microfluidic blood sample preparation for rapid sepsis

Microfluidic blood sample preparation for rapid sepsis diagnostics

Jonas Hansson

Licentiate Thesis in Biological Physics

Stockholm Sweden, 2012

ii Microfluidic blood sample preparation for rapid sepsis diagnostics

TRITA-FYS 2012:39 ISSN 0280-316X ISRN KTH/FYS/--12:39--SE ISBN 978-91-7501-428-9 KTH Albanova University Centre Cell Physics SE-106 91 Stockholm SWEDEN Akademisk avhandling som med tillstånd av Kungliga Tekniska Högskolan framlägges till offentlig granskning för avläggande av teknologie licentiatexamen i biologisk fysik, 15 juni 2012 © Jonas Hansson, 2012 Tryck: Universitetsservice US AB

Jonas Hansson iii

Abstract

Sepsis, commonly referred to as blood poisoning, is a serious medical condition characterized by a whole-body inflammatory state caused by microbial infection. Rapid treatment is crucial, however, traditional culture-based diagnostics usually takes 2-5 days. The overall aim of the thesis is to develop microfluidic based sample preparation strategies, capable of isolating bacteria from whole blood for rapid sepsis diagnostics.

Although emerging technologies, such as microfluidics and “lab-on-a-chip” (LOC) devices have the potential to spur the development of protocols and affordable instruments, most often sample preparation is performed manually with procedures that involve handling steps prone to introducing artifacts, require skilled technicians and well-equipped, expensive laboratories. Here, we propose the development of methods for fast and efficient sample preparation that can isolate bacteria from whole blood by using microfluidic techniques with potential to be incorporated in LOC systems.

We have developed two means for high throughput bacteria isolation: size based sorting and selective lysis of blood cells. To process the large blood samples needed in sepsis diagnostics, we introduce novel manufacturing techniques that enable scalable parallelization for increased throughput in miniaturized devices.

The novel manufacturing technique uses a flexible transfer carrier sheet, water-dissolvable release material, poly(vinyl alcohol), and a controlled polymerization inhibitor to enable highly complex polydimethylsiloxane (PDMS) structures containing thin membranes and 3D fluidic networks.

The size based sorting utilizes inertial microfluidics, a novel particles focusing method that operates at extremely high flow rates. Inertial focusing in flow through a single inlet and two outlet, scalable parallel channel devices, was demonstrated with filtration efficiency of >95% and a flowrate of 3.2 mL/min.

Finally, we have developed a novel microfluidic based sample preparation strategy to continuously isolate bacteria from whole blood for downstream analysis. The method takes advantage of the fact that bacteria cells have a rigid cell wall protecting the cell, while blood cells are much more susceptible to chemical lysis. Whole blood is continuously mixed with saponin for primary lysis, followed by osmotic shock in water. We obtained complete lysis of all blood cells, while more than 80% of the bacteria were readily recovered for downstream processing.

Altogether, we have provided new bacteria isolation methods, and improved the manufacturing techniques and microfluidic components that, combined offer the potential for affordable and effective sample preparation for subsequent pathogen identification, all in an automated LOC format.

iv Microfluidic blood sample preparation for rapid sepsis diagnostics

Sammanfattning

Jonas Hansson v

Change will lead to insight far more often than insight will lead to change Milton Erickson

vi Microfluidic blood sample preparation for rapid sepsis diagnostics

Jonas Hansson vii

List of publications

This thesis is based upon the following four papers:

I. J.M. Karlsson, T. Haraldsson, C.F. Carlborg, J. Hansson, A. Russom, and W. van der Wijngaart, Reliable manufacturing of complex PDMS microfluidic structures, Journal of Micromechanics and Microengineering, 2012, In print

II. J. Hansson, J.M. Karlsson, T. Haraldsson, W. van der Wijngaart, and A. Russom, Inertial particle focusing in parallel microfluidic channels for high-throughput filtration, proceedings for The 16th International Conference on Solid-State Sensors, Actuators and Microsystems (Transducers’11), 2011, Beijing, China

III. J. Hansson, J.M. Karlsson, T. Haraldsson, H. Brismar, W. van der Wijngaart, and A. Russom, Inertial microfluidics in parallel channels for high-throughput applications, 2012, Accepted

IV. S. Zelenin, J. Hansson, H. Ramachandraiah, S. Ardabili, H. Brismar, and A. Russom, Microfluidic selective cell lysis for bacteria isolation from whole blood, Manuscript

The contribution of J. Hansson to the different publications:

I. part of fabrication, experiments, and writing II. all design, fabrication, experiments, and major part of writing

III. major part of design, all fabrication, experiments, and writing IV. major part of fabrication, part of experiments and writing

Additional published conference proceedings not included in thesis:

i. S. Zelenin, H. Ramachandraiah, J. Hansson, S. Ardabili, H. Brismar, and A. Russom, Bacteria isolation from whole blood for sepsis diagnostics, proceedings for The 15th International Conference on Miniaturized Systems for Chemistry and Life Sciences (MicroTAS), 2011, Seattle, USA

ii. J. Hansson, P. Kolmskog, Y. Odenmark, and A-K. Högfeldt, Peer-learning – A concept for improving students’ knowledge in laboratory exercises, peer-reviewed proceedings for 7th International CDIO Conference, 2011, Copenhagen, Denmark

viii Microfluidic blood sample preparation for rapid sepsis diagnostics

Table of Contents

1 Introduction ........................................................................................................................................... 1

2 Background ............................................................................................................................................ 3

2.1 Sepsis ............................................................................................................................................. 3

2.1.1 Infected blood sample ............................................................................................................ 4

2.2 Diagnostics..................................................................................................................................... 5

2.2.1 Standard diagnostics ............................................................................................................... 5

2.2.2 New diagnostics ..................................................................................................................... 7

2.3 Microfluidics .................................................................................................................................. 8

2.3.1 Physics ................................................................................................................................... 8

2.3.2 Microfabrication .................................................................................................................. 10

2.4 Cell lysis ....................................................................................................................................... 13

2.4.1 Osmotic shock ..................................................................................................................... 13

2.4.2 Detergents ............................................................................................................................ 13

2.4.3 Selective cell lysis on chip ..................................................................................................... 14

3 Procedures and Methods ...................................................................................................................... 15

3.1 Traditional soft lithography .......................................................................................................... 15

3.2 Novel fabrication method ............................................................................................................. 16

3.3 Florescence microscopy ................................................................................................................ 17

3.4 Coulter counter ............................................................................................................................ 17

3.5 Bacterial culture ........................................................................................................................... 18

4 Results.................................................................................................................................................. 19

4.1 Fabrication results ........................................................................................................................ 19

4.1.1 Fabrication method development ......................................................................................... 19

4.1.2 Fabricated devices ................................................................................................................. 19

4.2 Inertial microfluidics .................................................................................................................... 21

4.2.1 Focusing ............................................................................................................................... 21

4.2.2 Filtration and sorting ............................................................................................................ 22

4.3 Selective cell lysis .......................................................................................................................... 23

5 Discussion ............................................................................................................................................ 25

Jonas Hansson ix

5.1 Fabrication ................................................................................................................................... 25

5.2 Inertial microfluidics .................................................................................................................... 25

5.2.1 Focusing ............................................................................................................................... 25

5.2.2 Filtration .............................................................................................................................. 26

5.3 Selective cell lysis .......................................................................................................................... 27

5.4 Integration ................................................................................................................................... 27

6 Conclusions ......................................................................................................................................... 29

6.1 Future studies ............................................................................................................................... 31

7 Acknowledgements ............................................................................................................................... 33

8 References ............................................................................................................................................ 35

Jonas Hansson 1

1 Introduction Sepsis is a potentially life threatening condition which is caused by infection in the blood stream with systemic effects. Rapid treatment is crucial. For every treatment is delayed, the mortality increases. Traditional diagnostics usually takes 2-5 days, in some cases up to 2 weeks (1). During that time the patients are treated with antibiotics, usually of a broad spectrum type. This might not be the best treatment and sometimes doesn’t work at all leading to increased spreading of antibiotic resistance in the long run. Faster diagnostic tools would thus be of great benefit.

Lab-on-a-chip (LOC) is a relatively new field using microfluidics and miniaturized sensors that have a potential to provide automated, rapid diagnostics at a low cost (2).

The overall purpose of this thesis is to find ways to decrease the time for sepsis diagnostics through development of methods for sample preparation that are fast and efficient and can isolate pathogens from whole blood by using techniques that could be incorporated in a LOC system.

The work can be divided in three main areas:

Manufacturing of microfluidic components Size based sorting by inertial microfluidics Selective blood cell lysis

These three areas are all parts of the development towards rapid diagnostics of sepsis which will be described in detail in this thesis. I will discuss and examine this in order to improve mine and others future work within the field of sample preparation for sepsis diagnostics.

Even though sepsis can be caused by different pathogens (bacteria, fungi or viruses) this work will be focused on bacterial infections since they are the most common, most aggressive and usually the most dangerous form of sepsis (3).

2 Microfluidic blood sample preparation for rapid sepsis diagnostics

Jonas Hansson 3

2 Background This chapter is provides background in three parts: sepsis including diagnostics, microfluidics including both inertial microfluidics and microfabrication, and cell lysis.

2.1 Sepsis In the EU alone, sepsis claims at least 135 000 lives/year and is responsible for 7.6 billion EUR/year health care costs (4). It is characterized by a systemic (whole body) inflammation as response to an infection. There are different levels of sepsis as defined 1990 (5) and updated 2001 (6).

Systematic inflammatory response syndrome (SIRS) is defined by two or more of: i. Abnormal body temperature (below 36oC or above 38oC)

ii. High heart rate (above 90 beats/minute) iii. Tachypnea (rapid breathing) iv. Alteration of white blood cell count (below 4000/ml or above 12000/ml)

Sepsis is defined as SIRS as a response to a confirmed or suspected infection. Severe sepsis is sepsis with organ dysfunction, hypoperfusion (low tissue levels of oxygen) or hypotension (low blood pressure). Septic shock is persistent hypotension together with organ dysfunction or hypoperfusion.

Following the medical records of 2731 patients Kumar (7) reported that effective antimicrobial treatment within the first hour of documented hypotension was associated with increased survival. Each hour of delay in effective treatment decreased the survival on average 8% for the following 6 hours. The cumulative effective treatment and survival rate are shown in figure 1.

Figure 1. Cumulative effective antimicrobial initiation following onset of septic shock associated hypotension and associated survival (7). (Image reprinted with permission © Lippincott Williams & Wilkins)

Faster diagnostics have the potential of saving lives, cut hospital costs and decrease the problem of antibiotic resistances developing (8).


2.1.1 Infected blood sample Human blood consists of 55% by volume plasma and 45% blood cells. The blood cells are divided in erythrocytes, leukocytes and thrombocytes. In addition to plasma and blood cells, the blood stream of a sepsis patient will also contain infectious bacteria, normally not found in healthy individuals. The most important properties of the blood components and bacteria are described below and summarized in table 1.

Table 1. Overview of the different components in an infected blood sample. Sources for thumbnail images are, from left to right: (9), (10), (11), (12), and (13).

Components: Bacteria Erythrocytes Leukocytes Thrombocytes Plasma

Common name - Red Blood Cells White Blood Cells Platelets -

Amount / mL 1-104 5x109 107 5x105 0,55 mL

Image

Shape Spheric, rod etc. Disc shaped Spherical Irregular Viscous fluid

Size 1-3 µm 6 x 6 x 2 µm 7-21 µm 2-3 µm -

Nucleus No No Yes No -

DNA Yes No (normally) Yes No Random debris

Wall Peptoglucan

layer Cell membrane Cell membrane Cell membrane -

Surface antigens Lipopolysaccharide, teichoic acids HLA, CD239 HLA, CD45 HLA, GPIIb Albumins,

immunoglobulins

Plasma consists of 92% water, 8% proteins and small amounts of nutrients, waste products and several other components.

Erythrocytes – also known as red blood cells, contain hemoglobin and distribute oxygen in the body. They are shaped like biconcave discs with a diameter of 6-8 μm and thickness of 2 μm with a cell volume of around 90 fL. Mature erythrocytes do not contain any nucleus, however during early phases they do. Humans have 4.2 – 6.1 x 106 erythrocytes/mL blood.

Leukocytes – also known as white blood cells, are a part of the body’s immune system. There are several types of leukocytes ranging in sizes of 7-21 μm. Of the different types of leukocytes most are nucleated, some have a multiple-lobed nucleus and some have no nucleus. Humans have 4-11 x 103 leukocytes/mL blood.

Thrombocytes – also known as platelets are involved in the coagulation of blood. They are irregularly shaped, 2-3 μm in diameter, with no nucleus. Humans have 2-5 x105 thrombocytes/mL blood. Bacteria are a large domain of prokaryotic microorganisms. They are typically a few microns in length and can have various shapes such as spheres, rods and spirals. Bacteria can be divided in two groups, gram positive

Jonas Hansson 5

and gram negative bacteria. Gram negative bacteria have a plasma membrane, a thin peptidoglycan layer and an outer membrane. Gram positive bacteria have a plasma membrane and a thick peptidoglycan layer. Bacterial strains causing sepsis include both gram positive and gram negative bacteria. The most common gram positive bacteria are Staphylococcus aureus and Streptococcus pneumoniae, while the most common gram negative bacteria are Escherichia coli, Klebsiella, Pseudomonas, and Enterobacter spp. The number of bacteria in a sepsis patient blood can vary greatly, cases of <1 CFU/ml (colony forming units per milliliter of blood) as well as over >1000 are found, but most common is somewhere in between these extremes (14).

As seen in table 1 there are several differences between the bacteria and the human blood cells. Some of these differences can be opportunities in the aim to isolate the bacteria from the rest of the blood while others will cause problems for bacteria isolation. One such problem is the very low number of bacteria compared to the high amount of blood cells. This means that: (i) several mL of blood have to be sampled in order to ensure enough bacteria in the sample and, (ii) it must be possible to detect very few bacteria. The detection can be facilitated either by the use of an amplification technique or an extremely sensitive sample preparation technique. If the subsequent analysis method depends on DNA sensing (such as PCR described in the Pathogen identification methods chapter) the leukocyte DNA present is a potential problem since it can act as a background noise to the bacterial DNA.

Three distinct opportunities for bacterial isolation can be identified: size and shape, membrane compositions and surface properties. Within this thesis size and shape differences will be utilized in a size based sorting technique (inertial microfluidics) and differences in themembrane composition will be utilized in selective cell lysis. Surface properties is outside the scope of this particular thesis.

2.2 Diagnostics

2.2.1 Standard diagnostics To handle large amount of blood samples generated in hospitals, routine procedures have been developed and are usually performed at specialized clinical laboratory departments. A typical routine procedure (15,16) is described below, specifically how it is performed at Karolinska Hospital in Solna, Stockholm, Sweden (see table 2 for overview).

The patient blood sample, typically 20-30 ml, is distributed in culture bottles containing various broth mediums. Aerobe and anaerobe culture bottles are always used. More specific cultures are also tested when e.g. fungicemia is suspected. The cultures are monitored for microbial growth by CO2 sensing. When a threshold value is reached the culture is considered positive. This takes 1-5 days for aerobic culture but up to 14 days for slow growing pathogens, after which the sample is considered negative.

The positive blood cultures are used to prepare gram stained samples for ocular inspection. Trained personnel check for properties such as: gram staining color, shape, and growth pattern. This information limits the possible bacteria species (providing input to the antibiotic treatment) and determines how to proceed in the identification procedure in terms of type of medium to use and which resistance tests should be performed.


Table 2. Example of routines in the clinical laboratory at Karolinska Hospital (17).

Jonas Hansson 7

Bacteria from the positive blood culture are then seeded on agar plates with several antibiotic patches for susceptibility testing. These are grown 24 hours for aerobic and 48 hours for anaerobic bacteria. For some suspected bacterial species additional methods are used. Minimum inhibitory concentration (MIC) is measured using a strip with an antibiotic concentration gradient on top of an agar plate. An automatic species and susceptibility test called VITEK is performed on certain types of bacteria. The bacteria sample is manually prepared and the automatic analysis is based on colorimetric measurements of bacterial growth in the presence of an array antibiotic screening creating a profile that is compared with a library. The VITEK analysis takes 5-10 hours. If Methicillin-resistant Staphylococcus aureus (MRSA) is suspected, a PCR-based method will be performed to identify MecA, the gene causing MRSA. This have to be done for reporting in accordance with the Swedish Communicable Diseases Act (18).

The total turnaround time to test results can vary from 2 to 16 days but normally 2-4 days. Even a hospital such as Karolinska Hospital in Solna is limited by trained laboratory personnel only being available during daytime which adds to the total turnaround time.

2.2.2 New diagnostics Emerging techniques to complement or improve the standard diagnostic routines are being developed for faster, more complete diagnosis. The technical development is mostly focused on the pathogen identification methods, but also on improving the sample preparation techniques. Below, new commercial kits and some promising research will be briefly described.

2.2.2.1 Pathogen identification methods Some of the most promising commercially available novel pathogen identification methods are based on polymerase chain reaction (PCR). PCR is a technique to amplify DNA sequences. Commercial kits are carefully designed to amplify the specific DNA sequences that can identify several bacterial species and antibiotic resistances. Detection of PCR products can be performed by e.g. fluorescence based DNA hybridization on a microarray (19), gel electrophoresis (20), real-time PCR (21) or even sequencing (22). Other methods have been used without the need for PCR, directly on positive blood cultures. Promising research for this is performed in areas such as bacteriophage based detection assays (23), proteomic based matrix-assisted laser desorbtion-ionization time of flight mass spectrometry (MALDI-TOF MS) (24), fluorescence in situ hybridization targeting naturally abundant ribosomal RNA (25), and DNA analysis using mass spectrometry (26).

2.2.2.2 Sample preparation When using PCR-based identification techniques, effective DNA purification is important. This process has three main challenges: (i) removing PCR inhibitors from the blood, (ii) removal of human background DNA and (iii) microbial DNA contamination (16). Removal of PCR inhibitors in commercial kits is usually performed while binding the DNA to a silica surface in a spin column format for easy washing. Removal of human DNA background can be dealt with by selectively dissolving (lysing) the blood cells and degrading the DNA before washing and lysing the bacteria as in the MolYsis kit (Molzym GmbH & Co. KG Bremen, Germany) (27). SIRS-Lab (Jena, Germany) removes background DNA by using a matrix that selectively


binds the pathogenic DNA, while the SeptiFast kit (Roche Molecular Systems, Branchburg, NJ) ignores the problem of background DNA. Contamination or carryover of DNA from other samples in the same batch is battled with with various cleanliness procedures (16). By performing the entire sample preparation in a separate enclosed system such as a LOC cross contamination could be decreased (28).

2.3 Microfluidics Microfluidics is the science and technology of systems that can handle small volumes of fluid in channels of ten to a few hundred microns in size (2). The most commercial successful application of microfluidic technology so far is the inkjet printhead. When applied to medical diagnostic applications microfluidics usually takes place on a lateral flow device or in a microchip where the fluids can be trasported, mixed, separated and processed in various ways. Microfluidic systems have the potential to offer automated analysis to a lower cost and in a shorter time compared to conventional technologies. Integration of both sample preparation and the sensor technology into one miniaturized system are referred to as µTAS (micro Total Analysis Systems) or LOC (Lab-on-a-chip).

Fluids in the microscale behave differently than we are used to in macroscale. In a microchannel ordinary water acts like a viscous fluid (such as syrup) would do in the macroscale – mixing two fluids becomes a challenge and interactions with solids are different compared to the macroscale.

Below the physical principles will be described as well as means of mixing, particle sorting, and how microfluidic systems can be fabricated.

2.3.1 Physics If a fluidic flow system is miniaturized by decreasing the length dimensions one order of magnitude (e.g. 1 cm becomes 1 mm), the surface area decreases two orders of magnitudes and the volume decreases three orders of magnitudes. The relation between these entities changes drastically e.g. the surface to volume ratio is increased which effects the physical properties. Physical effects dependent on length (e.g. diffusion), surface (e.g. friction), or volume (e.g. gravity) decreases accordingly. This means that friction will be more influential than gravity in miniaturized systems. In fluid mechanics, the dimensionless Reynolds number (Re = ρ v Dh/µ, where ρ is the density, v the flow velocity, µ the viscosity and Dh the characteristic length e.g. channel diameter in a circular channel) gives a measure of the ratio of the importance of inertial forces to viscous forces (29). High Re (>4000) will result in turbulent flows, low Re (<2300) will result in laminar flow, and for intermediate Re (2300<Re<4000) both laminar and turbulent flows are possible. In laminar flow layers of fluid with flow parallel to each other with no cross currents perpendicular to the direction of flow (the typical parabolic flow profile can be seem in figure 3). Typically Re in microfluidic systems are below 10.

2.3.1.1 Mixing As with any type of chemical interactions between fluids (or suspensions) lysing of blood cells require mixing of the blood with a lysis buffer. However, the laminar flow behavior in microfluidic systems makes proper mixing difficult. Since there is no chaotic mixing at these low Reynolds numbers, mixing is solely dependent on diffusion. The time for diffusive mixing grows with the square of the diffused distance. To decrease the

Jonas Hansson 9

diffusion length, various strategies for stretching and/or folding of the interface have been developed (30). To keep the blood and buffer mixing simple the method should ideally be passive (no external forces), provide effective mixing, and be able to manufacture with simple and reproducible techniques. The staggered herringbone structure fits that description (31). The herringbone mixer generates transverse flows by a steady axial pressure gradient induced by diagonal ridges along one of the channel walls (figure 2). This results in a steady chaotic flow where stretching and folding proceeds exponentially with travelled distance, hence effective mixing is achieved. Flows in staggered herringbone mixers also results in a decreased flow dispersion (31,32) making cell contact time with the lysis buffer more uniform.

Figure 2. Staggered herringbone mixer (31). (A) The mixing cycle composed of two sequential regions of ridges with switching asymmetries. The streamlines of the flow in the cross section are shown schematically above the channel. (B) Confocal micrographs of vertical cross sections of a channel as in (A). The folding of the fluid is illustrated by two streams of fluorescent solution injected on either side of a stream of clear solution. (Image reprinted with permission © AAAS)

2.3.1.2 Sorting Microfluidic size dependent sorting can be categorized in two groups, active and passive separation techniques. Active separation techniques use externally induced forces such as magnetic (33,34), dielectric (35), optical forces (36) or acoustic (37). Passive separation techniques rely purely on microfluidic phenomena and the fluid interaction with the geometries of the chip. Examples are deterministic lateral displacement (38), pinched flow fraction (39) and inertial microfluidics (40–42). Due to the high number of cells and large volume of sample that need to be processed in the case of sepsis diagnostic, unusually high volumetric throughput is required. Inertial microfluidics is a technique operating at high flowrates. The focusing mechanisms of inertial microfluidics are described below.

2.3.1.3 Inertial microfluidics Fluid pushed through a microchannel has the highest velocity in the middle of the channel and velocity close to zero along the channel walls. This is known as the parabolic flow profile (figure 3a). When a particle is introduced in this flow, the velocity of the fluid is different on each side of the particle. If this difference is large enough (dependent on the shape of the profile and size of the particle) the particle will start to rotate and be pushed due to a lift force, FLS, perpendicular to the direction of the flow, towards the channel wall (43). Close to the channel wall, an asymmetric wake will form around the particle resulting in an opposite lift force FLW, away from the channel wall. At a certain distance from the wall, equilibrium between FLS and FLW is reached, usually at around one fifth of the distance from the wall to the center of the channel. In a round


channel this will result in the particles focusing evenly along the wall at this equilibrium distance (figure 3b) (44). In a square channel four equilibrium positions (or attractors as Di Carlo describes them) will be present (45), one at the center along each channel wall (figure 3c). In a rectangular channel the probability of finding the particles at the shorter channel walls is very low resulting in only two attractors (figure 3d) (42).

Figure 3. Particle inertial focusing in flow through straight channels. (a) The shear induced lift force (FLS) and a wall induced lift force (FLW) acting on a particle flowing in a microchannel. Illustration of the cross-sectional equilibrium positions of particles flowing through different channel geometries: b) round, c) square, and d) rectangular.

For rectangular high-aspect ratio rectangular channel geometries (channel height >> width), the inertial forces push the particles over streamlines to two parallel streams (symmetrically located close to the side walls). This has previously been used to separate particles by collecting the focused particles through two side outlets (42,46). The possibility of using low aspect ratio (channel height << width), which enables collection of the focused particles in a single middle outlet, have been examined in Paper II and Paper III.

To further increase the throughput of inertial microfluidics, a rational approach is to use several channels in parallel. As the number of outlets grows with increased parallelization, sample collection and integration for downstream processing becomes cumbersome. To collect each fraction in a single dedicated outlet the channels will have to cross each other without mixing. This requires 3D microfluidic devices.

2.3.2 Microfabrication There are several techniques to manufacture microfluidic devices, including micromachining, soft lithography, hot embossing and injection molding (47).

Micromachining uses processes from the microelectronics industry such as photolithography based etching and wafer bonding (48). The first microfluidic chips were fabricated this way and were, just like electronic microchips, silicon based (49). The slow fabrication process, the need for a clean room facility and the limited imaging possibilities due to the opaque silicon are some of the reasons that most microfluidic chips nowadays are made out of polymers instead of silicon, both within academic research and commercial products.

Soft lithography is by far the most common polymer fabrication process for academic research. Its popularity is partly due to the low initial costs, short turnaround time and a wide variety of on-chip functionalities available. An elastomer is poured over a replication master, polymerized by heating or UV-light, removed from the master and bonded to a substrate to close the system (50) (this process is described in more detail in the methods section Traditional soft lithography below. The process is simple and has very good replication accuracy (47).

Jonas Hansson 11

Hot embossing has become more common over the last 15 years. It involves heating a thermoplastic sheet just above its glass transition temperature under vacuum and pressing it against a replication master to imprint microstructures. The process requires optimizations for each new pattern and requires some investments, but due to the short cycle time (4-15 min / wafer) it is a simple and cost effective alternative to microinjection molding (47). It has a large range of suitable materials but has problems with high aspect ratio features and can only replicate planar features.

Injection molding is the dominant replication process for plastics in general and most commercial microfluidic devices are manufactured this way. The high machine costs (>$75k), maintenance costs and tedious optimization process for each structure makes injection molding unsuitable for low volume production (e.g. for research) but the short molding time (30s to 2 min) makes it cost effective for large volume production (47). The process involves melting thermoplastic pellets and injecting the molten polymer into a cool, closed replication tool at high pressure.

Generally, soft lithography is used in the development phases, injection molding is used for mass production, and hot embossing could be used for both. The work in this thesis is in research stage and requires 3D microfluidic devices, making soft lithography the most suited manufacturing technique.

2.3.2.1 PDMS The most common elastomer used with soft lithography is polydimethylsiloxane (PDMS) (47). It is elastic, has excellent optical properties, resistant to acid and bases but not to solvents, is permeable to gases and to small molecules but not to water and it has advantageous surface chemistry for many applications. Multiple layers of PDMS channels enables mixers (31,51), valves (52), pumps (53) etc. to be built. Exceptionally high gas permeability properties are obtained with thin PDMS membranes which have been used in a wide range of applications such as separation devices (54), perfusion of live cells, gas sensors, fuel cells, bubble removal (55,56), and liquid pumping (57). Especially relevant in this thesis project is the bubble removal method developed to increase the performance of rapid, miniaturized PCR (55) and the possibility to integrate sequential microfluidic sample preparation. As a part of the fabrication process, the thin PDMS membranes typically have to be transferred from the replication master to a destination substrate. As thin PDMS membranes are fragile and easily ruptured the transfer and release processes have to be especially gentle. These challenges are addressed in the manufacturing techniques developed in Paper I. One of the techniques tested is using PVA as a release layer as described below.

2.3.2.2 PVA as release layer Materials that dissolve upon contact with suitable solvent have previously been used to fabricate fragile structures in traditional micromachining fabrication techniques both as sacrificial layer and release layer. Poly(vinyl alcohol) (PVA), a water soluble polymer that is resistant to solvents and has a large surface concentration of surface hydroxyl groups has previously been used as a release layer (58). This concept, illustrated in figure 4, is utilized in Paper I with PVA as a transfer carrier for gentle release.


Figure 4. Principle of using water soluble PVA as a transfer carrier for gentle release.

2.3.2.3 Vias production To create a high throughput, size based sorting device that is easy to integrate, 3D fabrication is necessary as stated previously. 3D fabrication techniques are also needed for many microfluidic functionalities such as mixers (30), valves (52), and pumps (53). A prerequisite to be able to build these complex multilayer structures is the ability to fabricate vertical interconnections (vias). Traditionally, vias in PDMS are fabricated manually by punching the holes with a needle. In order to automate the process and enable thinner, more densely spaced vias to be fabricated, the vias could be defined by taller structures in the replication master. To form holes at these taller structures the viscous prepolymer surface has in previous attempts been kept below the tall vias defining structures by various techniques. Attempts to remove excess prepolymer by blowing (59) or spinning (60) both result in an uneven top surface unsuitable for bonding to a second layer. Pressing a glass plate against the tall structures to squeeze away excess prepolymer (61) results in thin PDMS films being formed at the vias positions (even when using high pressure) which has to be removed manually (62). By immobilizing an amino silane on the glass plate that inhibits PDMS polymerization these thin films can be avoided (62). This method visualized in figure 5 is effective in creating densely spaced vias on areas <25mm2. At larger surfaces >250mm2 (such as the high throughput sorting devices within this thesis) the demoulding of carrier-PDMS stack from the replication master becomes challenging due to increased adhesion between the PDMS and the replication master as first encountered in Paper II. To solve this problem, techniques using flexible inhibition plates are developed in Paper II and III, while several solutions are tested in Paper I.

Figure 5. Principle of the inhibition process to produce through-hole vias in PDMS, as described by Carlborg et al. (62). (1) The PDMS prepolymer is casted on a mould and cover with an inhibition plate. During curing (2) Pt-catalysts binds to the amines which inhibits the PDMS polymerization between the inhibition plate and the tall mould structures while freely diffusing Pt catalyzes the polymerization in the rest of the PDMS. The cured PDMS is then transferred (3) to a destination substrate. As the inhibition plate is released (4) the uncured PDMS sticks to the inhibition plate creating open vias.

Jonas Hansson 13

2.4 Cell lysis Lysis is the destruction or dissolution of cells such as blood cells or bacteria. There are several mechanisms that can cause lysis e.g. osmotic shock, viral, enzymatic, chemical, abnormal temperatures, and mechanical disruption.

The bacterial cell wall consisting of a peptidoglycan layer makes bacteria more resistant to most kinds of lysis mechanisms compared to mammalian cells. We want to exploit that by finding a lysis method that will lyse the blood cells while keeping the bacteria intact. The method should be possible to integrate in a microfluidic device and not have any moving parts, temperature controllers, or other advanced functionality that increases the cost. This disqualifies temperature based and mechanical disruption. Viral and enzymatic lysis can be very sensitive to temperature and are both expensive alternatives reagent wise. Osmotic shock and some chemicals should potentially have the ability to selectively lyse blood cells while keeping the bacterial cells intact as explained below.

2.4.1 Osmotic shock The cell membrane is a semipermeable membrane selectively permeable to ions and organic molecules. Water can more rapidly cross the membrane through specialized channels called aquaporins (63,64). When the concentration of solutes (osmolality) such as ions and organic molecules are lower outside the cell than inside the cell (hyperosmolarity) we get an osmotic pressure, forcing the water into the cell in order to compensate the pressure. This leads to cell swelling and eventually, if the outside osmolality is sufficiently low, cell burst.

A mammalian cellular response to this can happen within seconds trying to maintain the concentrations to be compatible with cell function (65). Many bacteria, as opposed to mammalian blood cells are directly exposed to an uncontrolled environment which hyperosmolarity is a natural part of. Bacteria have developed a more intricate way of responding to hyperosmolarity e.g. metabolic reprogramming increasing their resistance (66).

2.4.2 Detergents Detergents are amphipathic consisting of both a hydrophilic group and a hydrophobic group (67). The hydrophilic head groups interact with the hydrogen bonds of the water molecules making it possible to dissolve other substances that would not normally dissolve in water. There is a large number of different detergents available (68). They can be classified based on the properties of their hydrophilic head group: non-ionic, ionic or zwitterionic. Non-ionic detergents are gentle and have the ability to break lipid-lipid and lipid-protein interactions but only limited ability to break protein-protein interactions. Ionic detergents can either be positively or negatively charged. They are strong detergents and are used for complete disruption of cellular structures and denaturation of proteins. Zwitterionic detergents have a net zero charge arising from an equal amount of negative and positive charges. They are used for their ability to protect the native state of proteins and prevent non-specific aggregation.

To be able to differentially lyse blood cells while keeping the bacteria alive we have opted for non-ionic detergents.


Triton X-100 and Tween 20 are both non-ionic detergents commonly used for solubilizing membrane proteins and have been used in DNA extraction as part of lysis buffers.

Saponin is a group of non-ionic detergents that are produced by extraction from different kind of plants. They have been used to lyse both erythrocytes (69) and leukocytes (70). When added to bacteria culturing broth, saponins from Quillaja bark has been shown to increase the growth of Escherichia coli in concentrations from 0.1-1% depending on manufacturer (71). The exact mechanism of cell lysis by saponins is not completely understood but they are known to interact with cholesterol in the cell membrane and make it permeable to macromolecules (72). One explanation is the formation of cholesterol-saponin complexes to pores, another is that they affect transport through the aquaporins. The increase in transport could influence the effect of an osmotic shock, which tested in Paper IV.

2.4.3 Selective cell lysis on chip The principle of selectively lysing certain cell types while keeping other cells intact has been used previously in microfluidic devices for leukocyte isolation from whole blood. In a chip using the staggered herringbone mixer (73) ammonium chloride mediated lysis was used to lyse all blood cells but not leukocytes. Pure DI-water based lysis was used to avoid leukocyte activation (74), a method later implemented in clinical settings obtaining similar or better results than traditional methods (75). Parichehreh (76) evaluated erythrocyte and leukocyte response to DI-water confirming erythrocyte depletion within 15 seconds. SooHoo (77) introduces a microfluidic cytometer for the characterization of cell lysis with the commercial lysis reagent Zap-OGLOBIN II developed specifically for erythrocyte lysis. They conclude it can achieve complete lysis in 0.7s under optimal conditions.

Jonas Hansson 15

3 Procedures and Methods This chapter describes the practical work done in the studies. Traditional device manufacturing are described, novel fabrication methods developed within this thesis are also described as well as methods to inspect the quality of the fabricated devices and evaluate their performance when tested on biological or artificial samples.

3.1 Traditional soft lithography Traditional soft lithography starts with a replication master being fabricated on which a polymer is cast, followed by assembly to produce a microfluidic chip. The steps of traditional soft lithography are described in more detail below.

Replication master manufacturing is done in several steps as illustrated in figure 6 (steps A1 to A5). A design is created in a CAD or graphics software, its negative is printed on transparent glass or plastic film to make a mask (A1). The height of the structures is defined by spincoating a negative photoresist (usually SU8) onto a silicon wafer to the desired thickness (A2). The photoresist film is densified by heating the wafer to evaporate the solvents. To define the structures the photoresist is selectively polymerized by UV-exposure through the mask (A3). For more complex structures, spinning and UV exposure (steps A2 and A3) can be repeated multiple times using different masks each time to add more layers of structures. The unpolymerized photoresist is removed in the developing step by dissolving it in a bath of developer solvent (A4). For improved mechanical properties the master can be heated again to further cross-linking of the photoresist. For easy demolding after casting the structures can be coated with a fluorocarbon film by C4F8 plasma deposition.

Figure 6. The replication master manufacturing process. (1) An example of a mask design negative. (2) Spincoating of SU-8 to desired thickness on a silicon wafer, followed by heating. (3) UV exposure of SU-8 through mask. (4) Developing of SU-8. (5) Finished replication master mould.

Casting and assembly (seen in figure 7, steps B1 to B7) starts by mixing a PDMS base and a curing agent to create a prepolymer that is poured onto the replication master (B1 and B2). Bubbles in the prepolymer are removed by placing it in a low pressure chamber. The PDMS is then polymerized by heating to 60-90 degrees C for a few hours (B2). The PDMS replica is demoulded (B3) and the casting process is finished. This can be repeated to create more devices. The back end processing starts by punching holes to create in- and outlet ports (B4). The system is closed by bonding the PDMS to a polymer or glass substrate by either clamping them or by O2-plasma (B5) creating a permanent bond (B6). If multiple devices are fabricated in the same


PDMS sheets they have to be diced, which can either be done before or after bonding. Finally tubing is connected to the in- and outlet ports (B7).

Figure 7. Traditional casting and assembly of microfluidic device. (1) Replication master mould. (2) PDMS poured onto the mould, cured by heating (70oC, >3 h), and (3) removed from mould. (4) manual punching of in- and outlets. (5) Oxygen plasma treatment of PDMS structure and destination surface if permanent bond (6) is desired. (7) Finished device with tube connected to in- and outlet ports.

3.2 Novel fabrication method The manufacturing procedure developed in this work for devices containing thin membranes and densely spaced vias are described here and illustrated in figure 8, steps C1 to C13.

PVA dissolved in water is spin-coated onto a PC sheet (C1) and the water is evaporated by heating (C2). An amine layer is created by submerging the PVA-PC sheet in a methanol solution of 3% AEAPS (aminoethyl-aminopropyltrimethoxysilane) and 1% MEMO (3-methacryloxypropyltrimethoxysilane) (C3), the access silane is washed off with methanol and the silanes are bonded by heating (C4), finalizing the transfer carrier. PDMS prepolymer is poured onto a replication master (C5), the transfer carrier pressed onto the PDMS prepolymer (C6) and polymerized by heating (C7). The flexibility of the transfer carrier is utilized when releasing the PC-PVA-PDMS stack from the replication master (C8). Oxygen plasma bonding is performed for a permanent bond between the PDMS and the glass slide (C9). Both the flexibility of the PC and the water solubility of the PVA are utilized for gentle removal of the PC sheet (C10) forming vias and non-ruptured membranes. A second, top, PDMS structure is plasma treated (C11), manually aligned and bonded on top of the first PDMS structure (C12) creating a 3D device containing both 3D structured channels and integrated membranes (C13).

The large high throughput particle filtration devices in Paper II and Paper III contain densely spaced vias but no thin PDMS membranes. For this purpose a slightly altered manufacturing process to make it more straightforward and faster was developed. No PVA was used making step C1 and C2 obsolete. In step C3 only one silane, AEAPS, was used. Step C10 was performed using simple dry peeling with no need to have water present.

Jonas Hansson 17

Figure 8. Process flow most suited for the manufacturing of PDMS structures containing both thin membranes and densely spaced vias. Steps marked with (*) can be excluded for faster vias manufacturing and (**) can be simplified.

3.3 Florescence microscopy Fluorescence microscopy is a widely used method within biological research. Usually a fluorescent dye is used to enhance part of the sample that is of specific interest, enabling it to be distinguished from the background. A fluorescent dye absorbs light of certain wavelengths and emits light of a longer wavelength. By the use of filters the emitted light can be imaged separately. A composite multicolored image can be reconstructed from several images from various colored dyes in addition to the normal brightfield image.

In Paper I, II, and III normal optical microscopy is used to inspect the produced structures and in Paper IV to count cells. Fluorescence microscopy is used in Paper II and III to image streams of particles and in Paper IV to evaluate mixing.

3.4 Coulter counter Coulter counter is an apparatus that measures the size and concentrations of particles suspended in an electrolyte. The technique is based on a microchannel or small aperture that the particle containing fluid is pumped through. The conductivity in the aperture is measured with an electrode on each side of the aperture.


As a particle passes through the aperture the conductivity changes as the non-conductive particle displaces the conductive electrolyte. The change in conductivity is proportional to the volume of the particle and the concentration of particles can be measured by pumping a specific volume while counting the number of particle signals. The suspension has to be diluted enough for the particles to pass the aperture one at a time. Cells are, due to their membrane potential, non-conducting particles and can therefore be counted by the coulter counter. One important application of the coulter counter is quick and accurate analysis complete blood count (concentration of erythrocytes, leukocytes and thrombocytes). This is a measurement procedure where the blood is diluted in different solutions in to reach measureable cell concentrations for respective cell type and remove the numerous erythrocytes when thrombocytes and leukocytes are to be measured.

Coulter counter is used in Paper II and III to count florescent particles and within the scope of Paper IV to count blood cells. The size of the aperture in the coulter counter available within this project is too large to be able to count cells smaller than spheres of 2 µm in diameter, making bacteria counting impossible.

3.5 Bacterial culture Bacterial culture is a method of growing bacteria by letting them reproduce in a culture media under controlled conditions. The culture can either be grown in liquid media or on the surface of a thin gelatinized agar-based medium, usually in a petri dish. There are numerous media types to culture different species of bacteria. The most commonly used bacteria is Escherichia coli which is classified into several different strains, some fast growing and non-pathogenic being very suitable for research. In these early development phases of methods to isolate bacteria from whole blood, working with clinical samples (from patients) is not desirable due to the highly variable and usually low number of bacteria in the samples. Blood from healthy donors spiked with bacteria culture provides a controlled and repeatable test sample. To evaluate the bacterial survival a specified volume of a treated sample is spread evenly on an agar medium petri dish and grown under controlled conditions. Each surviving bacteria will form a colony on the agar and can be counted in order to calculate the survival rate of the bacteria. The number of bacteria is then measured in colony forming units (CFU).

Jonas Hansson 19

4 Results

4.1 Fabrication results

4.1.1 Fabrication method development In order to enable manufacturing of both vias connection and thin PDMS membranes over large surface areas in 3D microfluidic devices, three prominent features were added to traditional soft lithography in this thesis: (i) controlled PDMS inhibition, (ii) flexible transfer substrate, and (iii) transfer substrate release by dissolution.

The inhibition level was controlled by diluting the Pt-binding silane AEAPS (aminoethylaminopropyltri-methoxysilane) with a silane without inhibitory function, MEMO (3-methacryloxypropyltrimethoxysilane), creating a lower density of Pt-binders on the surface of the transfer substrate. The effect of the silane mixture was tested by using different ratios of the two silanes. A correct level of inhibition to manufacture both densely spaced vias and thin membranes on the same chip was found at a surface activation mixture containing 1% AEAPS and 3% MEMO diluted in methanol. If only densely spaced vias are needed it was possible to use only AEAPS in methanol without any MEMO.

The flexible substrate tested was a 250 µm thick poly carbonate (PC) sheet. Its effect was tested by comparing it with thick PC sheets and glass plates. The flexibility of the transfer carrier was found to be necessary for proper demoulding as the stiff glass and PC carriers were difficult to remove, practically impossible for large area devices without breaking the replication master and/or the PDMS structures (in step C8, figure 8). The flexible 250 µm PC sheet could be removed by peeling resulting in successful low stress demoulding.

For transfer substrate release by dissolution, a layer of the water soluble of poly (vinyl alcohol) (PVA) was applied between the transfer substrate and the inhibition layer. Two different PVA sheet manufacturing techniques were tested: spin coating of PVA solution onto carrier producing a thin PVA layer, and lamination of PVA sheet onto carrier. The effect of the PVA dissolution release was tested by using three release techniques: submersion into water, peeling the carrier in the presence of water and peeling the carrier without the presence of water. It was found that the manufacturing of the transfer carrier was best done by spincoating the PVA. The release of transfer carrier from the PDMS was best done by peeling in the presence of water. If only densely spaced vias are needed, the PVA was redundant and instead using only the flexible PC sheet and dry peeling for efficient release.

4.1.2 Fabricated devices The manufacturing techniques developed were further tested by using them to manufacture two types of functional devices.

Devices for µPCR that includes bubble removal functionality and components for delivering reagents and sample was produced using the novel manufacturing technique for both thin PDMS membranes and densely spaced vias (figure 9). Thin PDMS membranes were successfully produced and their functionality has been


tested in another study (55). Vias enabled the use of integrated pneumatic valves that can be seen in figure 9 c-g.

Figure 9. PDMS device containing both thin membranes and densely spaced vias showing (a) the result after tuning the degree of inhibition to allow simultaneous fabrication of (b) through-hole vias and (c) a 35 μm thin, 1 cm2 large area membrane. This structure was assembled with a second layer, containing structures for pneumatic valving, shown in (d). (e) Microfluidic channels were visualized using dyed water introduced via the tubes. (f-g) The functionality of the fluidic connections between the layers was investigated using the integrated pneumatic valves.

Devices for high throughput particle sorting were manufactured using the simple and fast vias manufacturing technique. Devices using 4 and 16 channels in parallel were successfully produced; both with vias yield of 100% (figure 10). The inter layer bonds were capable of handling the pressure used at high volumetric throughput particle processing. The particle processing capabilities of these chips are further described below in the filtration and sorting section below.

Figure 10. Large area PDMS device containing densely spaced vias, designed for high throughput particle filtration. (a) A close-up photo of the bottom layer of a 16-parallel-channel PDMS device with open vias. (b) A 16-parallel channel, two-layer PDMS, interconnected through vias. The top PDMS layer is aligned and bonded onto the bottom PDMS layer. (c) A 4-parallel-channel, two-layer PDMS device showing the 3D-fluidic network leading to the two separate outlets. (d) A 16-parallel-channel filtration device filled with red dye. The distance from the inlet to the outlets is 32 mm and the total footprint area of the device is less than 550 mm2.

Jonas Hansson 21

4.2 Inertial microfluidics

4.2.1 Focusing Particles flowing in channels with intermediate aspect ratios were found to either focus in 2 cross sectional positions, 4 cross-sectional positions, or not at all, dependent on particle size (figure 11).

Figure 11. Fluorescence images of 2 μm (left), 10 μm (middle) and 24 μm (right) particles flowing at a flow rate of 200 μl/min in a channel with cross-section 70x50 µm the first 10 mm (from the inlet to the middle), corresponding Re=56. At the outlet the channel was wider for increased spatial resolution. The schematic cross-sections show the particle positions at the outlet of the channels. The 2 μm particles remain unfocused, while the 10 μm particles were focused into 4 streams and the 24 μm particles into two streams.

The focusing dependency to Re, using different flowrates while keeping the particle size constant, in channels of low aspect ratios (a.r.) , was studied at several positions along the channel. The results indicated that the number of focusing positions is also dependent on flow rate and can change along the channel length (figure 12).

Figure 12. Fluorescence images of 8 μm particles taken at different channel lengths (1-30 mm) in a channel with cross-section 80x30 µm under flowrates corresponding to Re of 30, 60, and 120. For Re=30 the particles focus to one 2 particle streams at approximately 20 mm, for Re=60 the particles are focused to 4 particle streams at approximately 15 mm and to 2 particle streams at approximately 25 mm, and for Re=120 the particles are focused to 4 particle streams at approximately 20 mm but are not able to focus into 2 particle streams within 30mm.

In figure 13a the intensity profile of three different particle sizes in figure 12 channels is plotted. The levels of focusing, measured as the fraction of particles located in the middle third of the channel, are calculated from


these types of plots and illustrated in figure 13b, as a function of time. This illustrates that particle migration towards the middle of the channel is faster at higher flowrates.

Figure 13. Analysis of focusing experiments from channel with cross-section 80x30 µm. (a) Fluorescence intensity profile of 5, 8, and 10 µm particles, baseline corrected and normalized with respect to total fluorescence and channel width. (b) Level of focusing, measured as the fraction of particles located in the middle third of the channel, potted against time for three different Re (based on images from figure 12).

4.2.2 Filtration and sorting The 4- and 16-paralell channel 3D devices were designed with each individual channel having a cross-section of 30x80 µm2 (tapering to 30x160 in the end for increased spatial resolution) and a total length of 20 mm. When a suspension mix with particles of 2 and 10 µm in diameter were pushed through the chips, the 2 µm particles remained unfocused while the 10 µm particles were focused in 2 positions in the middle of the channels, exiting through the vias to the top PDMS layer, and collected into outlet 1 (figure 14).

Figure 14. Particle filtration in parallel channel 3D devices. a) Fluorescence image of sorting 10 µm fluorescent particles in 4-parallel channel device. b) Filtration efficiency in 4-parallel channel device and in 16-parallel channel devices for 10 µm fluorescent particles.

The side outlets were reconnected in the bottom PDMS layer and collected in outlet 2. The flow resistances of the outlet channels were designed to collect a less volume from the middle fraction where the larger channel exits. For the 4-channel devices 25% of the volume was collected in outlet 1 and for the 16-channel devices 28% was collected. The 2 µm particles being evenly distributed in the channels had the same fractions as the rest of the fluid (25%-28% in outlet 1, 73-75% in outlet 2). The 10 µm particles, 95-97% could be

Jonas Hansson 23

fractionated through outlet 1 (figure 14). The flowrates used, optimized for focusing of 10 µm particles, were 0.2 mL/min per channel, corresponding to 0.8 mL/min for the 4-channel devices and 3.2 mL/min for the 16-channel devices. Additional flowrates in the range of 0.8-4 mL/min were tested a 16-channel device, all with filtering efficiencies >90%. Tests in single channels above these Reynolds numbers showed a decreased level of focusing.

4.3 Selective cell lysis A method for selectively lysing blood cells while keeping the bacteria intact was developed and implemented into a microfluidic device. Initially, bulk experiments were conducted to evaluate different lysis solutions to accomplish selective lysis of blood without lysing the bacteria. Based on the bulk experiments, saponin combined with osmotic shock was chosen for adaptation into a microfluidic system. A microfluidic device was designed for this two-step lysis procedure using a meander channel design with staggered herringbone structures on top, all fabricated in PDMS (figure 15a). The device was characterized for mixing by introducing fluorescein and blood cells and imaged by fluorescence microscopy. Complete mixing was achieved within 5 loops which correspond to 20% of the on-chip incubation time which can be controlled by varying the flowrate.

Figure 15. Microfluidic selective cell lysis. (a) Microfluidic chip filled with red and green dye (top, scale bar is 5 mm) and its mixing performance (bottom, scale bar is 250 µm) tested by flowing fluorescein and blood sample in the chip and fluorescence imaging at (from left to right) the blood inlet, 1st turn, 3rd turn, 5th turn, and 20th turn showing complete mixing at 5th turn. (b) Bacterial survival measured by bacterial plating survival following microfluidic lysis treatment of bacteria spiked blood using 2,5% Saponin treatment for 1 min and osmotic shock, only osmotic shock, and PBS as reference (no lysis).(c) Blood survival imaged by confocal microscopy of calcein stained lysates from experiment stated in (b). Scale bars are 20 µm.

The performance of the microfluidic selective cell lysis method is shown in figure 15 (b and c) where PBS control (no lysis), only osmotic shock (no saponin), and saponin and osmotic shock are compared. Both


osmotic shock and saponin treatment show >80% bacteria recovery while saponin is needed to completely lyse the blood cells. Long time survival was tested by further incubating the lysates in room temperature for 30 min showed bacterial survival >70%. Coulter counter analysis of the on-chip treated lysates indicated large amounts of cell debris in around 2 µm in size.

Jonas Hansson 25

5 Discussion

5.1 Fabrication The fabrication method presented in here represents the forefront in PDMS fabrication of thin structures combined with 3D fluidics networks, and constitutes a large step in the direction of making PDMS fabrication of complex microfluidic devices a routine pursuit. The method using PVA was also tested for whole wafer production, mostly with low yields as a result. This was especially true for manufacturing of vias structures as they result in an even inhibition plate, which is difficult to achieve over a large area for PVA. If whole wafer production of densely spaced vias is desired the method used in Paper I and II would be better suited as it doesn’t contain any PVA.

Although whole wafer production is highly desired in traditional microfabrication, it is less important for PDMS devices since the most common way to scale up fabrication of is by switching to injection moulding or hot embossing. Realistically this manufacturing technique would be best suited for application when either (i) semipermeable membranes are needed (such as degassing, PCR, cell cultures, lab-on-chip gas toxicity assays) or (ii) low volume production of thin membranes and/or 3D-devices (for research and development). For applications where the permeability of thin PDMS membranes is utilized, changing fabrication process would actually not be possible. In these cases the method of laminated PVA structures becomes useful as it resulted in successful whole wafer production of thin PDMS membranes.

The stickiness of the inhibited PDMS surface observed by Carlborg et. al. (62) and causing clogging in Paper II was not observed in the later development in Paper I and III. In the case of Paper I this is probably due to the controlled level of inhibition provided by using the inactive amine MEMO, while in Paper III, it is purely a case of parameter optimization. Specifically, an increased baking time and a decreased time from inhibition plate application to heat curing of PDMS.

5.2 Inertial microfluidics

5.2.1 Focusing In the particle focusing experiments, basic observations confirm previous work. We observe that small particles (a<0,1*Dh) (40) will not be focused and as particle size increases the focusing grows stronger since the lift forces are dependent on a4.

Other observations lead us to new discoveries. Previously 4 focusing positions were only found in square channels (40,45). In this thesis, however, we found that they can be found in channels of intermediate aspect ratio as well as in low aspect ratio channels. The focusing of particles previously only attributed to channel geometry (40,45,46) is in this work shown to also depend on particle size and flow rate (Re) and can also change along the length of the channel. The larger particle sizes are more prone to focus in two positions. Higher flow lead to focusing being less developed with respect to travelled distance (channel length) but more developed with time. Whether or not the higher flowrates (Re=120) in figure 3b will eventually be developed is only possible to speculate in. We do see an increased level of focusing along the channel (L). Either the


particles will (just like in the case of lower Re) eventually focus at the 2 attractors at the longer channel walls or the increased Re give rise to a shift in focusing properties making the particle streams locked at all the 4 attractors. At least in the latter case this would open up new opportunities for improved sized based particle sorting.

When designing a filtration or sorting device based on inertial focusing in straight channels several factor have to be considered. First, the size of the particles that we wish to focus will determine the possible cross sectional geometry of the channel. If a specific sized particle should avoid focusing the cross-sectional geometry has to be adjusted accordingly. For the desired high throughout, the working flow rate should be high. As our study suggests a higher flow rate would need a longer channel to accommodate enough flow time for sufficient focusing (this would be possible at least up to a certain flow rate). In extreme this will lead to large chips that will have to withstand high pressures due to the large pressure drops. Manufacturing techniques that can withstand high pressures as well as stiffer polymer materials should be taken into consideration.

5.2.2 Filtration The development of the multiple parallel channel filtering devices has been gradual. The first version, presented in Paper II showed a filtering efficiency of 82% for 24 µm particles filtered from 2 µm particles using a flowrate of 0.8 mL/min (4 parallel channels). The latest version presented in Paper III showed a higher filtering efficiency (>95%), for particles with smaller size difference (10 µm particles filtered from 2 µm particles), and uses a higher total flowrates (3.2 mL/min in 16 parallel channels). Focusing of smaller, 10 µm particles in 2 positions was enabled by a change in channel geometry from 50x70 to 30x80 µm2 resulting in a smaller Dh. The channel geometry is also one of the reasons for the improved filtering efficiency. Another is more optimized manufacturing parameters providing less sticky PDMS, therefore reducing problems with clogging that was present in the first version. The increased flowrates was obviously a result of further parallelization to 16 channels which is enabled by the improved fabrication technology.

This filtration device can be compared with previously published filtration devices. The only microfluidic separation or filtration devices capable of handling similar throughputs so far are the ones also based on inertial microfluidics, either in curved or straight channels. A device presented by Mach et al (42) also used multiple straight channels, but in a ring design. While using similar flowrates (also 0.2 mL/min in each channel) the footprint were larger and used a ring design not lending it for simple further scaling. It also relies on manual punching of the many outlets, and could therefore not be compared with our devices in terms of integration level provided by our manufacturing technology that allow integration with other lab on chip components. Spiral shaped channels can in a single spiral focus particles with flowrates in the mL range (41,78). The architecture however is not easily adoptable for parallelization and the focusing conditions are much more sensitive to flow fluctuations. The robustness to variations of flowrates reported in Paper III makes this device less sensitive to manufacturing imperfections, clogging and less dependent on an expensive pumping mechanism.

We use polymer microspheres in these filtration studies as a simple substitute for cells. The validity of polystyrene particles as a model for cells has previously been discussed (42,79). It has been shown that flexible

Jonas Hansson 27

cells, non-rigid particles (79), and non-spherical cells (erythrocytes) (42) focus closer to the channel centerline in the lateral position. Since our device divide the collected fractions cross-laterally this difference should not have an effect in biological samples. The cross lateral migration of non-spherical and non-rigid particles or cells have not been studied to the same extent but has indeed been shown to be able to focus in two positions (42) similar to the particles in this device.

The device can therefore with slight alterations of channel geometry be adopted for specific biological applications. In sepsis the larger blood cells could be removed while keeping bacteria and smaller cells unfocused. For subsequent PCR analysis the most crucial cells to remove are leukocytes (since they contain contaminating DNA). As leukocytes are the largest blood cells they would also be most straightforward to remove this way. Another potential application is isolation of circulating tumor cells which are larger than normal blood cells.

5.3 Selective cell lysis The use of bacteria spiked blood samples as a model for patient samples is a practical solution to achieve detectable amounts of bacteria in blood while developing this method. The specific use of Echerechia coli (strain BL21-A1) was not only a practical choice but it is also a sensitive bacteria, being a gram negative strain it has only a thin cell wall. Thus, if we manage to keep this alive after blood lysis treatment, most other types of bacteria would also be able to survive the same conditions. That being said, more bacterial strains will still have to be tested to further develop and test this technique.

While being a very fast technique when used on macroscale samples, the developed method is still limited by a low throughput when implemented in a microfluidic chip. Possible solutions here are to increase the channel geometries, use a parallelization approach and increase the flowrate. A decrease in saponin exposure time could have the benefit of both increasing throughput and increasing the bacterial survival.

The fact that only ~10% (~80% minus >70%) of the bacteria is lysed during the 30 min of exposure to low osmolality and ~1% of saponin suggests that this method could be integrated for further downstream processing without the need for an additional step to dilute the saponin and bringing the osmolality back to normal avoiding further bacterial dilution. The output lysate consists of 25% original whole blood is a comparably low dilution both compared to other microfluidic techniques and macroscale kits (16,74,80).

5.4 Integration The development within the three areas manufacturing, size based sorting, and selective cell lysis is not only connected by its purpose, sepsis diagnostics, but also in the sense that they could fit together in an integrated device.

Selective cell lysis and inertial microfluidics was first envisioned to be integrated by lysing of blood cells followed by inertial focusing of bacteria for up-concentration. While this still is a viable alternative it would require further improvement to decrease the size of the blood cell debris.


A more attractive alternative is to start by inertial filtration to remove blood cells (most importantly leukocytes) to remove human DNA contamination followed by either selective lysis if intact bacteria are needed for analysis, or by lysis and DNA extraction for subsequent nucleic acid based analysis.

The integration with these and other microfluidic components such as DNA extraction, µPCR, and DNA sensing for either a complete LOC system or a standalone automated sample preparation device could potentially be facilitated by the novel fabrication technologies developed within this thesis.

Jonas Hansson 29

6 Conclusions Within this thesis, I describe how we have:

Pushed the boundaries in manufacturing of complex PDMS devices by developing a fabrication technique that can manufacture both thin PDMS membranes and dense 3D structures within the same device.

Deepened our understanding of the inertial particle focusing phenomena by finding that the number of inertial focusing positions of particles are not only dependent on channel geometry, but also particle size, flowrate, and channel length.

Developed a particle-size filtration device for high throughput (3.2 mL/min) with high filtering efficiency (>95%). The device is robust in performance and is easy to integrate with other lab-on-a-chip components.

Developed a microfluidic device capable of selectively lysing blood cells while keeping >80% of the bacteria alive using a low dilution, high integration method

These novel findings, devices, and methods together have the potential to decrease the time of sepsis diagnostics. They can be incorporated in an automated sample preparation system for culture-free sepsis diagnostics, with subsequent pathogen identification, together forming a LOC system.

Jonas Hansson 31

7 Future studies The findings in this thesis give direction for future studies, especially in the areas of:

Further developing manufacturing techniques of microfluidic components specifically in the areas of high pressure resistant, rigid polymers and techniques for smother transitions towards mass-production.

Further investigations of cross-lateral inertial migration of particles and cells, especially in channels of intermediate aspect ratios.

Improving selective cell lysis by decreasing lysis time and increasing throughput and bacterial survival. Integration of several microfluidic blood sample preparation components in sequence or parallel.

Jonas Hansson 33

8 Acknowledgements The work within this thesis was funded by FP7 project InTopSens and IMI RAPP-ID. It would not be possible without the remarkable people around me. I would therefore like to give special attention to some of these people:

My supervisor Hjalmar for his scientific expertise and good support.

My co-supervisor Aman for his enthusiasm and great courage. Your level of commitment forming the Medical Microfluidics group has been monumental.

My co-authors for hard work and support: Mikael for a rewarding collaboration and your relentlessly good spirit, Tommy and Fredrik for manufacturing expertise, Wouter for macro perspective on micro, Sergey for microbiology expertise, Harisha and Sahar for our joined cell endeavors, joys, and healthy discussions.

My co-workers at Cell Physics for many interesting discussions and making this time so much fun: Ali K, Prem, Karolin, Marina, Björn, Bruno, Thomas, Athanasia, Per, Elin, Johanna, Karin, Zenib, Hattie, Susanna, Padideh, Jacob; and all Cell Physics people at KI and SciLife

All people at Microsystem Technology Lab making me feel at home during my intermittent periods there. The good spirit and productive fabrication discussions stands out.

All people involved in the InTopSens project for our collective efforts towards Sepsis diagnostics. Special thanks to the people at Mobidiag for commercial insights and hospitality in Helsinki.

All people at the School of Biotechnology that have been the new home of the Medical Microfluidics group over the last 6 months. Special thanks to: Helene for welcoming us into her group, the rest of the people in the Nanobiotechnology group for an instant team spirit, and everyone at plan 3 for transparent discussions and a positive atmosphere.

The nice people at Husläkarmottagningen Johannes and Blodcentralen Skanstull for providing their needle skills.

Special thanks to Ingemar Lundström for inspiring me to pursue academic research.

My family and friends for amazing support. Special thanks to Hasse, Anki and Berne for your proof-reading and feedback on the thesis. Fredrik, Maria, Peter, Eva, Laura and Max for discussions vital towards my Licentiate. My younger sisters Malin and Kajsa for trying to be like me ;-)

Jonas Hansson, Stockholm, May 2012

Jonas Hansson 35

9 References 1. Beekmann SE, Diekema DJ, Chapin KC, Doern GV. Effects of Rapid Detection of Bloodstream

Infections on Length of Hospitalization and Hospital Charges. Journal of Clinical Microbiology. 2003 Jul 1;41(7):3119–25.

2. Whitesides GM. The origins and the future of microfluidics. Nature. 2006 Jul 27;442(7101):368–73.

3. Reimer LG, Wilson ML, Weinstein MP. Update on detection of bacteremia and fungemia. Clinical Microbiology Reviews. 1997 Jul;10(3):444–65.

4. Hunter P. Sepsis under siege: A new understanding of sepsis might lead to the development of therapies to treat septic shock. EMBO reports. 2006 Jul;7(7):667–9.

5. Bone R, Balk R, Cerra F, Dellinger R, Fein A, Knaus W, et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest. 1992 Jun 1;101(6):1644–55.

6. Levy MM, Fink MP, Marshall JC, Abraham E, Angus D, Cook D, et al. 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Critical Care Medicine. 2003 Apr;31(4):1250–6.

7. Kumar A, Roberts D, Wood KE, Light B, Parrillo JE, Sharma S, et al. Duration of hypotension before initiation of effective antimicrobial therapy is the critical determinant of survival in human septic shock. Critical Care Medicine. 2006 Jun;34(6):1589–96.

8. Yang S, Rothman RE. PCR-based diagnostics for infectious diseases: uses, limitations, and future applications in acute-care settings. The Lancet Infectious Diseases. 2004 Jun;4(6):337–48.

9. Rocky Mountain Laboratories, NIAID, NIH. Scanning electron micrograph of Escherichia coli.

10. The National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), Noguchi, Rodgers, Schechte. A micrograph of red blood cells.

11. National Cancer Institute, Bruce Wetzel, Harry Schaefer. Scanning electron microscope image from normal circulating human blood. 1982.

12. Electron Microscopy Facility at The National Cancer Institute at Frederick (NCI-Frederick). A three-dimensional ultrastructural image analysis of a T-lymphocyte (right), a platelet (center) and a red blood cell (left).

13. Jonas Hansson. Sedimented blood in test tube from Blodcentralen in Skanstull. 2012.

14. Yagupsky P, Nolte FS. Quantitative aspects of septicemia. Clinical Microbiology Reviews. 1990 Jun;3(3):269–79.


15. Dellinger RP, Carlet JM, Masur H, Gerlach H, Calandra T, Cohen J, et al. Surviving Sepsis Campaign guidelines for management of severe sepsis and septic shock. Critical Care Medicine. 2004 Mar;32(3):858–73.

16. Mancini N, Carletti S, Ghidoli N, Cichero P, Burioni R, Clementi M. The Era of Molecular and Other Non-Culture-Based Methods in Diagnosis of Sepsis. Clinical Microbiology Reviews. 2010 Jan 11;23(1):235–51.

17. Hansson J. Study visit to Klinisk mikrobiologi at Karolinska Universitetslaboratoriet Solna, Stockholm, Sweden. 2012.

18. Anders Tegnell. Falldefinitioner – vid anmälan enligt smittskyddslagen. Sweden: Socialstyrelsen; 2012.

19. Tissari P, Zumla A, Tarkka E, Mero S, Savolainen L, Vaara M, et al. Accurate and rapid identification of bacterial species from positive blood cultures with a DNA-based microarray platform: an observational study. The Lancet. 2010 Jan;375(9710):224–30.

20. Bruns T, Sachse S, Straube E, Assefa S, Herrmann A, Hagel S, et al. Identification of bacterial DNA in neutrocytic and non-neutrocytic cirrhotic ascites by means of a multiplex polymerase chain reaction. Liver International. 2009 Sep;29(8):1206–14.

21. Casalta JP, Gouriet F, Roux V, Thuny F, Habib G, Raoult D. Evaluation of the LightCycler® SeptiFast test in the rapid etiologic diagnostic of infectious endocarditis. European Journal of Clinical Microbiology & Infectious Diseases. 2008 Dec 2;28(6):569–73.

22. Mühl H, Kochem A-J, Disqué C, Sakka SG. Activity and DNA contamination of commercial polymerase chain reaction reagents for the universal 16S rDNA real-time polymerase chain reaction detection of bacterial pathogens in blood. Diagnostic Microbiology and Infectious Disease. 2010 Jan;66(1):41–9.

23. Awais R, Fukudomi H, Miyanaga K, Unno H, Tanji Y. A Recombinant Bacteriophage-Based Assay for the Discriminative Detection of Culturable and Viable but Nonculturable Escherichia coli O157:H7. Biotechnology Progress. 2006 Jun 2;22(3):853–9.

24. van Baar BL. Characterisation of bacteria by matrix-assisted laser desorption/ionisation and electrospray mass spectrometry. FEMS Microbiol. Rev. 2000 Apr;24(2):193–219.

25. Forrest GN, Mankes K, Jabra-Rizk MA, Weekes E, Johnson JK, Lincalis DP, et al. Peptide Nucleic Acid Fluorescence In Situ Hybridization-Based Identification of Candida albicans and Its Impact on Mortality and Antifungal Therapy Costs. Journal of Clinical Microbiology. 2006 Sep 5;44(9):3381–3.

26. von Wintzingerode F. Base-specific fragmentation of amplified 16S rRNA genes analyzed by mass spectrometry: A tool for rapid bacterial identification. Proceedings of the National Academy of Sciences. 2002 Apr 30;99(10):7039–44.

27. Gebert S, Siegel D, Wellinghausen N. Rapid detection of pathogens in blood culture bottles by real-time PCR in conjunction with the pre-analytic tool MolYsis. Journal of Infection. 2008 Oct;57(4):307–16.

Jonas Hansson 37

28. Wen J, Legendre LA, Bienvenue JM, Landers JP. Purification of Nucleic Acids in Microfluidic Devices. Analytical Chemistry. 2008 Sep;80(17):6472–9.

29. Tabeling P. Introduction to microfluidics. Oxford: Oxford University Press; 2005.

30. Lee C-Y, Chang C-L, Wang Y-N, Fu L-M. Microfluidic Mixing: A Review. International Journal of Molecular Sciences. 2011 May 18;12(5):3263–87.

31. Stroock AD. Chaotic Mixer for Microchannels. Science. 2002 Jan 25;295(5555):647–51.

32. Stroock AD, McGraw GJ. Investigation of the staggered herringbone mixer with a simple analytical model. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences. 2004 May 15;362(1818):971–86.

33. Pamme N, Manz A. On-Chip Free-Flow Magnetophoresis: Continuous Flow Separation of Magnetic Particles and Agglomerates. Analytical Chemistry. 2004 Dec;76(24):7250–6.

34. Han K-H, Bruno Frazier A. Continuous magnetophoretic separation of blood cells in microdevice format. Journal of Applied Physics. 2004;96(10):5797.

35. Fiedler S, Shirley SG, Schnelle T, Fuhr G. Dielectrophoretic Sorting of Particles and Cells in a Microsystem. Analytical Chemistry. 1998 May;70(9):1909–15.

36. Wang MM, Tu E, Raymond DE, Yang JM, Zhang H, Hagen N, et al. Microfluidic sorting of mammalian cells by optical force switching. Nature Biotechnology. 2004 Dec 19;23(1):83–7.

37. Lenshof A, Ahmad-Tajudin A, Jarås K, Sward-Nilsson A-M, Åberg L, Marko-Varga G, et al. Acoustic Whole Blood Plasmapheresis Chip for Prostate Specific Antigen Microarray Diagnostics. Analytical Chemistry. 2009 Aug;81(15):6030–7.

38. Holm SH, Beech JP, Barrett MP, Tegenfeldt JO. Separation of parasites from human blood using deterministic lateral displacement. Lab on a Chip. 2011;11(7):1326.

39. Yamada M, Nakashima M, Seki M. Pinched Flow Fractionation: Continuous Size Separation of Particles Utilizing a Laminar Flow Profile in a Pinched Microchannel. Analytical Chemistry. 2004 Sep;76(18):5465–71.

40. Di Carlo D, Irimia D, Tompkins RG, Toner M. Continuous inertial focusing, ordering, and separation of particles in microchannels. Proceedings of the National Academy of Sciences. 2007 Nov;104(48):18892–7.

41. Russom A, Gupta AK, Nagrath S, Carlo DD, Edd JF, Toner M. Differential inertial focusing of particles in curved low-aspect-ratio microchannels. New J. Phys. 2009 Jul;11(7):075025.

42. Mach AJ, Di Carlo D. Continuous scalable blood filtration device using inertial microfluidics. Biotechnol. Bioeng. 2010 Jun;107(2):302–11.


43. Matas J, Morris J, Guazzelli E. Lateral Forces on a Sphere. Oil & Gas Science and Technology. 2004 Jan;59:59–70.

44. Segré G, Silberberg A. Radial Particle Displacements in Poiseuille Flow of Suspensions. Nature. 1961 Jan;189(4760):209–10.

45. Di Carlo D, Edd J, Humphry K, Stone H, Toner M. Particle Segregation and Dynamics in Confined Flows. Physical Review Letters. 2009 Mar;102(9).

46. Bhagat AAS, Kuntaegowdanahalli SS, Papautsky I. Inertial microfluidics for continuous particle filtration and extraction. Microfluid Nanofluid. 2008 Dec;7(2):217–26.

47. Becker H, Gärtner C. Polymer microfabrication technologies for microfluidic systems. Analytical and Bioanalytical Chemistry. 2007 Nov 8;390(1):89–111.

48. Becker H, Manz A. Microsystem Technology in Chemistry and Life Sciences. New York: Springer; 1997

49. Terry SC, Jerman JH, Angell JB. A gas chromatographic air analyzer fabricated on a silicon wafer. IEEE Transactions on Electron Devices. 1979 Dec;26(12):1880–6.

50. Xia Y, Whitesides GM. Soft Lithography. Angewandte Chemie International Edition. 1998 Mar 16;37(5):550–75.

51. Ng JMK, Gitlin I, Stroock AD, Whitesides GM. Components for integrated poly(dimethylsiloxane) microfluidic systems. Electrophoresis. 2002 Oct;23(20):3461–73.

52. Unger MA. Monolithic Microfabricated Valves and Pumps by Multilayer Soft Lithography. Science. 2000 Apr 7;288(5463):113–6.

53. Kim J-H, Na K-H, Kang CJ, Kim Y-S. A disposable thermopneumatic-actuated micropump stacked with PDMS layers and ITO-coated glass. Sensors and Actuators A: Physical. 2005 May;120(2):365–9.

54. de Jong J, Lammertink RGH, Wessling M. Membranes and microfluidics: a review. Lab on a Chip. 2006;6(9):1125.

55. Karlsson JM, Haraldsson T, Laakso S, Virtanen A, Maki M, Ronan G, et al. PCR on a PDMS-based microchip with integrated bubble removal. IEEE; 2011. p. 2215–8. Available from: http://ieeexplore.ieee.org/lpdocs/epic03/wrapper.htm?arnumber=5969303

56. Xu J, Vaillant R, Attinger D. Use of a porous membrane for gas bubble removal in microfluidic channels: physical mechanisms and design criteria. Microfluidics and Nanofluidics. 2010 Mar 24;9(4-5):765–72.

57. Eddings MA, Gale BK. A PDMS-based gas permeation pump for on-chip fluid handling in microfluidic devices. Journal of Micromechanics and Microengineering. 2006 Nov 1;16(11):2396–402.

Jonas Hansson 39

58. Addae-Mensah KA, Reiserer RS, Wikswo JP. Poly(vinyl alcohol) as a structure release layer for the microfabrication of polymer composite structures. Journal of Micromechanics and Microengineering. 2007 Jul 1;17(7):N41–N46.

59. Kang JH, Um E, Park J-K. Fabrication of a poly(dimethylsiloxane) membrane with well-defined through-holes for three-dimensional microfluidic networks. Journal of Micromechanics and Microengineering. 2009 Apr 1;19(4):045027.

60. Kartalov EP. Microfluidic vias enable nested bioarrays and autoregulatory devices in Newtonian fluids. Proceedings of the National Academy of Sciences. 2006 Aug 15;103(33):12280–4.

61. Anderson JR, Chiu DT, Jackman RJ, Cherniavskaya O, McDonald JC, Wu H, et al. Fabrication of Topologically Complex Three-Dimensional Microfluidic Systems in PDMS by Rapid Prototyping. Analytical Chemistry. 2000 Jul;72:3158–64.

62. Carlborg CF, Haraldsson T, Cornaglia M, Stemme G, van der Wijngaart W. A High-Yield Process for 3-D Large-Scale Integrated Microfluidic Networks in PDMS. J. Microelectromech. Syst. 2010 Oct;19(5):1050–7.

63. Preston GM, Agre P. Isolation of the cDNA for erythrocyte integral membrane protein of 28 kilodaltons: member of an ancient channel family. PNAS. 1991 Dec;88(24):11110–4.

64. Preston GM, Carroll TP, Guggino WB, Agre P. Appearance of Water Channels in Xenopus Oocytes Expressing Red Cell CHIP28 Protein. Science. 1992 Apr 17;256(5055):385–7.

65. Ho SN. Intracellular water homeostasis and the mammalian cellular osmotic stress response. Journal of Cellular Physiology. 2006 Jan;206(1):9–15.

66. Pichereau V, Hartke A, Auffray Y. Starvation and osmotic stress induced multiresistances. International Journal of Food Microbiology. 2000 Apr;55(1-3):19–25.

67. Helenius A, Simons K. Solubilization of membranes by detergents. Biochimica et Biophysica Acta (BBA) - Reviews on Biomembranes. 1975 Mar;415(1):29–79.

68. Detergents. AppliChem; 2008.

69. Hsiao LL, Howard RJ, Aikawa M, Taraschi TF. Modification of host cell membrane lipid composition by the intra-erythrocytic human malaria parasite Plasmodium falciparum. Biochem. J. 1991 Feb 15;274 ( Pt 1):121–32.

70. Mesnard R, Corvisier J, Sire J, Pouëdras P, Donnio P, Avril J. Continuous ambulatory peritoneal dialysis: evaluation of three new culture methods. Ann Biol Clin (Paris). 1993;51(7-8):697–700.

71. Sen, Makkar, Muetzel, Becker. Effect of Quillaja saponaria saponins and Yucca schidigera plant extract on growth of Escherichia coli. Letters in Applied Microbiology. 1998 Jul;27(1):35–8.

72. Baumann E. Hemolysis of human erythrocytes with saponin affects the membrane structure. Acta Histochemica. 2000;102(1):21–35.


73. Sethu P, Anahtar M, Moldawer LL, Tompkins RG, Toner M. Continuous Flow Microfluidic Device for Rapid Erythrocyte Lysis. Anal. Chem. 2004 Nov;76(21):6247–53.

74. Sethu P, Moldawer LL, Mindrinos MN, Scumpia PO, Tannahill CL, Wilhelmy J, et al. Microfluidic Isolation of Leukocytes from Whole Blood for Phenotype and Gene Expression Analysis. Analytical Chemistry. 2006 Aug;78(15):5453–61.

75. Russom A, Sethu P, Irimia D, Mindrinos MN, Calvano SE, Garcia I, et al. Microfluidic Leukocyte Isolation for Gene Expression Analysis in Critically Ill Hospitalized Patients. Clinical Chemistry. 2008 Feb 19;54(5):891–900.

76. Parichehreh V, Estrada R, Kumar SS, Bhavanam KK, Raj V, Raj A, et al. Exploiting osmosis for blood cell sorting. Biomed Microdevices. 2011 Jun;13(3):453–62.

77. SooHoo JR, Herr JK, Ramsey JM, Walker GM. Microfluidic Cytometer for the Characterization of Cell Lysis. Analytical Chemistry. 2012 Mar 6;84(5):2195–201.

78. Bhagat AAS, Kuntaegowdanahalli SS, Papautsky I. Continuous particle separation in spiral microchannels using dean flows and differential migration. Lab on a Chip. 2008;8(11):1906.

79. Hur SC, Henderson-MacLennan NK, McCabe ERB, Di Carlo D. Deformability-based cell classification and enrichment using inertial microfluidics. Lab on a Chip. 2011;11(5):912.

80. Lenshof A, Laurell T. Continuous separation of cells and particles in microfluidic systems. Chemical Society Reviews. 2010;39(3):1203.

microfluidic blood sample preparation for rapid sepsis

Documents