The Structure and Thermal Stability of Amylose ... - ACS Publications

Subscriber access provided by UNIVERSITY OF WISCONSIN MILWAUKEE

Article

The structure and thermal stability of amylose-lipid complexes: a case study on amylose-glycerol monostearate Bart Goderis, Joke Putseys, Cedric J. Gommes, Geertrui Bosmans, and Jan A. Delcour Cryst. Growth Des., Just Accepted Manuscript • Publication Date (Web): 10 Jun 2014 Downloaded from http://pubs.acs.org on June 11, 2014

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Crystal Growth & Design is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Crystal Growth & Design

The structure and thermal stability of amylose-lipid complexes: a case study on amylose-glycerol monostearate B. Goderisa,*, Joke A. Putseysb,°, Cédric J. Gommesa,c, Geertrui M. Bosmansb and Jan A. Delcourb a Katholieke Universiteit Leuven, Chemistry Department, Polymer Chemistry and Materials, Celestijnenlaan 200F, 3001 Heverlee, Belgium b Katholieke Universiteit Leuven, Laboratory of Food Chemistry and Biochemistry and Leuven Food Science and Nutrition Research Centre (LFoRCe), Kasteelpark Arenberg 20, 3001 Heverlee, Belgium c Université de Liège, Department of Chemical Engineering, Allée du 6 Août B6a, 4000 Liège, Belgium °Current address: DSM Ahead, Materials Sciences R&D, Urmonderbaan 22, 6167 RD Geleen, The Netherlands ABSTRACT Three different crystalline amylose-glycerol monostearate (GMS) complexes with increasing thermal stability can be distinguished: type I, type IIa and type IIb. All complexes consist of GMS-loaded amylose helices that pack hexagonally into lamellar habits. The complex melting points are proportional to the thickness of the lamellae and depend on the amount of water in the system. For type I complexes, SAXS experiments reveal folded amylose chains and a lamellar thickness governed by the presence of two stretched lipid molecules per amylose helix. In the conversion from type I to type IIa complexes, the short amylose chains unfold and assume a stretched conformation, which increases the number of aligned lipid molecules within the helices to four. In type IIb complexes another pair of lipid molecules is added. The derived quantitative relation between crystal layer thickness, water content and melting point for amylose-GMS complexes also predicts the melting points of other amylose-monoacyl glycerol complexes.

* Bart Goderis, Katholieke Universiteit Leuven, Chemistry Department, Polymer Chemistry and Materials, Celestijnenlaan 200 F, 3001 Heverlee, Belgium, phone: +32 16327806, fax: +32 16327990, [email protected] 1 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

The structure and thermal stability of amylose-lipid complexes: a case study on amylose-glycerol monostearate B. Goderisa,*, J.A. Putseysb,°, C.J. Gommesa,c, G.M. Bosmansb, J.A. Delcourb a

Katholieke Universiteit Leuven, Chemistry Department, Polymer Chemistry and Materials, Celestijnenlaan 200F, B-3001 Heverlee, Belgium

b

Katholieke Universiteit Leuven, Laboratory of Food Chemistry and Biochemistry and Leuven Food Science and Nutrition Research Centre (LFoRCe), Kasteelpark Arenberg 20, B-3001 Heverlee, Belgium c

Université de Liège, Department of Chemical Engineering, Allée du 6 Août B6a, B-4000 Liège, Belgium

°Current address: DSM Ahead, Materials Sciences R&D, Urmonderbaan 22, 6167 RD Geleen, The Netherlands * [email protected]

Abstract Three different crystalline amylose-glycerol monostearate (GMS) complexes with increasing thermal stability can be distinguished: type I, type IIa and type IIb. All complexes consist of GMS-loaded amylose helices that pack hexagonally into lamellar habits. The complex melting points are proportional to the thickness of the lamellae and depend on the amount of water in the system. For type I complexes, SAXS experiments reveal folded amylose chains and a lamellar thickness governed by the presence of two stretched lipid molecules per amylose helix. In the conversion from type I to type IIa complexes, the short amylose chains unfold and assume a stretched conformation, which increases the number of aligned lipid molecules within the helices 2 ACS Paragon Plus Environment

Page 2 of 46

Page 3 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


to four. In type IIb complexes another pair of lipid molecules is added. The derived quantitative relation between crystal layer thickness, water content and melting point for amylose-GMS complexes also predicts the melting points of other amylose-monoacyl glycerol complexes.

1. Introduction Amylose is one of the main carbohydrate fractions in starch. It is composed of glucose units mainly connected through α-1,4-bonds and is, as such, predominantly linear. This type of glucosidic bond induces a helix conformation of the chain, which - after assembly into double helices with a pitch of 21 Å - represents the building block for typical starch crystallites. In native starch, however, amylose is mainly amorphous and it is amylopectin – the highly branched glucose polymer present in starch next to amylose – which accounts for the starch crystallinity.1 Once the starch granules have melted (gelatinized) in aqueous media, long stretched amylose double helices, and on the longer run also crystals thereof, account for the stiffness of aqueous starch gels.2 In the presence of suitable ligands, however, the conformation of amylose changes. A more compact single, left-handed helix is formed with a pitch of 8 Å.3-5 Due to the spatial distribution of glucosidic hydroxyl groups, the exterior of this helix is hydrophilic, while its interior is hydrophobic.5,6 In this internal cavity the ligands can reside. This results in the formation of amylose-inclusion complexes. Complex formation of amylose with alcohols, flavor compounds, iodine, lactones… has been reported7-11, but complexation with lipids - mostly fatty acids or monacyl glycerols (MAG) - has been studied most profoundly. Amylose-lipid complexes (AMLC) can be endogenously present in starch12,13, or formed after gelatinization of starch in the presence of endogenous and/or added lipids, as is the case in starch-processing techniques, such as baking of bread14,15 or parboiling of rice.16 Most laboratory synthesis methods for AMLC involve mixing amylose and lipid solutions. 3 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 4 of 46

This generally results in amylose-lipid complexes with a spread in thermal stability.17 The synthesis procedure can be performed at different temperatures: at 60 °C or lower, it results in the formation of type I AMLC, whereas, at 90 °C or higher, type II complexes are obtained. Type I AMLC are said to consist of individual lipid-filled helical segments, randomly oriented in the medium.18,19 The low temperatures impose a high nucleation rate, leading to helices rapidly frozen into their position, and, hence, to complexes with low – if any – crystallinity.19 Given their presumed amorphous character, type I complexes are said to ‘dissociate’ rather than to ‘melt’ at temperatures below or around 100 °C, as determinable with e.g. Differential Scanning Calorimetry (DSC).18,20 When heating type I AMLC rather slowly at intermediate or low water contents (less than 70% water), after the dissociation endotherm, an exothermic and second endothermic event is often recorded in DSC.21,22 The exothermic transition indicates a reorganization of the dissociated complexes into more stable type II AMLC that melt upon further heating in the second endothermic signal.22 At higher water concentrations (> 70% water), only a single endotherm has been reported, pointing to the absence of reorganization processes.21,22 However, more thermostable complexes can still be obtained in this case by long time isothermal crystallization at temperatures above the type I dissociation point.21 The synthesis of complexes at higher temperatures (90 °C) directly leads to type II AMLC, with a perfection and crystallinity that exceeds what can be reached through the reorganization of low temperature synthesized type I complexes.23 The thus formed type II AMLC melt around 115 °C20,24, but can be annealed further to even more stable type IIb complexes by partial melting and recrystallisation.18,19 Accordingly, the originally formed type II AMLC with a melting point of 115 °C are often referred to as type IIa AMLC.

4 ACS Paragon Plus Environment

Page 5 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In contrast to for type I AMLC, there is no doubt that type II AMLC are crystalline, displaying a Wide Angle X-ray Diffraction (WAXD) pattern characterized by three strong reflections in the ranges 0.85 < q < 0.92 Å-1, 1.34 < q < 1.42 Å-1 and 1.52 < q < 1.59 Å-1, with q = 4πsinθ/λ being the modulus of the scattering vector, λ the X-ray wavelength and θ half the scattering angle.23 Such diffraction patterns are often referred to as VH-type patterns, with ‘V’ having a merely historical origin2 and ‘H’ referring at the fact that the complexes are hydrated.2 In VH-complexes, amylose assumes a left-handed helical conformation with 6 glucose units per turn. These helices stack parallel into crystals with – depending on whether the helices are arranged regularly or randomly in an up/down fashion – respectively orthorhombic3 or hexagonal25 symmetry. Both stacking modes produce the strong reflections typical for VH, but differ in position and intensity of the weaker WAXD maxima.25 Analysis of these solution-grown type II AMLC crystals by means of electron and Small Angle X-ray Scattering (SAXS) revealed the presence of lamellar crystallites26-29 with a thickness of 75 to 100 Å, depending on the material thermal history.22,30-32 Hydrolysis of the amorphous amylose chain fragments at the borders of the lamellar crystallites yielded amylodextrin chain fragments with a length equal to the crystallites’ thickness.33 Assuming that the hydrolysis specifically acts on the folds and chain ends at the crystal surfaces, this observation suggests that the polysaccharide molecules assume multiple chain folding, just as crystallizable synthetic polymers do. In these crystal habits, the helical orientation is perpendicular to the plane of the lamellae.10,26-29,31 Since the thermal history of the complexes affects the crystal thickness as well as the thermal stability, it seems logical to seek for quantitative relationships between morphological aspects and thermal properties. However, to the best of our knowledge such relations have not been 5 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 46

established so far. The morphological differences between type I and type II AMLC are particularly unclear and to some extent confusing. Inspired by knowledge on synthetic polymers that crystallize into chain-folded lamellar crystallites and knowing that lamellar thickening often accounts for the increase in thermal stability after a melting and recrystallization processes, it has been suggested that the mentioned AMLC recrystallization from type I to type II complexes may involve a lamellar thickening process.21,22 However, later on, this line of thought was abandoned as type I complexes did not reveal sharp VH-type WAXD reflections and were therefore often considered to be amorphous18,19, evidently casting doubt on the idea that these complexes would be organized in lamellar crystallites. In this article, it will be shown that type I AMLC based on short amylose chains and glycerol monostearate (GMS) are not amorphous, and that a detailed analysis of X-ray data strongly supports the notion of lamellar crystallites for both the type I and type II complexes of these components. Furthermore, it will be clearly demonstrated that differences in lamellar thickness are at the heart of differences in thermal stability for all studied type I and type II AMLC. This work is inspired by Gelders and coworkers who developed a semienzymatic synthesis method for type I AMLC in 2005.34,35 These complexes are composed of fairly monodisperse amylose chains with a length close to four times the lipid length.35,36 The structure and thermal stability of these complexes is studied as well as their transformation into type II complexes, upholding the working hypothesis that molecular monodispersity should result into AMLCs with well-defined morphologies and narrow thermal stability ranges. To that end, time-temperature resolved SAXS and WAXD measurements are used. For the interpretation of the SAXS data a model is used that accounts for details of electron density variations within the lamellae. This


Page 7 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


model allows describing the SAXS patterns up to relatively large scattering angles and goes beyond the more classical two-phase approach as used e.g. by Zabar et al.37

2. Experimental Section 2.1. The phosphorylase catalyzed semienzymatic synthesis of Amylose-GMS complexes All chemicals, solvents and reagents were of at least analytical grade and from Sigma-Aldrich (Bornem, Belgium) unless specified otherwise. The enzyme phosphorylase was extracted from potatoes and complexes were synthesized as described by Putseys and coworkers36, with 10 enzyme units potato phosphorylase (per 100 mL), a molar ratio of glc-1-P/primer of 200 and GMS as lipid. Two sets of complexes were synthesized, one being based on maltohexaose, the other on debranched glycogen as a primer. The paste resulting from the synthesis of the complexes and the subsequent centrifugation, contained approximately 21% dry matter (dm) AMLC. Among others, this paste was used for structural and calorimetric analysis. A part, however, was lyophilized and the free, uncomplexed lipid was removed from the amylose-GMS complexes with chloroform.36 For further X-ray and DSC studies, water was added to the lyophilized and purified samples to reach dry mass (dm) contents of 25 or 33% (w/w). 2.2. Time-temperature resolved SAXS and WAXD Time-resolved SAXS and WAXD measurements using synchrotron radiation were performed on the Dutch-Flemish (DUBBLE) beam line BM26B at the European Synchrotron Radiation Facility (ESRF) in Grenoble (France). Two different setups were used, involving a wavelength λ of 0.95 Å and of 1.03 Å for setup 1 and setup 2, respectively. Setup 1 was used for the materials based on debranched glycogen as a primer and setup 2 for the complexes synthesized from maltohexaose as a primer. 7 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 46

For measurement of the SAXS data, a two-dimensional multiwire gas-filled detector (13 by 13 cm area and pixel size of 250 µm) was placed after an evacuated tube at 3 m (for setup 1) or at 2 m (for setup 2). The reflections of silver behenate were used to calibrate the scattering angles38, expressed as a function of the scattering vector q, with q = 4πsinθ/λ, λ the wavelength and θ half the scattering angle. Setup 1 allowed measuring scattered intensities in the range 0.014 Å-1 < q < 0.32 Å-1 and setup 2 in the range 0.021 Å-1 < q < 0.44 Å-1. WAXD data were collected simultaneously at a position close to the sample on a twodimensional VHR CCD detector from Photonic Science (East Sussex, United Kingdom). The scattering angles were calibrated using a poly(ethylene) standard and also expressed in terms of q. The lowest q-value on the detector equals 0.86 Å-1 in setup 1 and 0.74 Å-1 in setup 2. In both setups the highest q-values go beyond the ranges that are most relevant for studying AMLC. Only the relevant data are displayed. The SAXS and WAXD patterns were normalized to the intensity of the incoming beam, measured by an ionization chamber placed downstream from the sample. The scattering patterns were corrected for the detector response and, in case of the WAXD, additionally for the dark current prior to azimuthally averaging the isotropic data using the home-made software ConeX.39 Both the SAXS and WAXD patterns were corrected for the scattering by an empty setup, taking into account sample and sample holder transmissions. The samples were presented in hermetically sealed DSC pans and the temperature was controlled by a Linkam HFS 191 Heating/Freezing stage (Surrey, United Kingdom). Scattering patterns were collected in consecutive time frames of 25 s for setup 1 and of 30 s for setup 2 while exposing the samples to three different temperature profiles in order to investigate the influence of high temperature isothermal annealing on the structure and thermal properties of the amyloseGMS complexes. 8 ACS Paragon Plus Environment

Page 9 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(i)

In the first profile, starting from 25 °C, the samples were heated to 140 °C (first heating), held at this temperature for 30 min, and subsequently cooled to 70 °C. After 1 min at 70 °C, they were reheated to 150 °C (second heating).

(ii)

In the second temperature profile, the samples were heated from 25 °C to 110 °C (first heating), kept at 110 °C for 30 min and cooled to 70 °C. After 1 min at 70 °C, they were reheated to 150 °C (second heating).

(iii)

In the third profile, the samples were heated from 25 °C to 110 °C (first heating), kept at 110 °C for 30 min, heated further to 120 °C and kept at this temperature for another 30 min before cooling to 70 °C. After 1 min at 70 °C the samples were reheated to 150 °C (second heating).

All heating and cooling ramps occurred at 4 °C/min. 2.3. Differential scanning calorimetry The three X-ray temperature profiles were also used for analysis with a DSC Q1000 (TA instruments, New Castle, DE, USA), calibrated with indium. The purified and lyophilized AMLC were accurately weighed (1.5 – 4.0 mg dm) into aluminium pans and water was added to a final dm content of 25 or 33% (w/w). The paste, containing non-lyophilized, non-purified AMLC, was also weighed in DSC pans (12.0 – 18.0 mg, i.e. ± 2.5 – 3.8 mg dm). The pans were then hermetically sealed. Transition peak temperatures, expressed in °C, and enthalpies expressed in J/g dm complex were determined using Universal Analysis 2000 software (TA instruments). Selected DSC curves are shown and discussed whereas numerical values for melting points and enthalpies are collected in Table SI-1 of the Supporting Information. These values are averages of at least three measurements.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 46

2.4. Modeling the SAXS patterns of AMLC A selection of background corrected SAXS patterns was interpreted as being the scattering from an assembly of randomly oriented one dimensional paracrystalline layer stacks with three electron density levels over the stack and smooth transitions in between.40 Describing the experimental data by classical 2-phase models was attempted but less successful. The experimental SAXS patterns were fitted to Eq. (1) I (q) =

C S ( q ) F ( q) 2 + A 2 4π q

Eq. (1)

with A representing a (constant) background, C a scaling constant and S(q) and F(q)2 the structure and form factor41 given by respectively Eq. (2)42 and (3)40. S (q) =

1 − exp ( − g 2 q 2 L2P )

1 + exp ( − g 2 q 2 L2P ) − 2 exp ( −0.5 g 2 q 2 L2P ) cos ( qLP ) n −1

F (q) = ∑ i =1

2 ( ρi − ρi +1 )  qli  sin   exp ( −0.5σ i2 q 2 ) q  2 

Eq. (2)

Eq. (3)

In this model the layer to layer distance varies according to a Gaussian distribution with an average value LP and a standard deviation σL. The parameter g in Eq. (2) is defined as g = σL/LP. F(q) is the scattering amplitude of the repeated motif with an electron density profile represented by ρ(x). This motif consists of n-1 superimposed block-like entities with electron density ρi and block length li on top of a constant electron density level, representing the nth electron density level. All distances li are defined symmetrically around the center of the motif and the constraint holds that li < li+1. The interfaces between the electron density levels are smoothed by transition layers with a thickness 3σ.43 In the fitting procedures, the number of electron density levels, n, was fixed at 3. Using a higher number of electron density levels was not useful as such fits returned layers with identical 10 ACS Paragon Plus Environment

Page 11 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


thickness, which effectively reduced the number of electron density levels to 3. Fitting was done by minimizing the sum of squared differences between the experimental and model curves on a logarithmic scale. The values of ρ1 and ρ3 were constrained to 0 and 1, respectively, which reduced the number of fitting parameters to 9. The outcome of the fitting procedure is discussed based on the corresponding electron density profiles, obtained via:

sin ( 0.5 NqLP ) 1 2 F ( q) cos ( qx ) dq ρ ( x) − ρ = ∫ 2π 0 0.5qLP ∞

Eq. (4)

with being the average electron density. N is an arbitrarily large odd number (typically 15) and represents the number of layers in the stack. The obtained electron density profiles do not account for paracrystalline distortions and, thus, represent the case in which all layer to layer distances are equal to the parameter LP. When inspecting such electron density profiles, only the shape is relevant given that the SAXS measurements were not conducted on an absolute scale. Integration following Eq. (4), was performed over the range 0 ≤ q ≤ 4 Å-1. All mathematical handling was done in Microsoft Excel and the built-in solver was used for fitting.

3. Results and Discussion 3.1. Morphological transitions: temperature resolved SAXS and WAXD Discussing the structural changes that occur during the three thermal programs is done through the SAXS and WAXD data of the lyophilized and rehydrated 25% dm amylose-GMS complex sample. The 21% dm paste and 33% dm sample data are considered as far as deviations are concerned from the 25% dm sample. The discussion in this paragraph is further limited to the samples based on maltohexaose as a primer since the behavior of materials from debranched glycogen is similar. 11 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 46

At 25 °C prior to any heating the 25% dm material displays a rather broad scattering maximum in SAXS and two broad peaks in WAXD as illustrated in Figure 1. During the first heating in temperature profile (i), the SAXS and WAXD maxima start disappearing at about 95 °C, which is the typical dissociation temperature of type I AMLC. Above the transition temperature, the patterns look like the one in Figure 1 at 140 °C: all mentioned maxima are gone with a featureless SAXS pattern being left, which is representative of a homogeneous mixture. The WAXD pattern only displays a broad maximum at high q–values due to the collective liquid-like scattering of the molecules in the mixture. Note that the maximum occurs at higher q-values than actually displayed in Figure 1, i.e. at q = 2 Å-1. During cooling down to 70 °C in temperature profile (i), another - better defined - scattering maximum appears in SAXS, whereas in WAXD the three reflections emerge that previously have been assigned to VH-type crystals. Following literature, this indicates the creation of type II amylose-GMS complexes. In Figure 1 the main reflections are labeled with the Miller indices for the hexagonal P6522 space group as suggested by Brisson et al25, given that some of the reflections expected for an orthorhombic type of stacking are missing. Note that the 110 reflection has a small shoulder at its high q-side due to the 101 contribution. Clearly, the patterns, characteristic of truly amorphous material (such as that at 140 °C in Figure 1), are very different from the patterns of the originally synthesized type I complexes (25 °C pattern in Figure 1), indicating that the type I complexes are not amorphous. Furthermore, the positions of the (for type I typical) broad maxima such as observed in the 25 °C WAXD pattern, coincide perfectly with the positions of the two major reflections of VH-type crystals. This suggests that also in type I AMLC, a VH-type crystal packing is present, but that the crystals are smaller or less perfect compared to in type II complexes with a well-developed VH-type 12 ACS Paragon Plus Environment

Page 13 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


scattering pattern. Clearly, representations of type I AMLC involving a completely random stacking of lipid containing amylose helices are not realistic. That the 300 reflection is missing makes sense, because crystal disorder tends to broaden higher order reflections more strongly so that ultimately they may dissapear.41

Figure 1: SAXS (left) and WAXD (right) patterns of lyophilized and rehydrated amylose-glycerol monostearate complexes (25% dm) at three selected temperatures while being subjected to temperature profile (i). The profile at 25 °C represents the scattering behavior prior to any heating. The pattern at 140 °C represents the situation after the first heating run whereas the 70 °C pattern illustrates the situation after cooling from 140 °C. The main reflections are labeled with the Miller indices of the P6522 space group suggested by Brisson et al.25

To better illustrate the morphological transitions, a grey scale was assigned to the scattered intensity values and all collected SAXS and WAXD are plotted together as in Figure 2. In this representation, higher scattering intensities have darker grey tones and the patterns progress with time from the bottom to the top, following the temperature profile as indicated in Figure 2. 13 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 14 of 46

Figure 2: Time-temperature resolved SAXS (left) and WAXD (right) patterns of lyophilized and rehydrated amylose-glycerol monostearate complexes (25% dm) when subjected to temperature profile (i), i.e. heating to 140 °C, holding at 140 °C for 30 min, cooling to 70 °C and heating to 150 °C.

Figure 2 shows how the type I AMLC signals disappear during the first heating run of profile (i) and how the type II typical reflections appear at the very end of the cooling segment. The type II SAXS and WAXD features disappear during second heating. The behavior of the 21% dm paste in profile (i) is qualitatively similar to that of the sample with 25% dm. In contrast, during the first heating run of the 33% dm sample and once the type I reflections have vanished, a very small amount of type II material is created, which melts again during further heating in the same run. During subsequent cooling and second heating, also in the 33% dm case, only the SAXS and WAXD signatures of type II complexes appear and disappear.


Page 15 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 3: Time-temperature resolved SAXS (left) and WAXD (right) patterns of lyophilized and rehydrated amylose-glycerol monostearate complexes (25% dm) when subjected to temperature profile (ii), i.e. heating to 110 °C, holding at 110 °C for 30 min, cooling to 70 °C and heating to 150 °C.

The recrystallization into type II complexes during the first heating is enhanced if isothermal steps are included, such as in profiles (ii) and (iii). In these cases, the recrystallization is more outspoken for the 33% dm sample and also occurs in the 25% dm sample, but remains absent in the 21% dm paste. Figure 3 illustrates how in profile (ii) the type I complexes of the 25% dm sample first melt and then recrystallize into type II AMLC at 110 °C. A short time window exists in between the two events where the material is truly amorphous. At 110 °C, the amount of type II material evolves to a certain plateau value. The growth of this fraction accelerates again half way during the subsequent cooling run (as discussed more quantitatively in section 3.5). This occurs at a higher temperature compared to when the samples were subjected to cooling in profile (i) (cfr Figure 2), likely as a result of the seeding by complexes that were formed at 110 °C. During treatment using profile (iii), the type II material in the 25% dm sample which was generated at 110 °C, melts partially, as can be clearly derived from the WAXD intensity reduction in going from 110 to 120 °C (Figure 4). Concomitantly, the SAXS patterns weaken 15 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 46

and shift to lower q-values, the origin of which will be discussed in section 3.3. The growing WAXD and SAXS intensities during subsequent cooling reveal the formation of additional type II complexes in this segment. A similar behavior is seen for the 33% dm sample. For the 21% dm paste sample, type II complexes are only formed during the cooling segment, not at 110, nor at 120 °C.

Figure 4: Time-temperature resolved SAXS (left) and WAXD (right) patterns of lyophilized and rehydrated amylose-glycerol monostearate complexes (25% dm) when subjected to temperature profile (iii), i.e. heating to 110 °C, holding at 110 °C for 30 min, further heating to 120 °C, holding at 120 °C for 30 min, cooling to 70 °C and heating to 150 °C.

3.2. Thermal transitions: DSC The morphological transitions as discussed above are also detectable by DSC. Complementary to the X-ray data, DSC delivers accurate temperatures and enthalpies for the thermal transitions and is able to very sensitively detect spreads in the thermal stability of different crystal populations. Figure 5 displays the DSC heating traces in the three different time-temperature profiles, for the 16 ACS Paragon Plus Environment

Page 17 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


samples with 33% and 25% dm and the 21% dm paste in the panels (A), (B) and (C), respectively. The most obvious transition peak temperatures are highlighted by open and solid bullets. The corresponding peak temperature values are listed in Table SI-1 of the Supporting Information. In section 3.6 the peak temperatures of the closed symbols are used in a further analysis. Again, the discussion primarily focusses on the complexes that are based on maltohexaose as a primer.

Figure 5: DSC heating runs (exo up) for (A) the lyophilized and rehydrated amylose-lipid complexes (33% dm), (B) the lyophilized and rehydrated amylose-lipid complexes (25% dm) and (C) the amylose-lipid paste (21% dm). From top to bottom in each panel: first heating to 140 °C; second 17 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 46

heating in temperature profile (i), i.e. after an isothermal phase at 140 °C and cooling; second heating in temperature profile (ii), i.e. after an isothermal phase at 110 °C and cooling; second heating in temperature profile (iii), i.e. after an isothermal phase at 110 and 120 °C and cooling. The peak temperatures (with values found in Table SI-1 of the Supporting Information) are highlighted by open and closed circles. The temperatures of the closed symbols are used as input for the analysis in section 3.6.

The first heating curve of the 33% sample (Figure 5A) exhibits two endothermic signals separated by an exothermic peak. This interpretation relies on the straight base line, drawn from the onset of the low temperature peak to the conclusion temperature of the high temperature peak. This base line allows interpreting the thermal behavior as being due to the melting of type I complexes (first endothermic peak, pI), the (partial) recrystallization into type II complexes (exothermic peak) and finally the melting of type II complexes (second endothermic peak, pII). This analysis is fully compatible with the X-ray data, revealing only type I complexes prior to heating, which melt and recrystallize partially into type II complexes during heating. Note that also Biliaderis and Galloway reported exothermic recrystallization effects for similar materials.23 For the samples with higher moisture content (Figure 5B and 5C), a single endotherm is recorded during the first heating run and hence only the melting of type I AMLC without conversion into type II AMLC. The exothermic peak recorded during cooling from 140 °C to 70 °C (data not shown) and which – given the X-ray data – is associated with the creation of type II AMLC, occurs at lower temperatures with increasing water content (lower dm%) in the sample. Interestingly, the endotherm in which these type II AMLC melt during the second heating run (the DSC curves labeled (i) in Figure 5) is bimodal for the 33 and 25% dm samples (with the low 18 ACS Paragon Plus Environment

Page 19 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


and high temperature peaks denoted as pIIa and pIIb respectively). The WAXD data are not indicative for complexes other than type II, given the VH-type patterns. It therefore seems that DSC is able to detect two different type II complexes with the least stables ones melting in pIIa and the most stable ones in pIIb. Following literature, the AMLC melting at pIIa and pIIb are referred to as type IIa and type IIb respectively. Note that the pII peak maximum recorded during the first heating run of the 33% dm sample lies in between the pIIa and pIIb maxima of this material’s second heating run. It seems that during the recrystallization in the first heating run type IIa as well as type IIb complexes are created, producing a peak at intermediate temperatures. Similarly, the pII peak during the second heating run of temperature profile (i) for the 21% dm paste sample is unimodal with a peak lying in between the positions expected for type IIa and type IIb AMLC (see discussion of profile (ii) DSC behavior below), suggesting that both AMLC are present. Furthermore, a very small endothermic peak exists around 97 °C, a temperature typical for melting/dissociation of type I AMLC. The fraction of type I AMLC is very low or non-existing in the 25 and 33% dm samples. Analysis of the DSC cooling behavior reveals that the creation of type II AMLC for the paste sample is so slow that it is postponed to the lowest temperatures at which - in a competitive mode - some type I complexes can be nucleated next to the type IIs. Indeed, for the 21% dm paste sample the majority of the AMLC (82%) is created isothermally at 70°C whereas for the 25 and 33% dm samples more than 75% is formed during the cooling run. Note that both pIIa and pIIb peak temperatures decrease with decreasing dm content during the second heating in profile (i). This behavior is also observed for pI in the first heating runs of the 33 and 25% dm samples. The pI peak of the 21% dm paste during the first heating run, however, deviates: pI is slightly bimodal, with the second mode being most intense and peaking at a higher 19 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 46

temperature than that of the 25% dm sample. The origin of this effect is unclear, but could be attributed to differences in sample history (preparation), leading to e.g. an uneven water distribution for the paste. After the first heating and subsequent cooling a water homogenization seems to have taken place since the pI peak of the 21% dm paste recorded in the profile (i) second heating run follows the general downward trend with increasing water content. Annealing at 110 °C followed by cooling, as in profile (ii), brings the pII endotherm during second heating to a melting enthalpy of about 20 J/g for all samples. This is significantly more than during the second heating run in profile (i) (see also Table SI-1 in the Supporting Information). The pIIa and pIIb shares for the 33% dm sample seem to be equally strong, which results in a peak with a single maximum in between the positions expected for type IIa and IIb complex melting. Note that the 33% dm sample synthesized from debranched glycogen as a primer, displays a more dominant pIIa share (data not shown). For that reason, the SAXS pattern of this material after the isothermal stay at 110 °C, rather than that of the maltohexaose-based material will be used to selectively reveal the type IIa AMLC morphology. Compared to in profile (i), the melting trace during second heating in profile (ii) for the 25% dm sample exhibits a larger type IIb share. Due to the creation of a high amount of type IIb AMLC, the second heating in profile (ii) of the 21% dm paste sample produces a bimodal melting peak. Since no exothermic heat is recorded during the isothermal stay at 110 °C, the amount of type II complexes formed at 110 °C for this sample is vanishingly small, corroborating the X-ray experiments discussed above. The creation of a (non-detectably) low fraction of type II nuclei at 110 °C is only revealed indirectly by an earlier crystallization (i.e. at higher temperatures) during cooling and the larger contribution of pIIb in the second heating run of profile (ii). Clearly, type I complexes are absent in the second heating run of the paste sample in profile (ii). Also for the 33 20 ACS Paragon Plus Environment

Page 21 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


and 25% dm samples, the exothermic peaks detected during cooling in profile (ii) occur at higher temperatures compared to when cooled in profile (i) (data not shown), suggesting that type II complexes formed during the isothermal period at 110 °C serve as nuclei for growth during the following cooling run. The melting enthalpy for the 33% dm sample during the second heating run in profile (iii) amounts to about 25 Jg-1, revealing that the creation of type II complexes is further enhanced by adding another isothermal period at 120 °C. For the 25% dm sample, no difference in enthalpy is seen compared to the value in profile (ii). In contrast, the melting enthalpy value decreases for the 21% dm paste sample compared to in profile (ii). For this sample, 120 °C is almost identical to the pIIb peak maximum, suggesting that the supercooling at 120 °C is too small to allow nucleation of type IIb complexes. For sure, type IIa AMLC cannot be created at 120 °C since the pIIa peak at this dm% is lower than 120 °C. All type IIa complexes that melt in the profile (iii) second heating run thus originate from the cooling run. For the 25% sample too, at 120 °C, only type IIb AMLC can be present since the melting point of the type IIa complexes at that particular dm (119.5°C, see also Table SI-1 in the Supporting Information) lies below 120 °C. The type IIa share in the second heating melting peak thus results from type IIa AMLC created during cooling. The selective presence of type IIb AMLC at 120 °C for this 25% dm material is exploited further below to selectively characterize the structure of type IIb AMLC at this dm% by means of SAXS. At 120 °C, type Ia as well as type IIb AMLC can be present or created in the 33% dm sample. 3.3. The SAXS based morphology of type II AMLC Based on the DSC discussion in section 3.2, it is clear that most often type IIa and type IIb AMLC are generated at the same time, except in a few cases which are chosen in the present 21 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 22 of 46

section to selectively analyze the corresponding SAXS patterns. The SAXS pattern at the end of the 110 °C isothermal segment in temperature profile (ii) for the 33% dm sample based on debranched glycogen is representative for pure type Ia AMLC, judging from the DSC second heating run (cfr supra). A representative case for type IIb AMLC is found in the SAXS pattern of the 25% dm sample (synthesized form a maltohexaose primer) at the end of the 120 °C isothermal segment in profile (iii). In Figure 6, the experimental SAXS pattern together with a best approximation obtained by fitting to Eq. (1) is shown for the mentioned type IIa representative. The agreement between experiment and fit is quite satisfactory, suggesting that the corresponding electron density profile (middle panel in Figure 6) is a good representation for the type IIa AMLC nanomorphology. Morphologies composed out of layer stacks are well accepted for type II AMLC.10,27,30-32 No absolute electron density values can be provided since the SAXS patterns were not collected on an absolute scale. Yet, the shape of the profile is correct although from a pure SAXS point of view also the profile obtained by mirroring the present one with respect to its average electron density would be equally valid. Given that earlier electron microscopy work revealed the existence of lamellar crystallites with a thickness of about 100 Å29 and considering that VH crystals have a density higher than that of water25, it was judged appropriate to represent the electron density profile as in Figure 6 and to interpret the largest, high density feature as being due to the AMLC. The sketch at the bottom of Figure 6 is a further interpretation of this electron density profile. Taking into account that the most frequently occurring amylose chains after synthesis have approximately 115 glucose units36, that in their helical conformation as encountered in VH-type crystals six glucose units are consumed per helical pitch, and that one pitch corresponds to 22 ACS Paragon Plus Environment

Page 23 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


8.0 Å3,25,32,44,45, it follows that one typical chain with a single helix conformation corresponds to a rod of about 153 Å. This length approaches the repeat distance of the electron density profile in Figure 6.

Figure 6: Top panel: Experimental SAXS pattern after 30 min at 110 °C in temperature profile (ii) (open circles) together with the fitted model SAXS pattern (full line); Middle panel: the associated electron density profile; Lower panel: sketch of a compatible stacked lamellar type IIa AMLC. The blue helices represent amylose helices and the grey features are stretched glycerol monostearate molecules. The green dots represent water molecules. Data correspond to the 33% dm sample based on debranched glycogen as a primer.

The sketch of the structure also clarifies that in the type IIa AMLC four monoacyl glycerol (MAG) molecules with a stretched conformation can be included within the helices that span the 23 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 24 of 46

high density feature of the electron density profile. The length (expressed in Å) of such a stretched MAG molecule, lMAG, can be computed46 from:

1.256 ( M − 1) + 7.8 = lMAG

Eq. (5)

with M being the number of carbon atoms in the aliphatic chain. In the present case, M equals 18 and lMAG thus equals 29.15 Å. Following this line of thought, the actual thickness of the AMLC, defined as lC, is thus equal to 116.6 Å (4 times lMAG). The low density features in between the AMLC are interpreted as water rich layers. Note that the thickness of these layers only accounts for - at most - one quarter of the stack repeat distance. For a 33% dm sample, this implies that a reasonable amount of the water must reside outside the layer stacks. The fact that the electron density declines at the borders of the complex layers is tentatively associated with the random occurrence of void spots, i.e. the occasional absence of lipid molecules at the crystal borders. The repeat distance is markedly larger for the type IIb complexes found in the 25% dm sample (Figure 7) than for the type IIa AMLC discussed above. The size of the high density AMLC feature increases so much that it allows incorporating another pair of GMS molecules, bringing the number of stacked GMS molecules across the lamellar thickness to six. Primarily, this structural difference is causing the difference in thermal stability between type IIa and type IIb AMLC as will be discussed further in section 3.6. Note that apparently very few - if any - voids are present at the borders of the type IIb AMLC, judging from the constant electron density across the crystal cross section, shown in the middle panel of Figure 7. The absence of lipid voids correlates with a narrow melting peak. Indeed, the pIIb peak in the DSC profile of the 25% dm sample of profile (iii) is remarkably sharp. In contrast, the pIIa peak in the second heating run for the 33% dm sample in profile (ii) (irrespective of whether it corresponds to that of the debranched glycogen or maltohexaose based 24 ACS Paragon Plus Environment

Page 25 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


material) is rather broad and it can be speculated that this is at least partially related to the distribution in degree of GMS fill-up. Another difference, with the profile type IIa electron density profile in Figure 6 is that the one in Figure 7 for type IIb AMLC has a more narrow water layer, implying that more water has to reside outside the lamellar stacks. A higher tolerance for water between the crystals in the stacks (as in Figure 6) seems to go together with a higher concentration of lipid voids. Finally, it can be pointed out that only a very small fraction of the amylose chains is long enough to span the layer thickness of type IIb complexes36. The low intensity of the VH reflections in the 120 °C segment of Figure 4 proofs that indeed the amount of this highly perfect type IIb material is very low.

Figure 7: Top panel: Experimental SAXS pattern after 30 min at 120 °C in temperature profile (iii) (open circles) together with the fitted model SAXS pattern (full line); Middle panel: the associated electron density profile; Lower panel: sketch of a compatible stacked lamellar type IIb AMLC, following a similar representation mode as in 25 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 26 of 46

Figure 7. Data correspond to the 25% dm sample based on maltohexaose as a primer.

3.4. The SAXS based morphology of type I AMLC Although different views exist on type I complexes18,19, the lamellar model outlined in section 2.4 was used for type I AMLC as well: the presence of (broad) WAXD reflections at q-values comparable to those of type II complexes points at a crystalline aggregation of the moieties in the type I complexes, rather than at a random, amorphous assembly. Since the 33% dm material synthesized from debranched glycogen as a primer was discussed in the context of type II complexes, it is convenient to also consider this sample in its type I form (as represented in Figure 8A). The SAXS-based type I AMLC structure of this material is compared to that of the 25% dm and 21% dm paste samples in Figure 8B and 8C, respectively. The latter two are based on the maltohexaose primer. Studying the type I AMLC as a function of the dm% is feasible at room temperature prior to any heating, since at that temperature only type I AMLC exist. For the type II complexes, the contributions of type IIa and type IIb SAXS patterns overlap quite often, which precludes a structural analysis more elaborate than presented in section 3.3.


Page 27 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 8: Experimental SAXS patterns at 25 °C prior to any heating (open circles) together with the fitted model patterns (full line) in the first row and in the second and third row, respectively, the associated electron density profiles and sketches of compatible stacked lamellar type I amylose-lipid complexes, following a similar representation mode as in Figure 7. Column (A) contains the data from the 33% dm sample based on debranched glycogen, column (B) and (C) the data from, respectively, the 25% dm and 21% dm paste synthesized with maltohexaose as a primer.

For the 33% dm material, the agreement between experimental and model SAXS pattern is very good, suggesting that the electron density profile in Figure 8A is realistic. The repeat distance within this electron density profile is about half as large as that of the type IIa AMLC in Figure 6. For the amylose chains to accommodate to stacks with a repeat distance of only half their molecular length, U-folded chains have to be accepted as sketched at the bottom of Figure 8A. Just as in type IIa complexes, the electron density is lower at the complex borders, likely as a result of lipid voids. In the conversion from type I to type II complexes, the amylose chains thus unfold and assume a stretched conformation.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 28 of 46

Figure 9: Morphological representations for type I AMLC, following a similar representation mode as in Figure 7. From (A) over (B) to (C) the dm% decreases and concomitantly the lamellar stacking order decreases as explained in the text. The long sides of the red rectangular boxes refer to the (lateral) size of the crystals as probed by WAXD, i.e. D120 (see section 3.6). The electron density profiles in the Figures 8B and 8C demonstrate that the stack periodicity increases with decreasing dm%, suggesting that the more water in the system, the further the complex layers are separated from each other by thicker water rich layers. Concomitantly, the lipid void regions are blurring out and the high electron density layers seem to become larger than the thickness expected for complexes made of folded amylose chains. These effects can be rationalized, assuming a progressive departure from the ideal lamellar model as illustrated in Figure 9. A projection of the electron density in a direction parallel to the layers will produce an electron density profile as sketched in Figure 8A only for a well aligned stacking of the helices as in Figure 9A (note that the lipid voids are omitted in all sketches although they should be present). At a poorer helix alignment, as e.g. in Figure 9B, the projection will naturally lead to a more blurred electron density profile and apparently thicker high density regions as in Figure 8B. The fact that the helices in Figure 9B are miss-aligned in patches of approximately 10 helices 28 ACS Paragon Plus Environment

Page 29 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


stems from an analysis of the WAXD patterns outlined in section 3.5. A very high degree of miss-alignment as in Figure 9C may ultimately lead to stacks of limited height and lateral dimensions, i.e. the loss of long range order. When interpreting such morphologies in terms of a model that assumes infinitely extended and parallel stacked lamella, disagreements between experimental and model SAXS patterns appear, in particular at the lowest q-values where contributions form the largest lengths scales can be found. Furthermore, besides a blurring of the model electron density profiles, also progressively higher values for the apparent paracrystallinity (i.e. the parameter g in Eq. 2) are obtained. Apparent refers to the fact that in principle both a stack height reduction and an increasing paracrystallinity contribute to the width of the SAXS peak41 but that for convenience all broadening effects are attributed to paracrystallinity in the adopted SAXS model. For sure, the loss of long range order and molecular orientation correlation as e.g. in Figure 9C will lead to the absence of birefringence as often reported for type I AMLC.23 On the other hand, the absence of birefringence does not mean that the materials are amorphous, nor that the crystals lost their lamellar habit. In fact, also for the 33 and 25% dm samples the long range order seems largely lost for the type I complexes given that rather high g parameters are obtained. The g values approach 0.4, which is the limit beyond which the paracrystalline model physically makes no longer sense.47 3.5. Morphological features extracted from WAXD All the WAXD patterns of the type II AMLC discussed in section 3.3 resemble that of the 70 °C pattern in Figure 1. At 25 °C, prior to any heating, all studied samples exhibit the two broad type I AMLC typical reflections as illustrated in the 25 °C pattern of Figure 1, which for the different dm% are indistinguishable. Accepting VH-type crystals with hexagonally arranged helices for 29 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 30 of 46

type II as well as for type I AMLC (recall section 3.1), an analysis was made of the 120 WAXD reflections (occurring approximately at q = 1.37 Å-1) as a function of temperature. The 120 reflections were separated from the patterns using linear sectors and approximated by a Gaussian profile. The center of the peak, qmax, was translated into the a (or b) unit cell dimensions using the relation

a=

2π qmax

28 3

Eq. (6)

This a dimension corresponds to the lateral helix-to-helix separation. The width of the peak was converted into an estimate for the crystal size, D120, seen in a direction perpendicular to the 120 reflecting planes using the Scherrer equation:

D120 =

λ β cos θ

Eq. (7)

In Eq. (7) λ is the X-ray wavelength, β the integral width of the peak expressed in radians and θ half the scattering angle. Note that D120 represents a minimum estimate for the lateral crystal size (the actual size must thus be somewhat larger) since no corrections were made for instrumental peak broadening and since broadening effects due to crystal disorder are not accounted for. The peak area serves as a relative measure for the amount of AMLC at a given temperature. The discussion focusses on the outcome of this exercise for the 25% dm sample during temperature profile (ii). These data are represented in Figure 10 and contain all features that are also relevant to the other samples and temperature profiles.


Page 31 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Figure 10: WAXD-based estimates for the AMLC lateral crystal size, D120 (open squares with values that can be read on the left ordinate – mind the axis break), the helix-to-helix separation distance represented by the lattice parameter a (crossed squares with values that can be read from the right ordinate) and the relative amount of AMLC probed via the 120 peak area (full squares on an arbitrary scale with the zero intensity coinciding with the horizontal black line within the graph). The data are collected during temperature profile (ii) for the 25% dm sample based on maltohexaose as a primer. The vertical lines delineate the different segments within thermal profile (ii).

While the 25% dm sample is heated from 25 to 110 °C (first heating), the broad type I AMLC typical reflection disappears due to melting as seen in the downwards trend of the black squares in Figure 10. Melting starts at about 78 °C, but progresses most rapidly in the range 95-100 °C where DSC displays its pI peak (see Figure 5). During the first heating, D120, increases continuously, even before melting starts. Since the crystalline content is not (significantly) increasing in this pre-melting range, the observed crystal growth must be established by a merging of existing smaller crystal fragments or by an (Ostwalt) ripening process in which the growth of larger crystals is accompanied by the dissolution or melting of smaller ones. This 31 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 32 of 46

increase of D120 continues in the melting range and may be enhanced by the earlier melting of the smallest crystallites. The distribution in lamellar lateral size thus contributes to the width of the melting range, besides perhaps differences in lipid void concentration or local water content. The invariance of the helix-to-helix distance (the unit cell parameter a in Figure 10) demonstrates that there is no distribution in helix packing efficiency that could contribute to the crystal stability distribution. Note that the a dimension is a bit larger (13.99 Å) than the 13.65 Å usually reported.25 Dividing D120 by 13.99 Å reveals that the lateral size of the type I AMLC corresponds to approximately 10 helices in a row. This leads to block like crystallites with dimensions as highlighted in Figure 9. The dimension along the helical axis results from the SAXS data interpretation. Assuming laterally isometric crystals 100 ordered helices reside in such blocks. At 110 °C type II AMLC are being created isothermally up to a certain plateau level (see the increase of the full squares in Figure 10). The crystals grown at the start are at least 750 Å wide, which for laterally isometric blocks and an a unit cell dimension of about 14.1 Å translates into assemblies of approximately 3000 ordered helices. This is a lot more than in the type I AMLC. Such laterally extended 116.6 Å thick (SAXS-based dimension) lamellae readily stack into high piles and may merge laterally into wider stacks with a high degree of orientation correlation and hence display birefringence. Interestingly, D120 decreases while the amount of type II AMLC is increasing. Since there is no driving force for crystals to become smaller with time, it seems that at first larger crystals are formed and later smaller ones, which on the average leads to a reduction. It is conceivable that a spatial confinement is set up by the creation of the earliest larger crystals, leaving only space for progressively smaller ones. When space is very limited and according to classical nucleation concepts, a larger supercooling might be needed for very small crystallites to nucleate. The cessation of the type II AMLC creation in the 110 °C segment may 32 ACS Paragon Plus Environment

Page 33 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


point to such a nucleation problem. Alternatively, this effect may originate from the steadily increasing water fraction in the non-crystalline regions while crystallization progresses (see also section 3.6). The packing density of the earliest formed larger type II crystals at 110°C is relatively poor, given the high unit cell dimension values. These values decrease to a plateau level, which is still significantly larger than the unit cell dimension of type I AMLC. As higher packing efficiencies are usually associated with higher thermal stability and since the opposite trend is observed here, one can conclude that the packing efficiency is not governing the difference in thermal stability between type I and type II AMLC. There exists, however, a clear correlation between the thermal stability and the crystal size in the direction of the helical axis (SAXS crystal dimension) as well as in the helix stacking direction (D120). In fact, one may wonder why the helix packing in type II AMLC is so poor. In their modelling work, Godet et al. demonstrated that the polar acid groups in amylose fatty acid complexes are preferentially excluded from the amylose helices.48 Nevertheless, they suggested that the incorporation of multiple fatty acids in the amylose helix cavity might be possible, but that this would result in distorted amylose chains. The expanded unit cells for type IIa and type IIb AMLC compared to for type I AMLC in the present case, indeed reflect disorder for systems in which some of the polar heads are supposedly included (see e.g. the Figures 6 and 7). The arrested crystallization of type II AMLC at 110 °C is triggered again during the ensuing cooling segment given the crystallinity increase seen there (full squares in Figure 10). The crystal unit cell parameter decreases during cooling and increases again during second heating. As this behavior runs parallel to the temperature profile it most likely reflects thermal contraction and expansion. The high noise in the unit cell parameter in the first heating segment prevents 33 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 34 of 46

deciding whether such a unit cell expansion is also relevant for type I AMLC. The D120 values (open squares, Figure 10) of the type II AMLC increase a little - if at all - during cooling although the crystallinity index (full squares, Figure 10) increases strongly. This behavior can be rationalized by the simultaneous creation of new small crystallites and the growth of the existing larger ones by which on the average hardly any change in D120 is observed. Most interestingly, the D120 values decrease during the second heating; at first rather moderately, but at high temperatures quite strongly. This observation is at odds with the idea that larger crystals should be more stable than smaller ones. With the smaller ones melting at the lowest temperatures, the largest ones should remain at high temperatures, which would lead to an increase, rather than a decrease in D120. From the SAXS and DSC experiments it was deduced that besides type IIa also very stable type IIb AMLC exist. It is suggested that during heating part of the type Ia AMLC melt and recrystallize into laterally less extended type IIb AMLC. This observation is taken as evidence for the fact that the helical axis dimension is prevailing over the crystal lateral dimension (D120) for what concerns its impact on the crystal thermal stability. Note that D120 is also only approximately 600 Å in the isothermal segment at 120 °C of temperature profile (ii) for the 25% dm sample (data not shown). In this 120°C segment only type IIb AMLC are present. 3.6. Relation between thermal stability and morphology From the discussion in the previous sections it follows that for the thermal stability of the different complexes (I < IIa < IIb) only two features are systematically relevant: i.e. the thickness of the complex layers on the one hand and the dm% and its morphological implications (larger water layer and increased level of lipid voids) on the other. The lateral size of the type II AMLC is larger than that of type I crystals, but seems to be of secondary importance for the stability in 34 ACS Paragon Plus Environment

Page 35 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


general since more stable type IIb AMLC are laterally less extended than less stable type IIa AMLC. The inferior role of the helix packing density stems from the observation that - although higher packing densities are often associated with a higher stability - the packing of the least stable type I AMLC is the highest. Differences in the crystal lateral size and helix packing density, therefore, at most contribute to the melting point distribution around the main transition, which is governed by the complex thickness and dm%. These observations are in line with Godet and coworkers49 who stated that the energy for helix unpacking is negligible compared to that required for helix dissociation and uncoiling. The DSC peak temperatures were considered for further analysis rather than the onsets for reasons that will become clear further below. In the Supporting Information, a three parameter equation is derived (using concepts suggested by Heck et al.50), to describe how the melting point, Tm, of a lamellar crystal relates to its thickness, lC, and dm%:

Tm =

 P3  1 −  P1 − P2 ( dm% )  lC  1

Eq. (8)

The parameters P1, P2 and P3 are functions of actual physical quantities:

P1 =

1 R + ∞ T vS ∆H

Eq. (9)

P2 =

R 100vS ∆H

Eq. (10)

P3 =

2σ U ∆H

Eq. (11)

T∞ is the theoretical melting point of complexes in absence of water and for infinite lC. The parameter vs is the molar volume of water, R is the gas constant, ∆H is the melting enthalpy per



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 36 of 46

unit of volume and σU is the surface energy of the top and bottom surfaces of the lamellae. The latter refer at the surfaces where the helix ends are facing the environment.

Figure 11: Graph relating the crystal thickness, lC, of amylose-GMS complexes to their melting points. Closed symbols refer at the 33% dm data, half open symbols to the 25% dm data and open symbols to the paste (21 dm) data. The straight lines through the experimental data are obtained by fitting the data to Eq. (8).

Eq. (8) implies that plotting the melting points as a function of 1/lC should reveal linear dependences at a given dm%. Figure 11 shows the melting points (in K rather than °C) of the type I, type IIa and type IIb AMLC highlighted in bold in Figure 5 as a function of 1/lC, with lC calculated from the product of lMAG (calculated using Eq. (5)) and the assumed number, N, of GMS molecules, i.e. two, four and six lipids in the type I (melting at the peak of pI), type IIa (melting at pIIa) and type IIb (melting at pIIb) complexes, respectively. Note that in this graph 1/lC is expressed in m-1 rather than in Å-1. The selected peak temperatures (bold in Figure 5) are judged to be the least influenced by the processing method or recrystallization phenomena or are best separated from nearby melting peaks. The straight lines in Figure 11 are obtained via a single least squares optimization procedure in which the parameters P1-P3 of Eq. (8) are varied 36 ACS Paragon Plus Environment

Page 37 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


without constraints to cover all data points at ones. Next, the straight lines are constructed via Eq. (8) with lC varying in a continuous fashion. The parameters resulting from the fitting procedure are listed in Table 1.

Table 1: Parameters (P1-P3) obtained by fitting Eq. (8) to the data represented in Figure 11. The parameters correspond to a fitting in which the data are expressed in K (melting point) and m (crystal thickness). parameter

value

P1

2.513 10-3

P2

3.435 10-6

P3

5.557 10-10

Table 2: Experimental and predicted [according to Eq. (8) and Eq. (5) and P1, P2 and P3 as in Table 1] melting points of 25% dm amylose-monoacyl glycerol type I complexes for different monoacyl glycerol lipid chain lengths, M [in Eq. (5)], with M = 10 for monocaprin, M = 12 for monolaurin, M = 14 for monomyristin, M = 16 for monopalmitin and M = 18 for monostearin. Lipid chain length M 10

Tm (°C) experimental Putseys et al.36

Tm (°C) experimental Tufvesson et al.50

-

75.6

79.4

12

94.3

82.2

86.4

14

95.6

88.6

91.2

16

99.0

95.5

96.3

18

100.6

100.1

100.0

Tm (°C) Predicted

To validate the combined use of Eq. (8) and Eq. (5), the melting point prediction of type I AMLC of shorter saturated monoacyl glycerol species at 25% dm is compared to experimental values as reported by Putseys et al.36 and Tufvesson et al.51 in Table 2. The assumption is upheld that N = 2 for type I AMLC. The predicted values occur in between the experimental values with the values by Putseys et al.36 being systematically higher and the ones by Tufvesson et al.51 being 37 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 38 of 46

systematically lower than the predictions. Lack of control over the dm% may well be at the origin of the discrepancies. All in all, the agreement is fair. Finally, it is instructive to comment on the physical meaning of the parameters P1-P3. The term before the brackets in Eq. (8) represents the melting point of infinitely thick crystals and reflects the dm% dependence. These morphology-independent melting points can be read from the intercepts of the straight lines in Figure 11 and decrease with decreasing dm%. This seems natural and is realized in Eq. (8) with the neglect of interactions between the solvent and the crystallizable species (see also Supporting Information). The dm% dependent melting point depression in Eq. (8) is thus governed by the entropy of mixing upon melting. Including interactions – which would make the denominator quadratic in dm% - was attempted but did not yield a better description of the data. Whether or not this simplified approach is justified, requires the analysis of data covering a wider dm% range. Based on Eq (9) and (10) a value for T∞ can be derived:

T∞ =

1 P1 − 100 P2

Eq. (12)

With P1 and P2 from Table 1 this yields T∞ = 461 K or 188°C. It is furthermore critical to realize that dm% in Eq. (8) actually refers at the dm% present in the solution at crystal melting. At the melting peak temperature, a fraction of the material is still in the crystalline state, by which the dm% in the solution is lower than the system overall dm%. In fact, for diluted systems, one can use the overall dm% only in relation with the melting end point (conclusion temperature).52 Furthermore, these end points should be corrected by extrapolation to a zero scanning rate.52 Using melting onset temperatures for sure is inappropriate. Given that end points are rather ill defined and that the melting peaks in diluted systems are approaching the


Page 39 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


end point52, it was decided to use melting peak temperatures as a best compromise. Moreover, the end points of type IIa melting are obscured by the melting of type IIb crystals. Using a too high value for dm% in the analysis will be compensated by a too low value for P2. With Eq. (10), this translates into a too high value for ∆H. With R = 8.31 Jmol-1 and vs ~1.8 10-5 m3mol-1 (for water) a value for ∆H of 1340 Jcm-3 can be computed, which is about 50 times larger than the value of about 25 Jg-1 based on DSC. Knowing that the derivation of Eq. (8) relies on a number of simplifications (see also Supporting Information) and having neglected the error involved in comparing mass based with volume based enthalpy values, this ∆H discrepancy is merely indicative. Moreover, the reported 25 Jg-1 accounts for the dm% fraction which is actually crystalline, whereas the melting enthalpy of a fully crystalline material is required in such a comparison. It is clear though that the gradual change of the solution composition upon melting contributes to the width of the melting transition. This effect may be more important compared to distributions in e.g. the crystal lateral size. The fact that part of the water may reside outside the lamellar stacks contributes to the complexity and calls for further research. The decreasing %dm in the solution that remains during isothermal crystallization at e.g. 110°C (see e.g. the discussion related to Figure 10) may explain why the crystallization eventually stops at a given level. Indeed, at a sufficiently low %dm (high water content) in the remaining solution the degree of supercooling (i.e. the difference between the depressed melting point and the isothermal crystallization temperature) may have decreased to such a level that crystallization from this water enriched solution is no longer possible, unless the system is cooled to lower temperatures.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 40 of 46

4. Conclusions Three different amylose-lipid complex structures with increasing thermal stability can be distinguished for the amylose-glycerol monostearate (GMS) samples discussed in this work: type I, type IIa and type IIb complexes. The increasing stability can be inferred from their increasing melting point. The complexes all have in common that they consist of aggregated amylose helices that are loaded with GMS molecules and arranged according to what is known as VH-type crystals. Type I complexes are thus not amorphous but clearly ordered, although their WAXD crystalline reflections are much broader than those of the type II complexes. Earlier claims on type I complexes being amorphous might be related to a lack of WAXD data quality or to type I AMLC being formed during heating in a DSC scan, starting from material that was truly WAXD amorphous prior to heating. For the type I complexes studied, one has to accept U-folding of the amylose chains and a complex layer thickness governed by the presence of two stretched lipid molecules per amylose strand. In the conversion of type I to type IIa complexes, the amylose chains unfold and assume a stretched conformation, which doubles the layer thickness and increases the number of aligned lipid molecules within the amylose helices to four. In type IIb complexes another pair of lipid molecules is added, increasing the complex layer thickness to a value six times the length of stretched lipid molecules. The stability of these complexes – probed by their melting point – is proportional to the thickness of the complex layers and depends on the amount of water in the system. The relation between the crystal layer thickness, water content and melting point can be rationalized in terms of a rather simple equation, based on thermodynamics. This equation seems to also predict the melting points of amylose monoacyl glycerol complexes based on fatty acids shorter than stearic acid. 40 ACS Paragon Plus Environment

Page 41 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In the past, it has been suggested that the higher melting point of type II complexes compared to type I AMLC was caused by the higher entropy, S, of type I complexes.23 In the reasoning behind this claim, clearly the entropy of a crystal (S) was confused with the melting entropy, ∆S, which is the difference in entropy between the crystalline and molten state. This bias on entropy was based on the observation that often the DSC based ∆H of type I and type II complexes are similar23,52,53 However, relying on an identical ∆H and the relation Tm= ∆H/∆S one would have to attribute a higher entropy difference (∆S) and hence a lower S for type I AMLC compared to type II AMLC because both type I and type II AMLC melt into an equal liquid state with identical liquid S.

Be that as it may, such considerations assume equilibrium conditions and neglect crystal size effects, which very clearly are the true origin of the differences in thermal stability between type I and type II AMLC.

5. Acknowledgements This publication is financially supported by the European Commission in the Communities 6th Framework Program, Project HEALTHGRAIN (FOOD-CT-2005-514008). It reflects the author’s views and the Community is not liable for any use that may be made of the information contained in this publication. This research was also conducted in the framework of research project G.0427.07, financed by the Fund for Scientific Research – Flanders (FWO Vlaanderen), Brussels, Belgium and is part of the Methusalem program “Food for the Future” at the K.U.Leuven. C.J.G is a researcher of the F.R.S.-FNRS, Belgium. The authors thank FWOVlaanderen for supporting the DUBBLE project.

6. Supporting Information Available 41 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 42 of 46

A table (Table SI-1) with the melting peak temperatures and enthalpies related to the graphs shown in Figure 5. A derivation of the equations 8-12 under the title: The melting of lamellar crystallites in the presence of a diluent. This information is available free of charge via the Internet at http://pubs.acs.org/.

7. References (1)

Buléon, A.; Colonna, P.; Planchot, V.; Ball, S. Int. J. Biol. Macromol. 1998, 23, 85-112.

(2)

Putseys, J.A.; Gommes, C.J.; Van Puyvelde, P.; Delcour, J.A.; Goderis B. Carbohydr.

Polym. 2011, 84 1141–1150. (3)

Rappenecker, G.; Zugenmaier, P. Carbohydr. Res. 1981, 89, 11-19.

(4)

Zobel, H. F. Starch-Stärke. 1988, 40, 44-50.

(5)

Putseys, J.A.; Lamberts, L.; Delcour, J.A. J. Cereal Sci. 2010, 51, 238-247.

(6)

Immel, S.; Lichtenthaler, F. W. Starch-Stärke. 2000, 52, 1-8.

(7)

Conde-Petit, B.; Escher, F.; Nuessli, J. Trends Food Sci. Technol. 2006, 17, 227-235.

(8)

Heinemann, C.; Escher, F.; Conde-Petit, B. Carbohydr. Polym. 2003, 51, 159-168.

(9)

Heinemann, C.; Zinsli, M.; Renggli, A.; Escher, F.; Conde-Petit, B. Food Sci. Technol.

2005, 38, 885-894. (10)

Rundle, R. E.; French, A. D. J. Am. Chem. Soc. 1943, 65, 558-561.

(11)

Valletta, R. M.; Germino, F. J.; Lang, R. E.; Moshy, R. J. J. Polym. Sci. Part A. 1964, 2,

1085-1094. (12)

Morrison, W. R.; Law, R. V.; Snape, C. E. J. Cereal Sci. 1993, 18, 107-109.

(13)

Evans, I. D. Starch-Stärke. 1986, 38, 227-235.

(14)

Krog, N.; Jensen, B. N. J. Food Technol. 1970, 5, 77-87.

(15)

Stampfli, L.; Nersten, B. Food Chem. 1995, 52, 353-360. 42 ACS Paragon Plus Environment

Page 43 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(16)

Derycke, V.; Vandeputte, G. E.; Vermeylen, R.; De Man, W.; Goderis, B.; Koch, M. H.

J.; Delcour, J. A. J. Cereal Sci. 2005, 42, 334-343. (17) Gelders, G. G.; Vanderstukken, T. C.; Goesaert, H.; Delcour, J. A. Carbohydr. Polym. 2004, 56, 447-458. (18)

Karkalas, J.; Ma, S.; Morrison, W. R.; Pethrick, R. A. Carbohydr. Res. 1995, 268, 233-

247. (19)

Biliaderis, C. G.; Seneviratne, H. D. Carbohydr. Polym. 1990, 13, 185-206.

(20)

Galloway, G. I.; Biliaderis, C. G.; Stanley, D. W. J. Food Sci. 1989, 54, 950-957.

(21)

Biliaderis, C. G.; Page, C. M.; Slade, L.; Sirett, R. R. Carbohydr. Polym. 1985, 5, 367-

389. (22)

Biliaderis, C. G.; Page, C. M.; Maurice, T. J. Carbohydr. Polym. 1986, 6, 269-288.

(23)

Biliaderis, C. G.; Galloway, G. Carbohydr. Res. 1989, 189, 31-48.

(24)

Biliaderis, C. G.; Page, C. M.; Maurice, T. J. Food Chem. 1986, 22, 279-295.

(25)

Brisson, J.; Chanzy, H.; Winter W.T. Int. J. Biol. Macromol. 1991, 13 31-39.

(26)

Yamashita, Y.; Monobe, K. J. Polym. Sci Part A. 1971, 9, 1471-1481.

(27)

Manley, R. S. J. J. Polym. Sci. Part A. 1964, 2, 4503-4515.

(28)

Yamashita, Y. J. Polym. Sci. Part A. 1965, 3, 3251-3260.

(29)

Buléon, A.; Duprat, F.; Booy, F. P.; Chanzy, H. Carbohydr. Polym. 1984, 4, 161-173.

(30)

Schoch, J. J. Am. Chem. Soc. 1942, 64, 2957-2961.

(31)

Whittam, M. A.; Orford, P. D.; Ring, S. G.; Clark, S. A.; Parker, M. L.; Cairns, P.; Miles,

M. J. Int. J. Biol. Macromol. 1989, 11, 339-344. (32)

Zobel, H. F.; French, A. D.; Hinkle, M. E. Biopolymers 1967, 5, 837-845.

(33)

Jane, J. L.; Robyt, J. F. Carbohydr. Res. 1984, 132, 105-118. 43 ACS Paragon Plus Environment


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 44 of 46

(34)

Pfannemuller, B.; Burchard, W. Makromolekulare Chemie. 1969, 121, 1-17.

(35)

Gelders, G. G.; Goesaert, H.; Delcour, J. A. Biomacromolecules. 2005, 6, 2622-2629.

(36)

Putseys, J. A.; Derde, L. J.; Lamberts, L.; Goesaert, H.; Delcour, J. A. Carbohydr. Polym.

2009, 78, 854-861. (37)

Zabar, S.; Lesmes, U.; Katz, I.; Shimoni, E.; Bianco-Peled, H. Food Hydrocolloid, 2009,

23, 1918–1925 (38)

Huang, T. C.; Toraya, H.; Blanton, T. N.; Wu, Y. J. Appl. Crystallogr. 1993, 26, 180-

184. (39)

Gommes, C. J.; Goderis, B. J. Appl. Crystallogr. 2010, 43, 352-355.

(40)

Hukins, D. W. X-ray Diffraction by Disordered and Ordered Systems. ed.; Pergamon

Press: Oxford, NY, USA, 1981. (41)

Goossens ,K.; Nockemann, P.; Driesen, K.; Goderis, B.; Görller-Walrand, C.; Van

Hecke, K.; Van Meervelt, L.; Pouzet, E.; Binnemans, K.; Cardinaels, T. Chem. Mater., 2008, 20, 157–168 (42)

Vignaud, R.; Schultz, J. M. Polymer 1986, 27, 651-658.

(43)

Ruland, W. J. Appl. Crystallogr. 1971, 4, 70-73.

(44)

Hinkle, M. E.; Zobel, H. F. Biopolymers 1968, 6, 1119-1128.

(45)

Takeo, K.; Kuge, T. Agric. Biol. Chem. 1971, 35, 537-542.

(46)

Karkalas, J.; Raphaelides, S. N. Carbohydr. Res. 1986, 157, 215-234.

(47)

Crist, B. J. Macromol. Sci., Part B, 2000, 39, 493−518.

(48)

Godet, M.C.; Buléon, V.; Colonna, P. Carbohydr. Polym. 1993, 21, 91-95

(49)

Godet, M. C.; Bizot, H.; Buléon, A. Carbohydr. Polym. 1995, 27, 47-52.

(50)

Heck, B.; Strobl, G.; Grasruck, M. Eur. Phys. J. 2003, E11, 117-130 44 ACS Paragon Plus Environment

Page 45 of 46

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


(51)

Tufvesson, F.; Wahlgren, M.; Eliasson, A.C. Starch-Stärke 2003, 55, 61-71.

(52)

Berghmans, H. In Calorimetry and Thermal Analysis of Polymers, Mathot, V.B.F., Ed.;

Hanser Publishers, New York, 1994; Chapter 8, pp 207-230. (52)

Kowblansky, M. Macromolecules 1985, 18, 1776-1779.

(53)

Putseys, J. A.; Derde, L. J.; Lamberts, L.; Östman, E.; Björck, I. M.; Delcour, J. A. J.

Agric. Food Chem. 2010, 58, 1939-1945.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 46 of 46

FOR TABLE OF CONTENTS USE ONLY

TABLE OF CONTENT GRAPHIC

SYNOPSIS Three different crystalline amylose-glycerol monostearate (GMS) complexes with increasing thermal stability can be distinguished: type I, type IIa and type IIb. They all consist of GMSloaded amylose helices that pack hexagonally into lamellar habits. The complex melting points are proportional to the thickness of the lamellae and depend on the amount of water in the system. The relation found between the crystal morphology and melting point for amylose-GMS complexes also predicts the melting points of other amylose-monoacyl glycerol complexes.


The Structure and Thermal Stability of Amylose ... - ACS Publications

Recommend Documents