Miniaturization in asymmetrical flow field-flow fractionation

(1)

1

Miniaturization in asymmetrical flow

field-flow fractionation

Thesis Bachelor project Chemistry

Liz Leenders

Amsterdam 09-06-2014 University of Amsterdam Analytical Chemistry Group Supervisors: Dr. W. Th. Kok

(2)

2

Samenvatting

Asymmetrische flow field-flow fractionatie (asymmetric flow field-flow fractionation, AF4) is een scheidingstechniek die veel gebruikt wordt in industriële, farmaceutische en biomedische doeleindes. Deze scheidingstechniek maakt gebruik van het verschil in diffusie coëfficiënt van de te scheiden proteïnes. Miniaturisatie van AF4 is interessant, omdat het op deze manier, door zijn kleine formaat, als testmethode buiten het ziekenhuis gebruikt kan worden voor analyse van macromoleculen in lichaamsstoffen. In dit project zijn één commercieel apparaat en enkele miniaturen van verschillende vorm getest en geoptimaliseerd. Verschillende hoogtes van scheidingskanalen zijn gebruikt, hoogtes van 190μm, 250μm, 350μm en 480μm voor het commerciële apparaat en 100μm, 200μm, 300μm, 400μm en 500μm voor de miniaturen. In de eerste stap van het project werd het commerciële apparaat geoptimaliseerd, hieruit bleek dat grotere (350μm/480μm) scheidingskanalen beter in staat zijn om een mengsel van proteïnes (BSA, apoferritin en thyroglobulin) te scheiden dan kleinere kanalen (250μm/190μm). Vervolgens is een miniatuur in dezelfde vorm als het commerciële apparaat getest en geoptimaliseerd. Ook hieruit bleek dat grotere kanalen (300μm/400μm/500μm) beter werken dan kleinere (200μm/100μm). Het kanaal met grootte 400μm gaf de beste resultaten. Het scheiden van de drie proteïnes met een miniatuur van deze vorm (Vorm 1) was mogelijk, maar er zijn een aantal instrumentele problemen die opgelost moeten worden om de miniaturen optimaal te gebruiken. De kleinere miniaturen (vorm 2 en 3) zijn niet in staat om het mengsel van proteïnes te scheiden. Analyse met Blue Dextran wees vervolgens uit dat het kanaal in de miniaturen niet goed op het membraan wordt gedrukt, wat leidde tot het lekken van het mengsel uit het kanaal in het apparaat. Hierdoor kan het mengsel niet geanalyseerd worden met UV/Vis. Verder onderzoek naar miniaturisatie van AF4 is dus nodig.

Abstract

Asymmetrical flow field-flow fractionation (AF4) is a promising separation mechanism that is used in industrial, pharmaceutical and biomedical applications. In AF4 the separation depends on difference in diffusion coefficients of the separated proteins. Over the past few years miniaturization of AF4 has been of great interest, because a miniaturized AF4 device could become a point of care diagnostic device (POCDD) for analysis of macromolecules in body fluids. During this project one commercial- and several miniaturized AF4 channels of different shapes were tested and optimized for the separation of proteins. Several heights of the fractionation channels were tested namely 190μm, 250μm, 350μm and 480μm for the commercial channel and 100μm, 200μm, 300μm, 400μm and 500μm for the miniaturized channels. First step of this project was the optimization of the commercially available channel. High spacer heights (350μm/480μm) provided better separation of the mixture of proteins (BSA, apoferritin and thyroglobulin) than small spacer heights (250μm/190μm). Spacer height 350μm gave the best results. After this optimization, a miniaturized channel with the same shape as the commercial channel was tested. Also in this case higher channel heights (300μm/400μm/500μm) were better able to separate the proteins than smaller ones (200μm/100μm). Channel height 400μm gave the best results. Last step in this project was testing three different shapes of miniaturized channels. Separation of the three proteins with the largest miniaturized channel (shape 1) was possible, however there are some instrumental issues that need to be solved to utilize the full potential of the miniaturized channels. The smaller miniatures (shape 2 and 3) were not able to separate the mixture of three proteins. Analysis with Blue Dextran showed that the channel was not pressed towards the membrane properly, and therefore part of the mixture leaked out of the channel into the device. In this way the mixture could not be analyzed properly by UV/Vis. Further research on miniaturization of AF4 is needed.

(3)

3

1. Introduction

Field-flow fractionation (FFF) is a collection of separation techniques suitable for macromolecules or analytes in the range of nano- and micrometers. FFF has some similarities with liquid chromatography (LC), but the difference between FFF and LC is in the separation mechanism for elements transported by the liquid-flow. In LC separation occurs via eluting species with the stationary phase of the column. In FFF the separation of sample components depends on the interaction with an externally applied force, which is perpendicular to the axial flow in a channel. The profile of an axial flow can be identified as a parabola with the highest flow velocity in the middle and a lower velocity on the channel walls. Because of molecular diffusion, the sample components take different positions across the parabolic flow-profile under influence of an external force and field. This results in elution with different velocities, leading to separation of the sample components.

Various types of external fields have been used in FFF, such as electrical fields (electrical FFF)1, temperature gradients (thermal FFF)2 and cross-flow (flow FFF)3. In all subclasses of FFF, the channel does not contain a stationary phase. Various solvents can be used as a mobile phase, including water and buffers. The physiochemical properties and applications of various FFF subtypes are listed in table 1.4 The focus in this thesis is on asymmetrical flow FFF and its strength in characterizing macromolecules, such as proteins or polymers, by difference in diffusion coefficients and size.

Table 1: Physiochemical properties and applications for various field-flow fractionation systems

FFF subtype Physiochemical properties Applications

Electrical Size, electrophoretic mobility Cells and organelles, bacteria and viral separations, characterization of emulsions, liposomes, protein adsorption

Thermal Size, thermal diffusion

coefficient

Separation of dissolved and suspended polymers, polymer and silica nanoparticle analysis Cyclical electrical Electrophoretic

mobility

Biopolymer separations and zeta potential measurements

Dielectrophoresis Dielectric permittivity, size

Cell separation and dielectric property measurements and cancer cell separation

(Asymmetrical) flow Diffusion, size Proteins, DNA, polymers, cells, micro and nanoparticles

Flow field-flow fractionation (FlFFF) was discovered by J.C. Giddings in late 1970’s3 and has become the most often used subtype of FFF. In FlFFF the field applied to the channel is a second flow which pushes the flowing components to a membrane on top of the wall of a thin rectangular channel. This membrane is permeable to carrier liquid, but does not permit the sample components to pass through. Diffusion of the sample components acts as a withstanding force, and therefore all of the

(5)

5 components take different locations in the channel. Separation occurs via differences in diffusion coefficients of the eluting elements. Particles with large diffusion coefficients (or smaller sizes) elute first; and particles with small diffusion coefficients elute later due to the parabolic profile of the channel flow.

FlFFF can be classified into three different variants based on the geometry of the channel. Firstly, the symmetric FlFFF, which has a flat channel having both upper and lower walls permeable, secondly, an asymmetric FlFFF (AF4) channel, which contains only one permeable wall on top of the channel. Thirdly a hollow fiber FlFFF, which is a relatively new technique, its channel is based on ceramic hollow fiber having a porous cylindrical wall.5

AF4 is the most often used technique, because of its promising features. Especially its possibility to separate particles with a wide range of sizes using variable cross-flow programs is promising. Today, AF4 is used in industrial, pharmaceutical and biomedical applications, as an alternative to size-exclusion chromatography.6

The principles of AF4 are discussed in detail in chapter 2. In short, the part of the axial flow which is pumped out through the membrane, acts as cross-flow. During sample injection a mobile phase flows in from both inlet and outlet of the channel and meets at a focusing point. The sample is introduced in the channel at the focusing point where it is focused for a short period. After the focusing step, the flow is introduced in the channel to start the elution, the sample components are separated in the channel and then detected by the UV-Vis detector.

Over the past few years miniaturization of separation techniques has been of great interest since it has many advantages as a sample volume reduction, a lower mobile phase consumption and a faster analysis.4 A miniaturized AF4 device could become a point of care diagnostic device (POCDD) for analysis of macromolecules in body fluids. POCDD measurements provide results rapidly and near the patient. This leads to major time savings because the samples do not travel to a laboratory and results can be collected immediately. During this project one commercial- and several miniaturized AF4 channels were tested for the separation of proteins. Obtained results were compared. The goal of this thesis is to examine the possibility of using miniaturized AF4 channels for the separation of macromolecules without losing the performance of a commercially available channel.

(6)

6

2. Principles of asymmetrical flow field-flow fractionation (AF4)

In FFF the separation is executed with a carrier liquid pumped through a flat channel. The channel is formed by a spacer between two walls (Figure 1)7. In a symmetric system the walls are both porous (Figure 1a). A second pump is used to flow the same carrier liquid in the perpendicular direction, through both walls of the channel. The macromolecular components of the sample stay inside the channel by a semi-permeable membrane on top of one of the porous walls.

Wahlund and Giddings8 proposed an easier system in 1987. This asymmetrical system (Figure 1b) contains only one porous wall. The in-going flow (Fin) in AF4 is split in two parts with help of flow

regulators or an extra pump. One part is flushed through the channel in the axial direction towards the detection side outlet, called the channel flow (Fout). The other part passes through the membrane and

the porous wall, called the cross-flow (Fc). The advantage of AF4 is the fact that the set-up requires

one pump or flow regulator less than a symmetrical system. All commercial instruments use the AF4 method.

Figure 1: Experimental set-up for (a) symmetrical and (b) asymmetrical FlFFF (Qureshi, R. N.; Kok, W. T7)

Separation in AF4 contains three different steps (Figure 2)7. The first step is called the injection (Figure 2a), in which a specific volume of the sample is introduced in the channel. This is done through an extra inlet port close to the channel inlet.

After this injection, a part of the in-going flow enters the channel through the channel inlet, and part is pumped into the channel from the channel outlet. This step is called focusing. In this way the sample components are concentrated on top of the membrane in a thin layer and focused close to the inlet port in a thin band.

The final step is the elution (Figure 2b). The carrier liquid enters the channel from the inlet side. A part of the liquid leaves the channel through the porous wall and the rest flows through the detector. The velocity profile of the channel flow is a parabola, velocity is high in the middle of the channel and low at the walls. The sample components are all concentrated in a thin layer on top of the membrane by the cross-flow, and then flushed towards the detector with the channel flow with different velocities, dependent on the position of the sample in the channel. Retention time of sample components only depend on differences in diffusion constant, and consequently on differences in molecular size.

(7)

7

3. Theory

3.1 Resolution

The resolution is referred to the separation ability of the system. It is the degree of overlap of two peaks. It is agreed that a resolution index Rs of more than 1.5 indicates a baseline separation between the peaks, and a resolution less than 1.5 indicates some degree of co-elution. A resolution of zero indicates that the peaks are completely overlapped. The resolution can be described mathematically by9:

(1)

In which Δtr is the retention time difference between the peaks and σ1 and σ2 are the standard

deviations of two peaks. The standard deviations in case of Gaussian peaks can be calculated using equation 2.9

√ (2)

3.2 Column efficiency

The number of theoretical plates and the plate height are mathematical concepts to predict column efficiency, these concepts are related to each other.

3.2.1 Number of theoretical plates

The number of theoretical plates N is an indirect measure of the width of the peak at a specific retention time, and can be calculated by10:

(

)

(3)

Where tr is the elution time of the peak and w1/2 is the width of the sample peak at half height.

3.2.2 Plate height

Another concept to predict column efficiency is the height equivalent of the theoretical plate H. In plate theory the plate height is given by the length of the column L divided by the number of theoretical plates N, as shown in equation 4.11

(4)

The plate height represents the length of the separation column that is necessary to generate a separation between two particles. The goal for an efficient column is to minimize H and maximize N. This is because the shorter each theoretical plate is, the more theoretical plates will fit in the length of the column. This translates to more plates per meter, which leads to higher column efficiency.

The total plate height contains several factors, such as instrumental effects Hi, non-equilibrium effects

Hn, polydispersity Hp and diffusion Hd, as shown in equation 5.

11

(8)

8 The diffusion coefficients for the particles used in this project are small because the particles have high molecular weights. The flow velocity is high, so the contribution of the diffusion to plate height can be neglected. The polydispersity is also negligible, because we are working in this project with fairly monodisperse macromolecules. The only effects to the total plate height taken into consideration are the instrumental and non-equilibrium effects, when optimizing the instrument. Instrumental plate height

The instrumental component of the plate height in field-flow fractionation systems depends on the channel geometry, the post-column volumes, the sample injection size, method, the instrument set-up and the fluidic connections. All of these elements are not easily expressed in a comprehensive theory, so they have not been mathematically examined. The instrumental plate height is usually measured experimentally.11

Non-equilibrium plate height

The non-equilibrium component of the plate height in field-flow fractionation systems depends on the channel thickness, the diffusion D and the flow velocity v, as shown in the complex equation 6.11

〈 〉

(6)

The function χ(λ) is traditionally represented by:

(7) The non-equilibrium term is the main contributor to the measured plate height. This effect is due to the differential axial movement of the zone components, as a result these components are located in different velocity streamlines across the channel thickness.11

3.3 Retention time

The relation of the retention of a compound and its molecular size in AF4 relies only on liquid flows and molecular diffusion. The retention time of a compound can be calculated using the simplified equation 8.7

(

) (8)

In which w is the thickness of the spacer (or height of the channel). Di is the diffusion coefficient of

the compound, and can be calculated using the Stokes-Einstein equation.

3.4 Standard deviation

The standard deviation of a peak of a compound with an adequate amount of retention can be calculated by the simplified equation 9.7

{ (

)}

(9) In which uC is the cross-flow velocity (the cross-flow rate divided by the area of the porous channel),

and w is the real spacer height of the channel. This equation does not take into account the instrumental parameters, but only the non-equilibrium plate height. In a specific fractionation the compounds are eluted as peaks with the same width.

(9)

9

4. Experimental

4.1 Reagents and samples

In this project three different proteins were used for the analysis, BSA, apoferritin and thyroglobulin, all provided by Sigma Aldrich. Phosphate Saline Buffer (PBS) (15mM) was used as mobile phase. In distilled water, 8.00g/L NaCl, 2.76g/L Na2HPO4∙2H2O and 0.63g/L NaH2PO4∙H2O were dissolved.

The pH of this buffer was adjusted to 7.4 with NaOH.

4.2 Instruments

All samples were analyzed with an Applied Biosystems 757 Absorbance detector (220/280nm wavelength used) equipped with an Agilent Technologies 1100/1200 series Isocratic LC system, Iso Pump, Degasser and a Wyatt Technology Europe GmbH Eclipse 2 separation system. For the commercial channel measurements an Eclipse AF4 mini-channel was used.

Data processing was achieved with software Clarity Lite Chromatography Station. For the final data treatment Microsoft Excel version 2010 was used.

The miniaturized channels are not commercially available, but are costum-made. In figure 3 the miniaturized channels are shown.

Figure 3: Miniaturized channel. Left: a. channel inlet, b. injection port, c. channel outlet. Right: Different parts of the device, membrane is placed on top of the black ring on the bottom part.

(10)

10

4.3 Asymmetrical flow field-flow fractionation

For the AF4 experiments 15µL of sample was injected with a concentration of approximately 1 mg/mL. For all experiments a blank was injected as well. The blank was PBS, which was the solvent for preparation of sample solutions.

Figure 4 shows a typical fractogram of BSA, apoferritin and thyroglobulin. As said in chapter 2, an AF4 run has three different steps. The first part of the fractogram shows the focusing part and sample injection, after a few minutes (in this case 5 min) the elution starts and the proteins are detected by the UV/Vis detector. BSA elutes first, because this is the smallest protein (66kDa). Apoferritin (444kDa) and thyroglobulin (660kDa) elute later. BSA and thyroglobulin have dimers, which are also shown in the fractogram. In the last part of the run there is no crossflow, so the amount of sample that is absorbed on the membrane elutes in this part.

(11)

11

5. Results

All results are obtained with UV-Vis detector at wavelength 280nm, except miniature of shape 1 with spacer height 100µm is obtained at wavelength 220nm. Resolution and number of theoretical plates are calculated using equation 1 and 3. The theoretical fractograms (figure 7,9,12,17,19,21 and 24) were made using equation 8 and 9, to calculate the retention time of the proteins and width of their peaks in the fractogram.

5.1 Commercial channels

First step in this project was to optimize the commercially available channels. Four different channels were tested, channels with spacer height 190µm, 250µm, 350µm and 480µm. All fractograms containing the best conditions for each channel were also compared to their theoretical fractogram.

Commercial channel with spacer height 350μm

The channel with spacer height 350μm was optimized first. Several measurements with different crossflows were tested, and resolutions and number of theoretical plates were calculated using equations 1 and 3. The peaks used for these calculations were the peaks of apoferritin and thyroglobulin, because BSA always elutes quite fast and is well separated from apoferritin and thyroglobulin. Crossflow 2.0ml/min gave a resolution of 1.49, which is almost baseline separation (resolution 1.5), as shown in table 2 and figure 5.

Table 2: Resolution and number of theoretical plates for various flow-conditions

Crossflow Fc (ml/min)

Detectorflow Fout (ml/min)

Resolution Rs Theoretical plates N

2.0 1.0 1.49 204

1.5 1.0 1.26 158

(12)

12

Figure 5: Fractogram of BSA + apoferritin + thyroglobulin. Experimental conditions: inj. vol. 15µL, mobile phase PBS, spacer height 350µm

In order to examine the effect of detectorflow on resolution, different detectorflows were tested and resolutions were calculated, as shown in table 3 and figure 6. Higher detectorflow led to lower number of theoretical plates and lower resolution, so crossflow 2.0ml/min and detectorflow 1.0ml/min are the optimum conditions for the 350μm spacer. The detectorflow was kept constant in every measurement, because this made it easier to compare the spacer height in chapter 5.1.1.

Resolution Rs Theoretical plates N 2.0 2.0 1.0 1.5 1.49 1.47 204 196 2.0 2.0 1.42 154 0,0 0,1 0,2 0,3 0,4 0,5 0 10 20 30 In te n si ty (m V) tR (min) 0,0 0,1 0,2 0,3 0,4 0,5 0 10 20 30 In te n si ty (m V) tR (min) 0,0 0,1 0,2 0,3 0,4 0,5 0 10 20 30 In te n si ty (m V) tR (min)

(13)

13

Figure 6: Fractogram of BSA + apoferritin + thyroglobulin. Experimental conditions: inj. vol. 15µL, mobile phase PBS, spacer height 350µm

The fractogram of the measurement with the best conditions was compared to the theoretical fractogram, as shown in figure 7. The theoretical fractogram doesn’t take dimers into account, so these peaks are not shown in the theoretical fractogram. The BSA-peak of the real fractogram appears to be similar to the theoretical one, the apoferritin and thyroglobulin elute slightly later than theory. This is due to the fact that the equations used for calculating the theoretical fractogram, only takes the non-equilibrium plate height (chapter 3.2.2) into account, not the instrumental plate height. In this case the peaks elute later than should be according to the theoretical fractogram, also their width is larger. This could be because of the fact that the instrumental plate height is quite large.

0,0 0,1 0,2 0,3 0,4 0,5 0 10 20 30 In te n si ty (m V) tR (min) 0,0 0,1 0,2 0,3 0,4 0 10 20 30 In te n si ty (m V) tR (min) 0,0 0,1 0,2 0,3 0,4 0 10 20 30 In te n si ty (m V) tR (min)

(14)

14

Figure 7: Theoretical fractogram of BSA + apoferritin + thyroglobulin. Experimental conditions: inj. vol. 15µL, mobile phase PBS, real spacer height 263µm, Fc=2.0mL/min, Fout=1.0mL/min

The conditions used for the first measurement with the channel with spacer height 250μm were the same as the optimum conditions for the 350μm. This measurement resulted in a resolution of 0.89, so there was still some degree of co-elution. In order to push the sample to the bottom of the channel more, more runs with higher crossflow were measured. After calculating the resolution, as shown in table 4, crossflow 3.0 ml/min seemed to be the best condition for this spacer height. The resolution for these conditions is 1.22, so there is still some degree of co-elution, as shown in figure 8, but this was the best result possible with detectorflow 1.0ml/min.

Resolution Rs Theoretical plates N

2.0 1.0 0.89 102 3.0 1.0 1.22 208 3.5 1.0 1.17 226 4.0 1.0 1.15 211 0,0 0,1 0,2 0,3 0,4 0,5 0 5 10 15 20 25 30 In te n si ty (m V) tR (min) Theoretical fractogram Real fractogram

(15)

15

The fractogram of the measurement with the best conditions was compared to the theoretical fractogram, as shown in figure 9. The peaks coming from the dimers are not shown in the theoretical fractogram, because these dimers were not taken into account. The experimental fractogram of the 250μm spacer looks very similar to the theoretical fractogram. Still the width of the peaks is larger than theory, but this is again due to the instrumental plate height.

0,0 0,5 1,0 1,5 0 10 20 30 In te n si ty (m V) tR (min) crossflow 2 detectorflow 1 0,0 0,5 1,0 0 10 20 30 In te n si ty (m V) tR (min) crossflow 3 detectorflow 1 0,0 0,5 1,0 1,5 0 10 20 30 In te n si ty (m V) tR (min) crossflow 3.5 detectorflow 1 0,0 0,5 1,0 0 10 20 30 Inte n si ty (m V) tR (min) crossflow 4 detectorflow 1 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9 1,0 0 5 10 15 20 25 30 In te n si ty (m V) tR (min) Theoretical fractogram Real fractogram

Figure 9: Theoretical fractogram of BSA + apoferritin + thyroglobulin. Experimental conditions: inj. vol. 15µL, mobile phase PBS, real spacer height 182µm, Fc=3.0mL/min, Fout=1.0mL/min

(16)

16 Commercial channel with spacer height 190μm

The first conditions used for the measurements with the channel with spacer height 190μm were the best conditions for the 250μm spacer. The fractogram didn’t show any peaks of the proteins, and the peak after focusing+elution (without crossflow) at the end of the fractogram was extremely large. Lower detectorflows were tested, and with detectorflow 0.5ml/min a peak in the fractogram appeared, but the proteins were not separated, as shown in figure 10. After this measurement higher crossflows were tested, but all of them gave a similar result, only the amount of sample left after the focusing+elution step (the large peak at the end of the fractogram) became larger. This is due to the fact that the sample components are pushed harder onto the membrane by a higher crossflow, and they might get stuck in the membrane. The channel with spacer height 190μm is not able to separate this mixture of proteins.

The channel with spacer height 480μm is closest in size to the 350μm spacer, so the optimum conditions for this spacer were tested on the 480μm spacer first. The resolution for this measurement was immediately above 1.5, so there was baseline separation already. Lower crossflows were tested in order to make the analysis as quick as possible. The crossflow 1.8ml/min also gave a resolution higher than 1.5, and the measurement was a lot faster than the measurement with crossflow 2.0ml/min and had a higher number of theoretical plates, as shown in table 5 and figure 11.

0,0 0,5 1,0 0 10 20 30 Inte n si ty (m V) tR (min) 0,0 0,5 1,0 0 10 20 30 In te n si ty (m V) tR (min) 0,0 0,5 1,0 0 10 20 30 In te n si ty (m V) tR (min)

(17)

17

Resolution Rs Theoretical plates N 2.0 1.8 1.0 1.0 1.68 1.58 111 228 1.6 1.0 1.43 52

The fractogram of the measurement with the best conditions was compared to the theoretical fractogram, as shown in figure 12. The width of the peaks in the real fractogram is much bigger than it should be, according to the theory. The width of the peaks is also larger than was the case in the 250μm and 350μm spacer fractograms, so the instrumental plate height of the 480μm spacer will be bigger than in case of the other two spacer heights.

0,0 0,1 0,2 0 10 20 30 In te n si ty (m V) tR (min) 0,0 0,1 0,2 0 10 20 30 In te n si ty (m V) tR (min) 0,0 0,1 0,2 0,3 0 10 20 30 In te n si ty (m V) tR (min)

(18)

18

Figure 12: Theoretical fractogram of BSA + apoferritin + thyroglobulin. Experimental conditions: inj. vol. 15µL, mobile phase PBS, real spacer height 398µm, Fc=1.8mL/min, Fout=1.0mL/min

5.1.1 Comparing commercial channels

As shown in table 6, the resolution for the channel with spacer height 480µm is the highest, there is baseline separation. Also the channel with spacer height 350µm has a resolution higher than 1.5, so also in this case there is baseline separation. The number of theoretical plates is highest for the channel with spacer height 480µm, which indicates more plates per meter in a separation column, which leads to higher column efficiency.

Table 6: Resolution and number of theoretical plates for various spacer heights

Spacer height w (µm) Crossflow Fc (ml/min) Detectorflow Fout (ml/min) Resolution Rs Theoretical plates N 190 3.5 0.5 - - 250 3.0 1.0 1.22 208 350 2.0 1.0 1.49 204 480 1.8 1.0 1.58 228

In figure 13 all fractograms for the different spacers are shown. The channel with spacer height 190µm is not able to separate the proteins. Table 6 indicates that the channel with spacer height 480µm is the best, but the fractogram shows that the separation takes almost 25 minutes. Compared to the channel with spacer height 350µm, which also had high resolution and number of theoretical plates, this is a very long separation. In order to use AF4 as a POC technique, the analysis should be as short as possible. The separation in channel with spacer height 350µm only takes 15 minutes, so this is the best spacer height for separation in the commercial channels.

0,0 0,1 0,2 0,3 0,4 0 5 10 15 20 25 30 In te n si ty (m V) tR (min) Theoretical fractogram Real fractogram

(19)

19

Figure 13: Fractogram of BSA + apoferritin + thyroglobulin, best conditions for every spacer. Experimental conditions: inj. vol. 15µL, mobile phase PBS, spacer height 190-480µm

5.2 Miniaturized channels

During the second part of the project several miniaturized channels of different shapes and dimensions (figure 14) were optimized and compared to the commercial channels. Five different miniaturized channels of shape 1 were tested, varying in channel height (100, 200, 300, 400 and 500 µm). In the optimization of the commercial channels, the channels with spacer height 350 and 480µm showed the best separation of the proteins. Because of that when testing the miniatures, the channel with height 400µm was tested first. For shape 2 and 3 only channel height 400μm was tested.

Figure 14: Shapes and dimensions of the different miniatures -0,1 0,4 0,9 0 10 20 30 In te n si ty (m V) tR (min) 190um -0,1 0,4 0,9 0 10 20 30 In te n si ty (m V) tR (min) 250um -0,1 0,4 0,9 0 10 20 30 In te n si ty (m V) tR (min) 350um -0,1 0,4 0,9 0 10 20 30 In te n si ty (m V) tR (min) 480um

(20)

20 Miniaturized channel shape 1 with channel height 400μm

Since the miniature is smaller in size than the commercial channel, the conditions for the measurements also need to be reduced. First a measurement was done with the same ratio in crossflow/detectorflow 2:1 as the best conditions for the commercial channel with spacer height 350μm. The resolution for these conditions was only 0.48, so lower crossflows were tested to make sure the sample is pushed to the membrane more. The resolution became higher (as shown in table 7 and figure 15), but there was still no baseline separation.

Figure 15: Fractogram of BSA + apoferritin + thyroglobulin. Experimental conditions: inj. vol. 15µL, mobile phase PBS, channel height 400µm

Crossflow 0.9ml/min gave the highest resolution with detectorflow 0.5ml/min, but to make sure this was the best resolution possible, lower detectorflows were tested, as shown in table 8. Although the run with detectorflow 0.5ml/min was shorter than the other ones (figure 16), detectorflow 0.2ml/min gave a much higher resolution and number of theoretical plates, so crossflow 0.9ml/min and detectorflow 0.2ml/min were the best conditions for channel height 400μm. The detectorflow was kept the same in every measurement, because this made it easier to compare the channel height in chapter 5.2.1 and 5.2.2. 0,0 0,1 0,2 0,3 0 10 20 30 In te n si ty (m V) tR (min) 0,0 0,1 0,2 0,3 0 10 20 30 In te n si ty (m V) tR (min) 0,0 0,1 0,2 0,3 0 10 20 30 In te n si ty (m V) tR (min)

(21)

21

The fractogram of the measurement with the best conditions was compared to the theoretical fractogram, as shown in figure 17. The width of the peaks is extremely large in the real fractogram, the proteins elute later than should according to the theory. The equation used for calculating the retention times in the theoretical fractogram was equation 8, which takes only into account the non-equilibrium plate height. In this case the instrumental plate height is large because the width of the peaks is large, larger than was the case in the commercial channels.

-0,1 0,0 0,1 0,2 0,3 0,4 0 10 20 30 In te n si ty (m V) tR (min) -0,1 0,0 0,1 0,2 0,3 0,4 0 10 20 30 In te n si ty (m V) tR (min) 0,0 0,1 0,2 0,3 0 10 20 30 In te n si ty (m V) tR (min)

(22)

22

Figure 17: Theoretical fractogram of BSA + apoferritin + thyroglobulin. Experimental conditions: inj. vol. 15µL, mobile phase PBS, real channel height 331µm, Fc=0.9mL/min, Fout=0.2mL/min

Miniaturized channel shape 1 with channel height 300μm

The channel height of this miniature is smaller than 400μm, therefore the crossflow should be higher than 0.9ml/min, as seen before with the commercial channels. First crossflow 1.0ml/min was tested, but the resolution was only 1.00. Higher crossflow was needed to make the resolution higher, and crossflow 1.3ml/min got the highest resolution possible (as shown in table 9 and figure 18), only there was still some degree of co-elution. Because the miniaturized channel is a small channel compared to the commercial channels, higher crossflows were not used. When using high crossflows, the channel could be leaking because the pressure in the channel becomes too high.

Resolution Rs Theoretical plates N 1.0 1.1 1.2 0.2 0.2 0.2 1.00 1.04 1.00 170 173 205 1.3 0.2 1.15 209 -0,1 0,0 0,1 0,2 0,3 0,4 0 5 10 15 20 25 30 In te n si ty (m V) tR (min) Theoretical fractogram Real fractogram

(23)

23

The fractogram of the measurement with the best conditions was compared to the theoretical fractogram, as shown in figure 19. Same as was the case in the fractogram of the 400μm spacer, the width of the peaks in the real fractogram is much bigger than should be according to theory, also the retention time is later than in the theoretical fractogram. This is again due to the instrumental plate height, which is in this case even larger than in the channel with height 400μm.

-0,1 0,0 0,1 0,2 0,3 0,4 0 10 20 30 In te n si ty (m V) tR (min) crossflow 1 detectorflow 0.2 -0,1 0,0 0,1 0,2 0,3 0,4 0 10 20 30 In te n si ty (m V) tR (min) crossflow 1.1 detectorflow 0.2 -0,1 0,0 0,1 0,2 0,3 0,4 0 10 20 30 In te n si ty (m V) tR (min) crossflow 1.2 detectorflow 0.2 -0,1 0,0 0,1 0,2 0,3 0,4 0 10 20 30 In te n si ty (m V) tR (min) crossflow 1.3 detectorflow 0.2

(24)

24

As said in the previous part, the crossflow for this channel height needed to be higher than for channel height 300μm. Crossflow 1.5ml/min was tested first, but the resolution was only 1.04, so there was some degree of co-elution. The number of theoretical plates was quite high already. Higher crossflows were tested, but this resulted in resolutions almost equal to 1.04, as shown in table 10 and figure 20. The reason for these similarities was the fact that the actual crossflow and detectorflow differed from the numbers set in the program during analysis, so the crossflows and detectorflows were almost similar in every measurement. In order to make the miniaturized channel a POCDD, the analysis should be short and the amount of mobile phase used should be as low as possible. Because the resolutions were all almost the same and the number of theoretical plates was high for every crossflow, the crossflow 1.5ml/min was the best condition for channel height 200μm.

Resolution Rs Theoretical plates N 1.5 1.7 2.0 0.2 0.2 0.2 1.04 1.05 1.07 249 236 253 -0,1 0,2 0,4 0,6 0,8 1,0 0 5 10 15 20 25 30 In te n si ty (m V) tR (min) Theoretical fractogram Real fractogram

(25)

25

The fractogram of the measurement with the best conditions was compared to the theoretical fractogram, as shown in figure 21. In this case the width of the peaks is still bigger than theory, but the width of the peaks is smaller than the width of the peaks in the fractograms of the 300 and 400μm spacer height. Also the retention times are faster, but still not fast enough according to theory. In this case the instrumental plate height is still quite large, but not as large as was the case in the 300 and 400μm channel height. -0,1 0,0 0,1 0,2 0 10 20 30 In te n si ty (m V) tR (min) _-0,1 0,0 0,1 0,2 0 10 20 30 In te n si ty (m V) tR (min) -0,1 0,0 0,1 0,2 0 10 20 30 In te n si ty (m V) tR (min)

(26)

26

Channel height 100μm is extremely small when compared to the other channels, so it is harder to separate the proteins, as seen before in the commercial channels. Proteins absorb UV light at 220nm due to the presence of double bonds within the carbonyl groups. Most proteins also absorb light at 280nm, which shows less impurities in the fractogram. In order to show higher signals (but possibly also more impurities), the UV/Vis detector was switched to 220nm. Crossflow 1.5ml/min was tested first, but this crossflow was too low and there was no separation of the proteins (as shown in figure 22). Higher crossflow did not change the rate of separation, so this miniature is not able to separate the proteins. -0,1 0,1 0,2 0,3 0,4 0 5 10 15 20 25 30 In te n si ty (m V) tR (min) Theoretical fractogram Real fractogram

(27)

27

The best conditions for the 400μm were crossflow 0.9ml/min and detectorflow 0.2ml/min, so these conditions were tested first. After a measurement of 45 minutes the proteins were still not detected by the UV/Vis detector. In order to make the run as short as possible, this takes too long. Lower crossflows were tested, and crossflow 0.6ml/min was the only condition in which the proteins were all detected after 45 minutes, as shown in figure 23. Lower crossflows gave the same results as crossflow 0.6ml/min, because the actual crossflow and detectorflow were similar to crossflow 0.6ml/min and detectorflow 0.2ml/min. The actual crossflow and detectorflow differed from the numbers set in the program during analysis. The resolution was 1.37 (as shown in table 11) which is quite good, almost baseline separation. Although the measurement takes too long, crossflow 0.6ml/min is the best condition possible for spacer height 500μm.

Resolution Rs Theoretical plates N 0.9 0.8 0.7 0.2 0.2 0.2 - - - - - - 0.6 0.2 1.37 180 -0,1 0,9 1,9 2,9 3,9 4,9 5,9 6,9 7,9 0 5 10 15 20 25 30 Inte n si ty (m V) tR (min) crossflow 1.5 detectorflow 0.2 crossflow 1.7 detectorflow 0.2

(28)

28

The fractogram of the measurement with the best conditions was compared to the theoretical fractogram, as shown in figure 24. Same as in the fractogram of the 300 and 400μm spacer, the width of the peaks is much bigger than theory, also the retention time is much later than is the case in the theoretical fractogram. In this case the instrumental plate height is even larger than was the case with channel height 400μm. -0,1 0,4 0,9 0 50 In te n si ty (m V) tR (min) crossflow 0.6 detectorflow 0.2 -0,1 0,4 0,9 0 50 In te n si ty (m V) tR (min) crossflow 0.7 detectorflow 0.2 -0,1 0,4 0,9 0 50 In te n si ty (m V) tR (min) crossflow 0.8 detectorflow 0.2 -0,1 0,4 0,9 1,4 0 50 In te n si ty (m V) tR (min) crossflow 0.9 detectorflow 0.2 -0,1 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0 10 20 30 40 In te n si ty (m V) tR (min) Theoretical fractogram Real fractogram

Figure 24: Theoretical fractogram of BSA + apoferritin + thyroglobulin. Experimental conditions: inj. vol. 15µL, mobile phase PBS, real channel height 462µm, Fc=0.6mL/min, Fout=0.2mL/min

(29)

29 5.2.1 Comparing miniaturized channels with shape 1

As shown in table 12, all resolutions are lower than 1.5, so there is some degree of co-elution in every measurement. The resolution for the channel with height 500µm is the highest, this means there is almost baseline separation. Also the channel with height 400µm has a relatively good resolution. The number of theoretical plates is highest for the channel with height 400µm, which indicates that the length of the separation column that is necessary to generate a separation between two particles is as short as possible in this channel height, if compared to the other spacer heights.

Table 12: Resolution and number of theoretical plates for various spacer heights

Spacer height w (µm) Crossflow Fc (ml/min) Detectorflow Fout (ml/min) Resolution Rs Theoretical plates N 200 1.5 0.2 1.04 249 300 1.3 0.2 1.15 209 400 0.9 0.2 1.28 228 500 0.6 0.2 1.37 180

Figure 25 shows the fractograms of all different channel heights. Channel height 100μm was not able to separate the proteins and therefore is not included in the figure. The 500μm channel height gave the highest resolution, but the measurement takes 45 minutes. In order use AF4 as a POCDD, a measurement should be as short as possible. The measurement of the 500μm channel height takes too long, so the 400μm channel height (which takes only 30 minutes) is the best spacer for the miniature with shape 1.

Figure 25: Fractogram of BSA + apoferritin + thyroglobulin, best conditions for every spacer. Experimental conditions: inj. vol. 15µL, mobile phase PBS, channel height 200-500µm

-0,1 0,0 0,1 0,2 0 20 40 Inte n si ty (m V) tR (min) 200um -0,1 0,0 0,1 0,2 0 20 40 In te n si ty (m V) tR (min) 300um -0,1 0,0 0,1 0,2 0 20 40 Inte n si ty (m V) tR (min) 400um -0,1 0,0 0,1 0,2 0 20 40 In te n si ty (m V) tR (min) 500um

(30)

30 Miniaturized channel shape 2 with channel height 400μm

The best spacer height for the miniature with shape 1 was the spacer height 400μm. The best conditions for this miniature were crossflow 0.9ml/min and detectorflow 0.2ml/min, so these conditions were tested first on the miniature with shape 2. As shown in figure 26, the proteins were not separated. The peak after focusing+elution (after 30min) is large, which indicates all sample components are pushed onto the membrane by the crossflow and got stuck in the membrane. The sample components stay inside the channel until the crossflow is stopped. Lower and higher crossflow were tested, but all of them gave the same results, no separation of the proteins.

Because the channel with shape 3 is very small, the flows also need to be low, otherwise the pressure inside the device will be too high. A few crossflows were tested (figure 27), but all of them gave fractograms with no signals for the proteins. In this case the peak after focusing+elution is quite small, so there is not much mixture stuck in the membrane. Blue Dextran (chapter 5.2.2) showed a large part of the mixture leaks out of the channel, leading to no signal from the UV-Vis detector.

-2,0 0,0 2,0 4,0 6,0 8,0 10,0 0 20 40 In te n si ty (m V) tR (min) -2,0 0,0 2,0 4,0 6,0 8,0 10,0 0 20 40 Inte n si ty (m V) tR (min) -2,0 0,0 2,0 4,0 6,0 8,0 10,0 0 20 40 In te n si ty (m V) tR (min)

(31)

31

5.2.2 Comparing miniaturized channels

The miniature with shape 1 was able to separate the proteins and channel height 400μm was the height with the best conditions possible, although the miniature did not give results as good as the commercial channel. Miniatures with shape 2 and 3 cannot be used for separation of the mixture of three proteins. A single run with only BSA and another run with only apoferritin were tested, and the results are shown in figure 28 and 29. Instead of just one peak expected, both proteins showed two peaks in the fractogram.

After this observation, a single run with Blue Dextran was measured, to see if all of the sample stayed inside of the channel and to see if the focusing in the channel went well. The Blue Dextran focused in almost half of the channel, so the sample is in fact insufficiently focused. Also the Blue Dextran leaked out of the channel (but stayed in the device), so this could be an explanation why the miniaturized channels were not able to separate the proteins properly.

0,0 10,0 20,0 30,0 40,0 50,0 0 10 20 30 40 In te n si ty (m V) tR (min) crossflow 0.1 detectorflow 0.1 crossflow 0.3 detectorflow 0.2

(32)

32

Figure 28: Fractogram of Apoferritin. Experimental conditions: inj. vol. 15μl, mobile phase PBS, channel height 400μm

Figure 29: Fractogram of BSA. Experimental conditions: inj. vol. 15μl, mobile phase PBS, channel height 400μm -0,2 0,4 0 5 10 15 20 25 30 In te n si ty (m V) tR (min) Apoferritin crossflow 0.5 detectorflow 0.2 -0,2 0,4 0 5 10 15 20 Inte n si ty (m V) tR (min) BSA crossflow 0.3 detectorflow 0.3

(33)

33

6. Conclusion

The commercial channel was optimized for separation of three proteins (BSA, apoferritin and thyroglobulin). For the fastest analysis close to baseline separation, the optimal spacer height and separation method were found. The best spacer height for the commercial channel is 350μm, the best conditions for this spacer are crossflow 2.0ml/min and detectorflow 1.0ml/min. The 480μm spacer gave a higher resolution than the 350μm spacer (respectively resolution 1.58 and 1.49) and a higher number of theoretical plates (respectively 228 and 204), but both resolutions show that there is baseline separation in the measurements and both spacer heights have a large number of theoretical plates. A single measurement with the 350μm spacer also takes 10 minutes less than a measurement with the 480μm spacer. In order to make an analysis faster and with lower mobile phase consumption to become a POCDD, the 350μm spacer is the best spacer height for the commercially available channels.

The best channel height for the miniature of shape 1 is 400μm, the best conditions for this channel are crossflow 0.9ml/min and detectorflow 0.2ml/min. All resolutions were lower than 1.5, so in every measurement there is some degree of co-elution. The 500μm channel height gave a higher resolution than the 400μm channel (respectively 1.37 and 1.28), but the number of theoretical plates was higher for the 400μm channel (228 and 180). A single measurement with the 500μm channel takes 45 minutes, which is 15 minutes longer than the 400μm channel. In order to make a measurement as short as possible this is too long, therefore the 400μm channel height (which takes only 30 minutes) is the best height for the miniaturized channel with shape 1.

The design of miniaturized channels with shape 2 and 3 at this stage of the research is not suited for separation of the mixture of three proteins.

In both the commercial channels and miniatures, the larger spacer/channel heights are best in separating the mixture of three proteins. The instrumental plate height for larger spacer/channel heights are also larger, so the peaks have a larger width than should be according to the theory. The smaller heights (190μm commercial and 100μm miniature) were not able to separate the chosen mixture of proteins. These spacer/channel heights are probably too small, so the focusing part will take place in a larger part of the channel and the separation doesn’t work well. In the miniaturized devices the channel is not pressed properly towards the membrane, so part of the mixture of proteins leaks out of the channel in the device. This way the proteins cannot be detected by the UV/Vis detector.

(34)

34

7. Discussion and Future Prospects

The miniaturized channels were not able to separate the proteins as properly as the commercial channels, and some of the miniaturized channels were not able to separate the proteins at all. After injecting Blue Dextran in the miniature channel, it leaked out of the channel but stayed in the device. More specifically, the Blue Dextran leaked outside the separation channel but it stayed inside the o-ring, which is used to seal the miniaturized channel. This is an explanation for the fact why the miniatures did not give good results, because the protein-sample probably also leaked out of the channel. There should be more pressure on the channel, to push the channel more towards the membrane. A solution could be some kind of glue, to make sure the membrane is attached to the channel and the sample stays inside the channel.

The focusing of the sample takes up almost half of the channel, so in fact the sample is not focused. The mixture of proteins is also not properly pushed towards the membrane (because of leakage). In all of the measurements in this project the focusing took at most 5-6 minutes, so e a longer focusing time (in the range of 10-20 minutes) could be a solution. In order to use AF4 as a POCDD it is not ideal to use long focusing times, because the analysis will take a lot longer.

A third solution in making the miniatures more efficient, is the use of another type of FFF. Frit-Inlet AF412 uses the frit-inlet injection technique with an AF4 channel. The AF4 method used in this project uses a channel flow that is divided in two parts, the crossflow and the axial flow. The driving force in bringing the sample towards the membrane, where the separation takes place, is created by the crossflow. Frit-Inlet AF4 is promising because of the fact that it utilizes a stopless sample injection technique with the conventional asymmetrical channel by implementing an inlet frit nearby the channel inlet, which reduces possible flow imperfections caused by the porous wall.12 Frit-Inlet AF4 does not require sample focusing and relaxation steps which are time consuming. This way it does not interrupt sample migration and valve switching, and thus could be a promising new method for protein separation.

(35)

35

Acknowledgements

Here I would like to thank Peter Schoenmakers for the opportunity to work on my project in his research group. Also very special thanks to Jana Králová and Wim Kok for supervising and helping me over the course of this project. Last but not least, a thank you note for the entire analytical chemistry group for all sociability and helping me out with anything.

List of abbreviations

FFF field-flow fractionation FlFFF flow field-flow fractionation

AF4 asymmetrical flow field-flow fractionation LC liquid chromatography

POCDD point of care diagnostic device

BSA Bovine Serum Albumin

(36)

36

References

1. Caldwell, K. D.; Kesner, L. F.; Myers, M. N.; Giddings, J. C. Electrical Field-Flow Fractionation of Proteins. Science 1972, 176, 296-298.

2. Giddings, J. C.; Smith, L. K.; Myers, M. N. Thermal field-flow fractionation. Extension to lower molecular weight separations by increasing the liquid temperature range using a pressurized system.

Anal. Chem. 1975, 47, 2389-2394.

3. Giddings, J.; Yang, F.; Myers, M. Flow-field-flow fractionation: a versatile new separation method.

Science 1976, 193, 1244-1245.

4. Sant, H. J.; Gale, B. K. Chapter 12: Microscale Field-Flow Fractionation: Theory and Practice. In

Microfluidic Technologies for Miniaturized Analysis Systems. Hardt, S.; Schönfeld, F. Eds.; Springer

US: New York, 2007; 471-521.

5. Lee, W.J.; Min, B; Moon, M.H. Improvement in Particle Separation by Hollow Fiber Flow Field-Flow Fractionation and the Potential Use in Obtaining Particle Size Distribution. Anal. Chem. 1999,

71, 3446-3452.

6. Qureshi, R. N.; Kok, W. T. Application of flow field-flow fractionation for the characterization of macromolecules of biological interest: a review. ABC, 2011, 399(4), 1401-1411.

7. Qureshi, R. N.; Kok, W. T. Optimization of Asymmetrical Flow Field-flow Fractionation (AF4).

LC-GC Europe 2010, 23, 18-25.

8. Wahlund, K. G.; Giddings, J. C. Properties of an asymmetrical flow field-flow fractionation channel having one permeable wall. Anal. Chem. 1987, 59, 1332-1339.

9. Inczédy, J.; Lengyel, T.; Ure, A. M. Compendium of Analytical Nomenclature [Online], Third edition; IUPAC, 1997; Chapter 9: Separations.

http://old.iupac.org/publications/analytical_compendium/TOC_cha9.html (accessed May 18,2014). 10. SHIMADZU (Shimadzu Corporation). Formula for Calculating the Number of Theoretical Plates. http://www.shimadzu.com/an/hplc/support/lib/lctalk/34/34tec.html (accessed May 18, 2014).

11. Sant, H. J.; Gale, B. K. Geometric scaling effects on instrumental plate height in field flow fractionation. Journal of Chromatography A 2006, 1104, 282-290.

12. Moon, M. H. FFF: Frit-Inlet Asymmetrical Flow. In Encyclopedia of Chromatography, Third edition. Cazes, J. Eds.; CRC Press: Boca Raton, Fl., USA, 2009; Volume 2; 860-861.

Miniaturization in asymmetrical flow field-flow fractionation