−415. (35) Zhang, L.; He, R.; Gu, H.-C. Oleic Acid Coating on the Monodisperse Magnetite Nanoparticles. Appl. Surf. Sci. 2006, 253, 2611−2617. (36) Huston, R.; Josephs, H. Buckling of Spherical Shells. In Practical Stress Analysis in Engineering Design, 3rd ed.; CRC Press: Boca Raton, FL, 2008. (37) Sen, D.; Mazumder, S.; Melo, J. S.; Khan, A.; Bhattyacharya, S.; D’Souza, S. F. Evaporation Driven Self-Assembly of a Colloidal Dispersion during Spray Drying: Volume Fraction Dependent Morphological Transition. Langmuir 2009, 25, 6690−6695. (38) Mirmehrabi, M.; Rohani, S. Measurement and Prediction of the Solubility of Stearic Acid Polymorphs by the UNIQUAC Equation. Can. J. Chem. Eng. 2004, 82, 335−342. (39) Happel, J.; Brenner, H. Low Reynolds Number Hydrodynamics, 2nd ed.; Prentice-Hall, Inc.: Upper Saddle River, NJ, 1973. (40) Wen-Ching, Y. Particle Characterization and Dynamics. In Handbook of Fluidization and Fluid-Particle Systems; CRC Press: Boca Raton, FL, 2003.

2553

dx.doi.org/10.1021/la3047565 | Langmuir 2013, 29, 2546−2553