Generation, dynamics and control of microbubbles in microdevices

(1)

Generation, dynamics and

control of microbubbles in

microdevices.

Davide Caprini

Dipartimento di ingegneria meccanica e aerospaziale Universit´a di Roma Sapienza

A thesis submitted for the degree of

Dottorato in Meccanica Teorica e Applicata (PhD) 2019

(2)

Tutor: Prof. Luca Marino 1.Advisor: Dr. Giorgia Sinibaldi

2.Advisor: Prof. Augusto Nascetti

1. Reviewer: Prof. Guido Bolognesi, Department of Chemical Engineering, Loughborough University.

2. Reviewer: Prof. Massimiliano Rossi, Department of Physics, Technical University of Denmark.

Day of the defense:

(3)

Abstract

The impact of microfluidic technologies in the field of life sciences has no-ticeably increased during the last decade, due to the need of developing innovative microfluidic platforms for biomedical applications. The present experimental work focuses on the development of microfluidic systems aimed at the investigation of the generation, dynamics and control of bubbles in microchannels. The dissertation, after an introduction on manufacturing techniques, is divided in three parts devoted to the explanation and discus-sion of the main topics addressed during the research activities.

The first part concerns on the break-up phenomenon of microbubbles in a T-junction microfluidic structure which is particularly relevant in several applications involving microbubble generation technology. The bubble cur-vature and velocity field during break-up are investigated using a smart microfluidic device that allows a double orthogonal view. Such device was designed, developed, and manufactured to offers the possibility to recon-struct a three-dimensional velocity field. Combining microPIV (Particle Image Velocimetry) and the double view, two planar velocity fields are obtained showing all three components of the velocity in their intersection line. Furthermore, the reconstruction of the three-dimensional velocity field is possible by spanning the two planes along the transversal axes of the mi-crochannel. The proposed chip could be used as microfluidic platform for T-junction break-up studies or as a benchmark for more sophisticated tech-niques to reconstruct 3D velocity field or 3D tracking. The system combined with dedicated optical elements and placed on the stage of a traditional in-verted microscope simultaneously yields a double orthogonal view without

(4)

the need of two synchronized cameras or more complex and expensive sys-tems.

In the second part of the thesis, an experimental study of ultrasonic Bessel beam generation is presented. Microbubbles combined with ultrasounds are widely exploited to image biological tissues due to their strong echogenicity. The bubble dynamics is modified by the interactions with the ultrasound field inducing secondary mechanical effects on tissues. Different ways to focus ultrasounds on the target zone can be adopted. In order to be com-patible with biological applications, ultrasound must be able to reach deep tissues without exceeding the maximum threshold of energy that induce damage. In this context, alternative systems to the traditional Gaussian fo-cused beam, like Bessel beam here proposed, can potentially target biomed-ical applications more efficiently. Here, a preliminary innovative large-scale annular structure was realized to demonstrate the scattering resistance and self-healing of an ultrasonic Bessel beam. An array of synchronous emitters is radially arranged. The emitters are tilted in order to generate a symmet-ric cone field which builds a constructive interference with the shape of a first order Bessel function. The generated beam, as in the case of light, is able to resist the scattering of tissues and to go in depth without focusing all the energy in a single point. The main outcome of this part has been to demonstrate that the ultrasound Bessel-function emission pattern leads to scattering resistance and self-healing capability, as expected. It is important to notice that the same ultrasound field generated in large scale could be transferred to smaller wavelengths, compatible with biological applications. The last part of the thesis deals with the study of microbubbles generation and control in microchannels and their possible application in drug deliv-ery. Microbubbles generated with microfluidic structures are widely used in biomedical applications due to the high generation rate and size control, which depends on the flow rates of the liquid and gaseous phases. In

(5)

par-ticular, an endothelial barrier in a microchannel (blood-vessel-on-chip) has been realized and the blood vessel permeability enhancement in presence of ultrasounds-mediated microbubbles was studied. The main goal of this part is the development of a microbubbles generator to be integrated to the blood-vessel-on-chip. The generator is characterized by the flow-focusing geometry and exploits a small orifice where gas and liquid converge and form microbubbles. As the reduced size facilitates gas diffusion in liquid, microbubbles are coated with biocompatible layer by adding lipids in the liquid phase resulting in bubbles of constant size over time. The microbub-bles formed at the orifice are then directed towards the endothelium and excited by a traditional ultrasound set-up. After sending cycles of sinusoidal bursts, the acoustic response is detected and evaluated through the analysis of sub- and ultra-harmonic generated by the microbubbles dynamics. The flow-focusing generator was adopted to investigate the microbubbles gener-ation and the dynamics when the bubbles production rate and size increase. In such conditions, the channel following the orifice is completely filled with bubbles that arrange in a ordered structure due to attractive forces, a pro-cess widely exploited in the engineering of biomaterials. The dissertation ends with a brief analysis of these bubbles structures, in particular aimed at the comprehension of the center of mass vibration and liquid phase viscosity effects.

(6)

(7)

Acknowledgements

I would like to acknowledge the Istituto Italiano di Tecnologia, iit CLNS@ Sapienza for the opportunity to take advantage of the laboratories and all the equipment that was necessary to achieve the results reached in this the-sis. I thank the center coordinator Prof. Giancarlo Ruocco for his valuable technical advice and for the contribution to the publication of one of the main works of the thesis. I thank Dr. Viola Folli for the availability shown during the correction and writing of the thesis in the last period. A big thank also to Prof. Augusto Nascetti for teaching me many of the micro-fabrication methods that I used to achieve the published results, and who initially directed me to follow this course of studies. I also thank Prof. C.M. Casciola, along with my tutor Prof. Luca Marino and Dr. Giorgia Sinibaldi, for all the work done during the three PhD years and who led us to achieve important results. I warmly thank my precious collaborators from the lab-oratory Ing. Giulia Silvani, Dr. Chiara Scognamiglio and Dr. Giovanna Peruzzi for their contribution on the biological part of the experiments.

(8)

(9)

Introduction

This thesis focuses on the development of microdevices to investigate bubbles gener-ation, dynamics and control. The dissertgener-ation, after an introduction on the role of bubbles in microfluidics and biomedical applications, describes the state of the art of microfabrication techniques and the main geometries used to produce bubbles. Then, the development of new experimental concepts are discussed: a new microfluidic chip designed to obtain two simultaneous views from orthogonal planes captured by a single camera; an experimental device developed to demonstrate of the self-healing and the scattering resilience properties of a non-diffracting zero-order acoustic Bessel beam; and the intgration of a microbubble generator with a vessel-on-a-chip used to study the change of permeability of an endothelial layer induced by ultrasound cavitation.

1.1 Bubbles and droplets in microfluidics

Two phase microfluidic systems can be found in most of the chemical processes where reactants and products in different phase are mixed at micrometer scale to maximize the performance by a specific and accurate control of the flow rates of the two fluid phases (152).

Two phase flows are generated when two partially miscible or immiscible fluids are placed in contact. The typical two phase flows are divided into two groups, the

(14)

1. INTRODUCTION

(42, 126) can be found in several applications involving reaction (113), mixing (121), emulsions (60), biomedicine (50) and material synthesis (143).

The numerous recent progresses in microfabrication have facilitated the development of functional elements, such as mixers, heaters, pumps, sensors and separators. In the last decades, the behavior of two phase flows in these microfluidic elements has been deeply investigated to maximize heating, mixing and emulsion processes (115).

An important classification of two phase flows in microchannels concerns the mecha-nism adopted to generate bubbles and/or droplets. Active and passive technologies are typically used in peculiar microfluidic structures. Compared to the active approach, which involves the use of valves and external forces, the passive approach takes ad-vantage of an important characteristic of microfluidics, i.e. the possibility to control the interface and capillary instability, thus generating bubbles/droplets. Therefore, external actuation is not required in the passive generation. Using passive methods, droplets/bubbles are typically highly monodisperse and the droplet size distribution variance of diameter can be as small as 1 − 3 % (73). This kind of monodisperse droplets/bubbles are more suitable for the applications above listed, compared to the generation in a bulk of emulsions or foams with large dimensions.

In microfluidics the flow is typically laminar and the diffusion effect between molecules is reduced. More important is the possible capability for precise control and manipula-tion of fluid flow rates. The flow in microchannels basically depends on three physical aspects: the channel geometry, the properties of both fluids and the flow conditions and can be described in terms of some characteristic dimensionless parameters (152). One of the most important dimensionless parameter is the Reynolds number (Re), defined as the ratio between inertial and viscous forces,

Re = ρDU

µ , (1.1)

where ρ is fluid density (kg/m3), U the characteristic velocity (m_s), D the characteristic

length scale in (m), and µ the dynamic viscosity (P a s). At micrometer scale the Reynolds number is very small (Re << 1), and only in some cases can be of order 1.

(15)

1.1 Bubbles and droplets in microfluidics

Another parameter that in microfluidics turns out to be very small and almost neg-ligible, is the Bond Number (Bo) which is the ratio between gravitational forces and surface tension forces,

Bo = ∆ρgD

2

γ , (1.2)

where ∆ρ is the difference of the two fluid density (kg/m3_{), γ the surface tension}

(N/m), and g the gravitational acceleration (m/s2).

Surface tension forces increase as the characteristic scale of the phenomenon decreases, so at micrometer scale it becomes comparable to the viscous forces. The ratio between these two forces is the Capillary number (Ca), defined as:

Ca = µU

γ . (1.3)

Other two important dimensionless numbers direct related on Ca and Re are the Ohne-sorge number (Oh) and the Weber number (W e). The Oh number reports the impor-tance of viscous force with respect to the inertial and surface tension forces, while the W e number compares the inertial effects and surface tension,

Oh = Ca Re 2 = µ (ρDγ)12 (1.4) W e = ReCa = ρU 2_D γ . (1.5)

The operating flow conditions of different phases are taking into account with other

parameters, such as the density ratio (α = ρc

ρd), viscosity ratio (β =

µc

µd) and the flow

rate ratio (φ = Qc

Qd) where Q is the fluid flow rate and the subscripts c and d represent

the continuous and dispersed phase, respectively.

The key issue for bubble formation and break-up processes is the dynamical evolu-tion of gas−liquid interface in confined spaces at low Reynolds numbers and capillary numbers (4). Typically the generation is controlled by manipulating the flow rates of the continuous and the dispersed phases (42).

(16)

1. INTRODUCTION

Figure 1.1: Typical structures for microbubble generation on chip and some exam-ples of two phase regimes (adapted from (152)). The dispersed phase is shown in or-ange. (A,a) Generic T-junction structure, (A,b) Cross-flowing(141), (A, c) Perpendicular flow(141); (B, a) Generic Flow-focusing structure, (B, b) Flow-focusing structure with in-clined channels(136) (B ,c) Dripping regime for Flow-focusing structure; (C, a) Generic Co-flowing structure, (C, b) Examples of Rayleigh-plateau instabilities at needles exit, (C, c) Co-Flow droplet formation in the dripping regime.

In the following a description of the typical configurations of microchannels used for the generation of droplets/bubbles is provided, with particular attention to the dimen-sionless parameters, see Fig.1.1. For the description of gas-liquid flows the liquid phase is indicated as the continuous phase while the gas phase is considered as the dispersed one.

(17)

droplets/bubbles (41). It consists of two orthogonal channels and droplets/bubbles are formed where the two channels meet (42). Two different kind of flows are pos-sible depending on where the dispersed phase is introduced:

1. Cross-flowing (Fig.1.1 A, b). The continuous phase is introduced in the horizontal channel and the dispersed phase flows through the orthogonal channel, so that droplets/bubbles are sheared off by the cross-flowing shear force (120).

2. Perpendicular flowing (Fig.1.1 A, c). The continuous phase is introduced from the perpendicular channel and the dispersed flow from the horizontal one.

Compared with the perpendicular flowing technique, the cross-flowing break-up technique produces droplets with a narrower range of droplet size (142).

For low Capillary number typical of microfluidic systems, evaluated for the

con-tinuous phase, i.e. Cac = µc_γUc < O(10−2), the build up of pressure upstream

of the incipient dispersed phase squeezes the liquid such that it breaks into droplets, and the length of the resulting slug scales with the flow rate ratio, φ (42). This is typically called squeezing regime. Above a critical capillary number,

i.e. Cac> O(10−2), droplet formation becomes shear dominated (32), where the

length of the resulting slug can be predicted by the capillary number, Cac. This

is called the dripping regime because of its similarities with the dripping regime in coaxial flows.

• Flow-focusing (Fig.1.1 B, a). In this geometry the continuous phase is conveyed through two side channels that converge at the outlet of the dispersed phase chan-nel so that the two lateral flows break the dispersed phase into droplets/bubbles. The process can be helped by passing the two phases from the different

chan-nels through an orifice. With the flow-focusing geometry the contact of the

(18)

1. INTRODUCTION

In this particular geometry the phenomenon is highly controlled by the value of surface tension compared to viscous forces, while inertia forces are almost negligible. For this reason, the transition from jetting (when the dispersed phase is in confined as a central jet), (Fig.1.1 B, b) to dripping (Fig.1.1 B, c) is even more sensitive to changes of the Capillary number (29).

• Co-flowing (Fig.1.1 C, a). In this geometry the continuous phase flows into

the main channel and the dispersed phase is fed through a axisymmetric nee-dle or a channel in the midnee-dle of the main channel and parallel to the flow (28). Droplets/bubbles in co-flowing geometries can be formed either right at the tip of the needle (Fig.1.1 C, b) or further into the main channel, at the end of a fluid stream (Fig.1.1 C, c), where the flow regimes are similar to the flow-focusing microchannels: dripping and jetting. Coaxial droplet formation around the needle introduced by the liquid flow rate is governed by the Rayleigh-Plateau instability that occurs immediately after the needle outlet. Therefore, the size of droplets/bubbles is therefore linked to the diameter of the needle and is stabilized by adapting itself to the jet in the continuous phase (124).

At higher flow rates, it is possible to identify a second class of the dripping to jetting transition by controlling the relative inertia of the dispersed phase to the capillary pressure. The inertia of the dispersed phase pushes the dripping location downstream, away from the capillary tip, so that a jet is formed at the tip itself. In this regime, the inertial forces and capillary forces are in equilibrium. In this condition, when the Weber number for the dispersed phase is above a critical

value, i.e. (W ed = O(1)), jets are formed. In the literature few papers, see

e.g. (18), extended the analysis to the jetting transition in coaxial devices. They

observed that the criteria W ed> O(1) for jet formation is valid only for Reynolds

number of the dispersed liquid phase Red> O(1). When Red< O(1) the Weber

number does not accurately capture the transition from dripping to jetting, but,

instead, jetting occurs when the Cad> O(1). As the flow rate of the continuous

(19)

Figure 1.2: Schematic of different two-phase flows regimes in co-flow, flow-focusing and T-junction microfluidic devices . Solid green arrows indicate the continuous flow direction and the solid red arrows the dispersed flow respectively (90).

A summary sketch for the possible regimes that can occur in the three structures above described is shown in Fig.1.2. These regimes are valid both for droplets and bubbles with some differences. Being the control of the flow rate of the liquid easier than the gaseous phase, that is very unstable, in the most gas-liquid cases the gaseous phase is dispersed in the liquid phase. In general, a microbubble contains a compressible gas and so, at given capillarity conditions, different bubbles dimensions are possible (45).

To understand the differences between the microbubble prediction models reported in literature with respect to the microdroplets are reported in the following for the simple case of a T-junction, with reference on Fig. 1.3.

(20)

1. INTRODUCTION

contour does not depend critically on the viscosity of the dispersed fluid. Both in the case of a viscous droplet and in the case of a gaseous bubble, when the dispersed fluid stops the continuous fluid during the inception in main channel, ∆p increases sharply upon decrease of the thickness of the film separating the interface from the walls of the channel. Thus the mechanism of breakup that is described for liquid-liquid systems should be applied equally well to liquid-gas systems. In this case, the length L of the

bubble produced in a T-junction should follow the same scaling law: L = d Qgas

Qliquid + w,

where Qgas is the flow rate of gas into the main channel given by Qgas= ∆p_R, where R

is the resistance of the flow in the channel, ∆p is the pressure gradient in the outlet

channel length, Lch, and w is the length of the bubble slug in the main channel during

the break-up. d shown the radius of curvature of the bubble interface with respect to the nearest vertex, see Fig.1.3. To a first order approximation, at low volume-fractions of bubbles in the downstream portion of the main channel, it is possible to assume that

R will scale as it would in a channel filled with the continuous liquid:R ∝ µLch

h2_w2 (42).

Figure 1.3: Sketch of the geometrical characteristics parameters of bubble pinch-off in a T-junction structure. The gas dispersed phase is shown in yellow (42).

The microbubble length can be influenced by three main effects:

1. The viscosity of the liquid phase. The length of the bubbles is inversely propor-tional to the viscosity of the continuous phase, which in this case is the liquid,

(21)

given that (L−w)_d = Qgas

Qliquid =

∆p

Qliquid R ∝

∆p

Qliquid µ. Some experiments in the

liter-ature are in agreement with the prediction that (L−w)_d scale inversely with µ, as

shown e.g. in (42).

2. The output channel length, Lch. Provided the dependence of the bubble length on

the viscous channel resistance R, Lch changes with the pressure drop by adding

sections of few centimetres to the outlet. According to the squeezing model, (L−w) d = Qgas Qliquid = ∆p QliquidR ∝ ∆p

Lch (because R ∝ µ) and then

(L−w)

d Lch ∝ ∆p.

Experiments with networks containing one or two added sections at the exit, confirm this behavior (42).

Figure 1.4: Example of scaling for the bubble length in T-junction. Qgas is

eval-uated by Qgas = ∆p_R, where R gives the estimate of the viscous resistance to flow

in the outlet channel (Lch = 2 cm for all the curves). (black ◦) Reference geometry

(h = 33 µm, w = 100 µm, wgas = 50 µm) at Qliquid = 0.417 µL s−1; (blue ♦) same

ge-ometry, Qliquid = 0.833 µL s−1; (green ) same geometry, Qliquid = 0.417 µL s−1,

con-tinuous liquid containing surfactant Tween 20 (2% w/w); (red /) and (grey .) w = 50 and 200 µm respectively, Qliquid= 0.417 µL s−1. The brown solid line gives the predicted

scaling (L−w)_d = Qgas

Qliquid. The fitting parameter d = 50 µm for all the curves besides the

one for w = 200 µm, for which d = 100 µm. Reproduced from Garstecki et al (42).

(22)

1. INTRODUCTION

formed in T-junctions with different widths (w = 50, 100, 200 mm) of the main

channel is reported in Fig.1.4 (42). Here the quantity (L−w)_d is shown as a function

of the ratio Qgas

Qliquid. The bubble lengths observed in experiment agree with the

predicted scaling shown in brown in Fig.1.4 for small bubbles and small bubbles volume fractions in the outlet channel. As the bubbles become larger, the model does not follow the volume changes due to larger pressure gradients than those provided by the model. This behaviour is related to the overestimation of the resistance of the large-volume channel of the gas phase.

1.2 Bio-engineering applications of microbubbles

Bubbles at micrometer and nanometer scales have great potential for applications in bio-medical and micro engineering fields. For example, ozone microbubbles are used in water purification and wastewater treatment processes, or dissolved oxygen is increased in water environments by supplying oxygen as microbubbles (76). Microbubbles are also used for transporting geogas in water (85, 129) and in food processes (48, 51). They play also a significant role in fabrication of solid foams such as bread and snacks, and in liquid foams, such as coffee and beer.

In biomedical applications microbubbles are widely used in combination with

ultra-sound (US), e.g. in the diagnostic, clots destruction and targeted drugs delivery

(37, 47, 61, 66). In these applications microbubbles are produced with particular char-acteristics to ensure long timelife and biocompatibility. In particular low poly disper-siveness is required at micrometric scale and specific production systems are adopted that allow the coating with biocompatible films (123). With this characteristics, mi-crobubbles are used as ultrasound contrast agents due to their high compressibility which enables them to efficiently scatter ultrasound (102). Exploiting microbubbles as contrast agents their surface can be functionalized with specific ligands that bind with specific receptors and this makes possible to visualize specific selected area (59, 64). Once the bubbles reach the targets and bind specifically, ultrasound can be utilized to increase contrast image in the area of interest.

(23)

1.2 Bio-engineering applications of microbubbles

As previously mentioned, microbubbles can be coated by an external shell. Coat-ing process prevents the diffusion of gas in surround liquid and makes them suitable for specific drugs, genes and oxygen delivery, that can be activated through the use of focused ultrasound (37). Bubbles, used as drugs carriers, are injected in the vas-cular system to reach a target site, where they are irradiated by sufficiently strong ultrasound to locally release the drugs. The delivery approach using microbubbles can be extended to targeted delivery based on electrostatic or hydrophobic interactions or ligand receptor binding system. For example, negative surface charges on lipid-coated microbubbles improve the surface attachment to the capillary endothelium (38). A layer of polyethylene glycol (PEG) on the bubble shell is often used as a protective layer of microbubbles reducing electrostatic and hydrophobic interactions between the lipid shell and the tissue, and therefore, inhibiting capillary retention of the bubbles.

Coated microbubbles have excellent mechanical properties, so they can be used as elementary particles for light materials or to improve structural efficiency in vehicles and equipment. Their flexibility and deformability make them necessary in the construction of sound-absorbing and impact-resistant materials. These materials, called functional materials, can be fabricated by direct assembly or using mechanically controlled bubble composites with matrix components, such as sludge, concrete, ceramics, polymers and metals (65). Microbubbles, suitably arranged in space with special polymer shells, are used in tissue engineering to create scaffolds, that are very important constitutive structures that support the proliferation of cells in the generation of biological organs tissues (107). Temporary structures for cell growth, have been developed thanks to processes of preparation of porous matrices, made by solvent casting, emulsion and subsequent freezing (75, 147). These methods allow to produce a series of scaffolds with desired porosity and different physical and mechanical properties.

However, the proliferation of a large number of cells is difficult due to the incompati-bility of the surrounding environment. Substantially, progress has been made both by coating the scaffold with various adhesive proteins such as fibronectin, collagen and vitronectin (9, 118) or soaking the scaffold with various growth factors like insulin growth factor (IGF), transforming growth factor-β (TGFβ), platelet-derived growth

(24)

1. INTRODUCTION

factor (PDGF), and fibroblast growth factor via adsorption or covalent linking (11). The main issue of these applications is the biological incompatibility with the solvents used to make scaffolds (39). However, in the literature it is shown that the scaffolds with growth promoters are more effective in cell ploriferation only in the first 1−2 days. To reduce the direct contact between growth factor and solvent used in scaffold fabrication, recent studies showed that albumin microbubbles (M B) can shield the encapsulated growth factors from solvent denaturation (88).

1.3 Ultrasound interaction with microbubbles

In this section the basic theory of ultrasound generation and propagation of the acoustic field are explained, with particular attention to the interaction between microbubbles and ultrasound waves.

Ultrasound-mediated microbubbles applications span from contrast imaging analysis to drug delivery (102). In fact under the effect of sufficiently intense ultrasound mi-crobubbles react with cavitation phenomena which are used in tissues treatment and dissolution of blood clots, or to enhance tissues permeation and promote drug delivery. Several applications adopt lipid−coated microbubbles to combine the dynamic effect induced by a low intensity ultrasound (US) and the presence of drugs in the transport medium either in vivo or in vitro localized delivery. This process is called sonoporation (135). Very similar is the electroporation technique (89), used as alternative method to sonoporation, that generates transient pores in the cell membranes allowing the de-livery of drugs, DNA and antibodies inside the cell (89). Actually the two techniques present differences based on physical reasons. By means of ultrasound approach it is possible to localize mechanical energy to enhance its effect. Electroporation instead can difficulty be selective in space and, moreover, the effects of sonoporation are not yet fully understood.

(25)

1.3 Ultrasound interaction with microbubbles

1.3.1 Ultrasound physical parameters

To introduce the basic physical parameters of ultrasound emission and the propagation of an acoustic field, a simple system of a source that generates a propagating pressure wave with a fixed frequency is considered (138).

With reference to Fig.1.5 a thin solid plane is placed at z = 0, with transversal size, perpendicular to the z direction, much greater than the wavelength (λ) of the sound wave. This plane oscillates sinusoidally with time, producing a sound wave.

Figure 1.5: Sketch of a ultrasound source where the plane vibrate in sinusoidal mode and its plane wave−front propagating along the axis z.

The displacement of this plane source with respect to z = 0 can be written as:

z(t) = Amcos(2πf t + Φ0) (1.6)

where f is the vibration frequency, Am > 0 the amplitude and Φ0 the initial phase

that fixes the initial condition (t = 0) of the source plane. In linear acoustic regime, a traveling pressure wave propagating along z direction in a medium is generated by the vibrating sound source. The pressure in the medium will vary as a function of z and t

(26)

1. INTRODUCTION

around the value the at atmospheric pressure. These pressure fluctuations p(z, t) can be written as:

p(z, t) = P0(z) cos(kz − ωt) = p0e−αzcos(kz − ωt) (1.7)

where P0(z) is the acoustic pressure amplitude which is a function of position z and

is equal to p0e−αz, with p0 = P0(0) the pressure amplitude at z = 0. The other

parameters are the angular frequency ω = 2πf , the wave number k = 2π_λ and the

attenuation coefficient of the medium α (138).

Frequency and wavelength are not independent for a sound wave; they are related by f λ = c, where c is called the phase velocity. In water or soft tissue, the phase velocity

at 20◦C is approximately equal to 1500 m/s.

The attenuation coefficient, α, of soft tissue is considered approximately an increas-ing linear function of the frequency in the MHz range. α describes the energy transfer to the medium mainly through absorption and scattering processes. During absorption, the acoustic energy is converted irreversibly into heat through viscous friction. Scatter-ing is a process in which the non-homogeneities diffuse sound energy to regions outside the original wave propagation path. Within the biological tissue or in aqueous suspen-sions of cells, many non−homogeneities exhist. If the density of non−homogeneity is high, multiple scattering effects can occur. In such cases the sound energy can scatter between different non−homogeneities back and forth several times before it has de-creased by absorption. In water, the attenuation coefficient α is often negligible and

the multiplying factor e−αz can be considered to be unity in Eq.1.7 (138).

Spatial resolution of emission and detection depends strictly on frequency. The higher the frequency, the smaller the wavelength, which translates into greater spatial resolution. On the contrary, the lower the frequency, the longer the wavelength, i.e. lower the spatial resolution. This consideration is very important when using ultrasound for imaging diagnostics. For example at 1 MHz the resolution is approximately 1.5 mm whereas it decreases down to 0.2 mm at 7.5 MHz.

(27)

Another important parameter is the penetration depth that is inversely proportional to the frequency. At 1 MHz, the penetration depth is in the order of 10 cm, while for frequencies higher than 10 MHz, the penetration depth is about 1 cm.

In conclusion, ultrasound with a frequency ≤ 1 MHz could reach tissues in depth, otherwise ultrasound with frequency >10 MHz is confined to the superficial region (138).

A plane or a surface where every point has the same phase is called a wave front. The acoustic wave that correspond to a set of planes as its wave fronts (see Eq. 1.7) is often called a non−focused plane−traveling wave. When the frequency f is above the typical human audible range (f ≥ 20kHz), this type of sound wave is called ultrasound (US).

The plane wave−front of wave described by Eq. 1.7 has infinite dimensions but in practice, a simple sound source is often a circular disk that exhibits a piezoelectric effect and has a finite radius a . US generated by this piston source has some differences from

plane-traveling wave. Infact it depends on the ratio a_λ. Under the condition a λ, the

sound wave in the far−field region operate as an ultrasonic beam with a circular cross section. Within the beam, particularly close to the beam axis, the acoustic pressure can be approximately described by Eq. 1.7.

Figure 1.6 shows the square of the pressure amplitude [p0(σ, z)/pRD]2, which is here

approximately proportional to the acoustic intensity, plotted vs the distance z from the piston sound source.

For z a, the pressure amplitude on the propagation axis can be simply written as: p0(z) pRD = sin1 2kz r 1 +a z 2 − 1. (1.8)

The term pRD = 2ρcU0 shows the amplitude of the acoustic pressure at the ”Rayleigh

distance” RD = a2/λ, which represents the boundary between the near and far field

regions.

In the near field where z < RD, [p0(z)/pRD]2varies with a complicated pattern; several

maxima and minima occur. In the far field, i.e. z > RD, [p0(z)/pRD]2 gradually

(28)

1. INTRODUCTION

Figure 1.6: Example of typical trend of [p0(z)/pRD]2 along the beam axis for plane

circular emitter. The number of peaks and RD value indicate the ”near-field” and depends from radius of emitter and frequency. (57)

to locate the sample at the Rayleigh distance instead of in the near field, since the emission is more uniform and easily controllable. When a plane traveling sound wave propagate into a liquid bulk or a tissue, several physical phenomena may occur. Since the wave propagation in the medium is accompanied by a pressure variation p(z, t) along the propagation axis z, a force is exerted in the adjacent tissue according to the local pressure gradient given by −∂p(z, t)/∂z.

By Newton’s second law, we have:

ρ0

∂v(z, t)

∂t = −

∂p(z, t)

∂z (1.9)

where ρ0 and v(z, t) are the mass density and the particle velocity, respectively.

Sub-stituting Eq.1.7 into Eq.1.9 and assuming αz 1 we derive:

v(z, t) ≈ p0 ρ0c

e−αzcos(kz − ωt) (1.10)

(29)

damping, where the acoustic pressure and particle velocity are in phase and their ratio is a constant and equal to the acoustic impedance, Z (138):

p(z, t)

(z, t) = ρ0c = Z . (1.11)

For a plane traveling wave, as described by Eq. 1.7 , the time-averaged rate at which work is done on the wave front, for unit area, is the acoustic intensity (I). I is given by the time average of the product p(z, t) and v(z, t) over the period T = 1/f , when α is negligible, I(z) = 1 T Z T 0 p0e−2αzcos(kz − ωt)v0cos(kz − ωt)dt = I0e−2αz (1.12)

where I0 = p202ρc is the intensity at z = 0. It is possible to evaluate also the the

acoustic power (W0), i.e. the total work done by source transducer for unit time, where

z = 0 at the source and the emitter area is πa2_,

W0= πa2I0. (1.13)

In most of the biological applications, the energy delivery can be controlled adopting a pulsed system that provides cyclically series of wave fronts. The energy is controlled

by tuning the pulse repetition frequency PRF and pulse duration time Tpd, where

P RF = 1/T0 with T0 the pulse period. The ratio between US on and US off is the

duty cycle, DC, that is equal to:

TpdP RF =

Tpd T0

. (1.14)

The average acoustic power delivered over a period T0 = _{P RF}1 is equal to W0DC.

1.3.2 Focused ultrasounds

Focused ultrasound allows the delivery of a large amount of energy in a small and deep region of interest. It was first demontrated in the 1940s, but it has been developed in the 1980s when the piezometric material fabrication techniques allowed spatial control of energy deposition (24). In biomedical field High Intensity Focused Ultrasound (HIFU)

(30)

1. INTRODUCTION

Figure 1.7: Illustration of a focused ultrasound field generated from a spherical trans-ducer. In the sketch are reported two section profiles of intensity with typical gaussian distribution. Near the focusing region there is an high gradient of energy confined in axial (z) and radial direction(r).

is used to avoid scattering and to go deep into the tissue. For example, HIFU are used in many in vivo experiments on animals, especially in treatments through the brain barrier (BBB) (122).

In the focused transducers the ultrasonic waves are directed using appropriate sys-tems of ”acoustic lenses” or, more easily, by suitably modeling the surface of the radiat-ing element such that the emission is virtually focused in a confined region. This type of focusing transducer is usually a piezoelectric ceramic whose front surface is nearly spherical of radius a, see Fig.1.7. The geometrical center highlighted as 0 .

The dimensions of the curved front surface have the following relations:

a = A sin α, h = A(1 − cos α), b2 = a2+ h2 (1.15)

where h is the depth of the surface and the distance OC the focal length. In sections along the focusing cone, the energy has an Gaussian distribution and the beam focus is located where there is a high energy density. Near the focus point there are thermal

(31)

and mechanical effects that could be harmful for the tissues. It is mathematically

demonstrated that the relation 2πh/λ2 reports the effectiveness of the focus that is

expressed as the ratio between the maximum intensity in focal point C and the average intensity at the radiating surface (95).

1.3.3 Free bubbles and coated bubbles

The interaction between microbubbles and ultrasound is usually known as acoustic cavitation. A bubble can be considered as a volume of gas surrounded by liquid and, when the dimension, i.e. the radius of curvature of the liquid-gas interface, is very small

( 10−3 m), the equilibrium pressure of the gas inside the bubble is high (Laplace law),

leading to gas diffusion in the surrounding liquid. To interact with ultrasound, the bub-ble must be stabub-ble over time. As already mentioned in the previous sections, coatings with lipid films or others biocompatible materials can be used to prevent diffusion and allow the bubble to survive over time. This kind of bubbles are already approved as

contrast agents for ultrasound diagnostics and imaging, see e.g. the SonoVue R_{, that}

are a second generation commercial ultrasound contrast agents composed of sulfur hex-afluor gas packaged from a single-ply phospholipid shell with diameter ranging from 2-7 µm (26). They are usually sold as lyophilized solutions and react with physiological solution for intravenous injection. The liquid solution containing microbubbles shows a strong contrast in terms of acoustic impedance with respect to the surrounding soft tissue (102). Obviously the dynamic response to ultrasound is modified by the presence of the biocompatible thin shell, as will be explained below. The presence of the shell introduces two parameters that are relevant to describe the microbubble mechanical

behavior in a sound field: the shell elastic parameter, Sp, and the shell friction factor,

Sf.

The nonlinear differential equation to describe nonlinear bubble oscillations was developed and subsequently modified by Rayleigh, Plesset, Noltingk, Neppiras and

Poritsky (30). By introducing Sp and Sf, the following equation may be used to

(32)

1. INTRODUCTION ¨ RRρ +3 2ρ ˙R 2_{= (P} 0+ 2γ R0 )(R0 R ) 3K_{− p} 0− 2γ R − 2Sp( 1 R0 − 1 R) − [ 4η R2+ Sf (4πR3₎] ˙RR + Pac, (1.16)

where R and R0 are the instantaneous and equilibrium bubble radius, respectively, P0

is the atmospheric pressure, K is polytropic exponent, and Pac is time varying applied

acoustic pressure, Pac = p0cos(2πf t). When the oscillations are of the order of MHz,

as in the diagnostic case, the predominant contribution is given by the viscous terms, so that the problem can be simplified and linearised (91). The equation 1.16 can be rewritten as:

m¨ + µ ˙ + s = Pac(4πr02), (1.17)

with = r(t) − r0 and where m, s, µ , and fr are the effective mass, stiffness, damping

constant, and resonance frequency. They are respectively given by

m = 4πr3₀ρ, s = 12πkr0 Po+ 2γ r0 + 8π(Sp− γ), µ = Sf+ 16ηπr0, (1.18)

so that the resonance frequency is given by

fr= ωr 2π = 1 2π s 3k ρr2 0 P0+ 2γ r0 − 2γ ρr3 0 +2Sp ρr3 0 (1.19)

where η and γ are viscosity and surface tension respectively. For Sf = 0 and Sp = 0

the resonance frequency of the simple free bubble is obtained. Here we do not report the mathematical analysis of the bubble behavior in the non-linear case, eq. 1.16 (79).

1.3.4 Inertial and stable cavitation

Acoustic cavitation can be calssified into two types. The first one is called ”inertial” or ”transient” cavitation and occurs when the acoustic pressure exceeds a critical threshold value that leads the bubble to rupture. Bubble collapse occurs when the stored energy is such to amplify the oscillation that becomes too large for the lipid coating. Moreover,

if the gas is characterised by a high specific heats ratio, k = Cp

(33)

reached at the collapse can generate highly reactive free radicals. This can induce tissue disruption and irreversible damages (12).

Figure 1.8: Bubble oscillation in presence of ultrasound. (Stable cavitation) The bubble dimension is modified by a variable acoustic pressure and the bubble surface integrity is preserved. (Inertial cavitation) The volume oscillation under a higher acoustic pressure is such to induce inertial effects and drives the bubble to collapse. The rupture is very violent to generate shock waves emission (12).

At low acoustic pressure ”stable” cavitation occurs, where bubbles oscillate by trans-ferring the radiated energy to the surrounding fluid. The lower energy levels stimulate the bubble at frequencies close to the natural frequency. The primary effect of these oscillations is the cyclic movement of the surrounding fluid which triggers microstream-ing and remixmicrostream-ing phenomena. If the oscillations are very large compared to the initial radius, non-linear phenomena can increase the recirculating efficiency of the surround-ing fluid. Microstreamsurround-ing is responsible for mechanical stresses on the biological tissue that are exploited during sonoporation.

During irradiation with ultrasound it is important to evaluate the effects of the acoustic field on tissues. When exposed to a sinusoidal pressure wave, tissues respond with a local compression and expansion movement. If the energy is such to make the movement non-reversible, a dispersion of energy in the tissue occurs. When an unstructured ultrasound beam excites an absorbing and non-reflective tissue, it behaves

(34)

1. INTRODUCTION

as if a constant force acts on it. This effective force is called the ”radiation force”

Frad =

W

c (1.20)

where W is the total power absorbed by the object. If the surface of the absorbent

object, Sa, is flat and a beam of area S impacts normally on it, the radiation pressure

averaged over Sa is _cSW_a.

Generally, the effects induced by US on tissues, are classified as thermal and mechanical (149). From the physical point of view, the thermal effect is due to the absorption of ultrasound by the tissue and the mechanical effect is usually associated with acoustic cavitation. Two parameters are used to quantify the transmission of energy under the two specifics forms, the thermal index (TI) and the mechanical index (MI). These parameters are typically used to evaluate the US level and the set-up safety.

The thermal index (TI) represents the increase of the tissue temperature with respect to the irradiated power and it is defined as the ratio of the acoustic power produced

by the transducer (W0) to the power required to increase the tissue temperature in the

beam area of 1◦C (Wdeg):

T I = W0

Wdeg

(1.21) For diagnostic applications, the thermal effect is not critical. Usually the temperature

increase is ≤ 1◦C for exposure duration up to 1000 min (149).

The mechanical index (MI) estimates the effect on the biological tissue of the ul-trasound radiation rarefaction peak (3) and is defined as:

M I = √pr

f (1.22)

where pr is the peak rarefaction pressure (negative value). A low value of MI indicates

a low probability of having a diagnostic effect, whereas a higher MI is associated to a very efficient ultrasound radiation system for diagnostic treatment.

(35)

1.3.5 Passive cavitation detection

One possibility to monitor and control the ultrasound treatment is to analyze the response spectrum signal of the irradiated microbubbles through Passive Cavitation Detection (PCD). A transducer is typically used to passively listen for sound scattered by cavitation events. Mainly this method consists in a time-domain acquisition and frequency-domain analysis, after a combination of proper filtering to isolate the contri-bution of different cavitation phenomena (100). In general, stable cavitation results in emission at subharmonics and ultraharmonics of the main excitation frequency, while in-ertial cavitation reveals broadband noise emission above the characteristic back-ground noise level (130). Quantifying both broadband noise and subharmonics/ultraharmonics emission allows to estimate the so-called cavitation dose, which provides information about the intensity of cavitation events (70). The bubble response is based on three main parameters such as the peak rarefraction pressure (PRP), the pulse repetition frequency (PRF) and the pulse duration (PD), which characterise a specific ultrasound application.

A typical example of experimental setup used to carried out PCD is shown if Fig.

1.9. The experiment is usually carried out in water, where the sample containing

microbubbles is immersed. The acoustic wave is generated by a function generator. The signal is then amplified and sent to an emitter piezo which irradiates the sample. The measuring setup is composed of a focused piezo receiver that picks up the signal from a portion of the sample. The signal is then amplified and digitalized for the analysis through a spectrum analyzer.

The microbubble response generates peaks at multiple and sub-multiple frequen-cies of the excitation frequency, odd multiples of the subharmonic. This contribution is given by the oscillating dynamics of microbubbles and therefore associated with the part of energy that involves stable cavitation. With appropriate filters, it is possible to remove the energy contributions due to the liquid bulk and obtain the only microbub-bles signal. For inertial cavitation, on the other hand, the violent bubmicrobub-bles collapse is accompanied by broadband emission.

(36)

1. INTRODUCTION

Figure 1.9: Schematic of a PCD setup. Typically it is composed by an emission sequence where an emitter piezo is driven by a function generator connected to an high power amplifier. The detection is obtained by a receiver piezo connected to an amplifier and an oscilloscope. The transducers are immersed in water to interact with the sample.

In the literature it has been observed that stable cavitation has a threshold in terms of pressure induced by ultrasound, dependent on PRP, that is defined as the minimum value of PRP to which ultra-harmonic are obtained (70). The experiments have shown that there is a stable cavitation peak at which the largest sub-harmonic and ultra-harmonic occur at a certain energy value. Then a decay that accompanies the growth of the inertial cavitation contribution appears. In Fig.1.10 a series of FFT sketch for coated microbubbles are shows, qualitatively reproduced from (70).

(37)

Figure 1.10: Sketches of the FFT of ultrasound dynamic response for liquid with (red) and without (black) microbubbles in stable (top) and inertial (bottom) cavitation. The higher pressure level of inertial cavitation is identified by a broadband energy contribution in the spectrum (70).

(38)

(39)

2

Microfluidic fabrication: state of

the art

2.1 Prototyping techniques for microfluidics

In this chapter, the microfabrication techniques for rapid prototyping of microfluidic devices are briefly explained. Particular attention is given to fabrication techniques in the biomedical and biological fields.

2.1.1 Micro-milling

Milling is the process that involves the use of a rotary tool to remove material from a stock called work-piece. The process involves a machine generating a series of controlled movements for the removal of excess parts from the full. The basic milling system, or mill, consists of three main parts:

• a worktable for positioning the work piece, • a cutting tool (most commonly an end mill),

• an overhead spindle for securing and rotating the cutting tool.

The positioning of the worktable and spindle are usually adjusted by hand with me-chanical levers and cranks, but now modern mills employ computer numerical control

(40)

2. MICROFLUIDIC FABRICATION: STATE OF THE ART

Figure 2.1: Basic components of a CNC milling process: a worktable (to provide motion in the XY − plane), a cutting tool (to remove material from the work piece), and a spindle to hold the cutting tool and usually provide motion along the Z-axis.

(CNC) with powerful electric motors that automate the process, reducing human error (6).

The worktable is controlled in position with high accuracy in XY plane, while a spindle controls the Z position and the rotation velocity during the process, see Fig.2.1. The cutting tool consists in a high performance part able to remove material from the work piece. The capabilities of the latest CNC generation support manufacture devices with dimensions ranging from few micrometers to several meters. This is also possible due to the great availability of cutting tools for all materials and different shapes. The main aspect that drives the continuous development of this manufacturing technique is the possibility of making the machine work directly from a three-dimensional CAD (Computer-Aided Drafting) model. In this approach there is a reduction in manufac-turing costs and times, especially in the prototyping phase.

The possibilty to go down with the production scale to a few micrometers has given birth to the term micromilling, giving the possibility to produce increasingly complex parts with microscopic resolution (7).

(41)

2.1 Prototyping techniques for microfluidics

• machining the mold used in subsequent fabrication steps, for example embossing or injection molds (93).

• machining microchannels and features directly into the final part. In this case, the greatest advantage is the processing time that allows to have simple realizations even a few minutes after sending the CAD file.

For widely used metals, such as aluminum or steel, milling machining has been supported over time by many technical information given on the machining character-istics of the tools in dedicated manuals and datasheets. Furthermore, the use has been facilitated by the possibility to program the machines with ever clearer user interfaces and supported by complete revisions for the optimization of the projects. In contrast, milling at microscale structures in non-traditional materials such as plastics or poly-mers, is much less characterized, especially in the context of microfluidic devices (20). Thus, there is a need to fill this gap in technical knowledge to determine the usefulness of micromilling in microfluidics.

2.1.2 Laser engraving

Other microfabrication techniques try to speed up the prototyping process. The most sophisticated technique is the direct engraving of the substrate by means of a laser beam. For experimental purposes, this technique is often adopted to modify the chan-nel design to optimize the microfluidic chip performance. Furthermore, for biological applications, this technique is useful to go down with dimensions working with complex substrates like glass.

The work principle of this technique is based on the possibility to allow a direct writing by a laser. The microfluidic network can be drawn in real time moving the writing laser directly from a CAD design file and thus considerable reducing the pro-totyping times, see scheme in Fig.2.2. A substantial feature that makes the use of this technique very interesting for microfluidic applications is the ability to engraving chan-nels with semicircular sections, which is not easy to do with lithographic techniques (22). When is important to obtain a different cross section respect to the traditional

(42)

Figure 2.2: Scheme of a typical laser direct-writing setup for glass micromachining. The high power laser beam is driven trough a mirrors system and focused on the substrate surface to obtain high energy intensity. Then the ablation occurs and the microchannel is formed. This approach is typically used on the glass surface.(22).

rectangular form, the focusing of the light energy on the substrate with high spatial resolution allows the specific cross section profile fabrication. In Figure 2.3 different cross sections for laser engraving process can be observed.

Figure 2.3: Example of cross sections for laser engraved microchannel obtained by mudul-ing the laser power and focusmudul-ing light angle. a) Rectangular cross section b) Semicircular cross section (22).

Laser direct writing has been used successfully for fabrication of microfluidic chips on Silicon (Si) and various plastic substrates. In particular, laser ablation has been recently used to fabricate a staggered ridge mixer and an artificial system on a Si sub-strate in (68). The realization of these two structures arises difficulties for conventional photolithography techniques. Laser direct writing has also been used for the devel-opment of plastic microfluidic chips, including polymethylmethacrylate (PMMA) and polycarbonate (PC) substrates (58, 74).

(43)

precision writing of glass substrates (22). Although these techniques need a etching and provide surfaces without crack and debris, the UV laser systems and nanosecond ultra fast/ femtosecond lasers are high expensive and are not widely available.

A less expensive possibility used for glass microfabrication is a CO2 laser. This kind

of laser has been used for the realization of microfluidic chips on various plastic ma-terials and its application to glass substrate writing has not yet been fully developed

(21). Glass have a high absorption of the CO2 laser energy (λ = 10.6 µm), so the

micromachining of this substrate is essentially a photothermal process. Although a superior surface quality has been reported in the microdrilling of borosilicate glass and

silica glass (78), fabrication of microfluidic chips without crack using a CO2 laser has

not yet been observed. Several studies are taking advantage of the high processing

speed and the ease of maintenance of the CO2 laser for the development of an ablation

system for the rapid prototyping of microfluidic glass chips. Specifically, these studies

investigates the use of the CO2 laser system for the ablation of quartz, borofloat and

pyrex substrates, which are widely used for biological chip and laboratory applications in general (49).

The flexibility of the laser writing system is demonstrated through its application to the fabrication of several complex structures and cell patterning known to present a challenge to traditional photolithography processes (49).

2.1.3 Chemical engraving

Glass micro manufacturing technology is increasingly important because more glass substrates are used to make MEMS devices. Glass has many advantages, such as good mechanical properties, good optical properties, high electrical insulation and it can be easily bonded to silicon substrates at temperatures lower than for fusion bonding. However, glass is a very difficult material to work with milling or laser techniques as explained before. To look for a quick solution to its incision for the prototyping of channels one approache is the chemical attack by an acid. The chemical attack at different concentrations of hydrofluoric (HF) acid is the most used method because

(44)

the process is fast and a large amount of glass wafers can be processed simultaneously (150).

Masks for glass etching in HF used during chemical development are normally pho-toresist and Cr/Au combinations (114). In the development processes it is however impossible to prevent the formation of small holes due to inaccuracies and defects of the mask. This problem is greater when the chemical bath is aggressive to obtain deeper incisions. Actually the etching depth has been limited to 50 µm. Another issue found for HF etching is rapid undercutting of the chrome (Cr) mask, which leads to a more rapid lateral etching of the glass than vertical etching. This poor aspect ratio which is generally (< 1) for the carved cavities. Many other materials, such as SU-8, anodically bonded silicon, polysilicon, silicon carbide, amorphous silicon, electroplated gold, and their combinations have been employed to solve these etching problems (10). The removal of SU-8 after the hard bake required by deep etching is very difficult and time consuming. Etching through a silicon mask is required before it is anodically bonded onto the glass. This silicon etching process normally takes more than 10 hours for a 500 µm thick 4 inches wafer. Deposition of silicon-based thin film masks on the glass has a risk of contaminating the facilities with sodium ions (150).

2.1.4 Xurography

Xurography is a prototyping technique that employs a knife plotter to structure thin foils. This technique uses a cutting plotter typical of cutting graphics in adhesive vinyl films in industry, see Fig.2.4. The plotters used for cutting are classified by two main characteristics, i.e. mechanical and addressable resolution. The mechanical resolution depends on the positioning accuracy of the motors while the addressable resolution is related to the minimum displacement of the programmable step size (80).

Three cutting methods have been classified, specific for different applications. Drag knife. The cutter uses a blade that can rotate to follow the cutting path of the feature as it moves relative to the material. This approach introduces a lateral force from the blade at the corners, which can break the tip when cutting harder or thicker substrates.

(45)

Figure 2.4: Rapid fabrication based on xurography patterning and lamination; a) Typ-ical sketch of Xurography plotter. A controlled worktable moves the Vinyl film and the overhead cutting tool removes the layer to obtain a microfluidic structure. b) Alignment with adhesion to substrate and final assembly of microdevice.

True tangential. The blade position is checked by a driven motor. When cutting the corners, the blade lifts completely from the material and rotates in the new direction. Linear strokes can be truncated to ensure that the material is completely cut from top to bottom at the corners of the features. This is useful when cutting thick materials.

Emulated tangential. A blade is used which can turn but is lifted only on the surface of the material before rotating on the tip when passing over the corners. This reduces the lateral force on the blade.

To make xurography applicable to more materials, a friction cutting device with the ability to cut material up to 1 mm thick is required. Test models with round, rectangular and angled characteristics ranging from 10 µm to 2mm in width were sent to a few cutting plotter manufacturers to verify the quality of their production. Each manufacturer cut test patterns in sample materials ranging from 50 to about 1000 µm thick. Cutting plotters with tangential blades were able to cut rectangles and square patterns better then device with rotational blades. However, those with swivel blades were able to cut circular features down to 50 µm in diameter better than tangential blade machines due to the continuous cutting nature of the rotational blade (80).

(46)

useful and inexpensive research tool. The saved time and prototyping costs will allow researchers to focus on the device development before moving on to more expensive fabrication techniques.

2.1.5 Soft-Lithography

The photolithography technique is explained before introducing the Soft-Lithography. This technique derives from the technologies used in microelectronics for the realization of nanometer and micrometer structures through photoengraving and chemical devel-opment. This technique allowed the realization of the first microfluidic structures for the first LOC (Lab-On-a-Chip) (151) and the first devices for biological experiments such as PCR for DNA (8). However, this very powerful technique has some limitations that do not make it suitable for the rapid prototyping of microfluidic devices to be used as tests in biological experiments.

As already mentioned, lithography was created to solve complex problems of micro-electronic manufacturing, where a level of cleanliness and precision are necessary that go beyond the requirements of the biological application. To carry out lithography, it is necessary to invest important capitals for the building of clean-rooms and devices for the production of high resolution photomasks. Moreover, this technique exploits spe-cific photoresists on spespe-cific materials that could not be compatible with microfluidic experiments in biological and medical fields. The mechanical and optical characteristics of these substrates do not have also the suitable transparency and flexibility.

The main steps of the method are here described. The first step is the design of a high resolution mask that allows the spatial selective exposure of a photoresist film spun on a substrate. The thickness of liquid resist is tuned with different spin velocity. Two intermediate bake treatment are necessary before and after the exposure to prepare the resist film to the chemical bath. At the end of this procedure, a microfluidic mold in cured resist on a substrate is obtained, see Fig. 2.5.

One method to combine the strong potential of photolitography resolution and the rapid prototyping is the development Soft-Lithigraphy. This process is basically based on printing, stamping and embossing with an elastomeric stamp, see Fig.2.6. Some

(47)

Figure 2.5: Microfabrication through Lithography, where a negative photoresist is used to fabbricate a mold on a substrate. a) The liquid high density resist is placed with pipette on a substrate (typically silicon or glass). Trough a spin coater a thick film of resist is formed on the surface of the substrate. b) After a first bake, the resist is selective exposed thanks to the high resolution mask. In a case of negative photoresist (for example SU-8) the exposed areas are ready to the polymerization effect in a second bake treatment. c) During the chemical bath the exposed area are resistant to the developer dissolution and the structure is ready for the final hard bake .

examples are listed here: microcontact printing (µCP) (63), replica molding (REM) (140), microtransfer molding(153), micromolding in capillary(54), solvent-assisted mi-cromolding (SAMIM) (56), phase-shifting edge lithography (105), nanotransfer printing (52), decal transfer lithography (23) and nanoskiving (144).

Soft-lithography is widely used also in biological applications with a particular type of silicone elastomer, i.e. PDMS (poly dimethyl siloxane)(81). This type of material is a valid alternative to the traditional silicon and glass most used in microelectronics and optics and not indicated for biological use.

Some peculiar characteristics of PDMS are listed belowe:

• It has a shear module of 0.25 MPa and a Young module of about 0.5 MPa (char-acteristic of a moderately rigid elastomer). This elastomeric character makes it possible to conform to a surface and obtain contact at the atomic level, with high reproducibility of the particular, ideal characteristic for forming and sealing microfluidic systems.

(48)

• It is no-toxic and biocompatible, in addition to its very low commercial cost. • It is optically transparent up to about 300 nm light wave length.

• It presents a low hydrophobicity, around 110◦ contact angle, but can be made

hydrophilic temporally with chemical-physical treatments of the surface. • It can be irreversibly bonded on a substrates with plasma treatments.

Figure 2.6: Schematic illustration of the main steps involved in softlithography (molding replica). The cured resist structure is replaced by a liquid polymer (PDMS) enclosure. After the curing of polymer, the replica is removed and bonded trough a plasma treatment on a glass substrate. After bonding, holes are made with a puncer to allow the inlet and outlet of fluids.

2.2 Capillary driven microfluidics

In some micrometric devices (autonomous microfluidic devices), the capillary effects are so relevant to avoid external pumps to drive the flow. In this case the manufacturing technology is aimed at obtaining the maximum flow rate and layout efficiency. So the design is developed to achive the best integration of the different components of the device.

(49)

2.2 Capillary driven microfluidics

One of the main goals of Lab-on-a-Chip (LoC) technology is the integration of all the functionalities of a laboratory, from sample preparation to the readout of the analytical response, in a single chip without the need for external equipments. Such kind of devices shall integrate analytical detection, on-chip sensors and microfluidic handling capabilities onto the same chip. Enabling the step toward portable systems suitable for Point-of-Care diagnostics while ensuring high analytical performances.

Most of the current microdevices, indeed, rely on external pumping systems for their operation. Apart from considerations about portability and power consumption, one of the main issues related to the need of external actuation is the presence of interfaces between macro- and micro-fluidics which represent the weak point of the whole system, strongly affecting its reliability when operated outside the laboratory environment. One possible approach for overcoming this limitation consists in relying on capillary forces to drive the flow.

By using particular structures as capillary pumps, directional valves and trigger junctions it is possible to control the flow rate and the timing (106). However, in order to exploit capillarity, hydrophilic channels are needed: this significantly limits the use of polydimethylsiloxane (PDMS), which is typically used for rapid prototyping of microfluidic chips, due to its hydrophobicity. Although PDMS can be chemically or physically treated to shown hydrophilicity, the result is not permanent (17). In addition, if physiological solutions are used, as it often occurs in bioanalytical protocols, more complex and expensive treatments are needed (44). In the paper Caprini et al (15), a technology solution is proposed for the fast and low cost realization, of autonomous capillary systems that can operate with saline solutions.

In microfluidic capillary systems the ”engine” is the superficial tension and therefore

the fluid motion is a function of surface wettability and adhesion force. By using

particular structures it is possible to implement a predefined sequence of microfluidic operations and modify the behavior of fluid in the network. Microfluidic network design can start from the basic equations that hold for a microchannel with rectangular cross section thus characterized by height (h) and width (w). In particular, it is possible to