Comprehensive two-dimensional hydrodynamic chromatography × size-exclusion chromatography with intermediate dissolution for the characterization of polymeric nanoparticles

(1)

MSc Analytical Sciences

Comprehensive two-dimensional hydrodynamic chromatography

× size-exclusion chromatography with intermediate dissolution for

the characterization of polymeric nanoparticles

By

Noor Abdulhussain

July 2017

48 EC

Supervisor

Examiner

Bob Pirok prof. P.J. Schoenmakers

(2)

- 1 -

Acknowledgements

This report is written to fulfill master research for the M. Chemistry study at the university of Amsterdam. Where I have worked in the department of Van 't Hoff Institute for Molecular Sciences (HIMS) at the university of Amsterdam. During my research I was supervised by Bob Pirok and Prof P.Schoenmakers

I would like to express my special thanks to those who gave me support and by giving me guidance and suggestion throughout this project. First of all, I want to thank my supervisors Bob Pirok and Prof P.Schoenmakers for giving me the opportunity to do this challenging project and their guidance and explanation throughout this internship. Their advices have helped me to understand more about this subject. Their effort throughout the internship has been invaluable.

I would like to thanks all the people from laboratory who gave support and advice. Also I want to thank all the interns for the great time.

Amsterdam, Noor Abdulhussain 26-7-2017

(3)

- 2 -

Abstract

Polymeric nanoparticles are quickly growing in a wide spectrum of different fields ranging from electronics to coatings and inks. A wide array of techniques is available to determine the different properties which correlate to their size, chemical composition, shape and size distribution based on different preparation techniques and the era of application. A comprehensive two-dimensional liquid chromatography (LC×LC) method has been developed for nanoparticle analyses. Whereby, the developed method consists of a combination of hydrodynamic chromatography (HDC) in the first-dimension, to separate the particles based on their size and ultra-high-performance size-exclusion chromatography (SEC) in the second dimension to separate the polymer molecules according to their molecular weight. When performing two-dimensional chromatography there are certain aspects that have to be considered. Several aspects that need to be taken into consideration during this study are the solvent incompatability, sample transformation (from nanoparticle to polymers), optimum parameters of the LC×LC- setup, possible adsorption effect of SEC separation and short analysis time of at least one hour.

A HDC×SEC method has been developed with intermediate transformation of separated nanoparticles by dissolving it online with THF into polymers. A mixer was incorporated to efficiently mix the first-dimension effluent with THF to transform the sample and the polymer fractions were focussed using stationary-phase-assisted modulation to enhance the signal intensity and to remove any remaining water from the first-dimension eluent. The usefulness of the developed system was demonstrated through a separation of polystyrene and polyacrylate nanoparticles. Which gave a good insight into the differences in molecular weight distributions of the polymers within 60 minutes.

(4)

- 3 -

1. Introduction

Polymer-nanoparticle (PNP) is a collective term to describe any kind of polymer-based NP but is particularly applied for referring to nanospheres and nanocapsules1_{. Nanoparticles are defined as}

colloidal, solid particles in the range from 10-1000 nm2_{. PNPs can be prepared from preformed polymers}

through various methods such as nanoprecipitation, dialysis, solvent evaporation, salting-out and supercritical fluid technology2–7_{. Another approach is to synthesize PNPs by polymerization of}

monomers by using various types polymerization methods such as emulsion polymerization2,8,9_and

mini-/micro-emulsion polymerization10,11_{. These methods are well described and have been compared}

in a detailed review by Prasad Raoa and Geckeler2_{. Polymeric nanoparticles are quickly growing in a}

wide spectrum of different fields ranging from electronics, conducting materials, biotechnology, coatings and inks1,2_{. Moreover, PNPs may have different properties which correlate to their size,}

chemical composition, shape and size distribution based on different preparation. Therefore, it is important to be able to fully investigate these properties accurately and efficiently.

Most used methods are microscopic-based techniques such as scanning electron microscopy (SEM), transmission electron microscopy (TEM) or dynamic light scattering (DLS)12–14_{. These techniques are}

able to identify individual nanoparticles and define their shape and size. Another array of analysis technique that can be used for nanoparticle characterization is size-selective based techniques. For example, field-flow fractionation (FFF), size-exclusion chromatography (SEC) and hydrodynamic chromatography (HDC)15_{. Moreover, some research groups have developed hyphenated techniques for}

nanoparticle analysis to obtain more information than is obtainable by an individual technique alone such as HDC-ICPMS16,17_{and AF4-ICPMS}18_{. Another technique to obtain more information for complex}

mixture analysis is by performing two-dimensional liquid chromatography (LC×LC). LC×LC is utilized by applying two different chromatographic systems that must have different selectivities for the sample components. The fractions obtained in the first-dimension are further distinguished through a modulator in the second-dimension column.

The aim of the project is that completely different and incompatible separation systems are combined into a single, highly efficient and extensively optimized instrument. Amongst the investigated applications is the analysis of complex polymeric nanoparticles encountered in coating formulations. A large number of sample dimensions are found in these complex samples such as the particle-size distribution, the chemical (surface) composition(s), charge and molecular weight. Therefore, an effective technique for the analysis of complex mixtures is comprehensive LC×LC is needed. In this study, hydrodynamic chromatography (HDC) is used to separate the nanoparticles based on size. Next, the modulator is used to dissolve the nanoparticle dispersions, yielding a homogeneous molecular solution. These molecules are then separated in the second dimension based on their molecular weight (SEC). However, a successful application of this technique faces a number of challenges.

Chromatography fundamentally is a physical process that dilutes the analyte before it is detected. In a second-dimension separation, this dilution process is compounded, resulting in low detection sensitivity19_{. It is required to concentrate the fraction from the first dimension either prior to or during}

the transfer to the second-dimension column. Moreover, in HDC an aqueous mobile phase is mostly used which is not compatible with the organic mobile phase in SEC analysis. Any remaining aqueous fraction may cause effects on the SEC analysis by potentially producing interaction in SEC mechanism. This requires an efficient mixing ratio of the two mobile phases and research in the effect of water in the second dimension. The goal of this study is to develop a comprehensive two-dimensional chromatography with HDC and SEC by addressing the mentioned challenges.

(6)

- 5 -

2. Background theory

2.1 High-performance liquid chromatography (HPLC)

Ultra-high-pressure liquid chromatography (UHPLC) uses the same separation principle as conventional high-performance liquid chromatography (HPLC). Through HPLC compound scan be separated, quantitated and identified which are present in the desolved sample. In this project the samples are polymers. The separation technique involves an injection of a small volume of the sample flows with the mobile phase into a column with porous particles (stationary phase). The sample molecules interact chemically or physically with the stationary phase, where the components of the sample are transported along the column at different retention time and get detected by a detector. Further advances in the HPLC instrument and column technology were made to achieve significant increases in resolution, speed and sensitivity in liquid chromatography. The use of a high-efficiency LC system holistically designed to accommodate smaller particles and very high operating pressure is termed ultra-performance liquid chromatography. In general, there are some of the characteristics of chemical components which can be used to create HPLC separation based on these characteristics. Such as the polarity of the samples, electrical charge or separations based on molecular size. Size-based separation techniques can be divided in several techniques such field-flow fractionation, size-exclusion chromatography and hydrodynamic chromatography. The last two technique are elaborately described in the next sections.

2.2 Hydrodynamic chromatography

Hydrodynamic chromatography (HDC) is a technique in which macromolecules and particles are separated according to size. HDC was first published in papers by, DiMarzio & Guttman, approximately five decades ago in which they described the theoretical fundamentals of the separation20_{. Later on,}

Small referred that the separation is known as HDC. HDC is a rapid and convenient method to obtain a fingerprint of the size distribution of particles. It is involved in a broad range of application such as polymer latexes21_{, DNA and proteins}22_{, engineered nanoparticles}16_{, biomaterials and much more}23_{. The}

term hydrodynamic is the key parameter for the driving source of the separation. The separation takes place in open capillaries (capillary hydrodynamic chromatography (CHDC)) or in a column packed with solid or non-porous beads (packed-column hydrodynamic chromatography (PCHDC)). The separation arises from a non-turbulent Poiseuille flow profile that develops in an open capillary or in the interstitial medium of a packed column24_.

The theories for PCHDC separation mechanism have been derived from theories that were used to describe the separation mechanism of particles in CHDC24_{. The fluid under laminar flow conditions in}

a tube develops a parabolic flow velocity, in which the faster streamlines tend to be in the centre line of the capillary and decreased toward the wall figure 1A. The parabolic flow profile can be described through Reynolds numbers. For packed capillary, the parabolic flow outline develop in the capillary at Reynolds numbers in the range of 1to100 (Re=dpρū/η, where dp is the diameter of the packing particles

ū the average linear velocity, η the viscosity, ρ the density, and the units are chosen such that Re is dimensionless), such that the fasters streamline develops in the center of the capillary24_.

Due to the fixed size of the particles, the centre of gravity of the particles can’t approach the capillary (radius, R) wall closer than their own effective radius (Ø) as shown in figure 1B. Large particles will migrate at the average velocities of the faster streamlines observed in the range R- Ø compared to the smaller particles. The smaller particles tend to migrate with the slower streamline close to the walls of the capillary.Therefore, this process results in a separation based on different particle size and the larger particle will elute earlier ( if the ratio λ = Ø /R is too large) from the capillary than the smaller particle does23_.

(7)

- 6 - In PCHDC the interstitial medium may be represented as a bundle of cylindrical capillaries with average radius in which HDC can be performed Figure 1C. Similar to SEC the elution order is same in HDC. The larger particles elute ahead than the smaller particles. However, the mechanism of separation of these two techniques is different as described.

The ratio of the effective analyte radius to the capillary radius is the key parameter in the HDC separation which determines the size based separation since particles with different values of λ will migrate at different velocities the capillary. For the HDC-effect to occur, the useful dynamic range for λ falls within the limits of λ > 0.4. However, Striegel described that particles may not always all fluid streamlines, mainly because of equation 6 and, hence, equation 7 (see Appendix A) are idealized equations in which analytes are considered as a non-rotating, impermeable hard sphere. On the contrary, it is important to take into account that when particles are occupied in a parabolic fluid, the side of the particle that is closer to the centre of the capillary is located on a higher streamline velocity than the side that is closest to the wall of the capillary. This can results that the particle rotates and therefore affects the translational velocity23_{. Moreover, particles may also have shapes that are non-spherical, may be permeable or both.}

To avoid these effects and non-idealities, a quadratic term (C) is added to equation 6 as shown in appendix A resulting in the following equation:

τ =(1+2λ−Cλ2₎−1 ₍₇₎

In literature several values of C have been proposed24,25_{, each value corresponds to a different situation}

of HDC retention. For a highly idealized exclusion, HDC retention that doesn’t incorporate any hydrodynamic effect the value of C would be zero. For a simple model based on a Poiseuille flow profile reﬂecting only the exclusion of the analyte center from the tube wall while taking into account the velocity proﬁle, the value of C would be one. Depending on the type of sample (e.g., hard sphere polymer or permeable), good solvent and temperature condition or effects such as rotation, C can vary between 1 and 524,25 _{. For polymer solution at a good solvent and temperature condition, a value of 2.698 is used}

for C.

Figure 1: Mechanism of separation in hydrodynamic chromatography (HDC), Arrows indicate the flow streamlines (A/B). Under laminar ﬂow conditions, the interstitial medium of a packed column may be considered as a collection of capillaries (tubes) in which HDC can be performed (C)24_.

(8)

- 7 - As mentioned before HDC is a convenient method for determination of particle size distribution for quantitative interpretation a calibration is needed in order to establish a relationship between the particle size and elution volume. A calibration of an HDC system is obtained by simply determine the values of a series of monodisperse polystyrene standards whose particle size has been thoroughly characterized by careful electron microscopy measurements. The calibration is then used to give a size information on the unknown particles. However, it is also possible to calculate the particle size of the eluting species, in theory, using the equations given before and in the appendix. The calculation for the particle size is first done by measuring τ and from τ the aspect ratio λ can be calculated. Thereafter, the effective radius of the particle can be obtained using the following equation, where RG is the radius of gyration.

∅ =√π

2 RG (8)

2.2.1 Effect of ionic strength of the mobile phase in hydrodynamic

chromatography

Besides being dependent on the diameters of the particles and the packing for HDC separation, it is been illustrated by small that HDC separation is also dependent on the eluent. As mentioned before in size-exclusion chromatography the mobile phase is important in order to avoid enthalpic interaction with the stationary phase. Similar case with HDC the occurrence of an electrical double-layer (EDL) around the particles and colloidal interactions such as the van der Waal’s attractive forces as well as colloidal repulsive forces must also be taken into account. The eluent for HDC in packed-column mostly consists of deionized water with additives such as formaldehyde or sodiumazide to prevent bacterial growth and contamination. Moreover, the beads should be made of a material that minimizes the enthalpic interactions between the beads and the analytes. Therefore, the beads should be ‘‘inert’’, in order to achieve this, salts and/or surfactants are added to the eluent to screen electrostatic repulsion between the analytes and the beads of the packing24_{. The addition of a surfactant is also used to stabilize the particles}26

The electrostatic repulsions arise from the overlap of the EDL’s around the particles as well as the capillary wall. The presence of electrolyte ions in the eluent has a strong effect on the interaction between the electrostatic repulsion and the van der Waal’s attractive forces. Small has emphasized in his work that the effectiveness of particle separating (Rf) was found to increase with decreasing ionic strength21_{. The separation factor is defined by the ratio of the average velocity of the particle to the}

average fluid velocity (highest volume). Under high ionic strength conditions, the EDL’s around the particles and the surface wall of the packing are compressed and van der Waal’s attractive forces dominate. Therefore, the particle will migrate at slower velocity streamlines near the wall resulting in an increase of particle axial dispersion and decrease in the average velocity of the particle. In contrast, at a low ionic strength of the eluent, the particle will be repelled from the surface wall of the packing and will be forced into the faster streamline resulting with an increase in Rf. The electrostatic repulsion determines how closely a particle will migrate near the packing beads at a given ionic strength.

2.2.2 Van Deemter Curve in hydrodynamic chromatography

A fundamental van Deemter equation has to be addressed in order to comprehend the bandbroading effect in HDC. The van Deemter Curve describes the column efficiency through equation(9), where H represents plate height and µ represents ﬂow velocity.

H = A +B

(9)

- 8 - The A term describes the contribution to band broadening from eddy dispersion, which is related to the multitude pathways that a molecule takes to find its way through a packed column. This term is also flow-velocity independent. The B-term corresponds to

the contribution of longitudinal diffusion which describes the dispersion of molecule within the column along the axis of the flow path. At a lower flow velocity, the contribution of this term increases because at lower velocities the analyte spends a longer time in the column and therefore has more time to diffuse. Moreover, the contribution of this term decreases once the analytes have small diffusion coefficients, Dm, and therefore decreases with increasing analyte size. The contribution is in essence very small for most macromolecules. The last term, C-term, contributes to resistance of the analyte to mass-transfer process and consist in three different form (see equation 9). The Csm term describes the resistance to

mass transfer delivered by the stagnant mobile phase inside the pores of the column particles. In PCHDC this term equals zero. The Cs term result from the resistance

to mass transfer caused by conventional liquid chromatographic sorptive processes, which are enthalpically dominated. These processes are negligible because the retention in ideal HDC the solution translational entropy of the analyte is most dominated, therefore Cs term equals zero. The last term

related to the mass transfer is Cm, which corresponds to resistance to mass transfer caused by the

mobile-phase effect in the interstitial medium. In PCHDC the separation occurs in this region and therefore Cm

is the only term contributing to the C-term. Striegel et al., illustrated in his work the next figure 2 in which van Deemter curve for HDC can be seen. Figure indicated that there is no upper limit to the flow velocity at which HDC separation may be conducted.

2.3 Size exclusion chromatography

Size exclusion chromatography (SEC) relies on the size of the polymer molecules in solution rather than any chemical interaction between particles and stationary phase. SEC is a technique for measuring the complete molecule distribution of polymers by separating them on the basis of their size. In addition, SEC also separates polymers into its own component parts, such as oligomers, monomers, additives. The mobile phase is a liquid so it is important to use the appropriate solvent to dissolve the polymers. When it gets dissolved the molecular chains of polymers coil up and become like a sphere of strings. The size of that sphere depends on the molecular weight of the polymer, the higher the molecular weight, the larger the sphere will become. In a sample solution may consist of different sizes of polymers spheres resulting in a size distribution. Once the solution gets injected, it will flow into the SEC column with the mobile phase. The SEC system is unique because the separations do not depend on any interaction between the sample molecules and stationary phase, unlike common chromatography systems. In order to prevent any interactions, a strong eluent has to be used. The mobile phase flows through the column, which consists of porous, rigid particles. The column is packed with different sized particles that have different sized pores made of silica molecules or polymers. The pore size can be controlled depending on the size of the molecules that have to be separated.

Figure 2: Reduced plate height h versus reduced velocity ν for packed-column hydrodynamic

chromatography (HDC), picture derived from Striegel et al.24

(10)

- 9 - The separation of the sample takes places inside the column consist of particles with known pore sizes. The pore volume and the pore-size distribution determine the molar mass range which can be separated, whereas the particle size determines the diffusion processes between the polymer molecules on their way through the column. When the dissolved polymer molecules flow through the column, they get separated based on their size. The polymer spheres that are larger than the biggest pores will tend to only pass between the particles in the column which leads to a short residence time. Polymer coils smaller than the smallest pores tend to permeate into the pores volume experiencing in a more complex pathway. Upon elution, the sample molecules are detected by the detector. In order to calculate molecular weight of a polymer sample, a

calibration with standard polymers with a known molecule weights should be measured first. The standards with a known MW-values, the peak molecular weight are determined. The calibration curve is plotted relating log (MW) at the vertical axis to elution time or volume on the horizontal axis.The molecular weight of the polymer is determined from the calibration curve by noting the elution time and then reading the molecular weight. Figure 3 illustrates a calibration curve, solutes that are too large to fit any of the pores can access only the interstitial volume, which corresponds to the exclusion limit of the column. The slope of a calibration curve may give information about

the resolution or peak separation27_{. A shallow slope corresponds to a relatively small change in M and}

relatively large change in V, two molecules with equivalent M are better separated with a low slope. In addition to the calibration curve, another effect, under high-stress conditions in the chromatographic column polymer chains can be deformed in a stressed shape instead of colloidal spheres resulting in affected SEC analysis. This phenomenon can be observed in the calibration curve as a strong shift in the exclusion limit, this phenomenon is known to be so-called slalom chromatography (SC). From the calibration, a molecular- weight distribution (MWD) can also be generated.

2.3.1 Core-shell columns

Significant advances have been achieved in SEC column technology which is the (re)-introduction of the application of core-shell particles in SEC analysis28,29_{. Core-shell particles consist of a solid core}

with porous silica or silica modified with C18 chains wrapped around it resulting in a typical diameter size of 2.6 µm. It is been established that chromatographic elution of small molecules using core-shell column is affected by both B term and C term of the van Deemter equation as described in section 2.2.2. Resulting in an increased efficiency and reduced plate height values due to the low pore volume in core-shell column29_{. Another feasibility of using core-shell particles as stationary is the fast analysis time}

compared to the conventional SEC columns (fully porous columns). This makes the core-shell column a fast, efficient and widely applicable columns for SEC analysis and more promising as second dimension separation in a two-dimensional liquid chromatography29_.

Figure 3: An example of a calibration curve derived from Y. Vander Heyden et al

(11)

- 10 -

2.4 Two-dimensional liquid chromatography

Two-dimensional liquid chromatography (2D-LC) is a technique that makes use of two different types of independent separation mechanisms. The columns are placed in series and coupled on-line to a modulator. The modulator includes a switching valve with loops which get filled with the fractions of the first-dimension and subsequently injected to the second dimension. The modulation time is referred to the rate that is needed for the valve to switch. Several modes can be used for 2D-LC, such as comprehensive two-dimensional liquid chromatography (LC×LC). Any multi-dimensional chromatographic separation is considered comprehensive in which all the molecules contained in the sample injected into the first column migrate through the next column30_{. Furthermore, heart-cutting, the}

first-dimension eluent is cut in fractions containing the compound(s) of interest and injected into the second dimension column. In stop-and-go 2D-LC, the mobile phase flow through the first-dimension column is stopped while the analysis of the transferred fraction is carried out on the second-dimension column, which makes easier the design of this second separation, increases the total analysis time compared to the previous approaches, but still avoids the storage of the collection fractions.

When performing two-dimensional chromatography there are certain aspects that have to be considered. Most important aspect is the degree of orthogonality between two different retention mechanism. An analysis is considered orthogonal if the two used separation mechanisms are independent of each other and provide different selectivities and retention profiles30_{. Analysis with high orthogonality represents}

a chromatogram in which the compounds are randomly distributed over the entire separation space. Several combinations have been applied in LC×LC31_{. Achievement of orthogonality not only depends}

on the separation mechanism but also on the separation condition such as solvent compatibility. If these solvent mixtures are not completely miscible, a serious difficulty arises. This is the case when the solvent of the first dimension is an aqueous solution while that used in the second dimension is an organic solvent that is not necessarily nor easily miscible to the aqueous solution. If a strong solvent from the first dimension get transferred to the second dimension in which the strong solvent is a weak solvent in that mode might cause peak broadening due to solvent strength and viscosity mismatch. This is particularly serious in SEC in which the mobile phase is selected to be an excellent strong organic solvent of the sample and mobile phase, so the sample components are not adsorbed on the stationary phase. If an aqueous mobile phase is used in first dimension and fraction are transferred into the second dimension when the strong solvent is used for reasonable separation, such as SEC, it might causing losses in second dimension column performance31_{. Other aspects that come with the transferred volumes}

are the maximally allowable injection volume through the second dimension column. The sample plugs are diluted when they migrate along the first dimension column, forming the injection bandwidth of the secondary column. Ideally, the primary bandwidths injected onto the secondary column shall be narrow enough to avoid deterioration of separation efficiency of the secondary column. A “refocusing” step is normally adopted to reduce the dispersion of bands and injection volumes on the secondary column Several approaches can be used to minimize this problem in LC×LC system example, partial vaporization, thermal modulation, on-column focusing at the head of the second dimension column and performing active modulation31_.

In this study active modulation is used to reduce the above-mentioned problems that occur in 2D-LC. In general, a key element between two separate dimensions of any LC×LC system is the modulator. The modulator is been implemented by using a switching valve, equipped with sampling loops. In active modulation, the switching valve is improved by replacing the loops by integrating trap columns. This type of configuration provides a focusing mechanism ensuring that the band broadening is reduced. In general, the trap columns in such a way that it match the selectivity of the second dimension. This is to allow the mobile phase of the second dimension to desorb the trapped analytes efficiently. Vonk et al. illustrated an approach in which using active modulation offered another additional benefit32_{. The}

(12)

- 11 - dimension strong cation exchange before transferring it to the second dimension reversed phase column. More interestingly is the use of a narrower second dimension column offering the use of capillary reversed phase columns operating with low flow rates which matched the mass spectrometry, and applicable to proteome analysis. Gargona et al. illustrated an active modulation set-up implemented with a makeup flow to the trap modulator31_{. The main reason for doing so was to reduce the elution}

strength of the first dimension column prior to sample loading on the trap column in the modulator interface. This set-up provides an improvement in trapping and reduction in dilution, band broadening and improving sensitivity.

A different point of consideration is the efficiency of sampling in order to maintain the first-dimension separation. This induces the requirement of the second-dimension to be fast, so as to complete the elution of one fraction before the next fraction from the first-dimension is transferred and injected. For this reason, it is important to obtain the best possible resolution in the shortest time possible. In the case of this study, the second-dimension is performed by using core-shell column allowing for fast analysis as described in section 2.3.

3. Experimental methods

3.1 Chemicals

Tetrahydrofuran (THF, non-stabilized and HPLC-S grade) was obtained from Biosolve B.V. (Valkenswaard, the Netherlands) and used for SEC analysis.Toluene was acquired from Biosolve B.V. (Valkenswaard, the Netherlands) and was applied as SEC marker. Polystyrene standards (PS) for determining calibration curves were obtained from Polymer Laboratories (now Agilent Technologies, Church Stretton, Shropshire, UK). For HDC the mobile phase was an exclusive aqueous eluent concentrate (Agilent, UK) formulated for use with a PL-PSDA type HDC column, which was diluted to the manufacturer’s specification with MiliQ water. Which consists of sodium dihydrogen phosphate (monohydrate) was obtained from Merck, Darmstadt, Germany. Sodium dodecyl sulphate, Brij® 35 non-ionic surfactant (30% w/v solution) and sodium azide were obtained from Sigma-Aldrich (Darmstadt, Germany).The 3000 series Nanosphere polystyrene nanoparticle standards were obtained from ThermoFisher Scientific (Bremen, Germany). The used particle diameters were: 900nm, 700 nm, 500nm, 220nm, 100nm, 50nm and 30 nm. The PMMA nanoparticle samples of 76nm and 59nm were kindly provided by DSM Coating Resins (DSM Waalwijk, The Netherlands).

3.2 Equipment

The method development and analysis in this study were carried out on an Agilent 1290 Infinity 2D-LC setup (Agilent, Waldbronn, Germany). The system utilized two Agilent 1290 Infinity Binary Pumps (Model G4220A), each equipped with a Agilent Jet Weaver V35 or V100 mixer (Model G4220-60006), one 1200 isocratic pump (Model G1310A), two Agilent 1290 Infinity Thermostatted Column Compartments (Model G1316C), of which each compartment was equipped with an Agilent 2-position 8-port valve (Model G4236A). The system also featured an Agilent 1290 Infinity Autosampler (G4226A) and two Agilent 1290 Infinity Diode Array Detectors (Model G4212A) with Agilent Max-Light Cartridge Cell 10 mm (Model G4212-6008, 10 mm, V(o) 1.0ul) flow cells for the first and second dimension respectively. All tubing connections were stainless steel. The system was controlled by Agilent OpenLAB CDS Chemstation Edition (Rev. C.01.04).

The HDC analysis was carried out on an Agilent PL-PSDA cartridge type-2 (800×7.5mm, 12µm) column. In the beginning of this study an Agilent PL-PSDA cartridge type-1 (800×7.5mm, 8µm) was used. For the second dimension, SEC analysis, three experimental were coupled in series. A core-shell (150×4.6mm, 3.6µm, 500Å and 328 Å) columns obtained from Phenomenex were coupled together with

(13)

- 12 - one core-shell column which was packed with XB-C18 modified silica particles (150×4.6mm, 3.6µm, 500Å). The modified column was used in front of the series. For the investigation of water on SEC the next column was used, a 150×4.6mm, 2.6µm, 98Å packed with XB-C18 modified silica particles. Additional parts were used for LC×LC modulation valve when traps were used. The traps were obtained from Phenomenex, SecurityGuard ULTRA V0=0.4μL) guard column holders were used in conjunction

with UHPLC C18 2.1mm i.d. SecurityGuard ULTRA cartridges. The traps were directly coupled to the valve. One end of the coupled traps was connected to the valve using a male-to-female tubing piece (Agilent, 70×0.12mm, m/f), whereas standard tubing was used to couple the other end of the valve (Agilent, 90×0.12mm).

3.3 Sample preparations

For SEC analysis a calibration curve is obtained by dissolving PS standards (Appendix B) in THF at concentrations of approximately 0.2 mg⋅mL−1 spiked with 400 ppm toluene as a marker. For the investigation of water influences on SEC a calibration curve was obtained by dissolving several PS standards in THF with 10% of water. For HDC analysis a stock solution of 1.0L buffer was prepared by dissolving 6.2g sodium dihydrogen orthophosphate, 10.0g sodium lauryl sulphate, 134mL Brij 35 non-ionic surfactant (30% w/v solution) and 4.0g sodium azide in 866.7mL MilliQ purified water. The stock solution was diluted 20 times with MilliQ purified water for use. Concentrations of the obtained PS and PMMA nanoparticles were reported as % solids. PS nanoparticles were diluted with HDC buffer 10x to a concentration of approximately 0.1% solids. PMMA nanoparticles were diluted with HDC buffer to concentrations between 0.1 and 0.5% solids.

3.4 Analytical procedure

Figure 4 displays the instrumental setup used for the LC×LC experiments in general. The parts that are coloured in light grey are the setup used for one-dimensional experiments. The first valve shown in figure 4 was by-passed to re-route flow to the waste during the flush program of the HDC column

.

For one-dimensional HDC, system A (light grey) was used in which one channel of the binary pump was connected to autosampler, which was coupled to the HDC or SEC column depending on the type of experiment and then linked to the DAD detector. For HDC experiments, the flow rate was set a 1 mL⋅min-1_{with a recorded pressure of ~80 bar using the HDC buffer as described in section 3.3 as the}

mobile phase. The analysis time was set at 15 minutes. The injection volume was 20 µl. For the

(14)

- 13 - investigations to the effect of a flush program in the HDC separation the next flow rate program was used; 1000 µL ⋅min-1_{from t=0.0 minutes to t=11.9 minutes, 40 µL⋅min}-1 _{from t=11.9 minutes to t=60.0}

minutes.

The same setup was also used for one-dimensional SEC analysis to generate a calibration curve with the three columns combinations and one trap as described in section 3.2 ( for a more realistic comparison). The flow rate was set at 3.0 mL.min-1_{with recorded pressure of 750-800 bar. For the mobile phase,}

100% non-stabilized THF was used, the analysis time was set at 5 minutes with an injection volume of 20 µl. For the investigation of the effect of water on SEC analysis the flow rate was set at 1mL⋅min-1

with an analysis time of 1.5 minutes and injection volume of 5 µl.

For studying the instrumental parameters for the performances in a LC×LC system such as dissolution process, traps and modulation, system B was used. The setup illustrated in figure 1 was used without the HDC column in order to investigate the SEC separation. The HDC-eluent from pump A was combined with THF (pump B) through the use of a T-junction. The solvent blend from the T-junction was then mixed by an Agilent Jet Weaver mixer in which the nanoparticles will be dissolved in the THF and to mix the two mobile phase solvents homogeneously together. For these experiments, the flow rates of pump A and B were varied according to table 1 and the second-dimension flow rate was set constantly at 3.3 mL⋅min-1 _{with a modulation time of 36 seconds. The injection volume was 20 µl. These}

experiments were performed using a 500 nm PS nanoparticle.

Table 1: varied flow rates used for pump a and pump b

Buffer THF Total ϕ(buffer) ϕ(THF) μL∙min-1 _μL∙min-1 _μL∙min-1

40 100 140 29% 71% 40 120 160 25% 75% 40 140 180 22% 78% 40 160 200 20% 80% 40 200 240 17% 83% 40 300 340 12% 88% 40 400 440 9% 91% 60 460 520 12% 88%

For LC×LC experiments the entire setup as illustrated in figure 4. Pump A is set at a flow rate of 40 µL⋅min-1 _{or with a flush programme as described before. The making-up flow (pump B) was set at a}

flow rate of 200 µL⋅min-1_{unless stated differently. The second-dimension pump C was set at flow rate}

of 3.3 µL⋅min-1_{. The total analysis time was 60 minutes and modulation time was 36 seconds. The}

recorded back pressure during the analysis ranged between 800-900 bars with one trap and with two traps the pressure ranged between 900-1000 bars.

(15)

- 14 -

4. Results and discussion

4.1 Hydrodynamic chromatography

In the beginning of this study, a 1 HDC column was used and due to technical difficulties, a type-2 HDC column was used to continue for experiments. However, it is interesting to investigate the differences betweens these columns. In this is shown figure 5 by the comparison of mixtures consisting of several nanoparticles standards analysed by type-1 and type-2 columns. Clearly, it is noticeable that with 500 nm particle size couldn’t be analysed on a type-1 column because of the smaller particle size of the packing compared to the type-2 column. Furthermore, the separation is better with a type 1 column as described by Small et al., as the particle size of the packing is reduced the separation factors increases resulting in an improved resolution of different particle sizes as the size of the packing is reduced. However, due to technical problems, the experiments in this study were continued with type 2 column.

Figure 6: Chromatogram of HDC separation of a mixture of PS nanoparticles (900 nm, 500 nm, 100 nm and 50 nm) by HDC Figure 5: Mixture of several nanoparticles standards analysed with two different HDC columns

(16)

- 15 - The HDC domain exploits a narrow domain between 0.8 < τ < 1.0, with τ=1.0 defined as the dead volume. This can be seen from figure 6 where the chromatogram of mixture separation is shown recorded with a flow rate of 1mL⋅min-1_{. The chromatogram shows that the first 12 minutes, which}

corresponds to an elution volume of 12 mL, are not helpful. Especially for LC×LC experiments when the flow rate is set very low and the outlook of waiting for this volume to elute is not practical. As explained in section 2.2.2 it is supported by the theory of van Deemter curve that there is no upper limit to the flow velocity at which HDC separation may be conducted and does not heavily depend on the flow rate. Therefore, the flow rate can be increased unpunished to flush through the first 11.9 mL of column volume as described in section 3.4 in the analytical procedure.

The output of the flush-program is plotted in figure 7 overlayed with the same chromatogram as in figure 6. Only the relevant section of figure 6 which was performed at a high flow rate of 1000 µL⋅min-1

compared with flush and flow rate of 40 µL⋅min-1_{of the same mixture. Comparing the elution volumes}

of both peaks it can be said that there are no significant differences. In order to investigate the performance and the reliability of the pump operating the flush programme a 900 nm particle size was repeated five times. Resulting in a standard deviation of 0.11 minutes at 40 μL·min-1 or 4.4 μL in terms of elution volume was found and it was concluded that the deviation was not significant (see Appendix C).

4.2 Influence of aqueous HDC buffer on SEC analysis

One of the challenges in this study is the solvent incompatibility. In HDC an aqueous mobile phase is mostly used which is not compatible with the organic mobile phase in SEC analysis. Any remaining aqueous fraction may cause effects on the SEC analysis by potentially producing interaction in SEC mechanism. Earlier research on the influences of SEC analysis has shown that at a percentage higher than 10% of aqueous buffer the SEC analysis is affected due to adsorption (see appendix D). It is also visible that for the higher water concentration the polymer peak has broadened and the peak for the dead time and water peak start to shift towards each other. Therefore, it is interesting to investigate the effect of the HDC buffer on the SEC calibration curve both in one-dimensional SEC analysis as well as in the two-dimensional analysis

.

Figure 7: Overlay of chromatogram section obtained using System A at 1 mL⋅min-1 and flush program ffollowed with the flowrate set at 40 µL⋅min-1 following

(17)

- 16 -

Figure 8: Calibration curves obtained with one-dimensional setup and two-dimensional setup analysed in ideal situation and with 10% HDC aqueous buffer

As shown in figure 8 the HDC buffer has non to little effect on the calibration curve of the SEC analysis. The same can be said for the calibration curve obtained in the two-dimensional analysis. It can also be seen that the gap between one-dimensional curves and two-dimensional curves is due to variations in the pump efficiency and effects related to the mobile-phase compressibility. As described by Pirok B.W.J. et al, the effect is clear at higher flow rates, where the effective flow rate is seen to decrease with increasing set flow rate29_{. Another effect that can occur under high-stress conditions in the}

chromatographic column is that polymer chains can be deformed in a stressed shape instead of colloidal spheres resulting in a strong shift in the exclusion limit as it can be seen for the red curve, this phenomenon is known to be slalom chromatography.

4.3 Comprehensive two-dimensional liquid chromatography

4.3.1 Effect of the traps, mixtures and columns

The HDC and SEC separations were coupled according to the setup shown in figure 4. The combination of the HDC and SEC separation required solving a set of challenges as described in section 2.4. In order to investigate the effect of all different parameters, such as the dissolution solvent composition, mixing volumes and trapping performance on the SEC separation, the same modulation domain was regularly selected (from t = 1.7 minutes to 2.15 minutes) and transformed to the τ-scale for

comparison as shown in figure 9 of the 500 nm PS nanoparticle.

(18)

- 17 - The chromatogram in figure 10A (blue line) is

the result of the LC×LC setup with loops. It can be clearly seen that relative to the dead volume signal (at τ=1), the signal of the polymer fraction was strongly reduced. This can be explained that the presence of water in the polymer solution is still complicated the SEC separation due to adsorption effects. To solve this dilution and adsorption problem the loops are replaced by traps. The effect of the replacement of the loops is shown in figure10A (orange line). The traps appeared to retain the dissolved polymers very well. The traps were replaced with very short connections towards the valve to minimize extraneous volumes. Due to short connections and small volumes, most of the aqueous buffer could be reduced, while the polymers were retained on the traps.

The Agilent Jet Weaver mixer has two features, a mixing volume of 100 µl and 35 µl. In this case, the first dimension aqueous buffer is mixed with THF to enhance the trapping operation. To study the effect of the mixing volume both features are used. The result is shown in the figure 10B where it is clearly seen that with a 100 µl volume mixture the yield is high and was used for all following experiments. However, more research is needed for the trap yield and the effect of mixing ratios on the trap.

4.3.2 SEC performance

As described in section 3.4 the second-dimension consist of three coupled SEC columns in series in order to obtain as much resolution as possible and no compounds elute in the exclusion volume which allows the use of overlapping injections. The first column was a non-modified silica column which was used to filter out any remaining HDC aqueous buffer from the fractions eluting from the first-dimension. The second and third- column was a C18 column with the same pore size. However, to gain more resolution in the low-molecular weight particles the possibility to replace the last column with a core shell column with smaller pore sizes (328 Å). This possibility was investigated by analysing 500 nm nanoparticle with both columns for comparison as shown in figure 11.

Figure 10: Separation of 500-nm PS nanoparticles analysed with LC × LC setup (see analytical procedure) A: No traps (blue line) and with traps (orange line). B: 35µL mixer (orange line) and 100 µL (blue line)

(19)

- 18 -

Figure 11: Effect of the pore size in the last of three SEC columns combination; 328-Å orange line) vs. 500-Å (blue line).

As shown in figure11 the orange line corresponds to SEC separation with the use of smaller pore size column. It can be seen that the chromatogram with a smaller pore size is not as smooth as the blue chromatogram corresponding to a 500 Å pore size column. This the results of a pore-size mismatch when using 328 Å column in combination with other two columns compared with the 500- Å column.

4.3.3 Effect of the mixing ratio

Another parameter to investigate is the mixing ratio of the making-up flow from pump B as described in table 1 in section 3.4. All possible flow rate as described in the table are performed and the obtained SEC separations are compared with each other. This is shown in figure 12 where all the chromatograms are overlaid. It can be clearly seen that the effect of the mixing ration causes some distortion in the SEC separation resulting into irregular peaks.

Figure 12 Overlaid SEC chromatograms of 500-nm PS nanoparticles obtained using different mixing ratios for dissolution according to table 1 (see section 3.4)

(20)

- 19 - To evaluate this distortion the use of the second-

derivative is applied. In order to avoid the noise signals, a Savitsky-Golay smoothening filter is applied to all chromatograms before the second derivate was calculated. The Savitzky-Golay (SG) smoothing method is known as a time-domain method. It is based on a moving polynomial fit which computes both the smoothed profile (fitted points) and the derivatives in a single step. Each inflection point in the second derivative curve corresponds to a concavity or irregularity in the SEC chromatogram as illustrated in figure13.

The amount of distortion is minimal for the ratio of 17% (HDC buffer/THF [17:83, v/v]) as well as for the ratio of 12% (HDC buffer/THF [12:88, v/v]).

However, it was mentioned before that more research was needed for the trap yield. Because an increase in THF under high-stress conditions in the chromatographic column causes an increase in the total flow rate which may increase the risk of polymer eluting from the traps. In order to calculate the trap yield, the chromatogram curve is integrated and the area was defined as the trap yield as shown in Table 2

Table 2 the trap yield at different percentage of mixing ratio ϕ(buffer) % Trap Yield

Area∙min 29% 11,849 25% 12,92 22% 13,495 20% 12,728 17% 13,949 12% 13,26 9% 13,346 12% 10,572

It can be seen that high amount of THF, which means lower percentages of the buffer, leads to good dissolution of the particle and high trap yield compared to lower amounts of THF. However, more THF will lead to better-dissolved polymers, but too much will render the traps useless due to a lack of retention, whereas too little will also endanger the traps by resulting in clogging of the traps. Compositions with more than 22% aqueous buffer resulted in occasional trap clogging. Above 30% aqueous buffer no signal was observed in the SEC dimension and almost every experiment resulted in clogging of the traps. The result in the table is that there is an optimum at 17% of buffer and 83% THF and this is also reflected by the consistency of the distribution curve and thus, appears to yield favourable results. This composition was used for following experiments in this study.

Figure 13 Second derivatives of chromatograms of all obtained mixing ratios for dissolution.

(21)

- 20 -

4.4 LC×LC analysis of PS and PMMA nanoparticles

A mixture of 500-, and 100-nm PS nanoparticles, all at concentrations of 0.5% (w/w) (see section 3.3 for sample preparation) was injected and analysed using the developed separation system with loops in the modulator. The resulting LC×LC (HDC×SEC) chromatogram is shown in figure 14. The corresponding information related to the particle-size distribution and the molecular weight distribution was tentatively added on the top and right axes, respectively.

As a first attempt, it can be seen that particles are dissolved and therefore separated in both dimensions. However, the signal intensity of the polymer is low compared to t0 signal. In this case, the polymer

signal is spread over 40 modulations. The low-intensity signal can be caused by the first column of silica, in which the water simple keeps sticking to the silica stationary phase and the signal gets slower as the modulation progresses. In this particular experiment, the volume added from pump B was 100 µl with 40 µl of HDC buffer from pump A. Increasing the volume to 150 µl of THF in a loop of 160 µl this could be a problem. Because the flow in the loop tends to have parabolic flow profile and therefore loop will be too small and therefore increasing the loop volume is an option. Another option could be to reduce the mixing ration to 100:20 µl (THF/HDC fraction). In combination with a higher flow rate in the first dimension to simultaneously increase the absolute concentration of polymer in the loops led to adsorption effect. Due these challenges the loops were replaced with a trap in order to avoid dilution problems and gain high signal intensity this resulted in the chromatogram shown in figure15. It clearly can be seen that the signal intensity increased. However, 100 nm particle has ridge distortions pattern which might be caused by asymmetrical trap connections and different pump pressure for each modulation

.

Figure 14 HDC×SEC chromatogram of a mixture of nanoparticles containing PS 500 nm and PS 100 nm recored from the 2D-setup without traps

(22)

- 21 -

Another mixture was analysed consist of 903-, 498- and 216-nm PS and 76- and 59-nm PMMA nanoparticles, all at concentrations of 0.1-0.5% (w/w). The mixture was analysed according to the final LC×LC setup described in section 3.4. The resulting chromatogram is shown in figure16. The calibration curve from the HDC separation was used to tentatively assign an indication of particle size to the top x-axis. It can be seen that the PS particles are well separated from each other and also from PMMA particles. However, the two PMMA particles are not good enough separated from each other. As explained in section 4.1 the particle size of the column has an effect on the resolution as well as the separation of the particles. So one solution to increase the resolution is the use an HDC column with a covering a range of smaller particle sizes. Also, it can be seen that the results of the SEC separation give a good insight into the differences in molecular weight distributions of the polymers. However, there are still some remaining challenges left to improve this chromatogram which will be described in the next section.

Figure 15 HDC×SEC chromatogram of a mixture of nanoparticles containing PS 500 nm and PS 100 nm recored from the 2D-setup with traps

Figure 16 15 HDC×SEC chromatogram of a mixture of nanoparticles containing several polystyrene standards and polyacrylate samples.

(23)

- 22 -

4.5 Remaining challenges and recommendations

As can be seen from the chromatogram in figure 15 there are some distortions due to the trap performance, which increases during the analysis time. Moreover, poorly dissolved nanoparticles can cause blockage of the traps and increase in back pressure. These issues might affect the trapping performance and most importantly the SEC separation. This depicts a weakness in the HDCxSEC system. In order to avoid distortions in trap performance, it is important that both traps are symmetrical as possible. The problem of the increased pressure can be solved by regenerating the traps by ultra-sonication. Another problem that occurred in this study that is related to experimental setup is the possible return of THF through the T-junction to the outlet of the first dimension column. Analysis has shown that the column has been dissolved in a hostile solvent. THF is obvious because the HDC buffer is a good solvent. But, the damage also occurred in the first of the two columns of the HDC column, meaning that the THF should be more than 7 mL that entered the column. A possible solution is the use of a check valve between the end of the first dimension column and the T-junction, which makes the flow only one way up. Another issue related to the experimental setup is the extraneous volumes. In the current setup, the first-dimension column is connected by a long piece of capillary tubing to the 100-L Jet Weaver V100 mixer and the exit of the mixer is again connected by a long piece of tubing back to the modulation valve. Ideally, the first-dimension column would be directly connected to an extremely low-volume mixer inside the column oven. In the present set-up, we observed significant peak tailing in the SEC dimension under UHPLC conditions.

Another discussion point is that it can be clearly seen in the chromatogram of figure 15 is the broad peak width in HDC, which causes some limitation in the HDC separation. the observed band broadening in the HDC separation significantly reduces the reliability of the data as presented in Figure 5 in terms of particle distribution widths. This can be solved be using HDC columns with smaller particle sizes or mathematically reducing the peak width. The latter is described by an earlier study by McGowan, G. R et al., in which an established data processing methods are used to calculate out the band broadening33

and will be elaborated in next section.

Ultimately, we aim to combine the HDC separation with other retention mechanisms, such as gradient-elution reversed-phase or normal-phase LC or ion-exchange chromatography, to characterise nanoparticles consisting of copolymers or charged polymers. We envisage that, depending on the type of application, different stationary phases can be used to retain the analytes.

4.5.1 Algorithm for band broadening reduction for HDC separation

McGowan, G. R et al described a general model, derived from the Pearson family of distribution (type VII), which was modified for the reduction of the band broadening in (one-dimensional) HDC analysis. The Pearson family of distribution which was designed by Pearson represents a system whereby every member the probability density function f (x) is composed of twelve families of distributions, all of which are solutions to the differential equation33_{.The algorithm of the model obtained from The Pearson}

type VII was modified by adding several parameters which result in the next equation:

𝑌𝑖 = [(1 + (𝑋𝑖− 𝑋0)2/𝑀(𝜎𝑛 + 𝐸(𝑋𝑖− 𝑋0))2]−𝑀+ 𝑉 ∗ 𝐴(1 + 𝐿2/𝑀𝐾2)−𝑀 (10)

Yi = height of peak at Xi

X0 = location of peak center

A = height of peak at X0 (maximum peak height)

M = Shape factor E = Asymmetry factor

σn = nominal sigma of peak (width) SF=snouting factor

J= integer of SF V=fraction of SF L= σ * (J/5+2) + Xi - X0

(24)

- 23 - In equation 10 the added parameters are the asymmetry factor, which corresponds to a linear change in the sigma of the peak as a function of direction and distance from the peak center, the snouting factor describes the foretailing on the peak and the shape factor changes the basic peak shape from modified Lorentzian to a Gaussian peak as M varies from 1 to ∞. These variables describing the shape of the peak. Most observed HDC peak shapes are being fitted very accurately through the modified Pearson type VII distribution. However, more research is needed in order to apply this method for LC×LC results.

(25)

- 24 -

5. Conclusion

The aim of this project is to comprehensively combine HDC and SEC together by overcoming the challenges such as solvent incompatibility, sample transformation and dilution problems. It can be concluded that an online separation system was developed whereby hydrodynamic chromatography (HDC ), to obtain the particle size distribution and size exclusion chromatography (SEC) to obtain molecular weight information, were comprehensively combined using active modulation and intermediate sample transformation. In the first dimension, the first 12 mL of the total 14 mL of the column was flushed with high flow rate to reduce analysis time. The remaining 2 mL which corresponds to the HDC separation was fractioned after dissolving with THF by using the mixer of 100 µl. The mixing ratio was set at 17% first-dimension aqueous buffer eluent/ 83% THF (v/v). This mixing ratio showed good yields of polymers and better trapping efficiency. To prevent any adsorption effect in the SEC separation, stationary-phase assisted modulation was applied whereby the loops were replaced by traps. For the second dimension, three core-shell columns were coupled and used at the high-pressure condition to obtain as much as possible resolution, reduce modulation times and apply subsequent injections. The developed system was demonstrated through a separation of polystyrene and polyacrylate nanoparticles and gives a good insight into the differences in molecular weight distributions of the polymers

.

(26)

- 25 -

References

1. Mallakpour, S. & Behranvand, V. Polymeric nanoparticles: Recent development in synthesis and application. Express Polym. Lett. 10, 895–913 (2016).

2. Rao, J. P. & Geckeler, K. E. Polymer nanoparticles: Preparation techniques and size-control parameters. Prog. Polym. Sci. 36, 887–913 (2011).

3. Lebouille, J. G. J. L., Stepanyan, R., Slot, J. J. M., Cohen Stuart, M. A. & Tuinier, R. Nanoprecipitation of polymers in a bad solvent. Colloids Surfaces A Physicochem. Eng. Asp.

460, 225–235 (2013).

4. Akagi, T., Kaneko, T., Kida, T. & Akashi, M. Preparation and characterization of biodegradable nanoparticles based on poly(??-glutamic acid) with L-phenylalanine as a protein carrier. J.

Control. Release 108, 226–236 (2005).

5. Anton, N., Benoit, J. P. & Saulnier, P. Design and production of nanoparticles formulated from nano-emulsion templates-A review. J. Control. Release 128, 185–199 (2008).

6. Allémann, E., Gurny, R. & Doelker, E. Preparation of aqueous polymeric nanodispersions by a reversible salting-out process: influence of process parameters on particle size. Int. J. Pharm. 87, 247–253 (1992).

7. Shariati, A. & Peters, C. J. Recent developments in particle design using supercritical fluids.

Curr. Opin. Solid State Mater. Sci. 7, 371–383 (2003).

8. Colombo, C. et al. PEGylated Nanoparticles Obtained through Emulsion Polymerization as Paclitaxel Carriers. Mol. Pharm. 13, 40–46 (2016).

9. Thickett, S. C. & Gilbert, R. G. Emulsion polymerization: State of the art in kinetics and mechanisms. Polymer (Guildf). 48, 6965–6991 (2007).

10. Landfester, K. Synthesis of Colloidal Particles in Miniemulsions. Annu. Rev. Mater. Res. 36, 231–279 (2006).

11. Mac??as, E. R. et al. Microemulsion polymerization of methyl methacrylate with the functional monomer N-methylolacrylamide. Colloids Surfaces A Physicochem. Eng. Asp. 103, 119–126 (1995).

12. Karlsson, L. S., Deppert, K. & Malm, J. O. Size determination of Au aerosol nanoparticles by off-line TEM/STEM observations. J. Nanoparticle Res. 8, 971–980 (2006).

13. Pabisch, S., Feichtenschlager, B., Kickelbick, G. & Peterlik, H. Effect of interparticle interactions on size determination of zirconia and silica based systems - A comparison of SAXS, DLS, BET, XRD and TEM. Chem. Phys. Lett. 521, 91–97 (2012).

14. Vippola, M., Valkonen, M., Sarlin, E., Honkanen, M. & Huttunen, H. Insight to Nanoparticle Size Analysis—Novel and Convenient Image Analysis Method Versus Conventional Techniques. Nanoscale Res. Lett. 11, 169 (2016).

15. Mori, Y. Size-selective separation techniques for nanoparticles in liquid. Bull. l’Institut Sci. Sect.

Sci. la Terre 32, 102–114 (2015).

16. Tiede, K. et al. Application of hydrodynamic chromatography-ICP-MS to investigate the fate of silver nanoparticles in activated sludge. J. Anal. At. Spectrom. 25, 1149 (2010).

17. Pergantis, S. A., Jones-Lepp, T. L. & Heithmar, E. M. Hydrodynamic chromatography online with single particle-inductively coupled plasma mass spectrometry for ultratrace detection of metal-containing nanoparticles. Anal. Chem. 84, 6454–6462 (2012).

(27)

- 26 - 18. Gray, E. P. et al. Analysis of gold nanoparticle mixtures: a comparison of hydrodynamic chromatography (HDC) and asymmetrical flow field-flow fractionation (AF4) coupled to ICP-MS. J. Anal. At. Spectrom. 27, 1532 (2012).

19. Schure, M. R. Limit of detection, dilution factors, and technique compatibility in multidimensional separations utilizing chromatography, capillary electrophoresis, and field-flow fractionation. Anal. Chem. 71, 1645–1657 (1999).

20. DiMarzio, E. A. & Guttman, C. M. Separation by Flow. Macromolecules 3, 131–146 (1970). 21. Small, H., Saunders, F. L. & Solc, J. Hydrodynamic chromatography A new approach to particle

size analysis. Adv. Colloid Interface Sci. 6, 237–266 (1976).

22. Peyrin, E. et al. Flow Rate Dependence on the Biopolymer Retention in Hydrodynamic Chromatography . Comparison Between the Behaviors of Proteins and Plasmids. 24, 1245–1252 (2001).

23. Dias, R. P. Size fractionation by slalom chromatography and hydrodynamic chromatography.

Recent Patents Eng. 2, 95–103 (2008).

24. Striegel, A. M. & Brewer, A. K. Hydrodynamic Chromatography. Annu. Rev. Anal. Chem. 5, 15–34 (2012).

25. Stegeman, G., Oostervink, R., Kraak, J. C., Poppe, H. & Unger, K. K. Hydrodynamic chromatography of macromolecules on small spherical non-porous silica particles. J.

Chromatogr. A 506, 547–561 (1990).

26. Small, I. & The, R. Hydrodynamic Chromatography--An Evaluation of Several Features. 79, 264–267 (1981).

27. Kostanski, L. K., Keller, D. M. & Hamielec, A. E. Size-exclusion chromatography - A review of calibration methodologies. J. Biochem. Biophys. Methods 58, 159–186 (2004).

28. Schure, M. R. & Moran, R. E. Size exclusion chromatography with superficially porous particles.

J. Chromatogr. A 1480, 11–19 (2017).

29. Pirok, B. W. J. et al. Size-exclusion chromatography using core-shell particles. J. Chromatogr.

A (2016). doi:10.1016/j.chroma.2016.12.015

30. Guiochon, G., Marchetti, N., Mriziq, K. & Shalliker, R. A. Implementations of two-dimensional liquid chromatography. J. Chromatogr. A 1189, 109–168 (2008).

31. Groskreutz, S. R., Swenson, M. M., Secor, L. B. & Stoll, D. R. Selective comprehensive multi-dimensional separation for resolution enhancement in high performance liquid chromatography. Part I: Principles and instrumentation. J. Chromatogr. A 1228, 31–40 (2012).

32. Vonk, R. J. et al. Comprehensive two-dimensional liquid chromatography with stationary-phase-assisted modulation coupled to high-resolution mass spectrometry applied to proteome analysis of saccharomyces cerevisiae. Anal. Chem. 87, 5387–5394 (2015).

33. McGowan, R.Gerald, Langhorts, M. Development and Application of an Integrated. 89, 69–74 (1997).

(28)

- 27 -

Appendix

A. HDC equations

The local velocity (µ) for a position r in a flow capillary of radius R is given by µ = 2ū[1 − (r

R)] 2

(1) For particles which quickly exchange positions between all the streamlines the average velocity is given by

ū = 2

R2∫ µ(r)dr

R

0 (2) The particle that exit the capillary of length L after a residence time tm, is given by

tm =L

ū (3)

For larger particles that are in the range of R-Ø ( see figure 1B), the average velocity ūp is given by

ūp = 2

(R−Ø)2∫ µ(r)dr

R−Ø

0 (4)

and the average residence time tp in the capillary is given by

tp = L

ūp (5)

By combining the ratio of the effective analyte radius to the capillary radius, equation 4 and equation 1

the next equations can be obtained

ūp = ū (1 + 2λ − λ2) (6) From equation 6 the residence time τ of the particle in the capillary can be described as following

τ =

tp

tm

=

1

(29)

- 28 -

B. PS standards

Sample PDS Mw log Mw tol - 92,14 1,964 PS01 1,11 580 2,763 PS02 1,1 980 2,991 PS03 1,05 1990 3,299 PS04 1,05 2100 3,322 PS05 1,04 2970 3,473 PS06 1,03 4920 3,692 PS07 1,03 6930 3,841 PS08 1,03 7000 3,845 PS09 1,05 9920 3,997 PS10 1,02 13880 4,142 PS11 1,02 19880 4,298 PS12 1,02 52400 4,719 PS13 1,03 70950 4,851 PS14 1,03 96000 4,982 PS15 1,03 126700 5,103 PS16 1,02 299400 5,476 PS18 1,02 52300 4,719 PS19 1,02 735500 5,867 PS20 1,03 1112000 6,046 PS21 1,04 1373000 6,138 PS22 1,05 2061000 6,314 PS23 1,03 2536000 6,404 PS24 1,03 3053000 6,485 PS25 1,04 3742000 6,573 PS26 1,07 7450000 6,872 PS27 1,13 13200000 7,121

(30)

- 29 -

C. Performance of the flush program

Figure 17: The performance and the reliability of the pump operating the flush programme by analysing 900 nm particle size which was injected five times.

Table 3 Results of the 900 nm particle (repeated 5 times)

Repetition Retention time (min) 1 20,07 2 20,19 3 20,26 4 20,40 5 20,08 Average 20.20 Std % 0.11

Comprehensive two-dimensional hydrodynamic chromatography × size-exclusion chromatography with intermediate dissolution for the characterization of polymeric nanoparticles

MSc Analytical Sciences