Extrusion Behaviour of Cohesive Potato Starch Pastes. II. Microstructure–Process Interactions

Journal of Food Engineering 66 (2005) 13–24

www.elsevier.com/locate/jfoodeng

Extrusion behaviour of cohesive potato starch pastes:II. Microstructure–process interactions

A. Cheyne a, J. Barnes b, S. Gedney b, D.I. Wilson a,*

a Department of Chemical Engineering, University of Cambridge, Pembroke Street, Cambridge CB2 3RA, UKb United Biscuits Research and Development, Lane End Road, High Wycombe, UK

Received 12 September 2003; accepted 21 February 2004

Abstract

The effect of shear imposed during isothermal ram extrusion of a mixture of potato starches on extrudate microstructure was

investigated by comparing the microstructure generated in the mixture with the microstructures generated by extruding its com-

ponents, namely native starch, potato granule and potato flake, individually. Hydration tests indicated that the fractions with pre-

gelatinised starch (granule and flake) absorbed free water rapidly, so that the native starch in the 40 wt.% w.b. water mixture was

extruded in an almost unhydrated, hard form. The three components exhibited markedly different extrusion behaviours, ranging

from visco-elastic effects to chronic dewatering and particle rupture. Electronic and optical microscopy, DSC and WAXS were used

to characterise microstructure changes across the length scales: the results indicated that the flake acted as an amorphous binder

which allowed the other components to retain their structure during the shear regime generated during extrusion.

� 2004 Elsevier Ltd. All rights reserved.

Keywords: Starch; Potato; Extrusion; Microstructure; Hydration; Paste

1. Introduction

The food industry often employs extrusion techniques

to generate solid or semi-solid products with specific

shapes or textures that would not be achievable via

other routes. The literature contains many studies of

extrusion of non-food soft solid materials (e.g. Benbow

and Bridgwater, 1993), a grouping which contains pastes

and concentrated suspensions. The microstructures of

the products are determined by the response of suchmaterials to local shear, temperature, and temperature

rise induced during extrusion. These interactions are

well known to be critical parameters in determining the

properties of many processed foodstuffs (Noel et al.,

1990). Development of new products and optimisation

of existing processes requires understanding the inter-

actions between ingredients, process parameters and

equipment design. The physical and chemical complex-ity of many foodstuffs, however, means that such rela-

tionships can rarely be predicted in advance. Key

mechanical parameters required for equipment design

*Corresponding author. Tel.: +44-1223-334-791; fax: +44-1223-331-

796.

E-mail address: [email protected] (D.I. Wilson).

0260-8774/$ - see front matter � 2004 Elsevier Ltd. All rights reserved.doi:10.1016/j.jfoodeng.2004.02.036

and operation principally involve the rheological re-

sponse of the material to deformation via extensionaland simple shear, while key product quality parameters

are related to the microstructure of the final form.

Linking these two types of data is not straightforward,

both for reasons of difficulty in quantifying the rheo-

logical response and in measuring appropriate micro-

structural features.

Starchy foods represent the major source of carbo-

hydrate in the human diet, estimated to comprise 80% ofthe global average calorie intake (FAO, 1999), and are

available in various forms featuring different extents of

pre-processing: (i) raw (native), e.g. potato or rice, fea-

turing only post-harvest treatment and storage; (ii)

as traditional foods, such as bread or pasta; (iii) as

modern consumer foods, e.g. snack foods, featuring

significant processing in order to achieve given textures,

shapes and tastes. The second and third categoriespresent considerable challenges for manufacturers, as

the rheology, chemistry and textural characteristics are

often intimately related to the extent and nature of

processing operations performed on the material.

Extrusion of starch pastes makes possible fine control of

product properties to be managed in large-scale pro-

cesses: starch-based food products therefore represent a

mail to: [email protected]

Nomenclature

Tonset onset set temperature of gelatinisation

C apparent shear rate

DH enthalpy change

v relative crystallinity

sw wall shear stress

sy yielding shear stress (of cells in cooked potato

flour)

14 A. Cheyne et al. / Journal of Food Engineering 66 (2005) 13–24

particularly important group of materials processed by

extrusion.

A significant body of work exists in the literature

concerning process-induced structural changes of both

food and non-food starches (Frazier et al., 1996). There

is little literature available, however, on pastes featuring

mixtures of native and pre-processed starches at inter-

mediate water contents, despite their common applica-tion in the food industry. This work is concerned with

the extrusion behaviour of a mixture of starchy potato

solids typical of those used to make snack food prod-

ucts. The key processing stage is ram extrusion at

ambient temperatures, such that extrusion cooking does

not occur. Product properties directly determined by the

microstructure were known to include the structural

stability and cooking behaviour of the ‘green’ extrudate,and the appearance, taste and ‘mouth feel’ of the final

snack.

This paper reports on studies of microstructure

development during extrusion: an associated paper

(Cheyne, Barnes, & Wilson, 2004) reports mechanical

characterisation and extrusion modelling. Further as-

pects of the extrusion of potato starch pastes, including

an investigation into the assessment of raw ingredientqualities, have also been reported elsewhere (Cheyne,

2000; Cheyne, Wilson, Barnes, & Sala€un, 2001).

2. Starch

2.1. Structure

Starch is produced as granules in almost all plant cells

as an inert store of energy. Starch in this granular state in

which it naturally develops, is referred to as ‘native’.Though native starches from different botanical sources

vary widely in structure and composition (Galliard,

1987), essentially all granules consist of shells of radi-

ally orientated polysaccharide chains centred on the

hilum. The layers, called growth rings, have alternating

crystalline and amorphous structure, typically about

100–500 nm thick. They consist of one or both of two

polysaccharide chains (with small amounts of lipids,proteins and trace amounts of organic and inorganic

substances), both composed of assemblies of glucose

molecules, but differing in bonding and structure.

Amylopectin, the major component of most starches,

is composed of branched chains of glucose. The mole-

cules are very large, with molecular weights varying

from 107 to 109 gmol�1, corresponding to about a mil-

lion glucose units. It exists only in the crystalline layers,

which are subdivided into semi-crystalline and semi-

amorphous lamellae (with a repeat distance of 9 nm

common to almost all starches). The former correspondto tightly packed arrangement of amylopectin double

helices, the structure that produces the characteristic

X-ray patterns in native starch. The latter correspond to

unaligned sections of amylopectin, forming molecular

bridges connecting the crystalline arrangement of heli-

ces.

Amylose consists of straight chains of a-glucose unitsand is smaller than amylopectin, of the order of 105 to106 gmol�1, which corresponds to a few hundred glu-

cose molecules. It exists in an amorphous state

throughout starch granules. The high degree of struc-

tural flexibility of this molecular configuration allows

amylose molecules to complex with other molecules,

notably lipids.

Potato starch granules are ellipsoidal, with a typical

dimension of between 10 and 100 lm. They containapproximately 23 wt.% amylose and have amylopectin

with B-type crystallinity (Imberty et al., 1991). The

granules are naturally low in lipids, but high in phos-

phates.

2.2. Processing

Cooking processes are essential to allow starch to be

metabolised by humans. They involve the action of heat,

moisture and often mechanical action, and are a subset

of the group of actions termed starch conversion

(Mitchell et al., 1997). The features of starch describedabove result in structural order at scales of scrutiny

between 10�9 and 10�4 m. Conversion involves the

destruction of this structure, starting from the highly

ordered native granules, finishing with depolymerisation

of individual amylose or amylopectin molecules and

molecular degradation of glucose monomers.

Conversion can take place through the action of heat

(Senouci and Smith, 1986), moisture (Svegmark andHermansson, 1991) or mechanical action (Yamada

et al., 1997) alone, however it is more common to apply

Table 1

Methods of investigating and measuring starch gelatinisation. G indicates a granule scale effect, M, a microstructural effect and C, a chemical effect

Method Gelatinisation identified by

G Microscopy (light and SEM) Granule swelling, folding and rupturing as water is absorbed (Hermansson & Svegmark, 1996)

M Measurements of birefringence Loss of birefringence as spherulitic organisation is disrupted (van Soest, 1996)

M X-ray diffraction Loss of diffraction pattern as crystallites melt (Le Bail et al., 1997)

C DSC A characteristic endothermic event (Stapley, 1995)

C NMR Loss of water mobility (short T2 times) as gel forms (Stapley, 1995)G Viscosity measurements A drop in suspension viscosity caused by granule rupture (Haase, Mintus, & Weipert, 1995)

C Iodine binding An increase in iodine binding capacity as amylose is released (Kudla & Tomasik, 1992)

C Enzymatic attack Increased enzymatic digestibility as amylose is released (Yeh et al., 1999)

A. Cheyne et al. / Journal of Food Engineering 66 (2005) 13–24 15

two or more of these together. For example, corn starch

snack foods are often extrusion-cooked as low moisture

pastes undergoing extensive shearing in screw extruders

at temperatures in excess of 150 �C (Yeh et al., 1999).This affects starch structure at all scales down to the

molecular level.One of the most commonly studied facets of con-

version is gelatinisation (Hermansson and Svegmark,

1996). This transformation involves the absorption of

water into granules at elevated temperature (approxi-

mately 60 �C for potato starch), followed by granuleswelling and eventual disintegration to form a homo-

geneous gel. This process, which destroys granule and

growth ring-scale structure (10�4 to 10�7 m), is com-monly exploited to thicken soups and sauces can be

identified and quantified using various methods, several

of which have been applied in this work (see Table 1).

A key question posed in extrusion processing is at

what length scales does the process affect the paste

microstructure. The model paste used here consists of a

mixture of native starch and pre-processed potato solids

at an intermediate water content (40 wt.%), which mixesto form a weakly cohesive dough. Such a dough con-

tains starch that has experienced a range of conversions

prior to forming. Each component has been tested sep-

arately here, so that the response of individual compo-

nents to processing could be used to interpret the

response of the blend. Industrial application of such

doughs involves a compaction stage (where the dough

charge is de-aerated) prior to extrusion through a multi-holed die plate of characteristic dimension 1 mm. The

extrudate is cut as it leaves the die plate, cooked, fin-

ished and packaged.

3. Experimental

3.1. Materials

The three materials examined in this work were

provided by United Biscuits Ltd., High Wycombe, UK,and are common ingredients of industrial starch-based

foods. They were refined native potato starch, potato

granule and potato flake. The latter materials are pre-

pared from steam-cooked and milled potato, and

therefore contain completely gelatinised starch, together

with small quantities of salt and mono-glyceride emul-

sifier, and potato cell components (predominantly cel-

lulose, but with trace elements of lipids and proteins;

Talburt and Smith, 1980).Fig. 1 shows SEM images of samples of each dry

powder. Despite the material similarity between potato

flake and granule, essential differences in form are evi-

dent in these images. During preparation, granule is

milled so as to preserve the cell wall integrity of indi-

vidual cells, which may be observed in the powder

sample. In contrast, flake is milled more severely so that

a large proportion of the cells are ruptured. The parti-cles evident in the image are therefore agglomerates of a

mixture of damaged and whole cells bound into a matrix

of the released, sticky gelatinised starch.

Native starch is the cheapest ingredient and is com-

monly used as a filler. The cooked potato solids, which

are more expensive, give taste and colour to the product,

but are also instrumental in determining handling and

processing properties of the paste. The level of extra-cellular gelatinised starch is crucial to determining paste

handling properties (e.g. cohesivity or compaction

behaviour) and the texture, appearance and stability of

green extrudates (i.e. the response of the paste to pro-

cessing). Balancing the three components in the paste

formulation is therefore extremely important. Table 2

summarises some material data for the paste ingredients

and Table 3 the formulations used in this work.

3.2. Mixing

The extrusion behaviour of individual components

and the mixed formulation were studied. The pastes

were prepared by 4 min of dry mixing and 4 wet (with

water at 40 wt.% w.b.) using a Kenwood planetary

mixer (Kenwood Ltd., UK) fitted with a K-beater,

operating at 40 rpm. The wet mixing time was deter-

mined in separate experiments by mixing for specific

intervals and then measuring the moisture content ofseveral samples taken from different locations through-

out the paste mass. The coefficient of variation did not

change appreciably after 4 min. These water contents,

Fig. 1. SEM images of paste components: (a) native starch, 1500�;(b) potato granule, 500�; (c) potato flake, 100�.

Table 3

Formulation of pastes used in this work

Paste Form Ingredient Mass frac-

tion (wt.%)

Blend Cohesive powder Native potato

starch

28.3

Potato granule 23.9

Potato flake 6.6

Salt 1.2

Reverse osmosis

water

40

Granule paste Cohesive powder Potato granule 60

Reverse osmosis

water

40

Flake paste Dough Potato flake 60

Reverse osmosis

water

40

Native starch

paste

Dense slurry Native potato

starch

60

Reverse osmosis

water

40

Table 2

Properties of paste ingredients used in this work

Ingredient Price ($/tonne) Modal particle size

range (lm)

Native potato starch 430 40–57

Potato granule 910 75–150

Potato flake 1010 250–425

Source: United Biscuits, High Wycombe, UK.


mixer and mixing conditions are also typical of values

used industrially.

3.3. Paste hydration

Starch, like many foodstuffs, becomes glassy at low

water contents. The level of hydration is therefore cru-

cial in determining its processing response. The cooked

potato materials appeared to absorb all of the added

water, whereas that in the native starch mixture ap-

peared to remain free of the solid. It was therefore ex-

pected that the moisture distribution in the blend would

be uneven and determined by competition between sol-

ids for free water. Hydration tests were performed in

order to gauge the relative rates of absorption of pow-ders: pre-weighed 10 g samples were contacted with

excess water for prescribed periods, filtered to remove

free liquid and weighed.

The results, summarised in Fig. 2, indicate two fac-

tors: (i) the initial rate of water uptake (important for

competitive adsorption of water in mixed pastes) and (ii)

the maximum quantity of water absorbed (which

determines the water content at which each paste be-comes saturated). Native starch absorbed water signifi-

cantly more slowly than the gelatinised starches. The

greater absorption rate and capacity of flake compared

to granule was expected because the unconfined gela-

tinised starch fraction in flake was able to swell exten-

sively.

These results indicated that flake and granule pastes

with water added at 40 wt.% would not be fully satu-rated (cf. 53 wt.% for granule: Pruvost, Corfield, King-

man, & Lawrence, 1998), whereas almost all the water

0

1

2

3

4

5

6

7

Wat

er A

bsor

bed/

Sam

ple

Mas

s (g

/g)

GranuleNative Starch Flake Blend

Fig. 2. Comparison of water uptake for the blend and individual

ingredients after: light grey, 0.5 min; white 2 min; dark grey, 4 min.

Error bars indicate the standard error.


would remain free in native starch pastes. Using the data

in Fig. 2 it was possible to predict water uptake in the

blend by summing contributions from each paste ingre-dient. These predictions matched the results within the

standard experimental variation (deviations of 1.2% after

half a minute, 8.3% after 2 min and 5.3% after four min),

however in all cases were lower than the experimentally

determined value. This may be explained by considering

the limitations of the filtration process to remove unab-

sorbed water. As for flake and granule, the blend would

also be unsaturated with 40 wt.% added water.Differences in adsorption behaviours of the three

components suggested that partitioning took place be-

tween them. The blend was therefore likely to consist of

hydrated, gelatinised starch and potato cell material

mixed with unhydrated, glassy native starch. This has

important consequences for the processing behaviour of

the blend, as the mechanical behaviour of the native

starch is strongly determined by its hydration state. Themicro-manipulation studies reported in the accompany-

ing paper (Cheyne et al., 2004) indicated that the hy-

drated starch exhibited rupture at compaction bursting

pressures of o(100 kPa), while the unhydrated form did

not rupture under the conditions employed in these tests.

The native starch will therefore be present in a mechan-

ically stable state and resistant to local deformation.

3.4. Extrusion

Extrudate samples were produced by ram extrusion

through a square entry, axi-symmetric capillary die

(length 48 mm, i.d. 3 mm) using an instrumented strain

frame configured for extrusion (SA100 Loading Frame

Twin Screw Machine, Dartec UK). The piston motionwas controlled at 10 mm s�1 over a stroke length of 100

mm through a 25 mm i.d. barrel. The extrudate plug flow

velocity was calculated to be 0.69 m s�1, corresponding

to an apparent shear rate, C ¼ 1850 s�1. These condi-tions were chosen to match those in typical industrial

processes. Mean extrusion pressures were logged.

All experiments were performed under ambient con-

ditions. As the hydrated materials changed properties

noticeably on storage, batches were discarded 6 h after

preparation.

3.5. Microstructural investigations

The following techniques were selected to assess dis-

ruption of starch structure at the scales of granule and

growth ring structure and crystallite organisation.(a) Light microscopy: Bright-field and cross-polarised

(XP) microscopy were performed using a Labophot

2 optical microscope (Nikon, UK). Differential Inter-

ference Contrast (DIC) images were produced using an

Optiphot (DIC) microscope (Nikon, UK). Resin-

embedded samples were fixed in 2.5% glutaraldehyde

solution and mounted using LR White Resin (London

Resin Company, UK). Three-micrometer sections werecut using a Reichert-Jung 2045 Multicut Microtome

(Leica, UK). Cryosections were produced by mounting

samples in Tissue Tec OCT Compound cryoprotectant

(Sakura Fine Chemical, Tokyo, Japan) and freezing in

liquid nitrogen for 2 min before 7 lm sections were cutin a cryostat (Bright Instruments, UK). Cryosections

were mounted in 3-in-1e mounting oil, or stained for

2 min with iodine vapour (IV, for starch), while resinsections were stained with toluidine blue (TB, for cell

wall material). Both normal and polarised (XP) light

sources were used to observe the sections. It should be

noted that sample freezing may cause distortions in

cryosections, and that soluble materials (such as pro-

cessed starch) may be severely affected by the resin

embedding process.

(b) Electron microscopy: Dried and gold sputter-coated samples were observed under vacuum conditions

using a JEOL JSM 820 Scanning Electron Microscope

(SEM, JEOL Ltd., Japan). To ensure that observed

features were not artefacts of sample preparation, fresh

samples were also studied using an ElectroScan Model

2010 Environmental SEM (ESEM, ElectroScan Cor-

poration, USA). Acceleration voltages and working

distances are noted individually in the figures.(c) DSC: Power compensated DSC experiments were

performed using a Perkin Elmer DSC-7 differential

scanning calorimeter equipped with an Intracooler II

(Perkin Elmer Corporation, USA). Temperature and

enthalpy parameters were calibrated using the melting

transition of indium (Tonset ¼ 156:60 �C, DH ¼ 28:45 J/g�19). The protocol was follows:

(i) Sample preparation: 10–15 mg of samples were

placed in a vapour-sealed LVC pan and hydrated

for 30 min with three times the mass of RO water.


(ii) DSC program: held at 20 �C for 1 min; heated to140 �C at 10 �C/min; held at 140 �C for 1 min;cooled to 20 �C at 200 �C/min; held at 20 �C for10 min; heated to 140 �C at 10 �C/min.

(d) X-ray diffraction: Wide angle X-ray scattering

(WAXS) experiments were performed a Bruker X-Ray

System (Bruker Analytical X-Ray Systems, USA). The

sample holder consisted of a 1 · 10 · 10mm slit comprisedof stainless steel with beryllium X-ray windows. The X-

ray beam power was 40 kV, 30 mA with k ¼ 1:54 �A.

Fig. 3. Extrudates from pastes at 40 wt.% initial water content, com-

prised of: (a) blend; (b) granule; (c) flake; (d) native starch (initial

solids); (e) native starch (extrudate with core); (f) native starch (final

phase). Scale in mm.

4. Results and discussion

4.1. Extrudate properties

Fig. 3 shows typical samples of extrudates produced

from each paste for further structural investigation. The

extrusion pressure required for each material is shown in

Table 4, as is an estimate of the wall shear stress, sw. Thelatter values were calculated by neglecting die entry

work and are therefore over-estimates: greater accuracy

would require a systematic rheological study, and it is

doubtful whether some of the materials (e.g. native

starch) could be successfully characterised. Extrudates

produced from the flake, granule and blend were of

consistent quality throughout an experiment. Flake ex-

trudates differed from the others by noticeable radialswelling and a rough extrudate surface. In contrast, the

native starch mix featured four stages in extrusion.

With native starch pastes, the initial movement of the

piston expelled water from the suspension. Such chronic

loss of liquid phase has been reported for a variety of

industrial extrusion systems dealing with pastes that

demonstrate poor liquid retention, and may be com-

pared to drainage of a saturated soil under an appliedstress (Rough et al., 2002; Terzaghi, 1943). A solid ex-

trudate was produced once sufficient water had been

removed to expose the solid matrix to the applied stress:

this quantity was found to be approximately 50% of the

initial value, so that the extruded native starch featured

a markedly lower water content than the initial slurry.

The first solid extrudate (Fig. 3(d)) was opaque, white

and irregular (and powdery when dry), however thisdeveloped into a more regular, tough cable of trans-

parent material containing an opaque, white core

(Fig. 3(e)). The core gradually shrank so that the final

extrudate was entirely transparent (Fig. 3(f)). The

extrusion force was relatively uniform for the stages in

which material shown in Fig. 3(e) and (f) were gener-

ated, and this value is used to estimate the wall shear

stress given in Table 4.Granule, flake and blend extrudates displayed

strength in tension, supporting their own weight over

varying lengths. Fresh granule extrudate was fragile,

unable to support itself over spans of more than 10–15

cm. Fresh flake samples were elastic and stronger, able

to support their own weight over spans of several me-

ters. Native starch, as described above, was very tough:

the transparent form was initially ductile but stiffened

after a few minutes. Extrudates of the blend featuredproperties between those of granule and flake. To

Table 4

Extrusion pressures and mechanical properties of dried extrudates (obtained from 3 point bending tests)

Sample Pextrusion (MPa) Estimated sw (MPa) Bending parameters

ry (MPa) E (MPa)

Granule 33 0.52 0.66 19

Flake 44 0.69 2.4 60

Native starch �70 1.09 20 110

Blend 21 0.33 2.6 490

Mechanical properties are averages from five repeat experiments.


quantify the distinct handling properties of the extru-

dates, Young’s Moduli, E, and uniaxial yield stresses, ry,were measured using 3 point bending tests (ASTM,

1991) in an ESH 2094 10 kN Servo-Hydraulic Testing

Machine (ESH, Brierly Hill, UK). The results, sum-marised in Table 4, did not fully reflect the substantial

differences between ‘green’ extrudates, however, because

samples had to be dried before testing.

Fig. 4. Images of granule extrudates: (a) SEM (15 kV, 80�); (b) TB-stained section (near surface); (c) IV-stained section (near surface).

4.2. Extrudate microstructures: single components

(a) Imaging: Fig. 4 shows cross-sections through a

typical granule extrudate, using different techniques.

The images reveal that the extrudate structure, which

did not vary over the cross-section, consisted of discrete

particles (cells), albeit strongly impacted, i.e. that the

particulate nature of the material was preserved during

extrusion. The integrity of individual cell walls couldalso be seen in the TB image (Fig. 4(b)), though the resin

embedding process caused significant disintegration of

the extrudate. IV images, (Fig. 4(c)) showed no pink and

blue mottling within cells, demonstrating that no amy-

lose and amylopectin separation had occurred. These

observations, together with the evident lack of binding

material even near to the extrudate surface, indicated

that the extrusion conditions were insufficient to damagethe intact potato cells. This was consistent with the

relative weakness of the extrudates. The strong impac-

tion of the particles upon each other showed the effects

of pressure, but there was no sign of damage to the

particles through shear. It can be said that the yielding

shear strength of the cells, sy, therefore exceeded theconditions associated with a wall shear stress, sw, of0.52 MPa. A more detailed description of the stress–strain conditions generated in the material can not be

made at this point for all the materials considered here,

as the associated rheology is not well defined.

Fig. 5 shows cross-sections of a typical flake extru-

date. In contrast to the granule, very little structure was

apparent in the SEM images, indicating that the mixture

of damaged cells and free starch had been homogenised

during the extrusion process. However, the TB and IVimages (Figs. 5(b) and (c)), which showed little effect of

staining and mounting––indicating good cohesivity––

revealed a complex structure composed of whole cells,

partially damaged cells and cell wall fragments bound

together by a matrix of gelatinised starch. The starch in

these samples revealed the mottling indicative of sepa-

ration of amylose and amylopectin through exposure to

Fig. 5. Images of flake extrudates: (a) SEM; (b) TB stained; (c) IV

stained.


shear. Again, these features were consistent with the

handling properties of the extrudates. In this case the

combined actions of pressure and shear had compacted

the paste, however there was little indication that ini-

tially intact particles were damaged. This indicates that

the cell yielding shear strength was not reached underthe flow conditions associated with a wall shear stress of

0.69 MPa.

Native starch extrudates featuring both translucent

material and white core (Fig. 3(e)) were studied in detail

as these contained the greatest variety of structure.

Cross-sections of such an extrudate are shown in Fig. 6.SEM revealed two regions of structure: a central core

(diameter �1 mm) of apparently undamaged nativestarch granules, and an annular region of visually

amorphous structure. When this annular region was

observed under XP light (Fig. 6(e)), however, it could be

seen that granular structure (indicated by Maltese

crosses) was preserved within the amorphous material,

except for a thin surface layer (�100 lm thickness) atthe surface, where any such structure had been de-

stroyed. In this case the shear stress distribution near the

wall (sw ¼ 1:09 MPa) was sufficient to rupture nativestarch granules. DIC images (Fig. 6(f)) showed that the

annular region consisted of a mixture of intact and

damaged starch granules, a state that may be compared

with that reported for starches subjected to high static

pressures. The extrudate, therefore, consisted of threeregions: a central core of predominantly undamaged

starch granules; an annular region of highly compressed

and partially damaged starch granules; and a surface

region of heavily sheared, amorphous starch. This sug-

gested that the starch had been compacted by the high

extrusion pressures, which had also caused partial

damage except at the centre, whilst shear had only been

significant in the die interfacial region.(b) DSC: Results for processed potato materials

showed no indications of gelatinisation (an irreversible

process), as expected. However, it was possible to dif-

ferentiate flake from granule by evidence in the former

of the complexing of mobile amylose (within the free

gelatinised starch) with emulsifier (added to the pow-

ders) (see Le Bail, Buleon, Colonna, & Bizot, 1997). This

result supports evidence from imaging that intact cellswere not damaged during extrusion.

DSC studies of the native starch (Fig. 7) and its ex-

trudates indicated increasing levels of conversion from

centre to edge of the extrudate. It was not physically

possible to separate the annular and surface regions

reliably, however, due to the thinness of the latter.

Compared to unsheared native starch (from Fig. 7,

DH ¼ 7:5 J/g, Tonset ¼ 59:2 �C), both the central andouter regions showed reduced gelatinisation endotherms

(on average by 10% and 75% respectively), indicating

that partial conversion had been effected throughout

the extrudate. In addition, samples from the outer re-

gions also displayed higher temperature endotherms

(Tonset ¼ 159:8 �C) characteristic of the formation ofamylose–lipid complexes. This evident mobility of

amylose indicated that the ordered structure of starchgranules in these regions had been severely disrupted. It

was therefore clear that structure in native starch had

been affected at all scales during extrusion.

(c) WAXS: Using this technique, some crystallinity

was detected in all samples bar the gelatinised starch.

Fig. 8 shows a typical result for unextruded native starch

showing the characteristic pattern for B-type crystallin-

Fig. 6. Cross-section images from extruded native starch: (a) SEM of a native starch extrudate; (b) SEM of material from the central region; (c) SEM

of material from the outer region; (d) the outer region of an extrudate, IV-stained; (e) image (d) using XP light; (f) DIC image of material in the

annular region.


ity, and a result for gelatinised starch with a character-

istic broad mound indicating amorphous structure. The

presence of crystallinity in cooked potato was presum-

ably due to retrogradation of the gelatinised starch

during storage (reformed crystallinity over periods of

weeks in non-crystalline amylopectin). This effect was

stronger in flake (with amylose/amylopectin separation)than granule, supporting Tester’s (2000) hypothesis that

amylose interferes with crystallite formation in amylo-

pectin.

There was no change in WAXS results for granule or

flake before and after extrusion, indicating that pro-

cessing had had no effect on starch structure at nano-

meter scales. For extruded native starch B-type

crystallinity was evident, though at reduced levels, inmaterial from both the centre and the outer regions.

50 54 58 62 66 70 74 78Temperature (°C)

Hea

t Inp

ut Native Starch

Extruded Starch (Central Region)

Gelatinised Starch

Extruded Starch (Outer Regions)

2 mW/mg

Fig. 7. DSC of native starch and extrudates. Scans have been offset on

the ordinate axis for clarity.

Inte

nsity

2 θ105 15

Fig. 8. X-ray diffraction patterns for native (black line) and gelatinised

(grey line) potato starch.

Table 5

Relative crystallinity, v, results (after Wakelin et al., 1959), calculatedusing peak area

Material Location v (%)

Native starch Un-extruded 100a

Extrudate core 70

Extrudate surface 61

Blend – 57

Flake – 58

Granule – 0.78

a By definition.

Fig. 9. Cross-section images of an extrudate of the blend: (a) IV-

stained; (b) under XP light.


This provides further evidence of the destruction during

extrusion of structure in native starch at length scales

from 10�4 to 10�7 m. It also indicated that the visually

unaffected starch from the central core had experienced

some level of conversion.

From these results, the relative crystallinities, v, of allmaterials were calculated (Wakelin et al., 1959), using

native starch as the crystalline reference and zero peak

height as the amorphous standard. Table 5 shows a

steady decrease in crystallinity, from unextruded native

starch to material from the outer layers of the native

starch extrudate, consistent with previous estimates of

conversion. These figures are also consistent with the

changes in enthalpy of gelatinisation noted previouslyand with the structural degradation in the outer material

of the extruded native starch identified by microscopy.

4.3. Extrudate microstructure: Blend

Fig. 9 shows a cross-section through a typical extru-

date of the blend. The structure, which was uniform

throughout, consisted of undamaged native starch

granules and intact potato cells enveloped in an amor-

phous binding material. The XP image (Fig. 9(b)) indi-cated that starch granules were present in their native

form even at the extrudate surface and therefore that no

surface shear layer was present. As in flake extrudates, it

was probable that the binder was derived from the free,


gelatinised starch initially present in ruptured potatocells rather than from damaged cells or native starch

granules. The low wall shear stress conditions (sw ¼ 0:33MPa) would preclude further damage.

DSC and WAXS results identified gelatinisation and

crystallinity in the extrudates. The former certainly

indicated the presence of un- or partially damaged na-

tive starch, i.e. the preservation of structure in the native

starch component. The presence of crystallinity in allingredients made interpretation of the WAXS results

impossible however.

5. Conclusions

Investigations of water uptake for each component

revealed strong differences between native starch and the

cooked potato materials. All pastes were mixed at the

same water content (40 wt.%), but chronic de-watering

of the native starch paste occurred during extrusion.

This meant that during deformation each componentwas hydrated to a level similar to its expected state in the

blend, namely unhydrated for the native starch, and

hydrated below the saturation point for the flake and

granule. The differences in the materials were apparent

in extrusion pressures required and the handling prop-

erties of the green extrudates, but these were not re-

flected in testing of dry extrudates.

Native starch extrudates exhibited increasing levels ofconversion as the extrusion (and de-watering) pro-

gressed and with proximity to the die interface (region of

high shear). These effects corresponded to a combina-

tion of increasing pressure and shear stress. Studies of

an intermediate extrudate revealed a core consisting of

visually undamaged granules, which had suffered some

structural degradation through pressure. A surface

layer, which had experienced shear stresses of the orderof 1 MPa, showed complete loss of granular structure

together with significant loss of molecular order and

crystallinity.

The microstructure of granule extrudates explained

their fragility, with little cellular degradation and bind-

ing material observed. This resistance to damage caused

increased extrusion pressures relative to the blend,

however. On the other hand, flake extrudates revealed agreat deal of binding material. The binder appeared to

have some visco-elastic properties, however deformation

of damaged cells resulted in increased extrusion pres-

sures compared to granule or the blend. In neither case

were the shear forces (associated with sw < 0:7 MPa)sufficient to damage intact cells.

The microstructure of extrudates made from the

blend revealed undamaged native starch granules andpotato cells, which had been protected during extrusion

by the presence of the amorphous binder. Despite

exhibiting the lowest extrusion pressures and shear

stresses, and therefore the least amount of particledegradation, sufficient binder was present (from the

flake) to give the green extrudates good structural

integrity. The balance of flake and granule in the for-

mulation, however, suppressed visco-elastic effects

associated with excess of binder, as occurred when flake

was extruded alone.

Acknowledgements

Funding for AC from the Biotechnology and Bio-

logical Sciences Research Council is gratefully

acknowledged, as is financial and technical support from

United Biscuits. Access and assistance with DSC, SEM

and ESEM at the Departments of Physics and Materials

at the University of Cambridge is also gratefullyacknowledged.

References

ASTM Standard C1161 (1991). In Flextural strength of advanced

ceramics at ambient temperature. Philadelphia, USA: American

Society for Testing and Materials.

Benbow, J. J., & Bridgwater, J. (1993). Paste flow and extrusion.

Oxford, UK: Clarendon Press.

Cheyne, A. (2000). Extrusion behaviour of starch-based food pastes,

PhD Dissertation, Department of Chemical Engineering, Univer-

sity of Cambridge, UK.

Cheyne, A., Barnes, O. J., & Wilson, D. I. (2004). Extrusion behaviour

of cohesive potato starch pastes. I. Rheological characterisation,

Journal of Food Engineering. doi:10.1016/j.jfoodeng.2004.02.028.

Cheyne, A., Wilson, D. I., Barnes, J., & Sala€un, Y. (2001). Modelling

of starch extrusion and damage in industrial forming processes. In

T. L. Barsby, A. M. Donald, & P. J. Frazier (Eds.), Starch structure

and functionality (pp. 8–26). Cambridge: Royal Society of Chem-

istry.

FAO (1999). Production Yearbook Volume 1998, FAO Statistics

Series No. 148, 53, Food and Agriculture Organisation of the

United Nations, Rome, Italy.

Frazier, P. J., Donald, A. M., & Richmond, P. (1996). Starch structure

and functionality. Cambridge, UK: Royal Society of Chemistry.

Galliard, T. (1987). Starch: properties and potential. London, UK:

Society of Chemical Industry.

Haase, N. U., Mintus, T., & Weipert, D. (1995). Viscosity measure-

ments of potato starch paste with the Rapid Visco Analyser.

Starch/St€arke, 47, 123–126.Hermansson, A.-M., & Svegmark, K. (1996). Developments in the

understanding of starch functionality. Trends in Food Science &

Technology, 7, 345–353.

Imberty, A., Buleon, A., Tran, V., & Perez, S. (1991). Recent advances

in the knowledge of starch structure. Starch/St€arke, 43, 375–384.Kudla, E., & Tomasik, P. (1992). The modification of starch by high

pressure. Starch/St€arke, 44, 253–259.Le Bail, P., Buleon, A., Colonna, P., & Bizot, H. (1997). Structural and

polymorphic transitions of amylose induced by water and temper-

ature changes. In P. J. Frazier, A. M. Donald, & P. Richmond

(Eds.), Starch Structure and Functionality. Cambridge, UK: Royal

Society of Chemistry.

Mitchell, J. R., Hill, S. E., Paterson, L., Vall�es, B., Barclay, F., &Blanshard, J. M. (1997). The role of molecular weight in the

conversion of starch. In P. J. Frazier, A. M. Donald, & P.

http://dx.doi.org/10.1016/jfoodeng.2004.02.028


Richmond (Eds.), Starch Structure and Functionality. Cambridge,

UK: Royal Society of Chemistry.

Noel, T. R., Ring, S. G., & Whittam, M. A. (1990). Physical properties

of starch products: structure and function. Norwich: AFRC Institute

of Food Research.

Pruvost, K., Corfield, G. M., Kingman, S., & Lawrence, C. J. (1998).

Effect of moisture content on the bulk and wall constitutive

behaviour of starch pastes. Transactions of IChemE: Food and

Bioproducts Processing, 76C, 188–192.

Rough, S. L., Wilson, D. I., & Bridgwater, J. (2002). A model

describing liquid phase migration within an extruding microcrys-

talline cellulose paste. Transactions of IChemE: Chemical Engi-

neering Research & Design, 80A, 701–714.

Senouci, A., & Smith, A. C. (1986). The extrusion cooking of potato

starch material. Starch/St€arke, 38, 78–82.Stapley, A. G. F. (1995). Diffusion and reaction in wheat grains. PhD

Dissertation, University of Cambridge.

Svegmark, K., & Hermansson, A.-M. (1991). Distribution of amylose

and amylopectin in potato starch pastes: effects of heating and

shearing. Food Structure, 10, 117–129.

Talburt, W. F., & Smith, O. (1980). Potato processing. New York,

USA: Van Nostrand Reinhold Company.

Terzaghi, K. (1943). Theoretical soil mechanics. NY, USA: Wiley &

Sons.

Tester, R. F. (2000). Amylopectin crystallisation in starch. In P. J.

Frazier, A. M. Donald, & P. Richmond (Eds.), Starch structure and

functionality. Cambridge, UK: Royal Society of Chemistry.

van Soest, J. J. G. (1996). Starch plastics: structure–property relation-

ships. Netherlands: Universiteit Utrecht.

Wakelin, J. H., Virgin, H. S., & Crystal, E. (1959). Journal of Applied

Physics, 30.

Yamada, T., Tamaki, S., Hisamatsu, M., & Teranishi, K. (1997).

Molecular change of starch granule with physical treatment: potato

starch by ball-mill treatment. In P. J. Frazier, A. M. Donald, & P.

Richmond (Eds.), Starch structure and functionality. Cambridge,

UK: Royal Society of Chemistry.

Yeh, A.-I., Wu, T.-Q., & Jaw, Y.-M. (1999). Starch transitions and

their influence on flow pattern during single-screw extrusion

cooking of rice flour. Transactions of IChemE: Food and Bioprod-

ucts Processing, 77C, 47–54.

Extrusion Behaviour of Cohesive Potato Starch Pastes. II. Microstructure–Process Interactions

Documents

Transcript of Extrusion Behaviour of Cohesive Potato Starch Pastes. II. Microstructure–Process Interactions