UNDERSTANDING GELATION SYSTEMS WITH DIFFERENT GELLING MECHANISMS IN FOODS QIUYANG XIA

Transcription

1 UNDERSTANDING GELATION SYSTEMS WITH DIFFERENT GELLING MECHANISMS IN FOODS By QIUYANG XIA A dissertation submitted to the Graduate School-New Brunswick Rutgers, The State University of New Jersey In partial fulfillment of the requirements For the degree of Doctor of Philosophy Graduate Program in Food Science Written under the direction of Dr. Qingrong Huang And Approved by New Brunswick, New Jersey OCTOBER, 2016

2 ABSTRACT OF THE DISSERTATION UNDERSTANDING GELATION SYSTEMS WITH DIFFERENT GELLING MECHANISMS IN FOODS By QIUYANG XIA Dissertation Director Dr. Qingrong Huang We explored several typical food gel systems to understand the structure-rheology relationship. The first one is zein protein in aqueous ethanol solution. We utilized rheology techniques, in effort to understand the aging phenomenon in zein solution. Time evolution of storage and loss modulus of zein solution with and without perturbation is tested. We conclude that the aging of zein solution follows a power law relation with time. The second system is BLG fibril and pectin complex. Fibrillar aggregates of betalactoglobulin (BLG) were prepared by heating at ph2 and 90 o C for different time period. Progress of protein hydrolysis was estimated by gel electrophoresis (SDS-PAGE). Morphology of fibril/pectin complex is probed by turbidity titration, atomic force microscopy (AFM) and small angle x-ray scattering (SAXS). Secondary structure of fibrils ii

3 with and without pectin complexation is studied by circular dichroism (CD). Our results suggest that the complexation process between BLG fibril and pectin is self-facilitated, results in system heterogeneity. BLG fibrils retained their linear structure at large length scale when associate with pectin. At local regions clusters and bundles are formed. The mixture of BLG-pectin and BLG fibril-pectin can be well separated domains that can be resolved by small angle X ray scattering. Addition of pectin can significantly change secondary structure of heated BLG. We validated microrheology technique in our lab. We compared the results from microrheology to that from bulk rheology in Newtonian-fluids and acrylamide gels. We found excellent agreement between bulk and micro- rheology in such classic conditions. We further used carrageenan solution, a complex fluid, and found significant difference between the two methods. The result shows that pores with size large than probe particles exist in carrageenan solution. We then used microrheology to test the gelation of KGM solution, and compared the results with bulk rheology. The effect of temperature, KGM concentration and calcium carbonate concentration are tested. We again found significant lower storage and loss modulus obtained from microrheology. The gelation rate of KGM gel was increased with increase of temperature and/or increase of coagulant concentration. The final storage modulus of the KGM gel on the other hand, decreases with increase of temperature and coagulant concentration. We found negative relation between inhomogeneity and storage modulus, which can be explained by cascade theory. We also found monotone iii

4 increase of storage modulus during heating time in microrheology experiment, further confirmed that the peak in storage modulus in bulk rheology is due to syneresis. The increase of coagulant concentration will improve the storage modulus of cured gel. Microrheology results suggested no significant relation between inhomogeneity and coagulant concentration. iv

5 ACKNOWLEDGEMENT This thesis was accomplished with the generous help of many people. I want to first thank my advisor, Dr. Qingrong Huang, for his guidance, providing financial support and experiment conditions. He is supportive when I chose research directions. He has also made valuable suggestions when my research was stuck in specific issue. Of course without his support I cannot finish this thesis. I also want to thank my committee members, Dr. Richard Ludescher, Dr. Chi-Tang Ho and Dr. Vikas Nanda. They all supported me in giving suggestions in research problems and providing experiment facilities. Next, I want to thank Dr. Yunqi Li, for his generous help in making research suggestions, developing data analyzing scripts and polishing manuscripts. I want to thank Dr. Peter Kahn, who not only generously provided me facility in CD experiment, but under his guidance I revived my interest in science. I want to thank Dr. Raymond Tu for his guidance in particle tracking experiment. I want to thank Dr. Kenneth Mcguinness for his friendship and support in TEM and CD experiment. My research cannot be accomplished without their help. I want to thank my friends and labmates, who go together with me for these years. I want to thank Dave Patrenka, Yakov and Karen for their generous help. Finally and most importantly, I want to thank my family for their unlimited help in these years. Whatever difficulties I face, they are still standing behind me. I owe you too much for what you have done. v

6 CONTENTS ABSTRACT OF THE DISSERTATION... ii ACKNOWLEDGEMENT... v List of illustrations... viii 1. Introduction and Background Information Definition of gel Categorization by link bonds Chemical gel Physical Gel Rheological properties of Gel Rheology Measurement of storage and loss modulus Glass Gelling system in food Polysaccharide gel Globular Protein based gel Kinetics of gelation Background information of zein Background information about beta-lactoglobulin (BLG) and BLG fibril Background information of pectin Coacervate Background information of konjac glucomannan (KGM) Microrheology Scientific rational and hypothesis zein form fractal clusters in aqueous ethanol Gelling kinetics and gelling point of KGM Complexation between BLG fibrils and counterions vi

7 3. Objectives Objective.1: Study the structure of zein solution in aqueous ethanol Objective.2: Study the aging of zein solution by rheology measurements Objective.3: Compare the rheological behavior of KGM in bulk and micro- rheology during the gelling process Objective.4: Produce and characterize microgel of BLG fibril/tpp and BLG fibril/pectin complex Materials and methods Materials Sample preparation Rheology Turbidimetric titration Atomic force microscope Circular Dichroism (CD) Optical microscope Small-Angle X-ray Scattering (SAXS) Particle tracking Results and Discussion Rheology of zein solution Determination of yield strain Frequency dependence of G and G at early time Time evolution under small perturbation Steady rate sweep test Conclusions of rheology properties of zein solution Calibration of optical microscopy microrheology Theory and rationale Calibration results Microrheology of Konjac Glucomannan gel Effect of temperature Effect of KGM concentration Effect of coagulant concentration Conclusions vii

8 5.4 Complexation of polypeptides from the hydrolysis of beta-lactoglobulin with negatively charge polysaccharide Hydrolysis of BLG Atomic force microscopy analysis Complex characterization Effect of heating time on heated BLG Effect of heating time on heated BLG/pectin interaction Effect of BLG/pectin mass ratio (R) Analysis of secondary structure Discussion Conclusions Future work zein aqueous ethanol solution BLG fibril and polyelectrolyte interaction microrheology technique applying microrheology in food biopolymers List of illustrations FIGURE.5-1 STRAIN SWEEP TEST OF 10% AND 20% ZEIN SOLUTION FIGURE.5-2 FREQUENCY SWEEP TEST OF 20% AND 35% ZEIN SOLUTION FIGURE.5-3 SCALING POWERS (A) AND PREFACTORS (B) OF G (SQUARE) AND G (CIRCLE) AS FUNCTION OF AGING TIME FIGURE.5-4 EVOLUTION OF G AND G UNDER CONSTANT PERTURBATION AS FUNCTION OF AGING TIME FIGURE.5-5 VISCOSITY OF 10% (A) AND 20%(B) ZEIN SOLUTION AS FUNCTION OF SHEAR RATE FIGURE.5-6 A: MSD VS T CURVE OF DEXTRAN T500 AT DIFFERENT CONCENTRATIONS. B: PLOT OF VISCOSITY FROM MICRORHEOLOGY AGAINST BULK VISCOSITY; DASH LINE: SLOPE OF 1 TO GUIDE THE EYE FIGURE.5-7 MICRORHEOLOGY RESULT FROM ACRYLAMIDE GEL USING 0.2ΜM AND 0.5ΜM PARTICLES FIGURE.5-8 : MSD VS T CURVE OF ACRYLAMIDE/NNAM GEL. FROM HIGH TO LOW, THE PORTION OF NNAM GRADUALLY INCREASES FIGURE.5-9 G AND G OF ACRYLAMIDE/BIS GEL. RESULTS FROM BOTH MICRO AND BULK RHEOLOGY FIGURE.5-10 G AND G OF ACRYLAMIDE/BIS GEL NEAR GELLING POINT, MEASURED BY MICRORHEOLOGY. TOTAL SOLUTE CONCENTRATION: 3WT% FIGURE.5-11 BULK RHEOLOGY TEST OF KGM SOLUTION WITH 0.1% COAGULANT FIGURE.5-12 MICRORHEOLOGY TEST 2% KGM SOLUTION WITH 0.1% COAGULANT FIGURE.5-13 NON-GAUSSIAN DISTRIBUTION OF DISPLACEMENT OF 1 SECOND INTERVAL FIGURE.5-14 MSD CURVE OF 2% KGM SOLUTION HEATED FOR 100 MINUTES WITH 0.1% COAGULANT viii

9 FIGURE.5-15 EFFECT OF KGM CONCENTRATION ON GEL FORMATION FIGURE.5-16 MICRORHEOLOGY RESULT OF KGM SOLUTION AT DIFFERENT KGM CONCENTRATION FIGURE.5-17 EFFECT OF COAGULANT CONCENTRATION ON GELATION OF 2% KGM SOLUTION FIGURE.5-18 SDS-PAGE OF HYDROLYZED BLG. LANE FROM LEFT TO RIGHT WITH DIFFERENT HEATING HOUR OF BLG: MARKER, 0H, 1H, 2H, 3H, 6H, 8H, 13H, 19H FIGURE.5-19 AFM IMAGE OF 6HOUR-HEATING BLG WITH DIFFERENT PECTIN TO BLG W/W RATIO FIGURE.5-20 TITRATION OF BLG HYDROLYSATE AND COMPLEX WITH DIFFERENT AMOUNT OF PECTIN FIGURE.5-21 CRITICAL PH AND TURBIDITY VALUES OF HEATED BLG/PECTIN MIXTURE FIGURE.5-22 SAXS PROFILE OF HEATED BLG AND HEATED BLG/PECTIN COMPLEX FIGURE.5-23 RADIUS OF GYRATION OF BLG/PECTIN COMPLEX WITH DIFFERENT HEATING TIME OF BLG FIGURE.5-24 CURVES OBTAINED BY SUBTRACTING SAXS SCATTERING CURVE OF COMPLEX; FIGURE.5-25 RADIUS OF GYRATION OF MIXTURE OF 6 HOUR HEATED BLG AND PECTIN, CALCULATED BY GUINIER PLOT (SQURE) AND GNOM (CIRCLE) FIGURE.5-26 KRATKY PLOT OF PECTIN/HEATED BLG COMPLEX AT DIFFERENT INITIAL CONDITIONS FIGURE.5-27 POSITION OF PEAK IN KRATKY PLOT AT A FUNCTION OF PECTIN/BLG MASS RATIO FIGURE.5-28 CD SPECTRA OF HEATED BLG AND PECTIN/HEATED BLG COMPLEX FIGURE.5-29 FRACTION OF RESIDUES IN SECONDARY STRUCTURE ix

10 1 1. Introduction and Background Information 1.1 Definition of gel Gel is defined as a space filling percolation network[1]. For chemical gel or strong physical gel, the strength of crosslink is very strong, which could well fit into the percolation scenario. On the other hand, if the crosslinking is not very strong, these cross-links could form and break dynamically with or without external stress. Such system is called transient gel because whether it behaves as a gel depends on the time scale we are looking at[2]. The characteristic time scale of transient gel is determined by the ratio between the friction force that slows down the relaxation of stress, and the elastic strength. At shorter time scale, the gel will store most of the energy exerted on it in elastic form, while at longer time, this energy will be dissipated by the relaxation process. Therefore, the definition of gel will be rather arbitrary, majorly in the choice of largest time scale. The definition of gel is further complicated by the continuous transition from ideal gel to ideal glass with respect to the change of attractive force between colloids and colloids volume fraction. Glass is the state kinetically arrested by the caging of elements by their neighbors, where an ideal glass is the glass state that the only interaction between colloidal particles is hard sphere repulsion. Network knots play especially important role in gels with strong linkage. For example, in a solution of linear polymer, increasing even a tiny amount of cross links could lead to significant gelation[1].

11 2 1.2 Categorization by link bonds The type of gel can also be categorized into chemical gel and physical gel, depending on the nature of the linkage that hold the infinite size network Chemical gel Chemical gel is formed by covalent bonds, which will not break due the physical thermal fluctuation. For example, acrylamide can polymerize into (poly) acrylamide in presence of ammonium persulfate (APS) and TEMED. Each acrylamide only has two reaction points, resulting in linear polymer. N,N'-Methylenebisacrylamide (MBAA) has four reaction sites. Adding MBAA into acrylamide polymerization system will introduce branch points. With enough concentration of branch point, a gel can be formed. An example of chemical gel in food is gluten in dough. When dough is mixed with water and kneaded, gluten will dissolve and partially defold. The defolded proteins will associate into polymers and further crosslink into films that contribute to the viscoelasticity of dough. In this case, disulfide bonds play crucial role in reinforcing the gel strength[3]. Another example in food is using transglutaminase to crosslink proteins. The ε-amino groups of lysine residues can form intra- and intermolecular ε-(γ-glutamine)-lysine bond in presence of this enzyme. Transglutaminase has been applied to meat and fish to improve the elasticity of the product [4, 5] and in soy proteins to create a stable gel [6]. The advantage of chemical gel is that it is very stable under environment change. The disadvantage of this kind of gel is slow curing and irreversibility.

12 Physical Gel Physical gel is formed by physical associations, including one or more of hydrogen bond, electrostatic interaction, hydrophobic effect, van der Waal s interaction etc.. Proper environment conditions, i.e., temperature, ph, ionic strength and polymer concentration need to be kept in limited range for such kind of gels to form. Temperature may be most notice factor among other condition. There are cold set gels and heat set gels. Cold set gels refer to suspensions that are fluid at high temperature, and cure into gel below a temperature threshold, and vice versa. Gelation of methylcellulose at elevated temperature is attributed to increased hydrophobic effect with increase of temperature. When temperature goes down, the gel will melt. The strength of such linkage varies greatly depending on how synergetic the junction zone of connection is based on the geometric nature of involved species. For example, a linkage formed by κ-carrageenan and potassium ion can be relatively strong due to ordered egg-box structure. Konjac glucomannan (KGM) forms another kind of gel. While KGM gel is formed under heat (a heat set gel) and alkaline treatment, involving chemical reaction of deacetylation, its junctions are formed by physical forces. Due to very high stability of KGM gel junction zone, KGM gel is irreversible once formed. The physical gel formation may also involve kinetic factors due to faster association compared chemical gel. One example is creating Ca 2+ -alginate gel by slow releasing of calcium ion from calcium carbonate particles [7]. Without slowed calcium release, lumpy and inhomogeneous gel will be formed.

13 4 On the other hand, there are so called transient gels, whose linkages are susceptible to thermal fluctuation. The understanding of this kind of system is far from complete, although extensive efforts have been paid during last few decades. The transient feature of crosslink leads to a relation between link strength and critical gel concentration. With increase of attractive force starting from pure hard core repulsion, the critical concentration of colloidal required to form a non-ergodic system first increase, then decrease, along the phase line from ideal glass on one end to ideal gel to the other end[2, 8, 9]. The interaction between particles may involve electrostatic interactions, steric interactions, van der Waals interactions, and depletion effect. Most commonly, depletion effect is used as controlling factor by introducing smaller non-absorbing polymers to induce attractive interaction between particles[8, 10]. The increase of critical concentration at narrow attractive force window could be explained by decrease of effective volume of colloidal[11]. The following decrease of critical concentration is due to formation of fractal clusters. Higher attractive force leads to larger cluster size. While the density of colloidal within a cluster decreases with increase of cluster size, larger cluster size means lower resulting concentration of colloidal when the clusters are close packed in the system. 1.3 Rheological properties of Gel Rheology The rheological properties refer to the response of an object to external strain or stress [12]. Strain describes the extent of deformation of an object, and stress describes the

14 5 force exerted on an object. The two simplest models for rheology are spring and Newtonian fluid. Spring follows Hooke s law F = kδx Where F is the exerted force, Δx is the distance of deformation from equilibrium position, k is spring constant. This model describes the solid nature of materials. In Newtonian fluid stress is proportional to strain rate, σ=ηγ where σ is stress, η is viscosity and γ is strain rate. Most food products exhibit more complex rheological behaviors. Food emulsions usually exhibit shear-thinning behavior, which can be described by a power law relation between stress and strain, with a power smaller than 1. Gel-like foods, such as jelly and pudding, simultaneously show a combination of viscous and elastic behavior. Some other semi-solid foods, such as ketchup and salad dressing, shows solid like behavior at static, and liquid like behavior when being forced to flow. These behaviors are crucial for processing, handling, quality, stability and application of food products Measurement of storage and loss modulus Storage modulus (G ) and loss modulus (G ) are used to quantify the elastic and viscous properties within a single object under shear, respectively. Modern rheology spectrometers use oscillatory shear strain with sine function to test the response of materials. For strain-controlled devices this is done directly, while for stress-controlled devices, this is done in pseudo-strain-controlled way. The stress measured by sensor can

15 6 be divided into two parts, one in phase with the phase strain and one 90 out of the phase of the strain. The addition of the two parts of stress results in another sine function with the same cycle time of strain, but δ difference in phase. The ratio between amplitude of stress and strain is defined as complex modulus G*=σ 0 /ε 0, and G = G*sin(δ), G = G*cos(δ). G defines the ability of material to store energy under shear force, and G define the dissipation of energy from shear. The loss tangent is tan(δ) = G /G, which describes the ratio between dissipated energy and stored energy. 1.4 Glass When binding energy between elements that forming the gel is in the order of k B T, the system will evolve with time. The process that particles rearrange to a lower energy landscape is called aging[13] When volume fraction φ of particle is high enough, a glass transition happens, which is defined similarly as gel point. Structurally, the difference between glass and gel is the length between interconnect junctions l comparing to the size of individual unit [14]. In gel l is much larger than a, while in glass l is close to a. For the system that zein molecules disperse in 70% aqueous ethanol, zein molecule should be considered charged [15] swollen [16] particle. Constitutive soft glassy rheology (SGR) model[17] was developed to illustrate the behavior of ideal glass. In SGR model, system is divided into small elements, each in its meta-equilibrium state in an energetic well. Either a generalized thermal fluctuation or applied shear force could excite an element to jump out from the energy well into a

16 7 new meta-equilibrium state. This model predicts that the aging process could be accelerated by exterior perturbation. 1.5 Gelling system in food Gelation has profound impact on the food processing (fixing the shape at proper time), food stability (suspension of ingredients in liquid media, or keeping the shape of the product), food texture (mouse feel), and encapsulation and/or release of flavors or active ingredients. Understanding the gelation process is therefore important to improve quality of food products. Typical building blocks of gels in a food are proteins, polysaccharides or emulsion droplets. Strong gel in food product involves jelly, candy, bread, meat product and fish product. In these foods, properties of gel control the mouth feel of product. In beverages, dressings and desserts, a very weak gel formed by food gums can hold ingredient from gravity precipitation. Gelation could also happen at interface, forming a 2-D network that can strengthen the enclosed object [18] Polysaccharide gel Polysaccharides are carbohydrate polymers whose monomers are sugar molecules. There are a variety of polysaccharides is food resources and some are extracted as food additives. The polysaccharides used to food are also called hydrocolloid because most of them are hydrophilic and perform as colloid in aqueous solution. A few exceptions are that polysaccharide can also wet oil. For example, gum Arabic exhibits surface activities and is frequently used in emulsions, although some argues that this property is due to proteineous moieties attached to the polysaccharide backbone [19]. N-Zorbit is

17 8 another example. This tapioca derived maltodextrin can absorb oil so that high content of fat can be thickened and stabilized for spray drying. Hydrocolloid conformation Generally, hydrocolloids are dispersed in water during food processing, and function in hydrophilic environment. The conformation of a hydrocolloid molecule in aqueous dispersion depends on many factors, such as the structure of the chain, the functional moieties on the chain, hydrocolloid concentration, and ph and ionic strength of the solvent etc.. Generally, at low concentration, hydrocolloids behave as swollen particles in water, although many of them are linear chain. Molecular weight (Mw), structure of branch, and net-charge are three most important factors that influence the size (Rg) of this swollen particle. The Rg and Mw usually have a power-law relation, Rg~Mw α. For very rigid chain, α can be close to 1, which is not totally in contrary with the swollenparticle picture, because the conformation of the chain also depends on the length scale we are looking at. Branching usually cause a decrease of Rg with constant Mw. For example, Mw of amylopectin is in the order of 10 7 to 10 8, while Rg of amylopectin is much lower compared to predicted value from linear chain model [20-22]. Charges of hydrocolloids are usually brought about by the substitute groups, such as amine (chitosan), sulfate (carrageenan) and carboxyl (pectin, alginate) group. The electrostatic repulsion between charges will make hydrocolloid chain stiffer. ph of solvent will influence the protonation and deprotonation of weak basic and acid groups, thus change the net charge, where sulfate group is not affected. Ionic strength will screen the

18 9 electrostatic interactions by forming a screening layer. The thickness of this layer is quantified by exponential decay of electrostatic energy potential. Mechanism of forming a polysaccharide gel The mechanism of gel forming of polysaccharides depends on the nature of the polysaccharide, while whether the gel will form depends on environmental conditions, such as concentration, ionic strength, counter ion, temperature and ph. The gelation process usually involves association between polysaccharide chains into branched bundles then into an infinite network. For most highly efficient gelling gums, the major force contribute to gelling is electrostatic interaction Globular Protein based gel Numerous proteins, including gelatin, whey protein, pea protein, egg white protein etc. are well known to gel after proper treatment. Generally, the gelation of protein solution involves denaturation, and subsequent association between protein molecules. The globular proteins, such as bovine serum albumin, beta-lactoglobulin, 7s pea protein can form gel with heat treatment. The result morphology of the gel depends on the ph of the solution. At ph near the isoelectric point (pi) of protein, the gel is coerce and particular. At ph far always from pi, a fibrillar gel will be obtained. Electrostatic repulsion is crucial in fibril formation stage.[23, 24]. Adamcik et.al reported that electrostatic interaction between fibrils is key factor in twisting of two or more

19 10 protofilaments into one fibril.[25] In above studies, electrostatic interaction is screened by changing ionic strength. 1.6 Kinetics of gelation The kinetics of gelation could be categorized as diffusion limited aggregation (DLA) or reaction limited aggregation (RLA). In the former case the binding energy between particles is much large than thermal energy k B T, which means (i) the particle will attach to each other once they meet, and (ii) the bond will not break when formed. Lu et al[8]suggested that if such strong attraction is in short range, the gelation is initiated by spinodal decomposition. In the latter case, particles need to overcome some barrier to bind to each other, which lead to more compact and more polydisperse structure[26]. In both cases, the volume fractions of clusters increase with cluster size due to the fractal dimension lower than 3. When arriving at the well defined gelling point t gel, the clusters are closed packed. This point can be experimentally determined by oscillation rheology that G ~G ~G*~ω u [27], where u is related to the fractal dimension of clusters. 1.7 Background information of zein Zein is a prolamine derived from maize. In nature, the major functionality of zein is storage. Zein composes 35% to 60% of total maize protein[28]. The hydrophobicity and biocompatibility of zein makes it a valuable material for coating, encapsulation and packaging[28]. Given 500 million tons of maize annual production but only 500 ton of zein annual production, zein still has great potential in food and material applications.

20 11 Zein is a mixture of different proteins. Different fractions of zein are classified by their solubility in different concentration of aqueous ethanol. [28]The major component of zein is α-zein. α-zein has two components Z22 and Z19, named after their apparent molecular weight in gel electrophoresis, with approximately equal amount of one another[29]. For α -zein, 50% to 90% aqueous ethanol is secondary solvent and pure acetic acid is primary solvent. Zein has more extended structure when dissolved in pure acetic acid[16], primarily due to increase of positive surface charge. The understanding of zein behavior in solution is still limited. High aspect ratio of zein has been indicated since 1930s by ultracentrifugation, rheology, birefringence and dielectric constant experiments[30]. Various methods, including FTIR and circular dichroism have confirmed that zein is rich in alpha helix[31]. Four recent models based on small angle X-ray scattering meet the above two limits [30, 32]. In Matsushima s model[32], zein has 9 helix units aligning parallel along the long axis. In Forato s model[30], a sample with only z19 content was used. They proposed that a series of alpha helices fold in the middle of the chain. Momany et al. constructed a super-helix model by molecular dynamics[31]. Li et al., also using a mixture of z19 and z22, proposed two possible models of z19 and one model of z22 based on aligning sequence similarity to structure[16]. They suggested that z19 could either be fully extended series of alpha helices or a barrel surrounded by helices domains. Zein forms aggregates in aqueous ethanol even at low concentrations.[33-35] The aggregate size ranges from 10 to 100nm depending on both ethanol and zein

21 12 concentration. The aggregation is partially attributed to the interaction between z19 and z22 components[36]. Accordingly, the well known gelation behavior of zein could be associated with such clustering phenomenon[8]. The gelling process of zein is also called aging, in contrast to the term curing used in chemical gels and some strong physical gels. Neither the structure of gel formed by zein, nor the mechanism of its aging is well understood. The rheology property of zein solution shows strong dependence on time. In aqueous ethanol, fresh zein solution is Newtonian fluid while aged zein solution is viscoelastic. Padua et al. studied the effect of solvent quality on rheology of fresh zein aqueous ethanol solution [37]. Their report showed that the rheology of fresh prepared 20% zein solution in aqueous ethanol does not change with ethanol content between 65%-90% ethanol. This result indicated that the friction between zein molecules or aggregates override that between solvent molecules in controlling viscosity of the solution, because the viscosity of aqueous ethanol drops about 20% when ethanol content increases from 65% to 90%[38]. Similar result was also shown by other reports[34]. In lower zein concentration regime, Fu et al. revealed generally a decrease of viscosity with increase of ethanol concentration from 60% to 90%[39]. The effect of zein concentration below 12% on viscosity could be fitted into a single log-linear model in aqueous ethanol solution[39]. On the other hand, viscosity of aged zein solution shows strong dependence on both ethanol concentration and time[35]. In acetic acid, a primary solvent, zein behaves like a flexible polyelectrolyte, and critical concentration was found at 4.3%[40].

22 13 Aged zein solution will form gel. Previous studies [37] suggested that disulfide bonds contribute to gelation with high content of gamma-zein, but alpha-zein was only indicated to form disulfide bond cross-linked dimer. Zhong et al. claimed that the concentration threshold of gelation could vary between 20% and 29%[35], and the gel was formed by swollen particles. This is consistent with the finding that zein form aggregates in aqueous ethanol[16]. However, it should be reminded that this gelling point is defined rather arbitrarily. The gelling point is time dependent. At long enough time, zein solution can even gel at a lower threshold between 3-6% [28] and researcher could define a subjective gel point that only have comparative meanings. Freshly prepared zein is in meta-stable state, and undergoes structural rearrangement in the time scale of several days. By observing slow kinetics, the rearrangement should happen in larger length scale compared to single zein molecule. 1.8 Background information about beta-lactoglobulin (BLG) and BLG fibril Bovine Beta-lactoglobulin is a major protein component of milk. BLG has 162 residues, molecular weight of Da, and isoelectrical point (pi) of 5.2[41]. BLG can selfassemble into fibrillar structure when heated up to denature temperature at ph beyond its isoelectric point [42]. The building blocks of BLG fibril are hydrophobic peptides of hydrolyzed product of BLG [43], which is different from the case of BSA[44]. Peptides are suggested to align in β-strands that orient perpendicular to the elongate direction of fibril[45]. Several parameters, such as protein concentration[46], temperature [46], ph [45], ionic strength[47], treatment time and shear force[48] will affect the morphology[49] and kinetics[43] of BlG fibril formation. The diameter of BLG fibril is

23 14 between 4nm to 10nm, and the length is between 1um to 10 um. The diameter of fibril is determined by number of proto-fibrils in the fibril bundle. Flexible proto-fibrils, and relatively more rigid double and triple helix of bundles have been reported[25, 50]. The rigidity of fibrils also depends on the electrostatic repulsion along the chain[50]. The length of fibril is determined by the progress of the growth of the fibril, as well as the scission of fibrils into shorter ones due to physical stress[51]. While reported to have diameter between 2nm to 10nm, amyloids are also suggested to have some features similar to linear polymers.[52] The persistence length of BLG fibril ranges between one hundred nanometers to a few microns, depending on the thickness of the chain and factors that affect electrostatic interaction [50, 53]. This value is close to stiff polymers. Meanwhile, BLG fibril can be either positively charged at low ph and negatively charge at high ph, thus can associate with other species carrying opposite charge. The polyelectrolyte features of protein fibril enlightens us to explore functionalities of BLG fibril that already been discovered in polyelectrolyte. Jones et al. studied the formation of BLG fibril in the presence of oppositely charge polymer[41]. Shimanovich et al designed protein fibril based microgel.[54] Humblet-Hua et al developed micro-encapsulation system based on ovalbumin fibril and pectin. Sagis et al used ovalbumin fibril to coat the surface of micro-capsule[18]. These studies have proven protein fibrils can preserve their high aspect ratio under various environments in application.

24 15 Due to high aspect ratio, BLG fibril can easily form gel at low concentration in solution [55]. The high connectivity, high rigidity of fibrils and electrostatic repulsion between fibrils contribute to the gelation phenomenon pure BLG fibril solution. The mechanism of gelling in this case can be understood as crowding of species with large effective excluded volume due to electrostatic repulsion, as in the case of Wagner glass[9]. This gel system has been extensively characterized by other authors [23, 45, 56, 57], so we explore the formation small gel particles, i.e. microgels in the suspension. The fibrils can associate with species that bear counter charge, depending on the side ph from pi. Jones et al showed carrageenan beads attach to the fibril backbone Jones[41]. Here we refer to the situation of chitosan, a positively charged polysaccharide, which can form nanoparticles by adding tripolyphosphate[58]. The complexation between fibrils and counterions is possible to form complex coacervate or microgel in the suspension with proper environment. 1.9 Background information of pectin Pectin is a polysaccharide derived from plant cell walls, giving structure to fruit and middle lamellae. Upon ripening, pectin is degraded by pectinase, and during this process fruit become softer. Most commercial pectin is originated from citrus peels. Pectin is a heteropolysaccharide majorly composed of α--(1-4)-linked D-galacturonic acid and methylated or amidated D-galacturonic acid. At neutral to basic ph, pectin bears negative charge, while at low ph, negative charges diminishes with decrease of ph. Therefore, the electrostatic interaction of pectin with positively charged molecules can

25 16 be tuned by change of ph. Pectin is widely used in jelly, candy, pudding, beverage and drinking yogurt as thickener, gelling agents and stabilizers. Pectin can be generally categorized into two groups: high methylated pectin (HM) and low methylated pectin (LM), depending on the degree of methylation at carboxyl group, with threshold at 50%. Methylation will decrease the charge density of pectin, making significant changes in the gelling properties. For LM pectin, a gel can be formed by adding calcium ions pectin solution. HM pectin can only properly form a gel in presence of high concentration of sugar and high acidity. This is due to the fact that the chains of HM pectin need to be deionized and dehydrated to associate with each other strongly[59]. The association of pectin with proteins was extensively studied. The interaction between pectin and gelatin[60], pectin and beta-lactoglobulin[61], pectin and casein[62], pectin and lysozme[63], pectin and bovine serum albumin[64] are reported. In these systems, pectin acts as polyions in the ph range between 3 and protein s PI to interact with positively charged proteins. Such association could be used to stabilize surface, increase viscosity, deactivate enzymes, or encapsulate compounds Coacervate Coacervate is a phenomenon refers to liquid-liquid phase separation due to electrostatic interaction between two species carrying opposite charges[65]. The two species are commonly both macromolecules, thus reduce the entropy loss of molecule fixation. Also

26 17 reports show that the driving force of species association in coacervation is gain of entropy due to release of counterions[66]. The major factors influencing formation of coacervate are ph, ionic strength, species mass ratio, structure and charge density of macromolecules. Due electrostatic interaction, low ionic strength is usually preferred in coacervate formation, while high ionic strength can sequester such association[67]. The maximal degree of association is achieved at ph and mass ratio value that result in neutral complex formation.[68] However, the complexation between two species can already occur when the two species carrying the same charges, if at least one of the species is zwitterion (e.g. protein). This is due to patches on the zwitterion surface with the opposite charge can emerge compared to the total net charge of the molecule.[69] Such association can form soluble complex but not enough to induce phase separation[70]. Coacervation has wide existence and application in food systems[71], such as controlling texture[72, 73], compound encapsulation[74, 75] and surface stabilization. It is also suggested that coacervation plays important role in biological function[69]. The coacervation phenomenon in polyelectrolyte-polyelectrolyte and protein-polyelectrolyte mixtures has been extensively studied during past two decades [76-78], which provided extensive knowledge of the structure of coacervates. The interaction of pathogenic fibrils with biological polyelectrolytes was studied [79-81], majorly focused on the condition of aggregation. The detail structure information in fibril/polyelectrolyte system is still sparse. The application coacervate in foods includes but not limited to: mimicking texture of fat, controlling release of small molecule

27 18 ingredients, encapsulating enzymes, improving viscosity and strengthening oil/water interface[82] Background information of konjac glucomannan (KGM) Konjac glucomannan (KGM) is a neutral polysaccharide derived from the tuber of Amorphophallus konjac C. KGM has a beta-1,4-d-glucose and D-mannose co-block backbone with molar ratio of 1:1.6. It also has partial substitution by acetyl group at C-6 position at rate of about 1 in 19. The branch structure is still varied among reports. The branch point is reported to be C-3[83, 84] or C-6 at frequency of 8%[85], or no branch at all[86]. This may reflect the diverse source of KGM. KGM is widely used in Japanese food either as major ingredient in jelly like food or additive in other kinds of food. KGM is well known to have high viscosity at the same concentration compared to other commonly used gums[84]. Its low calorie content also makes it very good source of dietary fiber[87]. KGM can gel at alkaline condition, usually with the help of elevated temperature. The mechanism of KGM gelation is still under debate. Generally, hydrolyzation of acetyl groups is essential in gelling process, although it could also gel at very high concentration[88] or in the presence of organic borate[89]. Such deacetylation could lead to bare long cellulose-like backbones that can form crystalline as junction zones[90]. However, further study also found complex involvement of hydrophobic effect of acetyl groups on gelation [91, 92]. The factors that affect gelation kinetics have been extensively studied, including temperature[93], KGM molecular weight[94], degree of acetylation[95], coagulant type[96], coagulant concentration[97], KGM concentration[88] and added salt[92]. So it makes KGM a good model system for our study in micro-structure of KGM gel. In this study, we conducted both bulk and micro-rheological experiment to characterize KGM gel, and compare

28 19 the result from bulk and small length scale to the reveal the evolvement of heterogeneity of KGM during gelation process Microrheology Understanding gelation of polysaccharides is crucial in food industry applications. Gelation involves formation of network of one phase in another liquid phase [1, ]. The gel formed can be characterized structurally or physically. Structurally, a gel can be categorized into particulate gel or fibrillar gel, depending on the morphology of its building block. The former is formed by association between particulate objects into fractal networks; the later topologically refers to linear strands inter-connected by branch points. Physically, gels has its characteristic response profile to external perturbations, either characterized by strain or stress, and/or their time derivatives, and is commonly referred to as viscoelastic. The term viscoelastic means a material generate stress in response to both strain (elastic) and strain rate (viscosity), and can be characterized by storage and loss modulus, respectively. Other categorization methods also exist such as by material of building block or nature of junction zones[98]. However, when the theories of both characters go down to micro-scale structure, the interest is mostly focused on bulk phenomenon. Even for applications of gums as stabilizer for particle suspensions, the environment around each particle is considered as a small piece of gel that has no difference to bulk gel[ ]. This simplified treatment could deviate far from what actually happens, since the bulk phenomenon of gel will not apply down to a small enough length scale, e.g., the pore size of gel[ ]. When the size of particle is larger than gel s pore size, its experience will be the same as in bulk gel. The degree of heterogeneity of gel could be large depending on the nature of polymer and kinds gelation process [ ]. When particle has smaller size than pore, it will go through the pore freely without hindrance. For application needs, we have to understand whether the pore size is small enough to suspend particles. The pore size of gel is

29 20 heterogeneous, which means with proper size of particle, the experience of particles at different location of gel will differ. During last two decade, microrheology technique has emerged to study the micro-scale viscoelastic properties of soft materials. Mason et al. [112]suggested a generalize Stokes- Einstein relation (GSER) that connected the viscoelastic properties of media to particles thermal movement in the media. Using this relation, we could probe the environment in close distance from a particle by characterizing movement of that particle. One direct and well established method is using camera to record particles location under optical microscope[113]. By this method, location of particles are tracked for a period of time and combined into trajectories. The mean square displacement (MSD) extracted from trajectories are used to interpret viscoelastic properties of local media. Data acquired by this method is usually treated statistically, which means large amount of data are needed and aggregated together. However, the heterogeneity of such data beyond normal distribution can still be used to characterize the local heterogeneity of the media. Another experiment method utilized dynamic light scattering (DLS)[ ] technique or diffusion wave spectroscopy (DWS) [ ], depending on the density of suspension, which are already used to characterize particle size in Newtonian fluid with known viscosity. This method is the converse use of DLS/DWS data in calculation, and allowed elastic part of media to be calculated by GSER. Therefore, the probing particles must be standardized and information of media can be obtained. This method has tremendously expanded the frequency range that can be rheological tested, but only a few spatially resolve heterogeneity analysis were reported [ ]. The advantages of both optical video and light scattering methods are that they are non-invasive to the sample. This is especially important for very fragile samples and studying the evolution of samples with time without perturbation. Another advantage of optical video microrheology includes very low need of sample volume,

30 21 low cost of facility, easy to maintenance and high sensitivity. In current setting only thermal fluctuation act as the driving force of stress, so no stress sweep test can be done, only frequency dependence of storage and loss modulus can be obtained. However, microrheology can be used to test very weak response of gel at low frequency, which is limited by sensor sensitivity and inertia in bulk rheology. 2. Scientific rational and hypothesis 2.1 zein form fractal clusters in aqueous ethanol To understand aging of zein solution, information about both 3-D structure of zein molecule and association between zein molecules are necessary. Although many authors proposed structures, there are two questions not answered: a. is the interaction between zein molecules specific or non-specific? b. Does the single molecule conformation change with aging? The second question may be trivial since the dynamic of single protein is fast. From knowledge of colloidal gel we could see that for low concentration of colloidal with symmetric spherical attractive potential to form a gel, the attractive interaction between particles should be strong enough[8]. On the other hand, from our preliminary data (shown in Preliminary Data) and literature report[16], the attractive interaction between zein is not strong enough. The critical gelling concentration of zein can be down to 5% with a month of aging. The net charge of zein is reminiscent of the situation of charged particles of proteins[56] or inorganic

31 22 particles[11], which tend to extend the aggregated structure. The asymmetric structure of zein further reduces the degree of freedom, because certain configurations of interaction could be favored to others. Therefore, we have the first two hypotheses: Hypothesis I: The association between zein molecules favors specific pattern Hypothesis II: Zein gel is formed by close packed fractal clusters 2.2 Gelling kinetics and gelling point of KGM While extensive studies have been done using bulk rheology to determine the gelling, non-invasive and local test is absent in our best knowledge. Hydrogen bonding contributes to gelation of KGM[124]. At high temperature, hydrophobic effect could also enhance the gel strength[125]. A crystallized junction zone is formed when the polymer is partially deacetylated [126]. The mechanism of KGM gelation is complex since both chemical reaction and rearrangement of polymer should take part in the process. Comparisons of microrheology to bulk rheology have been performed in various food systems. Caggioni et al.[105] studied both bulk and micro- rheology in gel of gellan gum. They found local heterogeneity which is a result of formation of polymer condensed microgel. Moschakis et al studied the gelation and local heterogeneity of barley betaglucan gel[127]. They found microrheology capable of capturing polymer aggregation at pre-gel stage, and the evolving of pore size with the maturing of gel. Microrheology has also been used to test the alternative mechanisms that hydrocolloid can stabilize emulsion[128]. For a system only comprising small molecules, the authors did not detect

32 23 any difference between bulk and micro-rheology[129]. Therefore, we make the third hypothesis. Hypothesis III: The local heterogeneity of KGM gel at gelling and pre-gelling stage can be probed by microrheology. The fact that local heterogeneity do occur in polymer gel is trivial, while only size of heterogeneity matters. Some remarks should be taken in this hypothesis. First, particle size of probes should play important role this experiment. When the particle size is smaller than pore size, local structure will be detected and vice versa. The pore size of the gel should depend on both the polymer concentration and the progress of gelation. Therefore, larger probe size will be suitable in detecting the gelling point, while smaller particles will be suitable to probe the pre-gel aggregation. 2.3 Complexation between BLG fibrils and counterions Here we explore the possibility of BLG fibrils to act as positively charged polyelectrolytes to form microgels. The necessary composition of microgel should be a long chain species and a multi-functional crosslinker. The parameters affect microgel formation will be type of counterion, the ratio of fibril concentration to crosslinker concentration, ionic strength and ph. With careful adjustment of these parameters, different morphologies can be obtained, which could be potential functional ingredients in foods. Here tripolyphosphate (TPP) will be chosen as small crosslinker and pectin as large crosslinker. Hypothesis IVa: BLG fibril can form microgel by complexation with TPP.

33 24 Hypothesis IVb: BLG fibril can form microgel by complexation with pectin. 3. Objectives 3.1 Objective.1: Study the structure of zein solution in aqueous ethanol SAXS spectrum can provide the radius of gyration and pair correlation function of the object. I plan to test my first hypothesis by study the structure of single zein molecule at different concentration. The pair correlation function should reflect the packing pattern of zein oligomer. I will first find the critical concentration at which the oligomers will form. The scattering pattern can be extracted by solving the set of equation =. = where i means the number of monomer is the cluster, c and I are molar concentration and scattering intensity. c z is the total concentration of zein monomer. The foot note 1 means scattering pattern of one object. We will first tentatively use dimer as the only oligomer. The scattering pattern of dimers of different possible configurations will be simulated using VMD and Crysol software. This procedure can only be done at low concentration, say, 1mg/ml to 15mg/ml, where larger structure is not formed. For higher concentration and larger structure, Ultra-small angle X-ray scattering (USAXS) will be used. The long range correlation proposed in my second hypothesis could be tested

34 25 in this region, if cannot be tested in the SAXS region. The effect of zein concentration, ethanol concentration and aging time will be tested. 3.2 Objective.2: Study the aging of zein solution by rheology measurements. The rheological properties of zein with different parameters, including aging time, zein concentration, ethanol concentration and temperature will be tested. The information about the structure of zein will be extracted from the relation between structure and rheology. Part of the measurements has been done on an ARES rheometer. The rest of these studies will be carried out on an ARES2000 rheometer. 3.3 Objective.3: Compare the rheological behavior of KGM in bulk and micro- rheology during the gelling process. This objective comprises two contents. The first is to compare the gelling point determined from bulk and micro- rheology. A series of measurements will be taken for the samples at different incubation times after preparation. The gelling time will be determined by the Winter-Chambon criteria G ~ G ~ G ~ ω α Samples with different degree of deacetylation (DD) and molecular weight will be tested. A range of solvent ph from 7 to 12 will also be tested. The second is to test the local heterogeneity. The MSD (mean square displacement) patterns of probes at different spots in the view of video will be individually analyzed and grouped.

35 Objective.4: Produce and characterize microgel of BLG fibril/tpp and BLG fibril/pectin complex. Different mass ratios of BLG fibril and TPP/pectin, say 1:5, 1:2, 1:1, 2:1 with constant total mass will be tested. Turbidity titration will determine the zone of association. The turbidity will rise with increase of both number and size of aggregates. The initial rise of turbidity indicates formation of soluble complex. Atomic force microscopy (AFM) will be used to characterize morphology of the complex. The height map of AFM is useful in determining the size of aggregates, while the phase map can better depict the shape of aggregates. Small angle X-ray scattering (SAXS) experiments will provide information about both size and inner structure of condensed phase. The size of aggregates can be characterized by Rg. This is complementary information to that obtained from AFM. SAXS also provides original information of molecules in the solution, preventing artifacts during deposition and drying in AFM sample preparation. The inner structure is reflected by fractal dimension, which can be extracted from the slope of I vs q curve in log-log plot. The radial distribution function can be extracted by available GNOM software. Guinier plot will be used to calculate Rg. Kratky plot will be used to characterize potential spherical morphologies. Other models, such rods, aggregates, micelles, etc will also be tried.

36 27 Circular dichroism (CD) measurements will be used to characterize possible change in secondary structure of fibril. Circular dichroism measures the difference in absorption of left and right hand light. Molecules need to have optically active chiral part to provide signal in this experiment. In proteins and peptides, the major chiral components are amides, aromatic side chains and disulfide bonds[130]. CD is especially good at identify characteristic signals of secondary structures. α-helix has negative bands at 208 and 222 nm. β- strand has negative band at 218nm and positive band at 195nm. Random coils have weak positive signal at 210nm and negative signal near 195nm.[131] These bands can be used qualitatively to determine the existence and change in secondary structure. Greenfield et al developed a CONTINll program that can quantitatively estimate the amount of secondary structures in protein from CD spetra[131]. This algorithm uses statistical methods to compare measured CD spectra with a pool of CD spectra of structurally well characterized proteins. We will use the available software package (CDPro2) to quantitatively calculated portion of α-helix and β-sheet. 4. Materials and methods 4.1 Materials Zein (Lot#: TLH5819) was purchased from Wako Pure Chemical Industries, Ltd. without further purification. It contains 2.3% moisture as specified by the manufacturer. Anhydrous ethanol was purchased from VWR Inc. All water used was deionized by a Milli-Q system, will be noted as water without further specification. Beta-Lactoglobulin (BLG) was kindly donated by Davisco Foods International. It was further purified through

37 28 centrifugation at 10,000g, ph4.9 for 10 minutes on 50 mg/ml BLG aqueous solutions. The supernatant was freeze dried and stored at -20 o C. Pectin was kindly donated by Danisco with the degree of methylation 31% as indicated by the manufacturer. It was further purified through dialysis at 4 o C with 12kDa cutoff for a week, followed by cation ion exchange resin to remove residue metal cation and freeze drying. Sodium chloride, Hydrochloric acid, sodium hydroxide and TPP were purchased from VWR International. 12KDa cutoff dialysis membrane was purchased from Sigma-Aldrich. Konjac glucomannan (KGM) powder was kindly donated by Hubei Yizhi Konjac Industry Co. Ltd. (Hubei, China) with purification procedure suggested by Xu et al.[132]. Briefly, KGM was suspended in 40% aqueous ethanol and heated to 48 C for 4 hours with reflux. The sample was then filtered through filter paper. The precipitate was washed two times with 40% aqueous ethanol at 48 C, then vacuumed dried at 80 C over night. Before each experiment, KGM solution was freshly prepared by dispersing KGM powder into cold water slowly with vigorous stirring, followed by 1 hour of continuing stirring and 1 hour incubation at 80 C. The solution was then cooled down for further use. Fluorescent beads with different size, surface coating and fluorescent labeling were purchased from Life Technology. Sodium carbonate was purchased from VWR International. All water used is DI water from Mili-Q system. Carboxylated fluorescent particles labeled with Nile red were purchased from Life Technology. The particles were washed with DI water before use. λ-carrageenan, acrylamide and dextran were purchased from Sigma-Aldrich. N,N'- Methylenebisacrylamide (NNMA), ammonium persulfate (AP) and Tetramethylethylenediamine (TEMED) was purchased from Bio-rad.

38 Sample preparation Zein solution was prepared in 70w/w% aqueous ethanol. Predetermined amount of zein was dissolved in 70wt% ethanol to 20ml total volume in a 50ml centrifuge tube, shook on a vortex for 30s, then ultra-sonicated for 3 minutes to get rid of bubble. Sample surface was covered by a layer of mineral oil to reduce evaporation. Hydrolysis of Beta-lactoglobulin 20mg/mL BLG aqueous solution was prepared by diluting 4% stock solution using DI water and allowed to equilibrate to room temperature. The solution was adjusted to ph2 using 6N HCl solution. The solution was incubated at 90 o C and stirred at 600 rpm in capped glass tube for different hours. After heating, these tubes were incubated on ice to stop hydrolysis. The final sample solutions were stored at -20 o C until further utilization. AFM samples 0.2mg/mL BLG fibril-pectin/tpp complex solution was deposited on a fresh peeled mica surface and then incubated for 1 minute. The mica surface is then washed by DI water with the same ph as the stock sample. Particle tracking samples The polysaccharide samples were prepared by dissolving desired amount of polysaccharide powder into DI water to form stock solution. The stock solution was mixed with proper concentration of probe suspension with volume ratio 9:1 to make

39 30 sample solution. To prepare gel samples, 30% acrylamide stock solution and 29% acrylamide/1% NNAM stock solutions were prepared. The mixing was made by mix proper volume of stock solutions, probe solution and DI water. After addition of 10% AP solution and TEMED, the mixture was immediately loaded to sample cell, and the gaps were sealed by UV-cure epoxy. Samples loaded with probes were filled into a homemade sample cell with about 0.5mm height and 30μl volume inside. Liquid samples were loaded to home-made sample cells. The upper side of the cell is a 0.17mm thick cover glass and the lower side is a rectangular glass slide. Two pieces of parafilm were sandwiched between the two pieces of glass. After filling sample, the two open sides of the gap were sealed with UV-cure epoxy. Semi-solid and very viscous samples were filled in a depressed glass plate, covered by 0.17mm cover glass and sealed with Vaseline. 4.3 Rheology Rheology test was performed on a strain controlled ARES system (TA Instrument). The geometry was 25mm diameter parallel plate. For oscillation test, a strain γ(t)= γ 0 sin(ωt) was applied to the sample. The ratio between measured stress σ(t) and strain γ(t) is complex modulus G*= G +ig. G is storage modulus which characterizes the elastic part of the sample response. The loss modulus G characterizes the dissipation of energy during the shear process. By changing the oscillation frequency ω, response at different time scale can be probed. We kept the gap at 1mm for higher concentration and 0.5mm for lower concentration. When doing the test, the edge of geometry was covered by mineral oil to prevent evaporation. For all samples to be tested freshly, 10 minutes preshear at 100s -1 was applied to eliminate any history. The linear response

40 31 strain (LRS) was determined by strain sweep test from 0.1% to 100% at 10rad/s frequency. The frequency sweep test was done at amplitude within the linear response region. The frequency range was from 0.1 rad/s to 100 rad/s. The effect of aging was evaluated in either perturbed condition or unperturbed condition. In the first method we incubate the sample on the geometry, and keep probing the modulus at small strain over time. In the second method, the samples were incubated in centrifuged tubes before test. 4.4 Turbidimetric titration Turbidimetric titration was carried out to study the complex of BLG polypeptides and pectin by collecting transmittance at 420nm using a Brinkmann Colorimetry. The device was first calibrated using DI water to 100% transmission. The probe of ph meter and probe for Colorimetry are dip below the top of sample solution. 1M HCl solution or 1M NaOH solution was added to the sample slowly drop by drop. The values were read after 30s equilibration after each addition of acid or basic. 4.5 Atomic force microscope The sample loaded mica slide was placed on J-mode probe on Multimode AFM. All experiments were done in tapping mode. A tip was placed on the device and calibrated for frequency and feedback signal. An offset in frequency was manually set to enhance the exerted force on the tip. 512x512 pixel pictures are captured with 1 second per line. A 30μmx30μm picture was scanned for each sample before smaller areas of sample are scanned. The captured images are flattened to get better view.

41 Circular Dichroism (CD) The CD spectra between 250nm and 194nm were collected using AVIV model420 circular dichroism spectrometer. 0.1mg/mL sample solution was analyzed in 1mm path length couvette. The threshold of dinode signal is 800. Eight curves are measured for each sample, with 10 seconds per point. Points are measure at 1nm interval. The curves with outstanding points are visually picked out. The final spectrum is average of rest of curves. Spectra were analyzed using CDS and sigma plot. Fraction of secondary structure was estimated using SELCON3 algorithm in CDPro2. Protein concentrations in CD experiments are determined by UV absorption at 208nm using 1cm path length cuvette. The extinction coefficient of BLG is 17600M -1 cm -1. These data are used in data processing with CDPro Optical microscope The images of particle tracking were captured on a Nikon T2000U microscope. 20x, 40x or 100x oil lens objectives were used depending on the sample. SimplePCI software was used in controlling the camera and capturing images. The microscope was equipped with UV light source and blue and green light filter for fluorescent experiments. A shutter was used to control the illumination of excitation light in fluorescent experiment. A QIClick camera was mounted on the microscope to capture digital images. 4.8 Small-Angle X-ray Scattering (SAXS) Small angle X-ray scattering experiment was carried out on beam line 18-ID-D in advanced photon source, Argonne. The wavelength of X-ray radiation was Å, and

42 33 a short exposure period of 1 s was used to acquire the scattering data. Coacervates were loaded between two layers polycarbonate tape. The q range of 2.8x10-3 Å -1 and 0.37Å -1 was used for all samples. Liquid samples were loaded by a pump to the quartz capillary calibrated in the path of X ray radiation. Solid samples were loaded in a square polycarbonate cell sealed with polycarbonate film. The final SAXS profiles were obtained through average of 8 to 15 measurements followed by the subtraction of tape or solvent background. The SAXS spectra are scattering intensity (I) as function of scattering factor (q). q is defined as q=4πsin(θ)/λ, where θ is the scattering angle, λ is wavelength of X-ray. Guinier plot is based on the equation I= I exp q R 3 So by plotting ln(i) against q 2 will yield straight line in the linear region if sample is monodisperse. The slope of this line is Rg 2 /3. The range that Guinier plot can be applied must fulfill the requirement of qrg<1.71. Kratky plot is plotting q 2 I(q) vs q. This plot is sensitive to globular structure and Gaussian chain. For globular compact particles, a peak will emerge in the intermediate range of q. If the object is flexible, the right side of this peak will move upward. 4.9 Particle tracking Samples loaded with probes were filled into a homemade sample cell with about 0.5mm height and 30μl volume inside. The upper side of the cell is a 0.17mm thick cover glass

43 34 and the lower side is a rectangular glass slide. Two pieces of parafilm were sandwiched between the two pieces of glass. After filling sample, the two open sides of the gap were sealed with UV-cure epoxy. The optical microscopy microrheology (OMMR) experiment was carried out on a Nikon T2000-U (Nikon) optical microscope. The video images were recorded by QIClick (QImaging) CCD camera. The magnification of objective was chosen depending on the time scale, viscosity and probe size. Generally, higher magnification is better to test shorter time scale, smaller particle, and thicker sample. This advantage of higher magnification is on the hand a trade off of data volume and short trajectory length. Given limited camera speed (number of bits that can be captured per second), larger region of interest (ROI) and higher resolution lead to lower speed. A sequence of images with equal time interval was captured for each sample. For most images there were 50 to 100 particles in each image. For highly thick samples, this number is smaller in order to enhance the spatial and temporal resolution. Meanwhile, multiple sequences of images were recorded for each sample to reduce the statistical error. All data processing steps were done using homemade Matlab codes. The first step of image processing is feature identification. The raw images were first cleaned by convoluting with a Gaussian function and a Box-Car function. The former can reduce unrelated smaller size noises; the latter can eliminate effect of uneven illumination. The size, shape, total intensity, peak intensity, and average intensity were used to screen for probe particles. The threshold values of these parameters were adjusted manually for

44 35 each image sequence. Preliminary peaks locations were found using a local maximum method. The precise location of each particle was determined by the average center weighed by intensity of each pixel within a user-defined distance from the preliminary peak. The precision of location using this method can be up to one tenth of pixel size. The trajectories were identified using a most likely-hood (MLH) method. Given an image with a set of particles with locations r i (i A), and its subsequent image with a set of particles located at r j (j B, all links between each element in a subset r i and each element in a subset r j was evaluated, and the set of links with highest likely-hood was chosen. Due to the popping of particles in the image sequence, a distance threshold was set to limit how far a particle can move within time interval between neighboring images. Particles without any particle within this distance in the former image were recognized as initiation of a new trajectory, while in the latter image case a trajectory will be terminated. The mean square displacement (MSD) of trajectories is defined as MSD(τ) = <[r(t+τ)-r(t)] 2 > Where τ is the relay time. The MSD curve was extracted from each trajectory and averaged over all trajectories in a sequence, or if applied, multiple sequences from the same sample and same measurement parameters. In order to avoid sequential correlation, in each trajectory with length N, only at most N/t pairs of data points were used to calculate MSD for each t. Here the unit of t is frame. The MSD(t) curves were also smoothed by convoluting with a Gaussian function before calculating storage and

45 36 loss modulus. This is necessary to get the first derivative robust. MSD is converted to complex modulus by GSER[133]: G ω=!" πα#$%& ' ω (Γ)*αω+[*βω/] (2) Where k B is the Boltzmann constant, T is absolute temperature, α and β are first and second order logarithmic derivative of MSD, respectively. The phase angle can be derived from second derivative of MSD. So storage and modulus are[133] G =Gω cos ω *β ω [πα β ωα ω& π 1(] (3) G =Gω cos ω *β ω [πα β ω[1 α ω]& π 1(] (4) Where α and β are local first and second order derivative of MSD. In order to characterize heterogeneity, we used non-gaussian parameter of distribution of displacement at shortest time interval. kτ= 3<[rt+τ rt]8 > 5<[rt+τ rt] > 3 The first item on the right hand side is the Kurtosis of distribution. For all Gaussian distributions, the kurtosis is 3. So the k value is zero if the distribution strictly follows Gaussian distribution. Usually large values will be obtained. The larger the k value is, the far this distribution is different from Gaussian distribution.

46 37 5. Results and Discussion 5.1 Rheology of zein solution Determination of yield strain Before frequency sweep test, the linear viscoelastic zone was determined using strain sweep test. The frequency was set at 10rad/s. Shown in fig.5-1 (A, B), at small strain, both G and G are independent of strain. When strain increased, G and G declined in a power-law fashion. The decay power of G is about one half that of G. G showed a peak near yielding strain, which diminished with increase of concentration. Both above phenomenon are universal on similar systems [ ]. To now the explanation of the physical origin of this peak is not conclusive yet. This peak is characterized in the hard sphere end of colloidal glass, which involves a cage breaking process under strain that beyond the local free volume for colloidal rattling[137]. This peak also decreases when the interaction between colloids becomes more attractive[138]. One of a simple physical explanation is given based on the fact that the relaxation time decreases with increase of strain rate [139]. This model captures both the phenomenon of peak in G curve and that the slope of G at large strain is about one half that of G. For 10% zein solution, it was noticeable that at large enough strain, all curves overlap into one. The steady state test also showed no significant difference in viscosity in the final curve between samples which only differ in aging time, indicating total destruction of higher order structure formed during aging.

47 38 figure.5-1 Strain sweep test of 10% and 20% zein solution. A: strain sweep test of 10% zein solution, diamond: G with t w =7, left triangle: G with t w = 7, right triangle: G with t w =9, hexagon: G with t w =9, square: G with t w =11, circle: G with t w =11, up triangle: G with t w =13, down triangle: G with t w =13; B: strain sweep test of 20% zein solution, up triangle: G with t w =4, down triangle: G with t w =4, square:

48 39 G with t w =7, circle: G with t w =7, star: G with t w =10, pentagon: G with t w =10, right triangle: G with t w =12, hexagon: G with t w =12, diamond: G with t w =15, left triangle: G with t w =15; C: plateau G (square) and G (circle) of 10% zein solution; D: plateau G (square) and G (circle) of 20% zein solution. E: yield strain of 10% zein solution; F: yield strain of 20% zein solution. Solid lines are apparent linear fits to guide the eye. For the linkages with finite strength that binding zein molecules together, the larger clusters will be loaded with higher shear force. When this force is stronger than the weakest cross section of a cluster, the cluster will break into smaller ones. We quantified the strength of gel using mean value at plateau in the linear viscoelastic region, shown in figure5-1 (C, D). We plotted the G and G value at plateau against aging time in a semi-logarithm form. To our surprise, both G and G are exponential function of aging time, since the evolution of colloidal gel usually leads to higher arrested system that slows down the evolution process. Such phenomenon was also observed in charge colloidal system [140]. We speculate that the major factor at this stage that driving aging process is the incorporation of more zein molecules into the stress bearing network, which is less affected by the kinetics compared to the rearrangement of clusters. With increase of aging time, the yield strain decreases in a power law fashion (figure5-1 (E, F)). This is attributed to increase of the density of connection. For a percolation gel, the yield strain is independent of molecular weight between crosslink[141]. While in a colloidal gel, the yield strain is a power-law function of colloidal concentration[142].

49 40 Interestingly, while yield strain decrease with aging time, it also increases with zein concentration, which suggest the change of cluster structure with different zein concentration. We speculate that the zein colloidal gel network composes highly coiled structure at higher concentration Frequency dependence of G and G at early time SGR model predicts that external perturbation will accelerate the aging of colloidal gel by activating colloidal particles to overcome the kinetic energy barriers[137]. In light of this, we tried to first minimize this effect while probing the time evolution of the gel. We studied the frequency dependence of 20wt % and 35wt% zein gel at different aging time. The strain was kept at 1% for all samples within linear viscoelastic region. In studying the aging in the first day of preparation, a dynamic frequency sweep test was applied at each 1 hour interval. The measuring time for each measurement was about 5 minutes. Figure.5-1 shows representative curves of these results. Both G and G increased with frequency within experiment range, which is typical for viscoelastic materials in a glassy region [143]. Both G and G increase with incubation time, while G increase faster than G. At 20% concentration, G is lower than G at measured range before 6 hours incubation. After 6 hours, G start to become higher than G. The slope of G and G decrease with incubation time, indicating a transition from viscoelastic fluid to gel. For all time points, the slope of G was larger than that of G. Possibly the region that G ~ω 2 and G ~ω exists at lower frequency beyond our experiment range. Fig.5-1A shows the points where G cross with G. With increase of incubation time, G and G curves cross at larger frequency. For higher concentration (35%), Both G and G are higher than in

50 41 lower concentration. Also, G become larger than G at earlier time and lower frequency. figure.5-2 Frequency sweep test of 20% and 35% zein solution. A: storage and loss modulus of 20% zein; B: storage and loss modulus of 35% zein; C-D: storage (C) and loss (D) modulus of 20wt% zein at different delay time, square: fresh,

51 42 circle: 3 hour, upper triangle: 6 hours, lower triangle: 11 hours, diamond: 17 hours; E-F: storage (E) and loss (F) modulus of 35wt% zein, square: fresh, circle: 4 hours, upper triangle: 8 hours, lower triangle: 12 hours, diamond: 19 hours. Solid lines connect points to guide the eye. Above evidences coincide with the evidences found in numerous colloidal systems[2, 144, 145]. In the study of Trappe et al., the particle clusters are modeled as a tenuous network. The G is purely controlled by cluster network, and G is controlled by solvent at high frequencies and by cluster network at low frequencies. We observed the same shape of G and G. The trends of G and G as a function of incubation time in our study are similar to the trend as a function of volume fraction and inter-particle potential in their study. We propose the picture of zein clusters growing while keeping single fractal dimension between the length scale of single protein molecule and the cluster. Because the fractal dimension is smaller than 3, the effective volume fraction increases while the size of clusters grow with time. The dynamic of diffusion limited aggregation is ;~ 4= >? 3@A BC DBC R is the size of cluster, C 0 is the initial monomer concentration, t is the time, df is the fractal dimension.

52 43 The G and G against ω plot is fitted by linear model with log10-log10 link. The slope in the fitting is therefore scaling factor of G and G against ω, and the intercepts are the prefactors in eq.1 and 2. G =k ω E ' (1) G =k ω E F (2) figure.5-3 Scaling powers (A) and Prefactors (B) of G (square) and G (circle) as function of aging time. We then plot all parameters in eq.1 and 2 against aging time (figure.5-3). The values of k 1 are not shown for short time due to high standard error. Surprisingly, two distinct regions are observed in plotting k 1 and k 2 against time in log-log form (figure.5-3b). The two regions crossed at about 4x10 4 seconds. In the first stage the slopes of G and G are 0.76 and 0.32 respectively. In the second stage the slopes of G and G are 3.8 and 2.5. This could be the critical point at which a percolate cluster is formed, although its volume fraction was still very low. Our proposed picture of zein solution is as below: zein molecules self-associate in to fractal clusters. As time goes such fractals grow larger.

53 44 The cluster creates fractal less than one at larger scale, therefore the occupation volume of clusters increase with increase of cluster mass, squeezing out free space in solution. At the critical point, clusters become close packed, further increase of cluster mass only lead to increase of fractal dimension. We notice that at this point, G is still about one fifth of G. Therefore, the cluster formed is quite soft, allowing mutual penetration. The stabilization of cluster from collapse is attributed to entropy induced repulsion. The aging of scaling factor is illustrated in figure.5-3a. The scaling of G is always higher than G, and both scaling factors never exceed the value 1. This value initially increases, then decreases with aging time Time evolution under small perturbation While colloidal system undergoes a gradual rearrangement spontaneously toward lower energy with time, SGR model suggests that external perturbation will accelerate this process by activate colloidal particles to overcome the kinetic energy barriers. The effect of perturbation was further analyzed for 20% zein solution. The fresh samples were first pre-sheared at 10Hz for 10 minutes, followed by oscillation test at 10rad/s and 1% strain for a period of time. Initially, G is smaller than G (figure.5-4). With increase of aging time, both G and G increased with accelerated rate. G increased faster than G at all time, and became larger than G at 1500 seconds, which is significantly shorter than obtained by small perturbation. And the G and G values are larger than tested in previous section. For short, small oscillation facilitated the aging process.

54 45 figure.5-4 Evolution of G and G under constant perturbation as function of aging time. Square: G of freshly prepared sample; circle: G of freshly prepared sample; up triangle: G of sample with 8 days static incubation; down triangle: G of sample with 8 days static incubation. Zein concentration: 20w/v%. In order to compare static effect of incubation and perturbation, we also tested the sample after 8 day of incubation in test tube (figure.5-4). A flat delay time was first observed. After the delay, both G and G increased sharply. This could be attributed to the lubrication between clusters by the solvent[146]. The freshly prepared zein solution first undergoes an aggregation stage, in which zein molecules associate to form clusters, and subsequent growth of clusters. After this stage, when the clusters are large enough and close enough to each other, an aging process similar to glass system happens. The dynamic of this stage is slow due to hindering between clusters. The system will be exploring for the state with lowest energy, which could take long time to complete. The

55 46 oscillate perturbation promotes the rearrangement process, which lead to faster increase of storage and loss modulus. The aged sample is relatively resistant to perturbation due to static aging. After about 3 hours of activation process, G and G grow quickly to about 2 orders of magnitude higher. Therefore, the activation process is more important when the clusters are formed and large, while clustering process is more important at earlier stage Steady rate sweep test We performed rate sweep test range from 0.1rad/s to 100rad/s. In contrast to frequency sweep test done below yield strain that do not break gel structure, the steady shearing force will destroy gel into smaller clusters, and change the orientation of clusters to be parallel to the shearing direction, leading to shear-thinning behavior that will only recover slowly with time. Figure.5-5A and figure.5-5b showed flow curves for 10% and 20% zein solutions at different aging time. The freshly prepared solution showed constant viscosity with respect to shear rate, i.e., Newtonian behavior. With increase of aging time, both viscosities of zein solution increased, as indicated by the zero shear viscosity shown in Figure.5-5 C and D. Therefore, we may not simply make relations between viscosity and loss modulus. Possibly for the stress bearing network, aging only changes connection between zein molecules at shorter length scale, while connections at long length scale are preserved in large strain. In contrast to our previous guess that large strain totally broke connection between clusters, above phenomenon

56 47 implies a hierarchy structure of this gel network, where later formed connections have shorter yield strain. We did not confirm this behavior in 20% solution, however, due to the limit of strain range we can probe. In contrast, when sheared at steady state, the network was totally broken, in which case only reminiscent clusters contribute to the viscosity. The higher density of connections leaded to stronger and larger cluster, which made the solution thicker. Similar to the previous result, 10% and 20% zein solution showed different form of aging kinetics. The zero shear viscosity of 10% solution showed a power-law increase with time, while that of 20% solution showed power law increase at shorter time and linear increase at longer time. Figure.5-5 A and B also revealed increase of slope of curves with increase of aging time, consistent with our previous study that with increase of zein

57 48 concentration, the degree of shear thinning increases [34]. figure.5-5 Viscosity of 10% (A) and 20%(B) zein solution as function of shear rate. A: 10% zein solution, square: fresh, circle: day 4, diamond: day 7, left triangle: day 11, right triangle: day 17; B: 20% zein solution, down triangle: fresh, up triangle: day 1, circle: day 2, diamond: day 3, left triangle: day 4, square: day 7; C: zero shear viscosity of 10% zein; D: zero shear viscosity of 20% zein Conclusions of rheology properties of zein solution We studied the rheological behavior of zein at different concentration and different aging time. The G and G curve against strain and/or frequency is consistent with

58 49 colloidal gel behavior. Such behavior was also found in globular protein solution [147]. The evidence cluster forming is also reported in [35]. In strain sweep test, G first showed a plateau, then decrease in a power law fashion, G also showed plateau at smaller strain, then showed a peak near the yielding strain. We observed overlap of G at high strain with same concentration are different aging time. Combined with the flow curve, we suggest that large strain breaks the gel in a hierarchy manner, and coincides with the reverse of aging process. With aging of zein solution, the frequency dependence of G and G tend toward a mature gel from viscoelastic fluid. The flow curves clearly indicate the growth of clusters even two weeks after preparation. By studying the perturbation effect, we found that both growing of cluster and cluster rearrangement contribute to the hardening of network. 5.2 Calibration of optical microscopy microrheology Abstract Microrheology method is calibrated with Newtonian, non-newtonian fluids and gel. Results from micro- and bulk rheology are compared. At low viscosity and low modulus, microrheology show good sensitivity, while bulk rheology can only measure a very narrow frequency range with current geometry setting. At high viscosity and high modulus, microrheology shows its limitation in probing small particle displacement, which is reflected by the low signal to noise ratio of the MDS vs. tau curve. At the

59 50 intermediate region where both micro- and bulk rheology gives good result, the two methods agree well within experiment error. The purpose of establishing this method is to extend the study of food gel systems to broader frequency ranges and smaller length scales Theory and rationale Microrheology is a recent emerged tool to study rheology properties at small length scale. By measuring the movement pattern of standard colloidal probes in the sample media, viscoelastic properties can be extracted either from the trajectory of probes under optical microscope recorded by a camera, or from the auto-correlation function measured by scattering method. The first advantage of this method is its non-evasive feature, which is especially suitable for weak gel systems. The second advantage is, as in optical microscopic method, the ability to probe rheological properties locally and/or at smaller length scale compared to macro-rheology. Third, this method needs only less than 100μl of sample, much sample saving compared to macro-rheology methods. The theoretical base of microrheology is generalized Stokes-Einstein relation (GSER). The stokes law states that the drag force of homogenous fluid on a moving spherical particle is proportional to the viscosity of this fluid, radius of the particle, and the velocity of the particle, written as (1) F = 6πηRv

60 51 where F is the dragging force, η is the fluid viscosity, R is the radius of particle, and v is the velocity of particle. Einstein relation made the connection between particle mobility to diffusion coefficient of this particle, given as (2) D = μ k B T where (3) μ = v/f k B is the Boltzmann constant, T is the absolute temperature. Combining equation (1)(2)(3) gives the Stokes-Einstein equation G = H IJ KLMN (4) From equation (4) we can calculate the viscosity of a fluid given diffusion coefficient of particle and the temperature. This is convenient for Newtonian fluid since its viscosity is independent of shear force. The diffusion coefficient could be deduced from the mean square displacement (MSD) of particle in the fluid as a function of time interval, (5) MSD (t) = 2*df*D*t Where MSD is defined as

61 52 MSD(t) = <ΔOP 2 (t)> (6) The viscosity is calculated by fitting MSD vs t curve to eq.5, put D value into eq.4. <> denotes the ensemble average to all possible values. ΔrP(t) is the displacement of particle during time interval t, df is the dimension. For 3-dimension image, df is 3, while for 2-dimension image, df is 2. GSER is developed to analyze non-newtonian fluid, relating the MSD(t) function to storage modulus G and loss modulus G. The inertia of particle, i.e. the net force exerted on the particle can be written as (7) W QRD = S T D U VD D RD XD The first item on the right hand side is the random force from the media. The second item is a convolution of the response function of the media ζ(t) with the velocity of the particle, therefore, this equation assumes linearity of the system. Using Laplace transform to simply the convolution part gives RYZ= [ \ ] ^*_` a\^*_^ (8)

62 53 where s is the Laplace frequency, ~ denotes the Laplace transform of original function. The velocity correlation function is given by multiply each side by v(0), then do the ensemble average, <R0RYZ>= c`[ ] d^*_`f e _^*a\ ^ (9) The stochastic thermal force F(t) is independent of velocity of particle, so <F*v> = 0; for each degree of freedom, the average energy is k B T/2, therefore the kinetic energy of particle at each dimension m<v(0) 2 >/2=k B T/2 [112]. These make eq.9 into <R0RYZ>= BC H IJ _^*a\ ^ (10) Given ms<<ζ(s), eq.10 can be simplified to <R0RYZ > BC H IJ a\^ =Xg > h? AiZ (11) where Mi(s) is the Laplace transform of mobility. <v(0)v(s)> can be related to the MSD as <R0RYZ>= Z AkG lz (12) so that l^ Mis ^F mno BC H I J (13) given mobility AiZ=6q@YZ; r (14)

63 54 Eq.13 and eq.14 give GSER MSDs l =!" =!" uπvw F ηyw uπvwxiw (15) where G(s) is Laplace transform of complex modulus. G and G can be obtained from G(s) using numerical method [112]. Alternatively, G and G can also be directly estimated from MSD(t)[148], G ω= G ω cos παω (16a) G ω= G ω sin παω (16b) where G ω =!" παc{ F ' ω e}[*~] (16c) and αs= Ec{ F e E ƒ/w (16d) Calibration results Microrheology in Newtonian and viscoelastic fluid The effect of particle size, polymer type and polymer concentration were studied. Three particle diameters, 200nm and 500nm were tested. All particle surfaces bear negative charge due carboxylated surface. A negative charged polymer (λ-carageenan), and

64 55 neutral polymer (dextran T500) were chosen in experiment. A series of concentrations were tested for each polymer in both bulk and micro- rheology. figure.5-6 a: MSD vs t curve of dextran T500 at different concentrations. b: plot of viscosity from microrheology against bulk viscosity; dash line: slope of 1 to guide the eye The viscosity of dextran solution was measured on strain controlled ARES rheometer (TA Instruments). The flows curve of dextran solutions showed Newtonian behavior (data not shown). The MSD vs t curves are shown in fig.5-6a. These curves were linear at small enough t which was limited by the data volume. Viscosity values were calculated using eq.4 and eq.5. We plotted viscosity obtained from microrheology against that from macro-rheology (fig.5-6b). A linear relation was obtained. However, the slope of this curve was slightly smaller than one, which is attributed to difference in temperature. Temperature has profound influence in this experiment since it not only affects the velocity of particle, but also the viscosity of solvent and association of solutes.

65 56

66 57 figure.5-7 Microrheology result from acrylamide gel using 0.2μm and 0.5μm particles. a,b: MSD vs tau curve of carrageenan at different concentration from 0.2μm (a) and 0.5μm (b) particle. c,d: Storage modulus measured from 0.2μm particle (c) and 0.5μm (d) particle. e,f: loss modulus measured from 0.2μm particle (e) and 0.5μm (f) particle. g: G and G calculated for 30mg/ml λ carrageenan using bulk rheology and microrheology. g: comparison of bulk- micro- rheology in 30mg/ml carrageenan solution. The microrheology results of λ-carrageenan samples were shown in fig.5-7(c-f). With increase of polymer concentration, both G and G increased, and their frequency dependence decreased. Two sizes of probes showed similar results, which indicated the mesh size of the solution to be smaller than 200nm. However, the macro-rheology did not agree with micro-rheology well (fig.5-7g). This is attributed to interaction between surface charge of probe and net charge of polysaccharide Acrylamide gel The total concentration of acrylamide and BIS is kept at 3% with variation in BIS concentrations. The MSD vs t curves are shown in fig.5-8. Below wt% BIS, the MSD vs t curves can be used to calculate storage and loss modulus, while above this concentration, the curves are highly noisy due limited spatial resolution, leading to reduced range of frequency from high frequency side (fig.5-9). Ideally, the MSD curve of pure fluid will be a power law function. For pure elastic sample, the MSD curve will approach a plateau at large t. The height of this plateau reflects the spring constant of

67 58 this material, since the elastic energy stored in this deformation for each strand is k B T. In fig.5-9 the storage and loss modulus from bulk and micro- rheology are drawn together. The results of micro-rheology agree well with that of macro-rheology within experiment error. The extremely noisy results from macro-rheology were truncated. From this figure we can see that optical micro-rheology is more sensitive at lower frequency where the stress response is low, and macro-rheology is better in the frequency range where dislocations cannot be precisely measured by optical microscope. These results also supported the influence of electrostatic interaction on micro-rheology method. We prepared a sequence of acrylamide/bis gels with the same total solute concentration (3 wt%) and different ratio of acrylamide to bis-acrylamide. This sequence crosses the gelling point, so we can observe the transition sol to gel. As shown in fig.5-10, G and G have the same scaling with frequency at wt% bis-acrylamide. This is an indicator of gelation suggested by Winter and Chambon [27]. At this point, the scaling of complex modulus keeps constant over all frequencies.

68 figure.5-8 : MSD vs t curve of acrylamide/nnam gel. From high to low, the portion of NNAM gradually increases 59

69 figure.5-9 G and G of acrylamide/bis gel. Results from both micro and bulk rheology. 60

70 61 Red lines: loss modulus from microrheology; Black lines: storage modulus from microrheology; Red dots: loss modulus from bulk rheology; Black squares: storage modulus from bulk rheology. figure.5-10 G and G of acrylamide/bis gel near gelling point, measured by microrheology. Total solute concentration: 3wt% Summary of calibration We successfully established the optical microscope based micro-rheology method. The parameters for measurements were optimized to give best result with current facilities. A charge particle probe can be reliably used in measuring neutral fluids and gels.

71 Microrheology of Konjac Glucomannan gel Effect of temperature figure.5-11 Bulk rheology test of KGM solution with 0.1% coagulant. (a): 2% KGM gel, evolution of storage modulus with time at different temperature. The strain is 1%, frequency is 1Hz. Squares: 50 C; circles: 60 C; triangles: 70 C. (b): frequency sweep test of 2% KGM gel, strain is 1%. Solid: storage modulus; hollow: loss modulus; square: 50 C; triangle: 60 C; diamond: 70 C. figure.5-12 Microrheology test 2% KGM solution with 0.1% coagulant. (a): 2% KGM gel, evolution of storage modulus with time at different temperature. Squares: 50 C; circles: 60 C; triangles: 70 C. (b): frequency sweep test of 2% KGM gel.

72 63 Solid: storage modulus; hollow: loss modulus; square: 50 C; triangle: 60 C; diamond: 70 C. We set coagulant concentration at 0.1wt/v%, KGM concentration at 2wt/v% to test the effect of temperature on dynamic of gelation. 50 C, 60 C and 70 C are tested and shown in fig.5-11 (bulk) and fig.5-12(micro). At higher temperature, the storage modulus stops to increase earlier than at lower temperature. However, the highest storage modulus at higher temperature is lower than at lower temperature, which is consistent previous studies[149] that slower gelation rate leads to better rearrangement of polymer, thus higher degree of junction zone formation. The gelation process of KGM is unique in the way that it involves slow change of backbone structure comparing to diffusion of molecule chains. We tentatively try to compare this mechanism with other polysaccharides. One example is gelation of alginate in presence of calcium ion. One widely reported technique to add calcium ion evenly distributed in alginate solution is by adding calcium carbonate particle and glucono-delta-lactone (GDL) into the suspension [ ]. GDL will gradually degrade and dissolve calcium ion from calcium carbonate particle to the solution. This process meets our requirement that reaction should be slow enough to allow polymer chain rearrangement, but the difference is that gelation limiting factor is fast diffusing ion not polymer which diffuses much slower. In the alginate case, changing ph is analogous to changing temperature in KGM gelation, where decrease of ph will accelerate the gelling process. Many reports showed that too

73 64 fast gelation of alginate leads to lump rather than uniform gel, while slow release of calcium ion can form strong and uniform gel [7, ]. Release of calcium ion is controlled by surface area of calcium carbonate particle, therefore increase of particle concentration is the same as rising temperature in the case of KGM gelation. Kuo et al [155]showed that in increase of calcium carbonate concentration or GDL concentration both leads stronger syneresis, and that increase of calcium release will decrease the gelation time. By rising temperature both increase the deacetylation rate and diffusion rate, but in the former is exponential effect, while in the latter is linear effect. So we expect the higher temperature should play major role in increase the deacetylation rate. On the other hand, it is also possible that the heterogeneity be coming from the uneven exposure to gelation environment, say, the temperature gradient during heating. Before heating process began, the MSD curve of KGM suspension was already nonlinear, reflecting pseudo-plastic nature of this gum. Therefore even without any further crosslinking, some structure already exists in the system under stress. As will be seen in later, some local heterogeneity can already be detected before gelation. Storage modulus measured from microrheology at short time of gelation is much lower than from bulk rheology, which reflects the pore size of KGM gel. With time going on, the storage modulus from microrheology does not rise very fast compared to bulk rheology. Storage modulus from microrheology showed different trend compared bulk rheology. The storage modulus only increase mildly with time, in contrast to the sharp increase observed in bulk rheology. Both storage and loss modulus in microrheology were significantly lower than measured by bulk rheology, suggesting pore size larger than the

74 65 probe even when the gelation has finished. The loss modulus was large than storage modulus in most frequency range even after long time of incubation, indicating free movement of particles among pores. We didn t observe the time point where storage modulus and loss modulus has the same scaling with frequency, which is defined as gelling point suggested by Winter-Chambon. They suggested that for fractal gel, the frequency dependence of storage and loss modulus should be the same, G*~G ~G ~ω α. This could be due to large time interval we used here. However, we do observe decrease of frequency dependence of storage modulus and loss modulus. And that the initial scaling factor of storage modulus is larger than that of loss modulus, and final scaling factor of storage modulus is smaller than loss modulus. Assuming continuous change of both scaling factor with time, we do expect a crossover time t gel to exist.

75 66 figure.5-13 non-gaussian distribution of displacement of 1 second interval. 2% KGM solution, 0.1% coagulant, probes with radius of 500nm and negative surface charge. The solutions were heated for 100 minutes. (a) distribution of displacement with 50 C treatment; (b) distribution of displacement with 60 C treatment; (c) distribution of displacement with 70 C treatment; (d) evolution of non-gaussian parameter with time at 50 C (square), 60 C (circle) and 70 C (triangle) We plot the normalized distributions of displacement and compare that with Gaussian curve (fig.5-13a-c). We found that heating at 60 C results in lowest non-gaussian distribution, as more clearly illustrated by the plot of non-gaussian parameter against temperature (fig.5-13d). In fig.5-13d, 70 C yielded much higher inhomogeneity than at lower temperature. The non-

76 67 Gaussian parameter before gelation is in the range of We compared these values with detected from DI water and acrylamide gel. The non-gaussian parameters from water and acrylamide gel are below 0.1 (data not shown). Therefore we can conclude that the non-zero non-gaussian parameter is not attributed to experiment precision or artifact, but the inhomogeneity of the system itself. It was also observed that the non-gaussian parameter at low level is not stable between samples in the same condition, so can be only treated qualitively rather than quantitatively. At 60 C, inhomogeneity first increase, then decrease with time. At 50 C, inhomogeneity kept increasing. At 70 C, the non-gaussian parameter first increased quickly to a high level, then fluctuated around. The origin of heterogeneity could be from rapid local multiple crosslinking or cyclization[108]. Penaloza et al. [161]suggested that the inhomogeneity of gel first increase, reaches a maxima, then decrease with increase of degree of crosslinking. These two effects can be generalized to uneven distribution of all reacted and potentially reactive sites and uneven distribution of reacted sites, respectively.[ ] Suppose uniform distribution of strands and crosslinkings, increase strands concentration will decrease the pore size. Supposed fixed strands concentration, more inhomogeneous distribution of strands will lead to larger pore size at some region and smaller pore size at other region. In our experiment, the strand concentration is fixed, and the degree deacetylation after gelation is expected to be no less than at lower temperature. There are two possibilities. The first is that heating at 70 C leads to more uneven distribution of junction zones. Combining this line with finding that 70 C heating yield lower gel storage modulus, heating at higher temperature caused more loops and danglings. The second is that the junctions zones are formed too fast that structurally prevent encounter of other strands to form more junction zones. In fig.5-13d we noticed that the non-gassian parameter at 70 C during heating is higher than that at lower temperature in all the time range. This is clearly supportive to the first

77 68 scenario, that actually more junction zones are formed at high temperature, but less strands are incorporated into the stress-bearing network. In the kinetic side, when temperature rises, the deacetylation process is accelerated, leading to faster formation of junctions zones. The increased heterogeneity is attributed to faster crosslinking compared to diffusion of polymers in the sol. Therefore we can conclude that the pore size is attributed to local heterogeneity. At 60 C, the heterogeneity is strongly influenced by degree of crosslinking. At short heating time, low degree of crosslinking causes increase of heterogeneity when number of crosslinking increases. At longer heating time, increase of crosslinking compensates the zones that lack of crosslinking, therefore decrease the heterogeneity. The later effect cannot be explained by distribution of potential junction zones. At 50 C, the heterogeneity keeps increasing with time to the longest time we studied. At this temperature, junction zones are formed at lowest rate compared to higher temperature. Other studies suggest that beyond 100 minute the storage modulus of KGM gel is growing slightly with time [93, 166]. The increase of heterogeneity is therefore attributed to the low degree of crosslinking. The result implies that the number of junction zones will still increase with time beyond 100 minutes, but at much slower rate. So the degree of crosslinking is also lowest at longest time we studied compared to higher temperature. These results are also consistent with previous reports by several other studies.[ ] In this case, we also speculate that increase of temperature should have similar effect as increase of coagulant concentration, which both increase the rate of junction zone formation. The differences between the two are that: (1) the coagulant will be consumed during gelation process, while temperature is controlled to be relatively constant; the effect of concentration of sodium carbonate can vary the ph of solution between 9.5 to 11[97] and (2) temperature also affects the diffusion rate of polymers, which is a controlling parameter of junction zone

78 69 formation and (3) it is also suggested that[170] the coagulant has a solubilizing role in KGM gelation, other than just deacetylate the chain. Since the heterogeneity, we may understand the calculated storage and modulus as effective G and G. To better understand the results from microrheology, we plot the MSD vs τ curve in log-log scale. As shown in fig.5-14, all MSD values increase with τ. We can see that after 100 minutes heat treatment, with increase of concentration, MSD values decreased. The curvature of MSD curve indicated the viscoelastic nature. At higher frequency, the slope of MSD curve becomes smaller. The trajectories are estimated by joining particles in consequential images together by max-likely hood method, with a given threshold of displacement beyond which the particles are not recognized as in the same trajectory. We can see that the area that trajectories can vary greatly at different location. MSD vs. t curve reflects the trend of particle mobility at different time scale. We can see that with increase of coagulant concentration, the area that probes move around in given time interval became smaller. A plateau value at long delay time indicates the elasticity of the material. The square displacement vs. t curves for individual particles show very large diversity, suggest the different environment each particle experienced. From trajectory image we can visually recognize regions with high and low viscosities. Some researchers applied F test to identify difference between trajectories. However, such method is particle based, so cannot cope with situations that a particle is moving through different regions in the period of measurement. Another problem of this method is that it is pair based, cannot to be extended to clustering. We also tested the effect of high concentration of coagulant in water on the movement particles. No effect was shown (data not shown), so we can attribute all the changes to KGM gelation.

79 70 figure.5-14 MSD curve of 2% KGM solution heated for 100 minutes with 0.1% coagulant. The temperatures are 70 C (square), 60 C (circle) and 50 C (triangle) Effect of KGM concentration To test the effect of KGM concentration on gelation dynamics of KGM solution under heat and alkaline condition, we performed a series of frequency sweep test at different heating time and KGM concentration. The strain of frequency sweep test in bulk rheology is set at 1%, which is within the linear viscoelastic region. The heating temperature is set at 50 C. The coagulant concentration is set at 0.2wt/v%, irrespective of KGM concentration. Three effects of increasing KGM concentration: (1) increase of polymer concentration, thus increase the viscosity of the

80 71 solution, making the molecular mobility lower; this will also increase association between molecules to form junction zones; it can also increase the density of the network (2) increase of acetyl group concentration, given the degree of acetylation fixed; this will increase the rate of acetylation, as shown above; (3) increase of total amount of acetyl groups; this is effective when the amount of added alkaline molecules is not much more than total amount of acetyl groups; when the reaction goes on, the decrease of alkaline concentration due to consumption will not be negligible. The first effect will affect the whole stage of KGM solution gelation, as modulation the diffusivities of molecule will affect the gelation process. The second effect will majorly take effect at early stage of gelation, not only on the dynamics, i.e. gelation time, but also on the annealing of network, to affect the final storage modulus of the gel. The third effect will mostly affect the final stage of reaction if the alkaline concentration is close or lower than the acetylated group concentration. The change of ph when 90% of sodium carbonate is converted to sodium hydrogen carbonate is about 1.8, which means the drop of reaction rate singly due to coagulant concentration can be nearly two orders of magnitude. Given our current experiment setting for example, 2wt/v% konjac glucomannan and 0.1wt/v% sodium carbonate, the acetylated group concentration is around 4.76x10-3 M, the sodium carbonate concentration is 5.65x10-3 M. If we increase the KGM concentration to 3wt/v%, the KGM concentration will exceed concentration of sodium carbonate. So with reaction going on, sodium carbonated will all be converted to sodium hydrogen carbonate, lose the ability to provide basic environment. Fig.5-16 shows the results from bulk rheology measurement. Initially, storage modulus increased quickly with time. With increase of KGM concentration, faster increase of storage modulus was observed. This is in agreement with previous reports. The first effect of KGM concentration as discussed above is not significant. Zhang et al. reported an decrease of gelation rate due to high molecular weight[166], and attributed the phenomenon to decrease of

81 72 molecular mobility. At the same time, they also attributed the higher final gel strength to higher connectivity of molecule chains. The effect of higher KGM concentration hinders molecule movement, but at the same time also causes molecules to closer to each other. In our observation the latter effect overcomes the former and leads to faster gelation rate. figure.5-15 Effect of KGM concentration on gel formation. The samples were heated at 50 C. 0.1% coagulant was added to each sample. (a): frequency sweep test of KGM gel after 100 minutes heat treatment. Strain is set at 1%. Solid symbols: storage modulus; hollow symbols: loss modulus; triangle: 1.5%; circle: 1.8%; square: 2%; diamond: 2.4%; right triangle: 3%. (b): evolution of storage modulus at 1Hz with time at different KGM concentration; square: 1.5% KGM; circle: 2% KGM; triangle: 3% KGM. We also observed increase of maximum storage modulus with increase of KGM concentration. The concentration dependency of gel modulus has been reported by numerous papers, yield various scaling factor ranging from 1.33 to 7[92, ], most commonly 2 for biopolymers. We obtained a scaling factor of 5.87, which is suggested to be in the low concentration region compared to critical gelling point [171, 174]. This result also implies that entanglement effect

82 73 may play important role in gelation. Several previous reports found a peak in plotting storage modulus against time [93, 94, 166]. They concluded that the origin of this peak is due to wall slipping of the gel in measurement process. They suggested that syneresis occurs due to fast association and cluster forming aggregation. The solvent is squeezed out from the network and forms a layer of polymer-depleted solvent. In our study we only found the storage modulus doesn t change at longest time studied. This is attributed to the lower temperature we studied compared to theirs. In fact, only at temperature higher than 70 C does the peak becomes pronounced [94, 166]. figure.5-16 microrheology result of KGM solution at different KGM concentration. 0.1% coagulant was added to each sample. The samples were heated at 50 C (a) storage (solid symbol) and loss (hollow symbol) modulus of 2% (square), 1.5% (triangle) and 3% (diamond) KGM solution heated for 100 minutes. (b) time evolution of non-gaussian parameter at 1.5% (square), 2% (circle) and 3% (triangle) KGM concentration. The microrheology results with different KGM concentration are shown in fig For all concentrations studied, storage modulus increased with time at initial stage of gelation. When the pore size is smaller than particle size, the result is expected to be close to bulk rheology. On

83 74 the other hand, if the pore size is larger than particle size, results from microrheology should be substantially smaller than bulk rheology, but rather closer to the sol. The pore size can be affected by the concentration of gel, degree of crosslinking, and heterogeneity of crosslinking. With gelation process going on, storage modulus emerged and rise to a detectable value. Storage modulus from microrheology is significant smaller than that from bulk rheology, indicating the pore size larger than particle size of probes. With increase of KGM concentration, storage modulus went higher at each time point of test. We can see that even at highest concentration, the frequency dependence of G and G did not show evidence of matured gel, which should be nearly flat along all frequency. This finding reflects the behavior of sols among the stress bearing network. We also find that the concentration dependency of G in microrheology at 1Hz is very close to 2, which coincides with what predicted by cascade model[175]. Also in microrheology experiment we didn t find drop of storage modulus within any heating time we studied. Even if the syneresis happens in our material, it may not be detected by the microrheology method with current procedure. Surely we can focus the objective on higher plane where the possible polymer-depleted layer forms. However, for microrheology experiment to be done properly, the probes we are going to observe need to be far enough away from the wall to avoid any surface effect. For now no algorithm to calibrate the surface effect has been reported for particle tracking microrheology to properly calculate the relation between mean square displacement and storage/loss modulus. And for current experiment setting, no reliable method is provided to accurately calculate the distance between all particles observed in an image and the wall.

84 75 At initial gelation stage, the heterogeneity of higher concentration rise first. At final stage of gelation, the heterogeneity (fig.5-16b) at higher concentration is lower than that at higher concentration. We explain this phenomenon as below. The reaction rate at each acetyl group is the same irrespective of the polymer concentration, given the same coagulant concentration and temperature. Therefore, higher polymer concentration leads faster total acetylation rate. During the growing of crosslinked zones, the heterogeneity will first increase due to increase of locally formed clusters. Further increase of crosslinking will occur more and more between clusters. This further progress will fill the gap between clusters and make the system more homogeneous. When gelation process completed, higher KGM concentration leads to lower heterogeneity. Consider that in our experiment setting the concentration of alkaline overdose that of acetyl groups, the result could be explained by higher homogeneity of reaction. Portions of network with lower crosslinking will have higher mobility, thus easier to rearrange to form more junction zones. It is also possible that increase of polymer concentration shorten the length scale at which heterogeneity spans due to higher density of crosslinking, making the size probed to be less detectable Effect of coagulant concentration figure.5-17 Effect of coagulant concentration on gelation of 2% KGM solution.

85 76 The samples were heated at 50 C. (A): frequency sweep test of KGM gel after 100 minutes treatment. Strain set at 1%. Solid: storage modulus; hollow: loss modulus; square: 0.05% coagulant; diamond: 0.1% coagulant; triangle: 0.2% coagulant. (B): microrheology test of KGM gel after 100 minutes heat treatment. Solid: storage modulus; hollow: loss modulus; square: 0.05% coagulant; circle: 0.1% coagulant; diamond: 0.2% coagulant. (C): time evolution of non-gaussian parameter with 0.05% (square), 0.1% (circle) and 0.2% (triangle) coagulant. We tested the coagulant concentration at 0.05% (which is insufficient to fully deacetylate all chains), 0.1% and 0.2%. The KGM concentration was set at 2wt/v%, incubated at 50 C for 0, 20, 40, 60, 80 and 100 minutes. Frequency sweep tests were done at 25 C with 1% strain, which is within the linear viscoelastic region. The results from frequency sweep test are shown in fig.5-17a. To our surprise, increase of coagulant concentration resulted in higher storage modulus after 100 minutes of heat treatment at 50 C. The increase of alkaline concentration will accelerate the decetylation process, similar to the effect of rise of temperature, but also solubilizing the polymer to be better exposed to solvent[170]. Similar results were also reported by Huang et al. [97], but their heat treatment condition was fixed at 100 C for 30 minutes, very different from what we used here. Higher alkaline concentration is also expected to be more resistant to consumption of alkaline molecules during deacetylation process. The effect of coagulant concentration can be explained by the finding from study the temperature effect at 50 C above. For lower coagulant concentration, as discussed, ph will drop more quickly with hydrolysis reaction going on, and the final degree of hydrolysis will be lower. For higher coagulant concentration, the increase of hydrolysis rate is not as significant as can be achieved by rising temperature. But at longer time, the drop of coagulant concentration will not cause too

86 77 much drop of ph that makes the hydrolysis reaction too slow. In previous section we already observed a decreased rate of increase of storage modulus with time at 50 C and 0.1% coagulant concentration. We performed microrheology experiment with the same set of samples. In fig.5-17b, we plot log-log plot of the storage modulus vs. frequency for different coagulant concentrations. Similar to result from bulk rheology experiment, higher coagulant concentration leaded to higher storage modulus. At each coagulant concentration, the storage modulus measured from microrheology is significantly lower than measured by bulk rheology. In fig.5-17c plotted the non-gaussian parameter for different coagulant concentrations. We can see that the heterogeneity at 0.05% coagulant concentration kept very low at all heating time, with a slight trend of increasing with heating time. This is attributed both effect of degree of crosslinking and low deacetylation rate. We can also see that the solubilization effect at this concentration is active. At short time, heterogeneity increase for all coagulant concentrations due to increase of crosslinking, with lowest coagulant concentration the slowest rate. At 0.2% coagulant concentration, the heterogeneity first increase, then decrease with heating time. The increase of heterogeneity at short time is due to increase of crosslinking Conclusions We tested the effect of temperature, KGM concentration and coagulant concentration on the gelation dynamics of KGM solution. Consistent with previous reports, we found faster gelation and lower final storage modulus at higher temperature. Using microrheology, we confirmed that the decrease of final storage modulus is due to increase of inhomogeneity of the gel. The decrease of heterogeneity is attributed to both higher degree of crosslinking and intermediate deacetylation rate. Higher KGM

87 78 concentration leads to both faster gelation and higher final storage modulus. The evolution of heterogeneity showed complex relation with KGM concentration. Unexpected, higher coagulant concentration results in higher final storage modulus. We found higher heterogeneity with lower coagulant concentration. This may partially explain the drop of storage modulus. By comparing the rheology result from bulk and micro- rheology, we found that storage and loss modulus measured from microrheology is about two orders of magnitude lower than from bulk rheology. This need to be noticed since if we want to create weak gels to prevent particles from phase separation, the actual stress on the particle, rather than the strength of whole system need to be considered. In these experiments, microrheology has been proven to be powerful and convenient tool to measure both the local environment and inhomogeneity. Many other systems can be tested by this technique in the future. 5.4 Complexation of polypeptides from the hydrolysis of betalactoglobulin with negatively charge polysaccharide Hydrolysis of BLG We prepare the BLG fibril by adjusting 2wt% BLG solution to ph2, and heating and stirring at 80 C for a series of time. SDS-PAGE experiment (fig.5-18) was performed to verify hydrolysis of BLG into short peptides. BLG hydrolysis progressed with time. The second lane in fig.1 shows band of BLG, whose residue molecular weight is 18400Da.

88 79 With increase of heating time, degree of hydrolysis increased. At 3 hours heat, about half of BLG molecules were hydrolyzed, resulted in two major bands, one of BLG and one of hydrolyzed product At 6 hour s heat, we observed more bands with smaller molecular weight, and no single band was prominent. After 13 hours of heating, almost all BLG molecules were hydrolyzed. Some light bands of high molecular weight molecule were also observed, which were attributed to BLG oligomers formed by intermolecular disulfide bonds. figure.5-18 SDS-PAGE of hydrolyzed BLG. Lane from left to right with different heating hour of BLG: marker, 0h, 1h, 2h, 3h, 6h, 8h, 13h, 19h Atomic force microscopy analysis To verify fibrillation of heat treated BLG, AFM images were collected for heated BLG. The diameters of fibrils were measured by cross section height of fibrils in the AFM images. Consistent with other studies[50], diameters of fibrils were between 3nm to 5nm, with periodic fluctuation along fibril length.

89 80 Formation of amyloid aggregates is shown in AFM images (figure.5-19). Fibril structure was already formed after 1 hour of heat treatment. We studied effect of pectin concentration on 6 hours-heat BLG sample. Addition of pectin induces large aggregates, which is attributed to electrostatic attraction. In fig.5-19(a), we can also observe some periodic fluctuation structure, which is attributed to twisted protofilaments[176]. For BLG pectin complex, partial aggregation is observed. With increase of pectin concentration, aggregation also increased. Two types of fibril were observed in R=0, 0.5, 0.1 and 0.2, with diameter of 4nm and 7 nm respectively. At R=0.5, fibrils with diameter of 10.1nm is observed and dominated the species.

90 81 figure.5-19 AFM image of 6hour-heating BLG with different pectin to BLG w/w ratio. Left up: BLG fibril; Left down: BLG fibril+1/20 pectin; Right down: BLG fibril+1/5 pectin; Right up: BLG fibril+1/1 pectin.