Peculiarities of Aqueous Amaranth Starch Suspensions - American

Biomacromolecules 2002, 3, 17-26

17

Peculiarities of Aqueous Amaranth Starch Suspensions Eckart Wilhelm Federal Center for Cereal, Potato and Lipid Research, Institute for Cereal, Potato and Starch Technology, 32756 Detmold, Germany

Thomas Aberle and Walther Burchard* Institute of Macromolecular Chemistry, University of Freiburg, Germany

Ru¨diger Landers Freiburg Materials Research Center, Freiburg, Germany Received December 4, 2000; Revised Manuscript Received September 12, 2001

Among the starches the granules from amaranth starch (a. hypochondriacus, amylopectin type) are singular because of their extremely small size of 1-3 µm and high uniformity. However, large spherical particles of 30-80 µm in diameter were observed from spray-dried amaranth starch by environmental scanning electron microscopy (ESEM) which exhibited a characteristic fine structure. After the mixture was stirred in cold water, the large particles disintegrate into small ones of 1-3 µm diameter. The particles from the stirred suspension were characterized by static and dynamic light scattering and viscometry. Almost the same diameters were found by the three techniques which were 1.76 times larger than those for the dry starch particles investigated by ESEM. The difference in size is explained by reversible uptake of water. A molar mass of Mw,granule ) 177 × 109 g/mol was measured for the granular particle. After dissolution in 0.5 N sodium hydroxide a molar mass Mw,amylopectin ) 104 × 106 g/mol of the amylopectin was obtained that agreed satisfactorily with that of waxy maize. Thus the granule consists on average of 1700 amylopectin molecules. Furthermore, rheological measurements were carried out with aqueous suspensions at room temperature. A shear gradient dependence was found for concentrations higher than 6% (w/v) of granules. At c > 19% reversible gel formation was observed with G′(ω) > G′′(ω) and a plateau over 2 decades. The zero shear viscosity as a function of c[η] shows behavior similar to glycogen and to latex particles. The granules, however, differ from common latex particles because of their capability to gel formation. Introduction Amaranth is an ancient crop which has been used by the native Indian cultures of Central and Latin America. Both, the leaves and seeds were used for food for 4000 years. More than 50 varieties are known today. Amaranth is cultivated in semiarid regions and even in warm mountain areas up to 12 000 feet. In the perisperm of the seeds, typical compounded starch particles are generated in the amyloplasts, approximately 50-90 µm in diameter. However, when suspended in water small single starch granules of 1-3 µm can be extracted from the agglomerates. Compound-starch particles are typical for most starch raw materials consisting of small granules. Amaranth starch is not well-known in the highly developed countries.1-7 Much higher starch yield per acre is obtained from common cereal grains, in particular with maize. The composition of the amaranth whole flour (a. hypochondriacus) was given by Paredes-Lopez1 et al. consisting of 62.0% starch, 15.7% protein, 7.3% lipid, and fiber. In food application, amaranth is becoming increasingly appreciated because of its high content in leucine, lysine,2 and other essential amino acids and valuable lipid components. The isolation of the protein and lipid fractions turned out to be

not as easy. However, by improved separation technology7 the amaranth starch resulted in a purity of less than 0.2% residual protein content. Recently, amaranth starch has raised the scientific interest of several authors,1-4 but the characterization was limited, so far, to common properties such as solubility in cold water, swelling ratio, freeze-thaw stability, water retention capacity, and aqueous paste viscosity. A characterization of physical-chemical properties has not yet been published for pure native amaranth starch.1-7 The extremely small granule size aroused our special interest because of two aspects. One is concerned with the morphology of these tiny granules and the influences on size and shape of the agglomerates, commonly denominated as “compound-starch” granules. The other motivation for the present investigation arose from the question to which extent such special structures can be determined in aqueous suspension by physical-chemical techniques which are common in colloid and polymer science. The agglomeration of starch granules has been a matter of interest for many years. For illustration agglomerates of starch granules from sweet corn maize are shown in Figure 1.8 The particles cluster together to minimize the surface and thereby form characteristic compound structures. The clustering can result already from the biosynthesis via hydrophobic

10.1021/bm000138+ CCC: $22.00 © 2002 American Chemical Society Published on Web 11/21/2001

18

Biomacromolecules, Vol. 3, No. 1, 2002

Wilhelm et al.

which may be useful to recollect. Thus, after the experimental part, we start our consideration with a short theoretical outline before displaying the experimental results. Experimental Section

Figure 1. Compound structure of granules from sweet corn.8 The smallest individual granules have diameters of 2-3 µm. Table 1. Median Particle Diameter D and Specific Surface Area of Native Amaranth, Oat, and Wheat Starch Granules from Laser Diffraction Particle Size Distributions Determined with the Malvern System 2600a

a

starch

D (µm)

specific surface area (m2/cm3)

amaranth-SGS oat-SGS wheat-SSG

1.6 7.9 6.5

5.194 0.833 0.976

Key: SGS, small granule starch; SSG, small size granules.9

interaction among membrane material from the amyloplasts. But it also can occur even when the membrane material was carefully removed. In the first case the agglomerates will hardly disintegrate in water because of the hydrophobic structure of the membranes, but in the second case dissociation is likely due to the high polarity of water. Small granule starches are known from various plants. Notably, the specific surface area of starches increases remarkably as the granule diameter decreases. Table 1 was taken from an earlier paper by one of us9 where the data from wheat small starch granules (a small size fraction from the whole granule distribution) are compared with smallgranule oat starch and amaranth starch. In the present contribution we extended the characterization made previously by other authors1-4 to spatial structure properties. This includes application of electron scanning microscopy, static and dynamic light scattering, viscometry, and determination of the particle size distribution. Common starch granules with diameters larger than 5 µm are far too large for a reliable structure determination by scattering techniques in aqueous suspension. However, the size of the amaranth starch granules is in the range of large latex particles which were prepared by emulsion copolymerization of styrene with divinylbenzene.10 Particles of this size could be analyzed in suspension by light scattering. It appeared of interest to us whether the same size is observed for the amaranth granules in the aqueous suspension as is measured by electron microscopy in the dry state. We were aware that we are approaching the limits of possibilities in light scattering which will be discussed in the course of this paper. These data are complemented by X-ray diffraction measurements and determination of the iodine binding capacity. Properties were found which differ remarkably from all other starches studied so far in our laboratory. The applied techniques require certain knowledge of the theoretical background

(1) Starch Isolation. (a) Milling Procedure. The isolation of the starch as described in the literature1,2,5 was modified as follows. In a first milling step the seeds of amaranth (a. hypochondriacus) were roughly broken to separate the big hull particles. The latter were sieved off from the starch perisperm fraction before the final flour milling was carried out. (b) Enzyme Treatment. To remove β-glucans and other viscous nonstarch polysaccharides (NSP) from the amaranth flour, a technical enzyme (SPECYME CP from Genencor) containing predominantly cellulase, β-glucanase, and arabinoxylase activities were added to the amaranth starch in a nonalkaline aqueous system.6,7 In the subsequent centrifugal separation the nonviscous slurry was separated into welldefined phases of protein, compound starch granules, and fine fibers, respectively. (c) Starch Separation (Detailed Prescription). A 200 g portion of amaranth flour of 77% starch content and 1 mL of SPECYME CP were suspended in water (1000 mL) and allowed to react for 30 min at 25 °C. The suspension was then mixed for 5 min in a Braun blender and was sieved off from the fibers using an ultrasonic vibrating screen (270 mesh). This starch/fiber wet extraction was repeated four times with 400 mL of water. The centrifugation separation of the combined starch extracts was carried out at 1500g for 15 min. The upper phase contained the protein fraction and was removed manually. The starch phase from the bottom was redispersed in 1000 mL of water, and the centrifugation procedure was then repeated twice. A highly purified starch, freed from fine fibers was obtained. Finally the starch was spray dried. (d) Spray Drying Conditions. A technical small scale spray dryer (NIRO, DK) preferably equipped with a spray disk was used. The experimental conditions were as follows: a disk speed of 22 000 rpm; air inlet/outlet temperatures 180 °C/80 °C to dry a 20% (w/v) aqueous amaranth starch suspension to approximately 8% humidity. In this process the singular amaranth granules of less than 2 µm in diameter agglomerate to globally shaped larger particles of about 30 µm median diameter. In accordance with the analyzed particle size distribution, hardly any singular starch particles were found in the air filter behind the cyclone separation. Agglomeration is typical for any small particles during drying due to the well-known physical forces to build interfaces, thus minimizing the large specific surface area of singular particles. (e) Yield. The procedure yielded 61% starch from the flour; protein content was 0.15%. At this point it might be appropriate to recall that amaranth starches of other varieties can vary in their composition of byproducts. Thus, slightly other behavior than reported here could be observed with other amaranth species. (2) Instruments and Methods. (a) Static Light Scattering (SLS). For static light scattering measurements a

Biomacromolecules, Vol. 3, No. 1, 2002 19

Aqueous Amaranth Starch Suspensions

modified SOFICA photogoniometer was used, fully computer driven (Baur Instrumentenbau, Hausen, Germany). The light source was a 2 mW HeNe laser (λο ) 632.8 nm). Measurements were performed for the aqueous amaranth suspension in an angular range from 25° to 125° in steps of 5°. A refractive index increment of dn/dc ) 0.151 mL/g was used. The concentrations were between 0.043 g/L and 0.43 g/L. (b) Dynamic Light Scattering (DLS). DLS measurements were performed with an ALV photogoniometer (ALV-Laser Vertrieb, Langen, Germany) equipped with an ALV 5000 correlator. A Krypton ion laser, Kr 2011 (Spectra Physics) was used with a wavelength of λο ) 647.1 nm. (c) X-ray Diffraction. The X-ray diffraction was received from powder of the spray-dried native granular starch using the Cu KR line of wavelength λ ) 0.154 nm. The diffraction pattern was made with a Kiessig geometry and consisted of a rotating anode, a collimator, followed by the sample that was placed in a distance of 160 mm from the detection plate. The diffraction pattern was first detected by an image plate and finally digitally scanned. The setup allowed determinations of distances between 0.37 and 6 nm. (d) Determination of Amylose Content. For determination of the amylose content the iodine binding capacity (IBC) of the amaranth starch was measured. The starches were titrated with a potassium iodide/iodate solution as described elsewhere.11 An automatic titratometer was used (Titroprocessor 636, Dosimat E65, Polarizator E 585, all products are from Metrohm, Herisau, Switzerland). An IBC of 20.5% (w/w) for pure amylose was taken as the basis for comparison with the amaranth starch. No detectable amylose content was found. (e) Viscosity Measurements. The viscosity measurements were made with an automatic Ubbelohde viscometer (Schott Mainz, Germany) at 20 °C in aqueous suspension with a capillary of 0.63 mm in diameter. (f) Environmental Scanning Electron Microscopy (ESEM). ESEM photomicrographs were performed by using a Electroscan 2020 electron microscope (Wilmington, MA) with a LaB6 cathode. A gaseous secondary electron detector (GSED) was used for signal amplification. The sample chamber contained water vapor of 6.3 T. Gun voltage was 25 kV. The instruments had no automatic image analyzer, but sizes of selected objects could digitally be measured on the screen. (3) Influence of Stirring. For a disintegration test the spray-dried material was suspended in cold water and stirred up to 10 days. Aliquots were taken every day and analyzed by a particle sizer (Malvern, Zeta Seizer 3). The initial bimodal distribution slowly changed to a monomodal one of much smaller size, and this distribution did not change after 3 days. Light scattering and viscosity measurements were made with the disintegrated starch suspension after 3 days of stirring. Theoretical Background The particular relationships of the analysis of light scattering data might be not well-known to a broader audience. We list here a number of theoretical expressions

without detailed comments which, however, are imperative for an understanding of the interpretation of the obtained data. Static Light Scattering. In static light scattering (SLS) the absolute intensity of scattered light Rθ from solutions is measured as a function of the scattering angle θ. In the notation of Debye12 the corresponding equation is given for dilute solutions as Kc 1 ) + 2A2c Rθ P(θ) Mw P(θ) )

Rθ 1 ) 1 + Rg2q2 + ... Rθ)0 3

(1)

(2)

where q)

4πn0 sin θ/2 λ0

(3)

K is the contrast factor, which depends on the refractive index increment dn/dc, the refractive index of the solvent n0, and the wavelength λ0 of the light source. Rθ is the Rayleigh ratio (absolute scattering intensity), c the concentration in g/L, Mw the weight average molar mass of the particles, Rg the radius of gyration, and A2 the second virial coefficient. The latter describes to a first approximation the interaction between the particles. The molecular parameters Mw, Rg, and A2 can be evaluated from the so-called Berry plot in which (Kc/Rθ)0.5 is plotted against q2 + kc, where k is a suitable constant which makes the measurements at different concentrations separately visible. The scattering vector q is related to the scattering angle by q ) (4n0π/λ)(sin θ/2), where n0 is the refractive index of the solvent. The shape of the molecule becomes detectable at large scattering angles. To emphasize this region, a so-called Kratky plot13 is often usefully applied. Here the particle scattering factor P(θ) is multiplied by (qRg)2 and plotted against qRg ) u. Note: The radius of gyration Rg has a dimension of nanometers, but q has a dimension of inverse nanometers. Thus u ) qRg is a dimensionless quantity. Special examples are given in the text. The Kratky plot allows clear differentiation between different molecular architectures, e.g., linear chains from branched ones, spheres from rods.14 Dynamic Light Scattering. In static light scattering, the time of recording is chosen sufficiently long (about 1 s) by which all fluctuations due to translational and internal mobility are averaged. In contrast, in dynamic light scattering (DLS) the scattering intensity is measured for very short intervals (about 10-6 s). From these intensities a time correlation function (TCF) is constructed, which is given by the following equation15 g2(t) )

〈I(0)I(t)〉 〈I〉2

(4)

where I(0) is the scattering intensity at a certain time t ) 0 and I(t) is the corresponding intensity at a short delay time later. The delay time is varied at least from 10-6 to 10-3 s but can be extended to seconds and minutes. A typical

20


Wilhelm et al.

Figure 2. Time correlation function (TCF) of scattered light from an amaranth starch granule suspension at concentration c ) 0.259 g/L and scattering angle of θ ) 90°. The shape of the TCF is represented by a single exponential decay that is solely determined by the translational diffusion of the particles. No angular dependence was observed. Both observations are characteristic of particles with negligible segmental motions.15b,17

example for such a correlation function from an amaranth suspension is given in Figure 2. The correlation function decays exponentially to a baseline (here normalized to 1). For uniform hard sphere particles without internal motion, the TCF is given by a single exponential g2(t) ) exp(-2Γ(q)t)

(5)

where the decay constant Γ is Γ ) Dq2

(6)

with D the translational diffusion coefficient. In his work on Brownian motion, Einstein16 noticed that the translational diffusion coefficient D is related to a hydrodynamically effective radius Rh Rh ) 6πη0/D

(8)

is a measure for the segment density in the particle.17 Viscosity. Starch fractions are commonly characterized by Brabender units,18 which are convenient in industrial quality characterization and are proportional to the suspension viscosity. Staudinger19 showed that for molecular (or particle) characterization the intrinsic viscosity is a more suitable quantity than the viscosity of the solution (or dispersion) at a certain concentration. This value is obtained by measuring the viscosity at different concentrations in comparison to the solvent viscosity. This leads to the specific viscosity ηsp. ηsp )

t - t0 t0

for hard spheres is given by the relationship [η] ) 2.5/d

(10)

where d ) m/V is a segment density in the swollen particle with m the mass and V the volume of the swollen particle. This relationship allows us to define a third radius Rη ≡

(

[η]M (10π/3)NA

)

1/3

(11)

All three radii will be used in the text. Results

(7)

This relationship is commonly called the Stokes-Einstein relationship. For hard spheres the hydrodynamic radius Rh equals the sphere radius R, but the radius of gyration Rg is only 0.778R. More generally the ratio p ) Rg/Rh

Figure 3. Berry plot, i.e., (Kc/Rθ)1/2 versus q2 + kc, from amaranth starch granule suspension after 3 days of stirring in cold water (room temperature). The concentrations were c1 ) 0.04 g/L, c3 ) 0.12 g/L, c4 ) 0.16 g/L, c5 ) 0.20 g/L, and c6 ) 0.24 g/L. Molar particle mass of the granule: Mw ) 177 × 109 g/mol, granule radius of gyration Rg ) 795 nm, corresponding to a hard sphere radius of R ) 1020 nm.

(9)

t is the time of the solution and t0 the time of the solvent of the same volume needed to pass through a capillary. Extrapolation of ηsp/c to zero concentration gives the intrinsic viscosity [η]. According to Einstein16 the intrinsic viscosity

Static Light Scattering. As already mentioned in the experimental part, the aqueous amaranth starch granule suspension was measured by static light scattering at 20 °C in a concentration range from 0.04 to 4 g/L. Figure 3 shows the result in the Berry representation (i.e., (Kc/Rθ)0.5 against q2 + kc). Because of the very high mass of the particles the commonly applied Zimm plot did not permit an accurate extrapolation to zero scattering angle. Even the Berry plot is accompanied in this case by a relatively large error for Mw and Rg, which lies in the range of about 10%. Such large particles are liable to sedimentation. However, the starch particles sediment only slowly, and static light scattering measurements could safely be performed. The molar mass of the amaranth starch particles was found to be Mw ) 177 × 109 g/mol with a radius of gyration of Rg ) 795 nm. Because of the very large radius of the particles, u ) qRg, values up to u ) 18 could be covered, which established information on the internal structure. For accurate determination of the shape of the amaranth particles, a Kratky plot was applied. Figure 4 shows the result in comparison to the theoretical curve for “homogeneously” branched structures in the limit of very large molar masses Mw.20-22 The measured points match very well the theoretical curve. Only in the high u range do small deviations become noticeable, probably resulting from the hairy region of outer chain on the surface of the particle. For comparison the fit of LS



Figure 4. Kratky representation of the scattering curves at c ) 0 from amaranth granules in water suspension (b) and of molecularly dispersed (O) amaranth starch in 0.5 N NaOH, respectively. The dashed lines correspond to theoretical curves of hyperbranched macromolecules of large size (C ) 0, homogeneously branched) and to the experimental curve for amylopectin with C ) 0.043. Note the agreement between molecularly dispersed amaranth starch with the dashed curve for amylopectin (waxy maize starch). C ) 1 represents linear chains and C ) 0 refers to homogeneously branched particles; a small value of C indicates a large number of branches per particle (see eq 13). Table 2. Molar Mass Mw and Radius of Gyration Rg of Amaranth Starch Granules in Aqueous Suspension in Comparison to the Correspnding Parameters of Molecularly in 0.5 N NaOH Dissolved Amaranth Starch Molecules Determined by Light Scattering amaranth starch in

Mw (g/mol)

Rg (nm)

aqueous suspension 0.5 N NaOH

177000 × 106 104 × 106

795 253

results from waxy maize starch21,23 (pure amylopectin, dissolved molecularly in 0.5 N NaOH) are also given in this Figure 4. These particles are very likely well organized aggregates of a large number of amylopectin molecules. To evaluate the molecular parameters of the amaranth starch, the spraydried powder was molecularly dissolved in 0.5 N NaOH and measured by means of static light scattering. The molecular parameters are listed in Table 2 and compared with the amaranth starch granules suspended in water. The molar mass Mw and the radius of gyration Rg for amaranth starch in 0.5 N NaOH have about the same values as those found for waxy maize starches21 and are thus considered as the molar mass and radius of gyration of the molecularly dispersed branched macromolecules. Also the shape seems to be the same.23 Comparison of the molar mass data obtained from the granules in water with those from the macromolecules in 0.5 N NaOH gave aggregation numbers of about 1700 amylopectin molecules per starch granule. The radius of the amaranth starch granule is only about 3 times larger than that for amaranth starch molecules in 0.5 N NaOH and indicates a strong swelling of the macromolecules when molecularly dissolved in NaOH. The density d ) m/V (mass per volume) of the macromolecules in the granule is about 55 times larger than that for the molecularly dissolved molecules in solution.

Figure 5. (a) Particle radius distribution of amaranth granules in suspension at the three different concentrations of c3 ) 0.173 mg/ mL, c4 ) 0.259 mg/mL, and c6 ) 0.43 mg/mL, derived from the TCFs in dynamic light scattering using the CONTIN program by Provencher24 (for further details, see text). (b) The same data as for c3, c4, and c6 in Figure 5a but in a magnified linear representation in a range from 760 to 1130 nm. Peak position Rh,peak ) 940 nm.

Dynamic Light Scattering. The dynamic light scattering measurements from granules were evaluated by the cumulant technique, from which the translational diffusion coefficient was obtained (see eq 6). No detectable angular dependence was observed, and a very weak concentration dependence was observed. Such behavior is characteristic for spherical particles. Applying the Stokes-Einstein relationship (eq 7), one finds for the hydrodynamic radii at room temperature values of about 900 ( 50 nm. At “elevated temperatures” the radius decreased slightly and reached a value of about 500 nm at 80 °C. In addition to the cumulant analysis the TCF was also analyzed by the CONTIN inversion program,24 by which the size distribution is obtained. Figure 5 shows the results of measurements at concentrations c ) 0.173, c ) 0.259, and c ) 0.431 g/L. For the first two concentrations, the shift of the peak with concentration is negligible. At about c ) 0.4 g/L, the effect of particle interaction becomes noticeable. The ratio of the two radii is F ) Rg/Rh ) 0.80 ( 0.06 (an error of 8% was estimated from those of 4% for Rh and 7% for Rg). This F value corresponds to hard-sphere particles. For amylopectin,25 glycogen,26 and dextran27 in aqueous

22


Wilhelm et al.

Figure 7. Shear gradient behavior of amaranth suspensions in cold water for various concentrations.

Figure 6. Double logarithmic plot of specific viscosity (η0(c) - ηsolvent)/ ηsolvent as a function of the scaled concentration c[η], where [η] is the intrinsic viscosity of amaranth particles in suspension. For comparison the data of molecularly dissolved glycogen from mussels are added.23 The dashed lines correspond to limiting power law behavior.

solution, the value is considerably larger and ranges from p ) 0.99 to p ) 1.3. Viscosity. Another structure-sensitive property is the intrinsic viscosity [η]. This is determined by measuring the reduced viscosity ηred )

ηsp η - η0 ) c cη0

(12)

at different concentrations and extrapolated toward zero concentration. A value of [η] ) 9.9 mL/g was obtained for the aqueous amaranth granule suspension. The viscosity was measured by capillary viscometry. At low concentrations the specific viscosity ((η - η0)/η0 ) ηsp) increases almost linearly with concentration. Because of interactions between the particles, the specific viscosity starts to increase strongly at higher concentrations. Figure 6 shows the result of the specific viscosity ηsp against c[η]. A point of intersection between the two asymptotic regions of low and concentrated suspensions (or solutions) can be estimated at ηspec = 1.0 and c[η] ) 0.8, corresponding to a concentration of c ) 8.1%. The two regions can asymptotically be described by power laws with exponents of s ) 1.02 and s ) 4.59 for the dilute and concentrated regions, respectively. For comparison the results from glycogen (from mussels) in aqueous solution are plotted, which has a somewhat lower intrinsic viscosity of [η] ) 6.9 mL/g.26 The mentioned highly branched glycogen molecules were molecularly dissolved and showed typical behavior of hard spheres.26 A shear rate dependence has been checked for concentrations between 1% and 10%. Up to 6% no shear rate dependence was detectable for shear rates of about 300 s-1. For 8% and 10% suspensions, i.e., just above the critical point of the intersection point in Figure 6, a weak shear gradient was obtained (Figure 7). Only the zero shear rate viscosity was used in Figure 6. Concentrations higher than

Figure 8. Frequency dependence of storage and loss moduli G′(ω) and G′′(ω) for 20% and 30% amaranth granule suspensions. At both concentrations a G′ plateau extends over nearly 2 decades, and G′(ω) > G′′(ω) is found. Both properties are characteristic for gels. At low frequencies G′(ω) decreases and indicates the onset of flow. Thus a reversible gel is formed via association, very likely by hydrogen bonds. The gel point (G′(ω) ) 0) can be estimated to occur at about c ) 19% (w/v).

10% no longer were measured. In a 20% suspension, the frequency dependence exhibits already a plateau for the elastic modulus G′, which is a clear indication for network formation. Figure 8 shows the result from 20% and 30% suspensions. These suspensions had already a white appearance. X-ray Diffraction Pattern. Starches from different origins often show different types of X-ray diffraction patterns, mainly of A or B type. We measured the diffractogram from the amaranth starch and compared this with a potato starch as a typical representative of the B-type crystalline modification. The characteristic Bragg peaks clearly show A-type crystallinity for the amaranth starch and the well-known B-type for potato starch28 (Figure 9). For clarity no example of an A-type crystal is shown in Figure 9; the characteristics can be taken from the literature.28 The diffractogram apparently indicates that no significant gelatinization has occurred on spray drying. Also partial gelatinization can be excluded as a result of the spray-drying procedure, because the water content of the individual starch granules is reduced to 8% in less than 0.1 s. This is much too fast for any starch



Figure 9. X-ray diffractograms from amaranth starch granules and potato granules. These diagrams agree well with those of A-type (amaranth) and B-type (potato) starches.28 Figure 11. ESEM photomicrograph (bar corresponds to 15 µm) from a particle selected from Figure 10. A fine structure becomes visible.

Figure 10. Environmental scanning electron microscopy (ESEM) photomicrograph from spray-dried amaranth granules (bar corresponds to 100 µm).

gelatinization even at 80 °C. Gelatinization is connected with pronounced swelling which for such large particles takes place at a very slow relaxation time. Environmental Scanning Electron Microscopy. Two types of experiments have been carried out. In the first we measured the particle size of the spray-dried amaranth starch powder. An ensemble of large spheres with a broad size distribution was observed as shown in Figure 10. At higher magnification the surface of the large spheres can be seen. This is shown in Figure 11. The fine structure corresponds to sizes of 0.7 to 1.3 µm in diameter. In a second series an aqueous suspension of 0.1% was stirred for about 3 days and then freeze-dried. The photomicrograph of Figure 12 demonstrates a statistical clustering of small, softly shaped polyhedral particles with diameters again of 0.7-1.3 µm. The well organized structure shown in Figure 11 does not correspond to a natiVe compound structure in amaranth seed, which would not dissociate in water into the observed small particles, but very likely it is an effect of the applied spray drying.

Figure 12. ESEM photomicrograph (bar corresponds to10 µm) from a freeze-dried amaranth suspension after 3 days stirring in cold water (room temperature). The loosely aggregated particles have the same size as those in the fine structure in Figure 10. A softly shaped polyhedral structure is recognized.

Discussion Shape and Size of the Particles. The ESEM photomicrograph (Figure 10) from spray-dried amaranth starch granules seems to indicate a well-organized supramolecular structure of spherical shape, which contains a number of smaller subunits. Similar photomicrograph and structures of amaranth starch as shown in Figure 10 were published by Zhao and Whistler5 a few years ago. They considered the supramolecular structure of a “popcorn ball”-morphology being formed by action of the residual protein and fat content in amaranth starch. However, stirring the large spheres for about 3 days in water at room temperature disintegrated this structure into small singular particles. In our view, the special aggregation to spherical supramolecular structures is an effect

24


Wilhelm et al.

of surface/interface interaction during the spray drying process enforced by the spin of the spray disk. The photomicrograph of Figure 11, from the stirred and freezedried starch exhibits only loosely aggregated small particles of 0.7-1.3 µm in diameter. Static light scattering, presented in Figure 4, makes clear that these spheres are not common compact spheres with a uniform internal density. The Kratky plot of the angular dependence resembles the behavior of so-called hyperbranched macromolecules with a very large number of branches (designated as homogeneously branched samples). Hyperbranched means that a structure is formed from a monomeric unit (anhydroglucose) in which the nonreducing end groups in C4 position and in C6 position exclusively can react with the reducing end group only. No other reactions (e.g., reaction between two reducing end groups) are possible. This type of branching differs significantly from the more common random branching in which all types of reaction occur. The corresponding particle scattering factor is given21 by the equation P(q) )

1 + (C/3)u2 [1 + ((1 + C)/6)u2]2

(13)

where u ) qRg and C is a constant that depends on the extent of branching. C ) 1 corresponds to linear chains and C ) 0 to the limit of very large numbers (>1000) of chains in a very large branched macromolecule. This similarity to hyperbranching may be interpreted by starch molecules in the semicrystalline granule being arranged in a special order and not in a random orientation. Such order can be described in the scattering theory by a well-defined space correlation function, in the present case approximately by that of the Debye-Bueche type.22 It is not our intention to overinterpret our findings because, as it is well-known, the angular dependence of scattered light allows no unique assignment to a special structure. Different structures can result in almost identical angular dependencies. The dashed line (amaranth suspension) is described by the equation20-22 P(u) )

1 [1 + (u2/6)]2

(13′)

which is the limiting function for hyperbranched structures21,23 for C ) 0 and is identical with the Debye-Bueche type of scattering but has to be considered here solely as a very efficient fit function. A radius of gyration of Rg ) 795 nm was found. This radius is considerably larger than that observed in electron microscopy. Assuming hard sphere behavior, this would give a diameter of 2.05 µm. However, we have to take into consideration that the ESEM photomicrographs were taken from the dried starch, whereas in light scattering the fully swollen particles were measured. This swelling causes an increase of the diameter, which in the present case would be small compared to synthetic crosslinked microgels. Still the data would indicate a fairly large swelling of the granules that does not coincide with the known water content in the granules, and additional effects have probably to be taken into account. The particles have a certain polydispersity that was measured by dynamic light

scattering using the CONTIN inversion program.24 In electron microscopy the number average of sizes is measured, but in static light scattering it is the z-average, which can be considerably larger. The hydrodynamic radius, which was determined by dynamic light scattering, corresponds very satisfactorily to the conclusion from static light scattering and gave a diameter of 1.8 µm. The ratio of F ) Rg/Rh with a value of F ) 0.8 agrees within experimental error with the value for hard spheres (F ) 0.775) which enable no mutual interpenetration. The molecular dimensions for the granules can be estimated also from the intrinsic viscosity. Applying eq 11, this gives a diameter of 1.94 µm, again in good agreement with the hydrodynamic diameter obtained via dynamic light scattering. The intrinsic viscosity can also be used to estimate the segment density, and thus the swelling of the particle, applying now Einstein’s eq 10. An apparent density of dapp ) 0.252 g/mL was found. The dry substance (containing 10% of water) has a density of d ) 1.369 g/mL. Thus a swelling ratio of (VH2O/VO) ) 5.43 is obtained. This corresponds to an increase of the radii as RH2O ) 1.758Rdry-substance‚ Taking Rdry-substance ) RESEM, we obtained with the data from ESEM the diameter of the “swollen” particle to be 1.23-2.11 µm. Our experimentally determined average dimensions from static and dynamic light scattering and viscometry are 1.801.94 µm, just in the middle of these values. For comparison: the starch granules from rice have diameters between 3 and 15 µm, maize between 4 and 28 µm, and potato starch between 10 and 100 µm.28,29 In the literature a swelling of only 40% in cold water is reported. Particle Size Distribution. One advantage of dynamic light scattering is that the time correlation function can be converted into a distribution of particle radii. This is done by the Laplace inversion, e.g., by the CONTIN program, developed by Provencher.24 This program is available online of the autocorrelator, a special computer which registers the time correlation function in dynamic light scattering. Figure 5 shows the result of such a conversion for three concentrations. Up to about 0.03% there is almost no change in the size distribution. The small shift in the maximum value is within the limits of experimental error. Normalizing these three curves with respect to the maximum position, one obtains the curve of Figure 5b. The combined curve has the advantage of being documented by a sufficient number of measured points. This permitted a fit of the curve by a Gaussian distribution with Rmax ) 928 ( 94 nm. The comparatively low standard deviation of 10% indicates a fairly narrow size distribution well comparable with those of latex particles of this size.10 Viscosity. A characteristic measure of the granule structure is given by the intrinsic viscosity. Polystyrene with a molar mass of Mw ) 106 g/mol has an intrinsic viscosity of [η] ) 1148 mL/g.30 Extrapolating to a molar mass of Mw ) 177 × 109 g/mol, the relationship would give the immensely high intrinsic viscosity of [η]PS ) 1.08 × 106 mL/g. Such high intrinsic viscosities are a result of free swelling due to excluded volume interaction among the segments in a chain. In molecularly dissolved starch free swelling is strongly inhibited by the many branching points in the macromolecule.



This restriction in swelling is much more pronounced for the semicristalline starch granules containing many double helices which are the basis for the partially crystalline granule structure. A restricted swelling can now take place only in the amorphous regions of the granule. Therefore, the intrinsic viscosity is with [η] ) 9.9 mL/g much smaller than for synthetic linear chains in a suitable solvent. A similar small value was found previously with molecularly dissolved glycogen from mussels26 with [η] ) 6.9 mL/g. This smaller value results from the fact that the branching density of glycogen is about 2 times larger than that in amylopectin. The amaranth granule suspensions as a function of concentration show viscosity behavior very similar to glycogen and correspond to that of spherical latex particles. However, it substantially differs from that of common linear synthetic polymers.31 The viscosity measurements could only be followed up to a suspension of 10% (w/v). At higher concentrations a weak shear rate dependence was found as shown in Figure 7. This shear gradient dependence was measured in a Bohlin-CS rheometer. With this instrument even suspensions of 20 and 30% could be measured. The frequency dependence of the storage moduli (Figure 8) clearly shows a plateau over 2 decades in frequency. Such a plateau is indicative for an elastic modulus of a gel. Evidently, this high concentrated suspension forms a physically reversible gel. We did not measure slightly lower and higher concentrations to find the correct gel point, where G′ ) 0. However, an estimation can be made by linear extrapolation of G′ against the concentration c that would give a gel point at about 19% (w/v). This, of course, is only a rough estimation. Conclusions Freeze-drying of amaranth starch suspensions in water caused no harm to the native starch structure. This is a striking effect and demonstrates an extraordinary stability of the granule structure. The high stability of amaranth starch granules may open new fields of application. The intention of the present study was as to whether in aqueous suspension the amaranth starch particles disintegrate into singular granules or remain aggregated in a compound structure. By use of the Malvern Particle Zeta Sizer 3, an initially asymmetric distribution with a tailing to 3-5 µm aggregates was observed. On gentle stirring in cold water for several days, these agglomerates broke up into singular granules and resulted in a symmetric distribution with a median diameter of 1.2 ( 0.2 µm (0.6 ( 0.1 µm radius). The analysis of the scattering data consistently proved hard sphere behavior of the particles in suspension with the following radii RSLS ) 1.026 ( 0.051 µm, RDLS ) 0.90 ( 0.05 µm, R[η] ) 0.997 ( 0.030 µm, corresponding to an average radius of Rav ) 0.97 ( 0.05 µm, which is significantly different from the ESEM radius, RESEM ) 0.50 ( 0.15 µm, for the dry starch (Table 3). This result could be explained by water uptake and ensuing swelling of the granule by a factor 5.4 in volume ()1.7583). In the literature28,29 reversible swelling at room temperature is reported to about 40%, only. This makes it likely that no

Table 3. Comparison of the Amaranth Starch Granule Average Sphere Radii (µm) As Determined by Static (SLS) and Dynamic (DLS) Light Scattering, Intrinsic Viscosity, and Environmental Electron Microscopy (ESEM)a radius (µm) SLS DLS viscometry

1.026 ( 0.051 0.90 ( 0.05 0.997 ( 0.030

radius (µm) averageb ESEM

0.97 ( 0.05 0.50 ( 0.15

a The dimensions in ESEM were measured digitally online on the screen for selected particles. An image analyzer was not available, and the size distribution could not be determined. b Average of light scattering and viscometric radii.

full dissociation of the aggregated granules was achieved. The surprisingly narrow and monomodal size distribution indicates stable clusters containing about 3-5 singular granules per cluster, but no remaining original clusters of about 30 µm in diameter. Such agglomerates could bind water in the interstices without causing significant swelling. The application of static light scattering to the small amaranth starch particles touches the limit of a reliable structure determination. In fact, the more precise Miescattering theory should have been used instead of the here applied Rayleigh-Gans approximation. Unfortunately, the Mie theory could not be applied. The size distribution of the amaranth particles could have been taken into account by time-consuming computer calculation. The result would still be not very reliable, because the Mie theory was developed for compact spheres with a uniform density distribution. However, the internal segment distribution of the present particles is described rather by the DebyeBueche22 space correlation function, and for this case the Mie theory has not yet been worked out. Nonetheless, the radius of gyration remains a correctly determined quantity. The hydrodynamic radius is not influenced by the Mie theory, and the same holds for the radius determined from the intrinsic viscosity. The good agreement of the independently measured radii seems to give sufficient evidence that these structural data represent the actual dimensions of the amaranth starch particles in aqueous suspension. References and Notes (1) Paredes-Lopez, O.; Schevenin, M. L.; Hernandez-Lopez, D.; CarabezTrejo, A. Starch/Sta¨ rke 1989, 41, 205. (2) Perez, E.; Bahnassey, Y. A.; Breene, W. M. Starch/Sta¨ rke 1993, 45, 215. (3) Singhal, R. S.; Kulkarni, P. R. Starch/Sta¨ rke 1990, 42, 5. (4) Bello-Perez, L. A.; de Leon, Y. P.; Agama-Aceredo, E.; ParedesLopez, O. Starch/Sta¨ rke 1998, 50, 409. (5) Zhao, J.; Whistler, R. L. Cereal Chem. 1994, 71, 392. (6) Wilhelm, E. C.; Themeier, H.; Lindhauer, M. G. Starch/Sta¨ rke 1998, 50, 7. (7) Wilhelm, E. C.; Themeier, H.; Mack, H.; Lindhauer, M.-G. In 6th Symposium on Renewable Resources and 4th European Symposium on Industrial Crops and Products, Nachwachsende Rohstoffe; Landwirtschaftsverlag: Mu¨nster, 1999; Vol. 14, pp 664-675. (8) Seidemann, J. Sta¨ rke-Atlas, Grundlagen der Sta¨ rkemikroskopie und Beschreibung der wichtigsten Sta¨ rkearten; Paul Parey Verlag: BerlinHamburg, 1966; p 186. (9) Wilhelm, E. In Plant Polymeric Carbohydrates; Meuser, F., Manners, D. J., Seibel, W., Eds.; Royal Society of Chemistry: Cambridge, U.K., 1993; p 180. (10) SIGMA Catalogue Biochemicals and Reagents for Life Science Research; Sigma-Aldrich Co., 1999; p 625. (11) Banks, W.; Greenwood, C. T. Starch and its Components; University Press: Edinburgh, Scotland, 1975.

26


(12) See: Huglin, M. B. Light Scattering from Polymer Solutions; Academic Press: New York, 1972. (13) Kratky, O.; Porod, G. J. Colloid. Sci. 1949, 4, 35. (14) Burchard, W. AdV. Polym. Sci. 1983, 48, 1. (15) (a) Brown, W. Dynamic Light Scattering; Clarendon Press: Oxford, U.K., 1993. (b) Berne, B. J.; Pecora, R. Dynamic Light Scattering; Wiley & Sons: New York, 1976. (16) Einstein, A. Ann. Phys. 1905, 17, 549. (17) Burchard, W.; Schmidt, M.; Stockmayer, W. H. Macromolecules 1980, 13, 1265. (18) (a) Anker, C. A.; Geddes, W. F. Cereal Chem. 1944, 21, 335. (b) Bloksma, A. H. Int. Union Food Sci. Technol. Symp., London 1977. (19) Staudinger, H. Arbeitserinnerungen; Hu¨thig Verlag: Heidelberg, 1961. (20) The notation “homogeneously branched” refers to the limiting behavior of large hyperbranched polymers when the effect of dangling outer chains becomes negligible compared to the large core of branched chains.21 In this limit the particle scattering factor is given by the equation P(q) ) (1 + u2/6)-2 with u ) qRg (see eq 12). An identical relationship was previously derived by Debye and Bueche22

Wilhelm et al.

(21) (22) (23) (24) (25) (26) (27) (28) (29) (30) (31)

on the basis of the space correlation function C(r,ξ) ) exp(-r/ξ), where ξ is a correlation length that for particles is correlated to the radius of gyration as ξ ) 61/2Rg. Burchard, W. Macromolecules 1972, 5, 604. Debye, P.; Bueche, A. J. Appl. Phys. 1949, 20, 518. Aberle, T.; Burchard, W.; Vorwerg, W.; Radosta, S. Starch/Sta¨ rke 1994, 46, 329. Provencher, S. W. Comput. Phys. Commun. 1982, 27, 229. Galinsky, G.; Burchard, W. Macromolecules 1995, 28, 2363. Ioan, C.; Aberle, T.; Burchard, W. Macromolecules 1999, 32, 7444, 8655. Ioan, C.; Aberle, T.; Burchard, W. Macromolecules 2000, 33, 5730. Guilbot, A.; Mercier, C. Starch. In The Polysaccharides; Aspinal, G. O., Ed.; Academic Press: London, 1985; Vol. 3, pp 210-283. Whistler, R. L.; BeMiller, J. N.; Paschall, E. F. Starch, Chemistry and Technology; Academic Press: Orlando, FL, 1984; Chapter III. Bhatt, M.; Jamieson, A. M. Macromolecules 1988, 21, 3015. Kulicke, W. M. FliessVerhalten Von Stoffen und Stoffgemischen; Hu¨thig & Wepf: Heidelberg 1986.

BM000138+

Peculiarities of Aqueous Amaranth Starch Suspensions - American

Recommend Documents