MASTER OF SCIENCE THESIS
SUNDSVALL, SWEDEN 2014
Development of an Expancel
Product through Optimisation of
Polymer Composition and the
Suspension Stabilising System
FRIDA BERGGREN
OFFICIAL REPORT
Supervisor
Dr. Magnus Jonsson
Examiner
Prof. Eva Malmström Jonsson
I
Abstract
Thermally expandable microspheres are spherical particles around 5-‐40 µm in size, consisting of a polymeric shell in which a blowing agent has been encapsulated. The microspheres are expanded upon heating, resulting in a particularly low density. Microspheres are therefore suitable to use as light weight filler or as foaming agent.
AkzoNobel is world leading in the production of expandable microspheres, which are
commercialised under the name Expancel. Sustainability is a great focus at AkzoNobel and two issues that AkzoNobel works with today is to develop products free from chlorine and Me1. The aim with this thesis has been to investigate whether it is possible to produce microspheres free from these chemicals and to see if they can be a more sustainable alternative to one of the existing Expancel grades.
In this study, the microspheres have been produced through free radical suspension polymerisation and analysed by measuring mainly the particle size and expansion properties. The polymeric shell was composed of the monomers acrylonitrile, methacrylonitrile, and methyl acrylate. The main focus has been to evaluate the silica-‐based stabilisation system, which stabilise the monomer droplets during the suspension polymerisation. The stabilisation is possible due to the formation of silica flocs that is adsorbed on the surface of the droplets. It has been investigating how different parameters, e.g. amount of stabiliser or mixing procedure, affects the formation of silica flocs and the stabilisation of monomer droplets.
For the silica-‐based system, it was found that the mixing order, stirring rate, and amount of stabilisers affect the formation of flocs. It was also seen that the amount of stabiliser affect the stabilisation of droplets, and that some stabilisers is more significant than others.
Microspheres without chlorine and Me1 have successfully been produced in laboratory scale (50 mL and 1 L). The expansion and size of the microspheres produced in this study was relatively similar to one of the existing Expancel grades. However, the reproducibility of polymerisations in 1 litre
Sammanfattning
Termiskt expanderbara mikrosfärer är sfäriska partiklar, ca 5-‐40 µm i diameter, som består av ett polymerskal som innesluter en drivgas. Mikrosfärerna expanderar när de utsätts för värme och erhåller då en mycket låg densitet. De är därför lämpliga att använda som fyllmedel då låg vikt är önskvärt eller som skummedel.
AkzoNobel är världsledande inom produktion av expanderbara mikrosfärer, som marknadsförs under namnet Expancel. Hållbar utveckling är en viktig fråga för AkzoNobel och två problem som de står inför idag är att utveckla produkter fria från klor och Me1. Målet med detta examensarbete har varit att undersöka om det är möjligt att framställa mikrosfärer fria från dessa kemikalier och om de framtagna mikrosfärerna skulle kunna vare ett hållbarare alternativ till en av de befintliga Expancel-‐ produkterna.
I den här studien har mikrosfärerna framställts genom suspensionspolymerisation som initierats av fria radikaler och de har analyserats främst genom att mäta partikelstorlek och
expansionsegenskaper. Polymerskalet bestod av monomererna akrylnitril, metakrylnitril och metylakrylat. I det här arbetet har det viktigaste varit att utvärdera det silikabaserade
stabiliseringssystemet som stabiliserar monomerdropparna vid polymerisationen. Stabiliseringen är möjlig eftersom silika bildar flockar som adsorberar på ytan av monomerdropparna. Olika
parametrar, exempelvis mängd stabiliseringsmedel och satsningsförfarande, har därför varierats för att undersöka vilken effekt det får på flockningen av silika och stabiliseringen av monomerdroppar. Satsningsordning och omrörningshastiget för stabiliseringssystemet samt mängd stabiliseringsmedel är några av de faktorer som påverkar bildningen av flockar. Det konstaterades även att mängd stabiliseringsmedel påverkar stabiliseringen utav monomerdropparna.
Fulländade mikrosfärer utan klor och Me1 har framställts i laboratorieskala (50 mL och 1 L) och partikelstorleken samt expansionsegenskaper är jämförbara med en av Expancels nuvarande
produkter. Dock har reproducerbarheten i 1 litersskala varit otillfredsställande.
III
List of Abbreviations
Tg glass transition temperature
LALLS low angle laser light scattering
Dv(0.5) median diameter of volume distribution
Dn(0.5) median diameter of number distribution
Span particle size distribution
TMA thermomechanical analysis
Tstart temperature where microspheres start to expand
Tmax temperature where maximum expansion is reached
TMA-‐dens density at maximum expansion
TGA thermogravimetric analysis
SEM scanning electron microscopy
TC Tinyclave
ACN acrylonitrile
MAN methacrylonitrile
MA methyl acrylate
NaOH sodium hydroxide
HAc acetic acid
LX colloidal silica
KO polyelectrolyte
Me2 Me2 salt
IP isopentane
IB isobutane
HLKIM LX-‐system mixed in the order HAc, LX, KO, Initiator 1, Me2
LKIMH LX-‐system mixed in the order LX, KO, Initiator 1, Me2 HAc
MKLIH LX-‐system mixed in the order Me2, KO, LX, Initiator 1, HAc
MLKIH LX-‐system mixed in the order Me2, LX, KO, Initiator 1, HAc
Table of Contents
1. Introduction ... 1 1.1 Background ... 1 1.2 Objective ... 2 2. Theory ... 2 2.1 Suspension Polymerisation ... 3 2.2 Silica System ... 5 2.2.1 Flocculation ... 5 2.2.2 Colloidal Silica (LX) ... 5 2.2.3 Polyelectrolyte (KO) ... 6 2.2.4 Me2 Salt ... 6 3. Experimental Methods ... 7 3.1 Materials ... 73.2 Flocculation of Silica: Floc Size and Turbidity Measurements ... 7
3.3 Synthesis of Microspheres in Tinyclave Reactor ... 7
3.4 Synthesis of Microspheres in 1 Litre Reactor ... 7
3.5 Characterization Methods ... 7
3.5.1 Particle Size Analysis ... 8
3.5.2 Turbidimetry ... 9
3.5.3 Thermomechanical Analysis ... 9
3.5.4 Thermogravimetric Analysis ... 9
3.5.5 Sieve Residue ... 10
3.5.6 Hot-‐stage Microscopy ... 10
3.5.7 Scanning Electron Microscopy ... 10
4. Results and Discussion ... 10
4.1 Influence of Mixing Parameters on Silica Flocculation ... 10
4.1.1 Mixing Order ... 11
4.1.2 Concentration of LX, KO and Me2 ... 13
4.1.3 Effect of Stirring ... 15
4.1.5 The Initiator’s Influence on Floc Size Measurements ... 15
4.2 Stabilisation of Toluene Emulsions ... 16
4.3 Polymerisation in Tinyclave ... 18
4.3.1 Particle Size and Expansion Properties ... 18
V
4.3.3 Homogenisation Speed ... 20
4.3.4 Effect of Me2 ... 20
4.3.5 Hydrolysis of KO ... 21
4.4 Polymerisation in 1 Litre Reactor ... 21
5. Conclusions ... 25 6. Future Work ... 25 7. Acknowledgements ... 26 8. References ... 27 9. Appendix ... 30
1. Introduction
AkzoNobel is world leading in the production of thermally expandable microspheres, a multi-‐ purpose additive which can be used in numerous applications. The patent was originally developed by Dow Chemical Co in the 1970’s and was taken on by AkzoNobel and further improved. The first order of microspheres was shipped in 1980, commercialised under the name Expancel. [1] The microspheres resemble tiny ping-‐pong balls with a diameter of 5-‐40 μm, and consist of a polymer shell that encapsulates a blowing agent. [2] When the microspheres are heated, the blowing agent will increase the pressure at the same time as the polymer shell will become soft and ductile and this causes the microspheres to expand. The microspheres can expand 60 to 80 times its own volume, and the expanded volume is retained after cooling. [3] Expanded microspheres have a particularly
low density (15-‐70 kg/m3) and are therefore suitable for use as lightweight filler. Microspheres also
offer exciting features like thermal insulation, sound insulation, increased solar reflection and increased friction on surfaces. The thermal expansion makes it suitable to use as an expanding agent or foaming agent, and it offers a more controlled and uniform foam structure when compared to other foaming agents. Examples of applications where microspheres are used are artificial leather, plastic foams, paper and board, wine stoppers and printing ink for wallpaper and textiles. [4]
1.1 Background
Sustainability is a great focus at AkzoNobel and the company constantly works to reduce its environmental impact. [5] As a result, AkzoNobel was ranked as number one in the Dow Jones Sustainability Index (DJSI) in the Materials Industry Group (previously the Chemicals Supersector) in 2013 and has been top three for the last eight years. [6] It is thus of great interest to continue to improve the formulations of AkzoNobel products and replace chemicals which have a negative impact on the environment with more sustainable alternatives. The demands from public authorities constantly increases and it is important for AkzoNobel to always strive to minimise their
environmental footprint.
The polymer that makes up the shell in the microspheres is composed of a number of constituents called monomers. One of the monomers used in several Expancel grades is vinylidene chloride (VDC), which is polymerised into PVDC. The hazardous risk with PVDC is that hydrochloric acid (g) is formed when the plastic is decomposed, e.g. if it is exposed to fire. [7]
The metal ion Me1 contributes to the stabilisation of the monomer droplets during the formation of
microspheres. [8] Me1 is relatively harmless compared to other pollutants, but further processing might generate environmental issues. [9-‐11]
AkzoNobel has performed much research to evaluate different chlorine free monomers to replace VDC in the existing grades of Expancel. It has been discovered that a shell made from acrylonitrile (ACN), methacrylonitrile (MAN) and methyl acrylate (MA) gives similar properties as the shell in an existing grade in terms of expansion. [12-‐13] This monomer combination is therefore used in this study.
Attempts to replace Me1 with other metal ions have been made and AkzoNobel has successfully replaced Me1 with Me2 in a number of their products. The next step is to investigate if Me1 can be replaced in further grades in which Me1 is still used. [14-‐15]
1.2 Objective
The purpose of this project was to investigate whether it is possible to produce microspheres free from Me1 and chlorine and evaluate whether these microspheres can be a more sustainable alternative to one of the existing products of Expancel. The main focus has been to study the stabilisation system, which is a mixture of chemicals that ensures the formation of microspheres. The stabilisation system normally contains Me1, but it has in this study been replaced by Me2. It is not fully understood how the stabilisation system interacts with the monomer droplets and another aim of this work has therefore been to obtain a deeper understanding of the stabilising system. This has been accomplished by investigating how different parameters affect the stabilisation and the formation of microspheres.
In this study, the microspheres have a polymer shell of ACN, MAN and MA, which has been proven to give expansion properties similar to the existing grade. One challenge has been to tailor the stabilisation system so to suit the selected monomer composition.
2. Theory
Microspheres consist of a thermoplastic polymeric shell, which surrounds a core with a volatile hydrocarbon. When the microspheres are heated, the hydrocarbon vaporises and the internal pressure is increased in the microsphere. At the same time, the polymeric shell becomes soft and
ductile as it reaches its glass transition temperature (Tg). The microspheres start to expand when the
internal pressure of the hydrocarbon gas exceeds the yield strength of the polymer and the decrease in density is substantial since the mass remains the same while the volume increases tremendously, see Figure 1. The hydrocarbon works as a blowing agent, and the expansion is controlled by the type
and amount of encapsulated blowing agent and the Tg of the polymer. The expansion continues as
long as the internal pressure exceeds the yield strength of the polymer shell, or until the shell breaks, or becomes so thin that the hydrocarbon diffuses through the shell, causing the microspheres to decrease in volume. [3]
Figure 1. Schematic illustration of expandable microspheres prior to and after expansion. The size of Expancel microspheres varies between 5-‐40 μm (unexpanded) and 20-‐120 μm (expanded), depending on the grade.
The glass transition temperature (Tg) is a property that is unique for polymers and it is fundamental
in polymer technology, since the polymers’ properties are dependent on it. It is defined as the temperature where large scale segments (20-‐50 atoms long) in the polymer backbone start to move co-‐ordinately. The mobility of the polymer chains is very restricted at low temperatures, and the
polymer will be stiff and brittle below its Tg, almost like glass. This state is therefore called glassy
Unexpanded Density ~ 1000 kg/m3 Expanded Density ~ 30 kg/m3 Ø40 μm 0,1 μm 2 µm Ø12 μm Heat
state. When the temperature is elevated, the chains are able to move around more freely. This will make the polymer soft and flexible and it also loses its shape stability. The polymer is therefore said
to be in its rubbery state when the temperature is over the Tg. The stiffness (E-‐modulus or Young’s
modulus) is decreased by a factor of 1000 when the polymer is heated from the glassy state to the rubbery state. [16]
Some of the properties related to the composition of the polymeric shell are Tg, chemical resistance
and barrier properties. The microspheres synthesised in this study have a shell that consists of a co-‐ polymer of acrylonitrile (ACN), methacrylonitrile (MAN) and methyl acrylate (MA). ACN is the major component and is used because of its excellent barrier properties and chemical resistance, which is due to its semicrystalline structure and high cohesive strength. [3] The barrier properties are very important for the expansion of the microspheres since they determine how much of the blowing agent that is lost through diffusion through the polymer shell which is detrimental for the expansion.
MA is added to lower the Tg and as a result making the shell more ductile. Another way to alter the
properties of the polymer shell is to introduce a cross-‐linker, which decreases the mobility of the
polymer chain and increases the Tg. The structure becomes denser and this will increase the shell’s
chemical resistance. Cross-‐linking of the shell is known to have a large effect of the expansion
properties, especially on Tmax, the temperature when maximum expansion occurs. [17-‐18]
The expansion properties of the microspheres can be altered by using different hydrocarbons as blowing agents. The temperature at which the microspheres start to expand is related to the boiling point of the hydrocarbon; a lower boiling point will give a lower expansion temperature and vice versa. Figure 2 illustrates some common blowing agents, their boiling points, and vapour pressure.
isobutane b.p. -‐11.7 °C v.p. 20 bar (100 °C) isopentane b.p. 27.7 °C v.p. 7 bar (100 °C)
Figure 2. Chemical structure, vapour pressure (v.p), and boiling point (b.p) of common blowing agents used in microspheres.
The amount of encapsulated hydrocarbon has been shown to affect the expansion. A too low amount of hydrocarbon does not give a sufficient internal pressure, although a too high amount of hydrocarbon is not good either. A too high internal pressure will make microspheres pop like balloons, or make the polymeric shell too thin, causing the hydrocarbon to diffuse through the shell. Branched hydrocarbons, as the two examples above, are often used since they are more prone to remain in the polymeric shell because of steric hindrance. [3]
2.1 Suspension Polymerisation
The microspheres are produced via free radical suspension polymerisation. The system is composed of two immiscible phases, where the monomers, cross-‐linker, volatile hydrocarbon, and radical initiator compose the organic phase, and water together with stabilising agents compose the water phase. The phases are mixed and the organic phase is broken into droplets by vigorous stirring, see Figure 3. For safety reasons, the initiator is added to the water phase but migrates to the organic phase upon mixing, where it starts the radical polymerisation. When the emulsion is heated, the initiator forms radicals that initiate the polymerisation. Each droplet behaves as a miniature reactor,
and the reaction can therefore be considered as a bulk polymerisation in terms of kinetics. Compared to bulk polymerisation, the viscosity is generally lower since 70% of the total volume is water. The water allows for good heat transfer so that the excess heat generated from the exothermic polymerisation can be removed easily, which reduces the risk of Trommsdorff effect (auto-‐acceleration). A good temperature control during the reaction is important to ensure a safe process and stable product properties. [3, 19-‐20]
Figure 3. Schematic illustration of suspension polymerisation. Figure adopted from [21].
Figure 4. In situ encapsulation of the volatile hydrocarbon. The organic phase (1), containing monomers, cross-‐linker and
the volatile hydrocarbon, forms drops which is stabilised by the stabilising system (2). The radical initiator (I•) starts the
polymerisation and the monomers are converted into polymers (3). A core/shell structure starts to form, with the volatile hydrocarbon (4) in the centre. The polymers fuse into a polymer shell (5), encapsulating the volatile hydrocarbon (6).
The encapsulation of the volatile hydrocarbon is done in situ during the polymerisation, see Figure 4. As long as the reaction is thermodynamically controlled, the interfacial tensions between different phases will determine the morphology of the polymer particles. The driving force of the system is to minimize the interfacial energy, and for a core/shell structure, the interfacial energy is minimized when
𝛾!" > (𝛾!"+ 𝛾!")
where γwo, γwp and γop are the interfacial tensions of the water/oil interface, water/polymer interface
and oil/polymer interface, respectively. [22] Ultimately, this means that water and oil is less prone to be in contact with each other at the same time as the newly formed polymer is able to be in contact with both water and oil.
The initial size of the monomer droplets is an important quantity since it controls the size of the final microspheres. The size of the droplets is controlled by stirring speed, volume fraction of organic phase, and amount and type of stabiliser. The sizes of the droplets are dependent on the balance of
1 2 3 4 5 6 I•
droplet break-‐up and droplet coalescence. [23] The monomer droplets will increase in viscosity during polymerisation and since coalescence still occurs during this stage, there is a risk that the microspheres will agglomerate. In Expancel microspheres, the particle size is controlled by inorganic particles, which adsorbs to the monomer/water interface and prevents the droplets from
coalescence. One of the stabilising systems consists of colloidal silica and it is further discussed
below, whereas another system uses Mg(OH)2. [3] Depending on the composition of the
microspheres and the intended application, either the acidic silica system or the alkaline
Mg(OH)2-‐system is used.
2.2 Silica System
Colloidal silica is the main component in the silica system. Previous studies [24-‐25] show that in order for the silica to stabilise the monomer droplets adequately, the silica particles needs to be bundled together into flocs. The process is called flocculation and it is widely used in the water treatment industry to remove fine particles from water. [26]
2.2.1 Flocculation
The flocculation takes place due to ionic interactions between colloidal silica (further in this report called LX) and a polyelectrolyte (KO). The pH is set to approximately 4.5 in the experiments conducted in this work, which means that KO is positively charged while LX is negatively charged. Ionic species in solution with opposite charges attract each other [27], hence KO adsorbs on the surface of LX. This will cause a neutralisation of the surface charge, which lowers the ionic repulsion between LX particles, allowing them to come closer to each other and flocculate. If KO is oriented so that hydrophobic areas are produced, these areas can form micelles with hydrophobic areas on other particles. [28] The particles are held together by so called hydrophobic-‐hydrophobic interactions, which form due to strong surface tension forces. If the KO chain is long enough, another possible flocculation mechanism is formation of bridges between adjacent LX particles. [33] However, most of the KO chains are probably too short to form bridges and the flocculation is believed to be mainly due to charge neutralisation and hydrophobic-‐hydrophobic interactions. 2.2.2 Colloidal Silica (LX)
Colloidal silica consists of solid spherical silica particles, which are approximately 12 nm in diameter. [29] The solution is stable against settling since the particles are small enough not to be dragged to the bottom by gravity. The surfaces of the particles are covered with silanol groups, which make the particles hydrophilic, see Figure 5. Depending on the pH, the silica particles can either be neutral or negatively charged. The net surface charge of silica is zero at pH 2 and this is defined as the
isoelectric point. The silanol groups become deprotonated at pH higher than 2, which gives a negative surface charge. The negative charge will affect the distribution of ions in the solution and an electrical double layer will form. Positive ions in the solution are attracted to the negative charge and form a mono-‐layer on the silica surface that partially compensates the surface charge. This layer is referred to as the Stern layer. A diffuse region is formed around the Stern layer, where the ions are distributed according to the influence of electrical forces and random thermal motion. This layer is referred to as the diffuse layer. [27]
Figure 5. Illustration of the surface of colloidal silica.
2.2.3 Polyelectrolyte (KO)
Polyelectrolytes are polymers carrying either positively or negatively charged groups. [30] In this study, the silica particles are flocculated by addition of a polyelectrolyte (KO) made through polycondensation of adipic acid and diethanolamine, see Figure 6. Branched products are possible, since diethanolamine has three functional groups and is able to form both ester bonds and amide bonds. [31] These bonds are sensitive to hydrolysis and KO will over time decrease in molecular weight. However, the hydrolysis reaction is slow enough why the shelf life of KO is sufficient under normal conditions. [32] KO is mores sensitive towards alkaline conditions compared to acidic conditions since the hydrolysis is catalysed by high pH. [39]
Figure 6. The polyelectrolyte KO consists of adipic acid and diethanolamine and the chain can have various length and structure. This structure is just an example which illustrates that both ester bonds and amide bonds are possible. The charge of KO is very important for the flocculation process. The charge will be positive at acidic conditions due to protonation of the hydroxide groups and negative at alkaline conditions due to deprotonation. The net charge is also dependent on addition of salt, since ions with opposite charge will have a screening effect on the charges on KO.
2.2.4 Me2 Salt
Me2 salt is added to the LX-‐system since it subsidies the flocculation of LX, it is, however, not clear in what way. One theory is that Me2 reduces the electrical double layer, which makes KO adsorb more effectively to LX. [33] From empirical studies, Me2 is known to reduce the sieve residue and the risk of agglomeration during polymerisation. Previous work has shown that Me2 flocculates faster than Me1. [33-‐34]
3. Experimental Methods
The LX-‐system has been evaluated by monitoring the floc size and turbidity over time. The LX system has also been evaluated by polymerising in small reactor vessels called Tinyclave (TC) and 1 litre reactors.
3.1 Materials
Acrylonitrile, methacrylonitrile, methyl acrylate, isopentane, isobutane, cross-‐linking agent (XA), sodium hydroxide (1 M, aq), acetic acid (10 wt %, aq), silica (40 wt %, aq), polyelectrolyte (KO) (34 wt %, aq), Me2 solution (15.1 wt %, aq), Me2 solution (1.51 wt %, aq), sodium bisulfite (20 wt %, aq), initiator 1 (25 wt %, aq), initiator 2 (40 wt %, aq).
All chemicals were provided by AkzoNobel and used as received.
3.2 Flocculation of Silica: Floc Size and Turbidity Measurements
The stabilisation system was prepared by mixing deionized water (120.0 g), NaOH (4.7 g), 1.51 wt% Me2 solution (0.4 g), KO (1.2 g), LX (7.1 g), initiator 1 (1.1 g) and HAc (7.6 g). The solution was stirred with magnetic stirrer (IKAMAG REO) at 500 rpm for 40 min. Floc size analysis was performed after 5, 10, 20, 30, 40 min of stirring and the turbidity was measured after 15, 25, and 45 min of stirring.
3.3 Synthesis of Microspheres in Tinyclave Reactor
In a typical experiment, the organic phase was prepared by mixing XA, ACN, MAN, MA and IP in a 50 ml flask with lid and put into a refrigerator. The stabilisation system was prepared by mixing
deionized water (17.0 g), NaOH (0.9 g), Me2 solution (0.8 g, 1.51 wt %), KO (0.2 g), LX (1.4 g), initiator 1 (0.2 g) and HAc (1.5 g) into a glass E-‐flask followed by stirring for 30 min. The stabilisation system was mixed with the organic phase (10.9 g) and emulsified using a Silverson high shear mixer for 45 s at 11500 rpm. Polymerisation was performed in a 50 ml glass reactor (Tinyclave from Büchi) in a shaking water bath at 56 °C for 20 h.
3.4 Synthesis of Microspheres in 1 Litre Reactor
In a typical experiment, the organic phase was prepared by mixing XA, ACN, MAN, and MA in a 1000 ml flask with lid and put into a refrigerator. The stabilisation system was prepared by mixing
deionized water (419 g), NaOH (16.9 g), Me2 solution (1.4 g, 15.1 wt %), KO (4.3 g), LX (25.4 g), initiator 1 (4.0 g) and HAc (27.4 g) into a 1000 ml flask followed by stirring for 30 min. Thereafter, the stabilisation system, IB (52 g) and the organic phase (157.2 g) were emulsified with an Ultra Turrax® IKA T18 at 6500 rpm for 30 s. The speed was ramped to 21500 rpm during 30 s and mixed at that speed for 45 s. The suspension was added to the reaction vessel and the stirring speed was set to 500 rpm, after which polymerisations was performed as in Tinyclave reactor.
To remove monomer residues, sodium bisulphite (20 w%) (60 g) was added to the slurry after the polymerisation.
3.5 Characterization Methods
The flocs have been evaluated by particle size analysis and turbidity. The microspheres have been evaluated by particle size analysis, TMA, TGA, hot-‐stage microscopy and SEM.
3.5.1 Particle Size Analysis
The size of the microspheres and the silica flocs are measured by a Malvern Mastersizer 2000 equipped with a dispersion unit Hydro 2000SM. The instrument analyses the particle shape and size by means of light scattering (LALLS) from a He-‐Ne laser. The instrument has a light sensitive detector, which consists of many elements to be able to detect the light scattering from many different angles. The light scattering pattern is recalculated into volume by means of density and refractive index for the sample and the solution it is dispersed in. Based on this, the diameter of a sphere with
equivalent volume can be calculated. By this action, the particle size is described with only one number (the diameter) even though they have an irregularly shape, see Figure 7. [35]
Figure 8. A typical size distribution curve, which defines the parameters Dv,n(0.1), Dv,n(0.5), Dv,n(0.9), and the span.
Particle size can be presented in several different medians and means. In this report, the particle size
is mainly analysed using volume median (Dv(0.5)) and number median (Dn(0.5)). Volume and number
distributions can be very different so it is important to study both quantities. Dn(0.5) can in some
cases give a better value of what size that is the most common, especially if the sizes are very different. This is because if all the particles where lined up according to size, the size on the one in
the middle would be Dn(0.5). Dv(0.5) gives a better picture of where the volume of the particles are.
This means that Dv(0.5) is sensitive to larger particles while small particles will only have a small
impact on the result. The particle size distribution is determined by the span, which is defined as Equation 1. Hence, a narrow peak is corresponds to a low particle size distribution. A typical size distribution curve in which the parameters are defined can be seen in Figure 8.
Figure 7. Schematic illustration showing how the size of an irregularly shaped particle is calculated. The size and shape of the particle is measured by means on light scattering and the light scattering pattern is recalculated into volume. The size of the irregularly shaped particle is reported as the diameter of a sphere with equal volume. By this method, the size can be represented with only one number, even though the particle has an irregular shape.
span
Dv,n(0.5)
𝑠𝑝𝑎𝑛 =𝐷!,! 0.9 − 𝐷!,!(0.1)
𝐷!,!(0.5)
Equation 1
This equipment is developed for solid particles, hence measurements on the silica flocs is
problematic. In this study, refractive index and density for solid silica has been used since the data for flocculated silica is unknown. The floc size has in this study been reported as 6-‐24 μm which is too high, since the flocs are believe to be in nano size. The equipment has still been used with the motivation that the floc sizes can be relatively compared.
3.5.2 Turbidimetry
Turbidimetry determines the amount of cloudiness in a solution by means of measuring the
transmission of light through the sample. When a light beam enters the cuvette with the sample, the light can either be transmitted through the cuvette or scattered in various directions. The amount of scattered light depends on the number of particles and the size of the particles. Small particles scatter the light more compared to large particles. The amount of scattered light, or turbidity, is measured in Nephelometric Turbidity Units (NTU).
3.5.3 Thermomechanical Analysis
Thermomechanical Analysis (TMA) determines the expansion properties of the microspheres. A small sample of microspheres is placed in an oven and heated according to a temperature program. A sensitive probe is in contact with the sample and the probe measures the height difference of the
sample as it is heated. Tstart is defined as the temperature where expansion starts and Tmax is defined
as the temperature where the maximum probe displacement is measured, see Figure 9. TMA-‐
density is the density of the microspheres at Tmax as calculated from the volume change of the
sample.
Figure 9. A typical TMA thermogram defining the parameters Tstart, Tmax, and maximum probe displacement. [18]
3.5.4 Thermogravimetric Analysis
TGA evaluates the thermal stability of the microspheres and is used to determine the amount of encapsulated blowing agent. The analysis is performed by placing a small sample on a sensitive balance and the change of sample weight is recorded when the temperature is increased. The first weight change will be when the blowing agent is diffusing through the collapsing microspheres. Further heating of the sample causes thermal degradation of the polymer. Finally, the sample is oxidatively combusted when air is let into the system, leaving only ash as residue.
Figure 10. Typical TGA-‐curve of a microsphere. The first slope corresponds to the vaporisation of blowing agent, the second slope is where the polymer decomposes and the third slope is the combustion of the sample, leaving only ash. 3.5.5 Sieve Residue
After polymerisation, the microsphere slurrys were filtered through a 63 μm sieve and rinsed with water. The amount of sieve residue was determined by visually evaluating how much of the slurry that was captured by the sieve. The sieve residues are relative and graded from 1-‐5, where 1 represent no sieve residue and 5 the highest amount. A large sieve residue indicates that the slurry contain a lot of large particles, agglomerated or in other ways defect microspheres. The aim is therefore to obtain as low sieve residue as possible.
3.5.6 Hot-‐stage Microscopy
Optical hot-‐stage microscopy was used to study the expansion behaviour, using a Zeiss optical microscope at 100x magnification equipped with a hot-‐stage equipment from Leitz Wetzlar (Germany). The temperature was controlled by manually regulating the power supply, while monitored by a thermometer attached to the hot-‐stage set up.
3.5.7 Scanning Electron Microscopy
Field-‐emission scanning electron microscope (FE-‐SEM) images were recorded on a Hitachi S-‐4800 FE-‐SEM. The samples were mounted on a substrate with carbon tape and sputtered for 150 s at 40 mA with gold/palladium (Cressington 208HR).
4. Results and Discussion
The stabilisation system is crucial in suspension polymerisation, because it is a prerequisite for formation of individual particles and a stable emulsion. Both the size and the shape of the microspheres are dependent on the stabilisation. In this work, experiments have been conducted with the LX-‐system to better understand the flocculation of silica and how different experimental parameters affect the floc size. Different types of stabilisation systems have been evaluated by polymerising in Tinyclaves (TC) and 1 litre reactors.
4.1 Influence of Mixing Parameters on Silica Flocculation
The flocculation of silica is a complex process and it is not fully understood how KO and Me2 interact with LX to form flocs. The flocculation depends on many parameters e.g. pH, concentration of flocculants, stirring rate, and mixing order. [24, 36] It is therefore desirable to develop a better understanding of how these parameters effect the flocculation. In this part of the study, the effect of various changes in the LX-‐system has been evaluated by measuring the floc size and the turbidity of the dispersion over time.
Sa m p le w e ig h t
4.1.1 Mixing Order
A prerequisite for flocculation is adsorption of KO and Me2 on to the silica particles due to ionic interactions. The surface charges of LX, KO and Me2 are pH dependent and it is therefore interesting to see if the flocculation will be different if NaOH and HAc are added to the suspension in different orders. Different addition orders of LX, KO and Me2 have also been tested. The conducted
experiments and the addition order of the chemicals can be seen in Table 1.
Table 1. The components in the stabilising system were added in different orders to investigate if the addition order has an effect on the floc size. The stabilising system were named HLKIM, LKIMH, MLKIH and MKLIH.
Addidtion
order HLKIM LKIMH MLKIH MKLIH
1 Water Water Water Water
2 NaOH NaOH NaOH NaOH
3 HAc LX Me2 Me2
4 LX KO LX KO
5 KO Initiator KO LX
6 Initiator Me2 Initiator Initiator
7 Me2 HAc HAc HAc
As can be seen in Figure 11, the size distribution of silica flocs is very dependent on the addition order. In HLKIM, HAc is added before LX, KO and Me2 adjusting the pH to 4.4. The flocculation then starts immediately after addition of KO, and since the flocculation is so fast, it is not obvious how the addition of Me2 contributes to the flocculation. Me2 forms small cationic hydroxide particles at pH 4.4, and ionic interaction with the negatively charged silica is therefore one possible explanation. [37]
Figure 11. Differences in size distribution of silica flocs depending on the addition order of components in the system. (―)
HLKIM, (―) LKIMH, (―) MLKIH and (―) MKLIH after 20 min of flocculation.
In LKIMH, the pH is 12.4 prior to the addition of LX, KO and Me2. A pH this high will result in negatively charged LX since a majority of the silanol groups will be deprotonated. KO will also be negatively charged due to deprotonated functional groups. Me2 is believed to form uncharged
hydrated oxides [37] that can be observed in the suspension as small particles. These particles are probably the reason to why the colour of LKIMH is darker compared to HLKIM, see Figure 12. Due to the charges of LX, KO and Me2, the flocculation does not start until addition of HAc. This seems to have an effect on the size distribution, since the size span is more narrow for LKIMH compared to HLKIM. This may be due to better dispersion of LX, KO, and Me2 since the chemicals have time to mix before flocculation. It can be that KO and Me2 more easily can interact with LX since they are close to each other when HAc is added, resulting in smaller, denser flocs.
Figure 12. Silica suspensions after 30 minutes of flocculation. LKIMH (to the left) where HAc is added after LX, KO and Me2 has a darker colour compared to HLKIM (to the right) where HAc is added before LX, KO and Me2.
What appears to be even more significant is when Me2 is added, since LKIMH and MLKIH have completely different size distributions and the only difference between them is when Me2 is added. This difference in size distribution is unexpected. It seems unlikely that LX, KO and Me2 are able to interact with each other due to their charges, so it is strange that the order of LX, KO and Me2 have such a large impact on the size distribution. This suggests that the interactions between LX, KO and Me2 are not only dependent on ionic interactions. Even small changes in the addition order seem to have an effect on the floc size, which can be seen by comparing MLKIH and MKLIH where the only difference is the order of LX and KO.
4.1.2 Concentration of LX, KO and Me2
The amount of LX, KO and Me2 affects the flocculation and the following experiments aim to
elucidate how these factors affect the floc size by a 23 fractional design in the experimental setup.
Two levels and a centre point were chosen for each factor, giving a total of nine measure points. The experimental design can be seen in Table 2 and Figure 13.
Table 2. The effect of LX, KO
and Me2 on flocculation was examined using a
23 factorial design. The weights are stated in
amounts per TC and are referred to as Low (L), medium (M), or high (H).
Trial
Factor
LX [g] KO [g] Me2* [g] LLL 0.7 0.1 0.2 LHL 0.7 0.4 0.2 LHH 0.7 0.4 1.4 LLH 0.7 0.1 1.4 HHH 2.1 0.4 1.4 HLH 2.1 0.1 1.4 HLL 2.1 0.1 0.2 HHL 2.1 0.4 0.2 MMM 1.4 0.2 0.8 *15.1 wt% Me2-‐solutionThe floc sizes of the conducted stabilising systems are summarised in Figure 14. Except from HLL and HLH, the differences in floc sizes are small. HLL and HLH have a considerable larger floc size as compared to the other samples and both samples have a high level of LX and KO. It is interesting to see that by increasing the KO level, the floc size decrease drastically (compare HLH with HHL and HLH with HHH). This is unexpected since previous work [33] state that higher levels of KO will flocculate LX more. Another trend is that high levels of Me2 result in larger flocs, which can be seen when comparing LHL with LHH and HLL with HLH.
Figure 15 and Figure 16 illustrates flocculation experiments where KO and Me2, respectively, have been varied while the other parameters were kept constant. These diagrams give a better picture of how the floc size changes when one parameter is varied. It seems like a higher amount of KO results in an increase of smaller flocs (~ 4-‐10 μm), the same trend that was seen in Figure 14.
A change in the size distribution can also be seen in Figure 16, when Me2 is varied. Higher levels of Me2 seem to result in an increase of larger flocs (~ 40 μm) and a decrease in smaller flocs (~ 10 μm). The same trend was also seen in Figure 14.
Me2 LX KO + High + High + High -‐ Low -‐ Low -‐ Low
Figure 13. The flocculation experiments were conducted according
to a 23 fractional design. The factors are the amounts of LX, KO and
Me2. Two levels and a centre point were chosen for each factor.
Figure 14. Floc sizes for different compositions of the stabilising system measured after 20 minutes of flocculation. Components were added in following order: HAc, LX, KO, initiator, Me2.
Figure 15. Differences in size distribution of silica flocs with various amounts of polyelectrolyte (KO), (―) 0.1 g, (―) 0.2 g
and (―) and 0.4 g. Other components were added in amounts according to the recipe described in section 3.2.
Figure 16. Differences in size distribution of silica flocs with various amounts of Me2, (―) 0 g (―) 0.02 g (―) 0.08 g and
(―) 0.14 g. Other components were added in amounts according to the recipe described in section 3.2.
Figure 12. Floc sizes for different compositions of the stabilising system measured after 20 minutes of flocculation. Components were added in following order: HAc, LX, KO, initiator, Me2.
Figure 13. Differences in size distribution of silica flocs with various amounts of polyelectrolyte (KO). Other components
were added in amounts according to the recipe described in section 3.2. (―) 0.1 g, (―) 0.2 g and (―) and 0.4 g.
Figure 14. Differences in size distribution of silica flocs with various amounts of Me2. Other components were added in
amounts according to the recipe described in section 3.2. (―) 0 g (―) 0.02 g (―) 0.08 g and (―) 0.14 g.
0 5 10 15 20 25 HHL HHH MMM LLH LLL LHL LHH HLL HLH Dv(0.5) 5.9 6.2 6.4 6.7 7.0 7.2 8.1 13.2 24.4 Dn(0.5) 1.8 2.1 2.6 2.6 2.5 2.6 3.0 3.1 2.4 Fl o c Si ze [ μ m]
4.1.3 Effect of Stirring
Previous research [33] has shown that the floc size is time dependent and that the floc size increases over time, since the silica flocs continue to grow as long as there are adjacent silica particles and flocculating agents nearby. The growth of flocs has also been studied in this work but the
experiments showed quite surprising results, since the flocs tend to decrease in size over time, see Figure 17. The same trend is seen for the number mean floc size. One theory that can explain this is that the flocs are broken by shear forces during stirring.
Figure 17. Floc size (Dv(0.5)) as a function of stirring time for a selection of samples. Components were added in
following order: HAc, LX, KO, initiator, Me2.
Considering that floc breakage is a possible explanation for the results discussed above, it was motivated to further study how the stirring rate affects the flocculation. It can be seen in Figure 18, that the stirring rate has a remarkable effect on the floc size. Higher stirring rates results in lower floc sizes, which confirm that the flocs are sensitive to shear forces. The floc size fluctuates at 150 rpm, which might indicate that flocs are able to break and reform.
Figure 18. MLKIH flocculated at different stirring rates (i.e. shear forces) as a function of stirring rate.
4.1.5 The Initiator’s Influence on Floc Size Measurements
The initiator contains emulsifiers that might have an effect on the flocculation. In previous studies on the flocculation, the initiator has been left out. [34] The idea of adding the initiator was that it would generate a more true value of the floc sizes, which would better correspond to the floc sizes during polymerisation. Figure 19 illustrates how the floc size will change if the initiator is left out of HLKIM. It can also be seen that the initiator itself can be detected by the Malvern instrument.
5 6 7 0 10 20 30 40 50 Dv (0. 5) [um ]
Sarring ame [min]
LLL MMM HHH 0 5 10 15 20 25 0 10 20 30 40 Dv (0. 5) [μ m]
Sarring ame [min]
500 rpm 150 rpm
