Accepted Manuscript
Title: Infrared spectroscopy as a tool to characterise starchordered structure- a joint FTIR-ATR, NMR, XRD and DSCstudy
Author: Frederick J. Warren Michael J. Gidley BernadineFlanagan
PII: S0144-8617(15)01156-XDOI: http://dx.doi.org/doi:10.1016/j.carbpol.2015.11.066Reference: CARP 10585
To appear in:
Received date: 12-10-2015Revised date: 25-11-2015Accepted date: 26-11-2015
Please cite this article as: Warren, F. J., Gidley, M. J., and Flanagan,B.,Infrared spectroscopy as a tool to characterise starch ordered structure- ajoint FTIR-ATR, NMR, XRD and DSC study, Carbohydrate Polymers (2015),http://dx.doi.org/10.1016/j.carbpol.2015.11.066
This is a PDF file of an unedited manuscript that has been accepted for publication.As a service to our customers we are providing this early version of the manuscript.The manuscript will undergo copyediting, typesetting, and review of the resulting proofbefore it is published in its final form. Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers thatapply to the journal pertain.
Page 1 of 26
Accep
ted
Man
uscr
ipt
Highlights
A larger sample set (61) of starches than has been used in previous studies
Full structural characterisation using a range of techniques
A non-linear, hydration dependent response occurs in the infrared spectrum
Complex changes in the infrared spectra of starch with increasing structural order
Highlights (for review)
Page 2 of 26
Accep
ted
Man
uscr
ipt
Infrared spectroscopy as a tool to characterise starch ordered 1
structure- a joint FTIR-ATR, NMR, XRD and DSC study 2
Frederick J. Warren †, Michael J. Gidley
†, Bernadine Flanagan
†* 3
†Centre for Nutrition and Food Sciences, ARC Centre of Excellence in Plant Cell Walls, 4
Queensland Alliance for Agriculture and Food Innovation, The University of Queensland, St. 5
Lucia, Brisbane, Queensland 4072, Australia 6
7 8 9 10
11 12 * Corresponding author. Centre for Nutrition and Food Sciences, ARC Centre of Excellence 13
in Plant Cell Walls, Queensland Alliance for Agriculture and Food Innovation, The 14
University of Queensland, St. Lucia, Brisbane, Queensland 4072, Australia 15
16
17
Title Running Head: Infrared spectroscopy to characterise starch ordered structure 18
Keywords: Starch, crystallinity, polymer structure, FTIR-ATR, NMR, XRD 19
20
*Manuscript
Page 3 of 26
Accep
ted
Man
uscr
ipt
21
Abstract 22
Starch has a heterogeneous, semi-crystalline granular structure, and the degree of ordered 23
structure can affect its behaviour in foods and bioplastics. A range of methodologies are 24
employed to study starch structure; differential scanning calorimetry, 13
C nuclear magnetic 25
resonance, X-ray diffraction and Fourier transform infrared spectroscopy (FTIR). Despite the 26
appeal of FTIR as a rapid, non-destructive methodology, there is currently no systematically 27
defined quantitative relationship between FTIR spectral features and other starch structural 28
measures. Here, we subject 61 starch samples to structural analysis, and systematically 29
correlate FTIR spectra with other measures of starch structure. A hydration dependent peak 30
position shift in the FTIR spectra of starch is observed, resulting from increased molecular 31
order, but with complex, non-linear behaviour. We demonstrate that FTIR is a tool that can 32
quantitatively probe short range interactions in starch structure. However, the assumptions of 33
linear relationships between starch ordered structure and peak ratios are overly simplistic. 34
Graphical abstract 35
36
Page 4 of 26
Accep
ted
Man
uscr
ipt
37
1. Introduction 38
Starch is one of the most important biopolymers in the world. It is the main source of energy, 39
in the form of exogenous glucose supply, in the human diet, and is a key feedstock for a 40
number of industries such as bioethanol production and biodegradable plastics manufacture 41
(Butterworth, Warren & Ellis, 2011; John, Anisha, Nampoothiri & Pandey, 2011; Shen, 42
Worrell & Patel, 2010). Starch granules are made up of two glucose polymers, amylose (an 43
essentially linear polymer composed of α-1→4 linked anhydro-glucose residues, typically 44
reported to have a molar mass ∼ 105–10
6 g mol
−1) and amylopectin (a highly branched 45
polymer consisting of short chains of α-1→4 linked anhydro-glucose residues, interspersed 46
with 5-6% branch points of α-1→6 linkages, with a molecular weight perhaps as high as 107–47
109 g mol
−1) (Gidley et al., 2010). These two polymers are arranged into a complex semi-48
crystalline granular arrangement, based on aggregates of double helices formed primarily 49
from adjacent branches of amylopectin molecules. 50
The ratio of crystalline to amorphous material in the native starch granule varies greatly, from 51
10% to 50% crystallinity, depending on the botanical origin of the starch (Lopez‐Rubio, 52
Flanagan, Gilbert & Gidley, 2008). Further alterations in structure can occur as a result of 53
subsequent processing of the starch. For example, the most common processing step is 54
gelatinisation, where the starch is heated in excess water at a temperature between 50 and 55
70°C. The starch undergoes a cooperative transition whereby water enters the amorphous 56
regions of granules, causing them to swell and destabilise the crystalline regions. This results 57
in rapid swelling of the granule, melting of the crystallites and a complete loss of ordered 58
structure (Bogracheva, Wang, Wang & Hedley, 2002; Cooke & Gidley, 1992; Perry & 59
Donald, 2002; Waigh, Gidley, Komanshek & Donald, 2000). Other processing steps which 60
Page 5 of 26
Accep
ted
Man
uscr
ipt
may increase or decrease starch molecular order include annealing (Tester & Debon, 2000), 61
extrusion (Brümmer, Meuser, van Lengerich & Niemann, 2002) and enzymatic digestion 62
(Lopez-Rubio, Flanagan, Shrestha, Gidley & Gilbert, 2008). For all processing applications 63
involving starch, measuring the degree of ordered structure is important, but doing so is far 64
from facile. As different analytical methods measure ordered structure in different ways, and 65
may measure subtly different aspects of starch structure, it is informative to compare and 66
contrast analytical methods. In the present paper, four of the most common analysis methods 67
are used; x-diffraction (XRD), differential scanning calorimetry (DSC), 13
C cross-68
polarization magic angle spinning nuclear magnetic resonance spectroscopy (13
C CP-MAS 69
NMR) and Fourier transform infrared spectroscopy with attenuated total internal reflectance 70
(FTIR-ATR). 71
Diffraction based methods, such as XRD, are the only methods available for the assessment 72
and quantification of long-range crystalline order in starch. Crystalline order in starch is 73
conventionally determined using a two-phase model, introduced by Sterling (Sterling, 1960) 74
and Nara (Nara, Mori & Komiya, 1978), where an amorphous halo is subtracted from the 75
ordered parts of the scattering pattern, and the ratio of the two phases is used as an estimate 76
of long range crystalline order. A more nuanced peak fitting approach has recently been 77
suggested (Lopez‐Rubio, Flanagan, Gilbert & Gidley, 2008), taking into account incoherent 78
scattering from crystalline regions, but due to its simplicity and robustness, the two-phase 79
model is still commonly employed. 80
While XRD is the only method that can measure long-range crystalline order, there are a 81
number of methods available to measure short-range double helical order in starch. DSC 82
measures the enthalpy change associated with gelatinisation of starch, as gelatinisation 83
involves the disruption of ordered regions in the granule, and therefore the magnitude of the 84
enthalpy change is proportional to the amount of ordered structure. Early work assumed that 85
Page 6 of 26
Accep
ted
Man
uscr
ipt
the enthalpy change was predominantly the result of crystallite melting, and was in essence a 86
direct measure of long-range crystalline order (Biliaderis, Page, Maurice & Juliano, 1986; 87
Donovan, 1979; Liu, Lelievre & Ayoung-Chee, 1991). More recent studies, however, have 88
indicated a more complex picture, in which the contribution from disruption of crystalline 89
order is only one small component, with contributions to the overall enthalpy change also 90
coming from swelling, hydration and disruption of short-range double helical structure, and 91
that gelatinisation is a kinetic event that is dependent on both heating rate and water content 92
(Bogracheva, Meares & Hedley, 2006; Cooke & Gidley, 1992; Gidley, 1992). The enthalpy 93
change associated with starch gelatinisation, under standard conditions, can therefore be used 94
as an indicator of ordered structure, but any further inferences regarding long-range 95
crystallinity should be drawn with caution. 96
In the mid-1980’s Gidley and Bociek (Gidley & Bociek, 1985) noted that the 13
C CP-MAS 97
NMR spectra of starch shows differences in chemical shift displacements due to changes in 98
the conformation of the glucan chain. These changes were particularly pronounced at the C1 99
and C4 positions, and were observed between amorphous starch polymers and double helices, 100
and between A and B type crystalline starch. This observation was subsequently exploited to 101
allow direct quantification of the proportion of double-helices in starch, giving a direct 102
measure of short-range order (Cooke & Gidley, 1992). This method has subsequently been 103
refined through use of peak fitting procedures and, more recently, chemometric approaches to 104
give a reliable estimate of ordered structure (Flanagan, Gidley & Warren, 2015; Tan, 105
Flanagan, Halley, Whittaker & Gidley, 2007). It has been consistently noted that the 106
proportion of double-helices in starch is greater than the amount of crystallinity, suggesting 107
that there are a significant proportion of double helices not in crystalline register (Cooke & 108
Gidley, 1992; Lopez‐Rubio, Flanagan, Gilbert & Gidley, 2008). 109
Page 7 of 26
Accep
ted
Man
uscr
ipt
More recently, FTIR-ATR has been suggested as a method to measure short range order in 110
starch, and even as an “infrared crystalline phase index” (Capron, Robert, Colonna, Brogly & 111
Planchot, 2007; Sevenou, Hill, Farhat & Mitchell, 2002; Van Soest & Vliegenthart, 1997). 112
FTIR-ATR has some notable advantages over other commonly used methods for measuring 113
starch structure- the equipment is relatively cheap, simple to use and maintain, and offers 114
highly flexible sample presentation. The ATR accessory gives the flexibility to analyse starch 115
samples in a range of physical forms, particularly in the hydrated state, making it possible to 116
observe spectral differences due to the level of hydration. Using microspectroscopy, these 117
advantages can also be applied to microscopically-localised regions within samples, in 118
contrast to the other methods which give bulk average data. The FTIR spectra of starch 119
typically shows bands at 2900-3000 cm-1
(C-H stretching), 1100-1150 cm-1
(C-O, C-C and C-120
O-H stretching) and 1100-900 cm-1
(C-O-H bending). It is not possible to assign bands in the 121
spectra of starch unambiguously, as the absorbance bands are overlapping and poorly 122
resolved (Van Soest & Vliegenthart, 1997). The bands in the region 1100-900 cm-1
have been 123
shown to be sensitive to changes in starch structure, in particular bands at 1000, 1022 and 124
1047 cm-1
have been widely studied, using a range of samples including retrograded starches 125
(Wilson, Kalichevsky, Ring & Belton, 1987), mixtures of starch and amorphous maltodextrin 126
(Van Soest & Vliegenthart, 1997), enzyme hydrolysed starches (Sevenou, Hill, Farhat & 127
Mitchell, 2002) and acid hydrolysis residues (‘lintners’) (Capron, Robert, Colonna, Brogly & 128
Planchot, 2007). From these studies, the band at 1022 cm-1
seems to increase in more 129
amorphous samples, while the bands at 1000 and 1047 cm-1
become more defined in more 130
crystalline samples. This has led to the adoption of the band ratios at 1022:1000 cm-1
and 131
1047:1022 cm-1
as measures of short-range ordered molecular structure in a range of different 132
studies (Bello-Pérez, Ottenhof, Agama-Acevedo & Farhat, 2005; Brümmer, Meuser, van 133
Lengerich & Niemann, 2002; Iizuka & Aishima, 1999; Rubens, Snauwaert, Heremans & 134
Page 8 of 26
Accep
ted
Man
uscr
ipt
Stute, 1999; Smits, Ruhnau, Vliegenthart & van Soest, 1998). Using chemometric 135
approaches, this procedure has been taken further, demonstrating the hydration sensitivity of 136
the 1022:1000 cm-1
intensity ratio and the possible relationship between crystallinity 137
measured by XRD and FTIR band ratio (Capron, Robert, Colonna, Brogly & Planchot, 2007). 138
Capron and co-workers (Capron, Robert, Colonna, Brogly & Planchot, 2007) found the 139
absorbance at 1047 cm-1
to be essentially independent of the degree of structure in the starch. 140
They linked the hydration sensitivity of the 1022:1000 cm-1
band ratio to the model liquid-141
crystalline polymeric model of starch structure proposed by Waigh and co-workers (Waigh, 142
Kato, Donald, Gidley, Clarke & Riekel, 2000; Waigh, Perry, Riekel, Gidley & Donald, 143
1998). This led to the suggestion that the change in the 1022:1000 cm-1
band ratio upon 144
hydration of crystalline (but not amorphous) starches is the result of a nematic-smectic 145
transition. Warren and co-workers (Warren, Perston, Royall, Butterworth & Ellis, 2013) 146
recently investigated the spectral changes that occur during in situ hydrothermal 147
gelatinisation of starch. Their findings showed that changes in the spectra during 148
gelatinisation can accurately determine the gelatinisation temperature reflecting the loss of 149
ordered structure. The intensity ratio of the bands at 1022:1000 cm-1
proved a useful tool to 150
monitor loss of structure, but a detailed appraisal of the spectra suggested greater complexity, 151
with an apparent peak position shift during gelatinisation. 152
In the present study, we examine a wider range of starches than has previously been used in 153
any single structural study of starch, including native wild type, mutant, amorphous, and 154
extruded starches, both in the dry and hydrated states. Using this extensive sample set, we 155
aim to develop a detailed understanding of the changes that occur in the FTIR spectra of 156
starches with different levels of molecular structure using chemometric analysis tools in 157
addition to peak ratio analysis, to improve the use of FTIR-ATR as a tool to understand 158
Page 9 of 26
Accep
ted
Man
uscr
ipt
starch ordered structure. The starch samples are also analysed using 13
C CP/MAS NMR, 159
XRD and DSC to explore short-range and long-range molecular order in starch. 160
2. Materials and Methods 161
2.1. Materials 162
Commercially available starches: regular maize, gelose 80 (high amylose) maize and mazaca 163
waxy maize were purchased from National Starch (Australia), waxy maize was bought from 164
Penford (Australia) and potato starch was purchased from National Starch and Chemicals 165
(UK). All commercially obtained starches were used without further treatment. All other 166
starches were gifts from collaborators and were from a range of botanical sources, including 167
native barley and barley mutants (Higgins et al., 2013; Regina et al., 2012) and wild rice and 168
rice mutants (Butardo et al., 2011). Due to the limited amount of mutant rice samples, whole 169
flours were used rather than isolated starches. Additional starch samples were produced under 170
different drying conditions (Zhang et al., 2014), and using extrusion to add variation to the 171
sample set (Zhang, Dhital, Flanagan, Luckman, Halley & Gidley, 2015). Wheat starch 172
(Cerestar, cv. GL04) and pea starches (WT, r and lam (Wang, Bogracheva & Hedley, 1998)) 173
were gifts from Prof. T. Bogracheva and Prof. C. Hedley (formerly of the John Innes Centre, 174
Norwich, UK), waxy rice starch (cv. Remyrise) was a gift from Dr. P. Rayment (Unilever, 175
UK); these starches are described in detail elsewhere (Warren, Royall, Gaisford, Butterworth 176
& Ellis, 2011). Native starch powders from tapioca (Penford, Ausralia), regular maize 177
(Penford, NZ), waxy maize (Tate and Lyle, Decatur, IL), high amylose maize (HylonVII, 178
Penford, Australia), potato, and wheat (both commercial material, supermarket, Sydney, 179
Australia) were gifts from Dr. E.P. Gilbert (ANSTO, Australia) (Doutch & Gilbert, 2013). 180
Page 10 of 26
Accep
ted
Man
uscr
ipt
Amorphous starches were prepared following the method of Gidley and Bociek (Gidley & 181
Bociek, 1985) with minor adjustments. Briefly, starches were boiled in water (1% w/v) for 30 182
minutes and precipitated with ethanol or rapidly frozen in liquid nitrogen and lyophilised. 183
Moisture contents were measured by weighing approximately 200mg of sample into a 184
crucible, drying the sample for 18 hours in an oven at 105°C. The resultant weight of the 185
sample was recorded and the difference in weight is taken as the moisture content. 186
The amylose contents of the samples were measured according to the iodine binding method 187
of Knutson (Knutson 1986; Knutson and Grove 1994; Knutson 2000). In this analysis starch 188
samples were dissolved overnight by shaking in a 6 mM iodine solution which was made by 189
first dissolving iodine in 9 parts DMSO then adding 1 part water. Once the starch had 190
dissolved, the solution was diluted in 8 parts H2O. After 30 min the iodine-amylose complex 191
formed a stable colour and the absorbance of the starch samples and amylose standards was 192
read spectrophotometrically at 600 nm against the reagent blank. 193
2.2. Fourier transform infrared spectroscopy 194
All spectra were collected using a PerkinElmer Spectrum One FTIR spectrometer fitted with 195
a PerkinElmer UATR single bounce ATR accessory with a diamond crystal. Data were 196
collected by an attached computer running PerkinElmer Spectrum 6 software. Dry starches 197
were equilibrated at laboratory humidity (50 % RH) and clamped directly onto the crystal for 198
analysis. Hydrated starches were made into a slurry by mixing with deionised water (50% 199
w/w), and added to the top of the crystal. Spectra of hydrated starches were obtained without 200
applying pressure. 32 spectra were obtained and co-added for each sample at a resolution of 4 201
cm-1
. A background spectrum was obtained by collecting 32 co-added scans following 202
cleaning of the crystal with a mixture of ethanol and water. The spectra were vector 203
Page 11 of 26
Accep
ted
Man
uscr
ipt
normalized by dividing each spectrum by the standard deviation of its absorbance values 204
(with mean subtraction), prior to subsequent analysis (SNV normalisation). 205
2.3. Wide Angle X-Ray Diffractometry 206
X-ray diffraction measurements were performed with an X’Pert Pro X-ray diffractometer 207
(XRD) (PANalytical, Almelo, the Netherlands) operating at 40 kV and 40 mA with Cu Kα 208
radiation (λ) at 0.15405 nm. The scanning region was set from 2θ = 3 to 2θ = 40 with a step 209
interval of 0.02 and a scan rate of 0.5/min. The crystalline peak area and amorphous area 210
were separated by PeakFit software (Version 4.12, Systat Software Inc., San Jose, CA, USA) 211
following the method of Lopez-Rubio, Flanagan, Shrestha, Gidley and Gilbert (2008) 212
Relative crystallinity was calculated as the ratio of the crystalline peak area to the total 213
diffraction area. 214
2.4. 13
C CP/MAS Nuclear Magnetic Resonance Spectroscopy 215
All starches were analysed by 13
C cross-polarized magic angle spinning (CP/MAS) nuclear 216
magnetic resonance (NMR) spectroscopy, using a Bruker MSL-300 spectrometer (Bruker, 217
Billerica, MA, USA) at a frequency of 75.46 MHz. Depending on availability, up to 200 mg 218
starch was packed in a 4-mm diameter, cylindrical, PSZ (partially stabilized zirconium oxide) 219
rotor with a Kel-F end cap. When as little as 50 mg of starch was available, the rotor was 220
packed with Teflon tape above and below the sample to maximize signal and ensure correct 221
spinning. The rotor was spun at 5 kHz at the magic angle (54.7°). The 90° pulse width of 5 μs 222
and a contact time of 1 ms were used for all starches with a recycle delay of 3 s. The spectral 223
width was 38 kHz, the acquisition time was 50 ms, 2 k points were recorded, and zero-filled 224
to 4 k before transformation. Exponential adopization resulting in 20 Hz of line broadening 225
was used. At least 1000 scans and up to 2000 scans were accumulated for each spectrum. 226
Spectral acquisition and interpretation methodology as described by Tan, Flanagan, Halley, 227
Page 12 of 26
Accep
ted
Man
uscr
ipt
Whittaker and Gidley (2007) and Flanagan, Gidley and Warren (2015) were used to quantify 228
the double helices, single helices, and amorphous conformational features. 229
2.5. Differential Scanning Calorimetry 230
All DSC data were obtained using a TA Instruments Q2000 instrument, using Tzero 231
hermetically sealed aluminium pans. The sample chamber was purged with nitrogen gas at a 232
rate of 40 mL/min. The instrument was calibrated for temperature using indium and tin 233
standards. Samples were prepared in triplicate by accurately weighing approximately 5 mg of 234
starch with 40 mg of deionised water into a pan, mixing, sealing and leaving overnight to 235
equilibrate. Samples were heated from 10 to 95°C at a rate of 5°C/min. Subsequent 236
thermograms were analysed using TA Instruments Universal Analysis software to obtain 237
gelatinisation enthalpies, as described elsewhere (Bogracheva, Wang, Wang & Hedley, 238
2002). 239
2.6. Data analysis 240
Preparation of figures, statistical analyses and linear regression fitting was carried out using 241
Sigma Plot 13 (Systat, USA). Principal component analysis (PCA) was carried out using 242
Unscrambler X 10.3 (Camo, Norway). 243
3. Results 244
3.1. Analytical data 245
DSC, XRD and 13
C CP/MAS NMR data were obtained for a total of 61 starch samples. The 246
mean DSC enthalpy of the sample set was 9.0 J/g with a range from 0 to 17.6 J/g. 13
C 247
CP/MAS NMR spectra were analysed to calculate total helical order for each of the samples. 248
The mean helical order in the sample set was 30.0 % with range from 4 to 49 % total helical 249
order. XRD data were analysed to obtain relative crystallinity, and the sample set had a mean 250
Page 13 of 26
Accep
ted
Man
uscr
ipt
of 25.8 % with a range from 0 to 51 % relative crystallinity (see Supporting Information 251
Table S1 for full data sets). 252
3.2. Fourier transform infrared spectra 253
Spectra were collected for the starch sample set used in the present study, and are shown in 254
Figures 1 A and B for the dry and hydrated samples, respectively. The hydrated samples 255
showed much more pronounced peaks at 3000-3700 cm-1
, 2000-2250 cm-1
and 1500-1750 256
cm-1
, as well as a pronounced baseline slope below 900 cm-1
as a result of adsorption by 257
excess water. The dry samples showed a large number of adsorption bands in the region 258
1700-1200 cm-1
presumably arising from minor components (protein, lipid) in the starch, as 259
well as the CH- stretch peaks at approximately 2900 cm-1
, which were masked by the large 260
water adsorption bands in the hydrated samples. The amide I and amide II peaks at 1640 and 261
1530 cm-1
are particularly prominent in the rice flour samples, which have a lower degree of 262
purity, and reflect their slightly elevated protein content. The major adsorption bands arising 263
from starch can be observed in the region 1200-1000 cm-1
, arising from C-O, C-C and C-O-H 264
stretching and C-O-H bending, as discussed in the introduction. A visual inspection of the 265
spectra indicates that there is significant variation between the different spectra in the region 266
1200-1000 cm-1
, and that the hydrated starches showed significantly more variation in this 267
region than the dry samples. Both the hydrated and dry sample sets were subjected to PCA 268
and the loadings for the first and second principal components (PC) are shown in Figure 2. 269
For the hydrated starches, the loadings for the first and second PC’s (Figure 2A and B) show 270
that the majority of the variance in the spectra occurs in the region 1000-1030 cm-1
. The same 271
can be observed for dry samples (Figure 2C and D), although the loadings for PC 2 (Figure 272
2D) also show a number of peaks in the region 1700-1200 cm-1
. This confirms the findings of 273
Capron et al. and Sevenou et al., (Capron, Robert, Colonna, Brogly & Planchot, 2007; 274
Sevenou, Hill, Farhat & Mitchell, 2002) using an extended data set, that the 1000-1022 cm-1
275
Page 14 of 26
Accep
ted
Man
uscr
ipt
region is the main source of variation in the spectra of starch, and that there is significant 276
water sensitivity in the spectra. 277
278
279
Page 17 of 26
Accep
ted
Man
uscr
ipt
3.3. Relationship between peak height ratios and ordered structure from other analytical 284
methods 285
As previously discussed, a number of authors have suggested that the peak ratios 995:1022 286
and 1045:1022 cm-1
may be useful indicators of ordered structure in starch. In the present 287
study, these two peak ratios have been systematically investigated for correlation against data 288
from DSC, XRD and 13
C CP/MAS NMR, to investigate if they can be used as predictors of 289
ordered molecular structure in starch (Table 1 and, Supplementary Information Table S1 and 290
Figure S1). In the dry state, there are correlations between both peak ratios and all three 291
analytical methods, although these are all weak correlations. The root mean square error of 292
calibration (RMSEC) (Table 1 and 2) for all of these models are large, representing errors of 293
greater than 25 % for the average sample in the calibration set. 294
Table 1. Fitting parameters for correlations between selected FTIR peak ratios and measures 295
of starch ordered structure for hydrated starches 296
Analytical
method
Ratio r2
(Calibration) RMSEC P value
DSC 995:1022 0.68 2.83 J/g <0.001
NMR 995:1022 0.48 9.53 % <0.001
XRD 995:1022 0.48 8.61 % <0.001
DSC 1045:1022 0.001 5.02 J/g 0.81
NMR 1045:1022 0.006 13.25 % 0.57
XRD 1045:1022 0.006 11.92 % 0.57
297
Table 2. Fitting parameters for correlations between selected FTIR peak ratios and measures 298
of starch ordered structure for dry starches 299
Analytical
method
Ratio r2
(Calibration) RMSEC P value
DSC 995:1022 0.42 3.64 J/g <0.001
NMR 995:1022 0.31 11.34 % <0.001
XRD 995:1022 0.39 9.50 % <0.001
DSC 1045:1022 0.44 3.17 J/g <0.001
NMR 1045:1022 0.59 8.32 % <0.001
XRD 1045:1022 0.48 8.53 % <0.001
300
Page 18 of 26
Accep
ted
Man
uscr
ipt
For the hydrated starch samples, the peak ratio 1045:1022 cm-1
showed no correlation with 301
any of the analytical methods employed in this study. The 1022:995 cm-1
peak ratio showed a 302
correlation with all three analytical methods, and the best correlation (r2 0.69) for any of the 303
peak ratios in either the hydrated or dry states was observed for the relationship between the 304
hydrated starch 1022:995 cm-1
ratio and DSC enthalpy for the same starch samples. To 305
further investigate the relationship between these peak ratios and measures of ordered 306
structure, heat map plots were generated (Figure 3) in which the spectra have been ranked 307
according to DSC enthalpy from high to low, and then plotted in series, with high absorbance 308
values shown as “hot” colours (reds and yellows) and low absorbance values shown as “cold” 309
colours (greens and blues). 310
311
Page 19 of 26
Accep
ted
Man
uscr
ipt
312
The hydrated starch spectra (Figure 3a) show a pronounced peak position shift, with the most 313
intense peak moving from around 1000 cm-1
in starches with a high order (above 14 J/g) to 314
1020 cm-1
in starches with low order (less than 10 J/g).The bi-plot from the PCA analysis of 315
the hydrated starches (Figure 4) confirms this, showing that the high and low ordered starches 316
divide into two groups. The dry starches show a similar, but much less pronounced, peak 317
position shift. 318
319
Page 20 of 26
Accep
ted
Man
uscr
ipt
4. Discussion 320
In the present study we have subjected a very wide range of starch samples to a range of 321
commonly used analytical methods (XRD, NMR, DSC and FTIR-ATR). This was carried out 322
with the aim of exploring the use of FTIR-ATR as a general method for the analysis of starch 323
ordered structure, and to build calibrations allowing direct quantification of starch ordered 324
structure. 325
The speed and flexibility in sample presentation of FTIR-ATR make it an attractive 326
alternative to conventional analysis methods, and it has received wide use, but few authors 327
have attempted to validate FTIR-ATR as a quantitative method (Capron, Robert, Colonna, 328
Brogly & Planchot, 2007). The starch samples used in the present study cover the full range 329
of levels of ordered structure that would be commonly encountered in the study of starch 330
structure, from fully amorphous hydrothermally treated or extruded starches through to native 331
tuber starches with levels of crystallinity measured by XRD to be as high as 51%. The 332
samples come from a wide range of botanical origins, including genetic mutants, allowing a 333
thorough assessment of the expected variation in FTIR-ATR spectra. 334
It is clear that large variability exists across FTIR-ATR spectra of different starches, and 335
Figure 3 highlights this variation. A key observation is that in the hydrated state there is a 336
pronounced peak position shift from around 1022 cm-1
to around 1000 cm-1
between the low 337
and high ordered starches, and that this change is far less pronounced for starch in the dry 338
state, where all but the most amorphous samples have a mean absorbance peak at around 339
1000 cm-1
. Thus, for starch in the hydrated state, it is possible to use the 1000:1022 cm-1
peak 340
ratio to indicate (but not to fully quantify) whether a starch has high or low levels of ordered 341
structure. The peak position shift from 1000 to 1022cm-1
is the most dramatic change in the 342
spectra as a result of changes in ordered structure, far more so than those at 1045 and 1080 343
Page 21 of 26
Accep
ted
Man
uscr
ipt
cm-1
, which have been suggested by some authors to be alternative indices of ordered 344
structure (Rubens, Snauwaert, Heremans & Stute, 1999). This is observation in good 345
agreement with previous work (Warren, Perston, Royall, Butterworth & Ellis, 2013) 346
demonstrating that when native (and therefore ordered) starch in excess water undergoes 347
hydrothermal treatment to an amorphous form there is a distinct peak position change from 348
1000 to 1022 cm-1
. 349
It is possible to speculate on the reasons for this peak shift occurring, and as to why it may be 350
more pronounced in hydrated samples as opposed to dry samples. It is well established that 351
hydration of the starch granule leads to significant structural rearrangements (Waigh, Gidley, 352
Komanshek & Donald, 2000; Waigh, Kato, Donald, Gidley, Clarke & Riekel, 2000) at longer 353
length scales (10-100 nm) than are directly probed by infrared spectroscopy. From the data in 354
the present paper, and from previous work (Capron, Robert, Colonna, Brogly & Planchot, 355
2007), these starch structural rearrangements have a significant impact at the scale of 356
individual chemical bonds probed by infrared spectroscopy. It is interesting to note that for 357
FTIR spectra collected from hydrated samples, a considerably better correlation was seen 358
between the infrared spectra and DSC enthalpy, than NMR and XRD data (Table 1 and 3). 359
The DSC enthalpy change associated with starch gelatinisation is dependent, in a complex 360
manner, on a range of short range molecular interactions which are altered upon heating in a 361
hydrated environment (Cooke & Gidley, 1992). It may be speculated that the FTIR spectra of 362
hydrated samples reflects these short range interactions, rather than the longer range 363
associations which result in the formation of helices (detected by NMR), and the arrangement 364
of these helices into crystalline arrays (detected by XRD), although clearly structures 365
detected at all these length scales contribute to the overall structure of the starch granule. 366
5. Conclusions 367
Page 22 of 26
Accep
ted
Man
uscr
ipt
The present study uses a wider range of starch samples than has been available in previous 368
studies to compare different measures of starch ordered structure, with the aim of probing the 369
relationship between FTIR spectra of starch in the dry and hydrated states, and other 370
measures of starch ordered structure. It has been confirmed that in the spectra of hydrated 371
starch, the main difference between starches with high and low degrees of order is a peak 372
position shift from 1000 to 1022 cm-1
, but using the wide range of starches available in this 373
study it has been shown that there is a more complex relationship between infrared spectra of 374
starches with different degrees of order than has previously been appreciated. There is a clear 375
and distinct shift in the spectra, in particular of hydrated starches, as starch molecular order 376
increases, however this is a non-linear transition. While this complexity means that FTIR 377
spectra cannot be used directly in place of alternative quantitative measures of starch 378
structure, we demonstrate the feasibility of using FTIR to characterise short range molecular 379
interactions in starch, so long as care is taken in the analysis of the spectra. 380
6. Acknowledgements 381
We acknowledge the facilities, and the scientific and technical assistance of the Australian 382
Microscopy & Microanalysis Research Facility at the Centre for Microscopy and 383
Microanalysis for XRD measurements, and the NMR facilities at the Centre for Advanced 384
Imaging at The University of Queensland. We would like to thank Barbara Gorham from the 385
Centre for Nutrition and Food Science at The University of Queensland for her assistance in 386
measuring the starch moisture contents.387
Page 23 of 26
Accep
ted
Man
uscr
ipt
388
Bello-Pérez, L., Ottenhof, M.-A., Agama-Acevedo, E., & Farhat, I. (2005). Effect of storage time on the 389 retrogradation of banana starch extrudate. Journal of Agricultural and Food Chemistry, 53(4), 1081-390 1086. 391 Biliaderis, C. G., Page, C. M., Maurice, T. J., & Juliano, B. O. (1986). Thermal characterization of rice 392 starches: A polymeric approach to phase transitions of granular starch. Journal of Agricultural and 393 Food Chemistry, 34(1), 6-14. 394 Bogracheva, T., Wang, Y., Wang, T., & Hedley, C. (2002). Structural studies of starches with different 395 water contents. Biopolymers, 64(5), 268-281. 396 Bogracheva, T. Y., Meares, C., & Hedley, C. L. (2006). The effect of heating on the thermodynamic 397 characteristics of potato starch. Carbohydrate Polymers, 63(3), 323-330. 398 Brümmer, T., Meuser, F., van Lengerich, B., & Niemann, C. (2002). Effect of extrusion cooking on 399 molecular parameters of corn starch. Starch‐Stärke, 54(1), 1-8. 400 Butardo, V. M., Fitzgerald, M. A., Bird, A. R., Gidley, M. J., Flanagan, B. M., Larroque, O., 401 Resurreccion, A. P., Laidlaw, H. K., Jobling, S. A., & Morell, M. K. (2011). Impact of down-regulation of 402 starch branching enzyme IIb in rice by artificial microRNA-and hairpin RNA-mediated RNA silencing. 403 Journal of Experimental Botany, 62(14), 4927-4941. 404 Butterworth, P. J., Warren, F. J., & Ellis, P. R. (2011). Human α‐amylase and starch digestion: An 405 interesting marriage. Starch‐Stärke, 63(7), 395-405. 406 Capron, I., Robert, P., Colonna, P., Brogly, M., & Planchot, V. (2007). Starch in rubbery and glassy 407 states by FTIR spectroscopy. Carbohydrate Polymers, 68(2), 249-259. 408 Cooke, D., & Gidley, M. J. (1992). Loss of crystalline and molecular order during starch gelatinisation: 409 origin of the enthalpic transition. Carbohydrate Research, 227, 103-112. 410 Donovan, J. W. (1979). Phase transitions of the starch–water system. Biopolymers, 18(2), 263-275. 411 Doutch, J., & Gilbert, E. P. (2013). Characterisation of large scale structures in starch granules via 412 small-angle neutron and X-ray scattering. Carbohydrate Polymers, 91(1), 444-451. 413 Flanagan, B. M., Gidley, M. J., & Warren, F. J. (2015). Rapid quantification of starch molecular order 414 through multivariate modelling of 13C CP/MAS NMR spectra. Chemical Communications, 2015, 51, 415 14856-14858. 416 Gidley, M. (1992). Structural order in starch granules and its loss during gelatinisation. Gums and 417 stabilisers for the food industry, 6, 87-92. 418 Gidley, M. J., & Bociek, S. M. (1985). Molecular organization in starches: a carbon 13CP/MAS NMR 419 study. Journal of the American Chemical Society, 107(24), 7040-7044. 420 Gidley, M. J., Hanashiro, I., Hani, N. M., Hill, S. E., Huber, A., Jane, J.-L., Liu, Q., Morris, G. A., Rolland-421 Sabaté, A., & Striegel, A. M. (2010). Reliable measurements of the size distributions of starch 422 molecules in solution: Current dilemmas and recommendations. Carbohydrate Polymers, 79(2), 255-423 261. 424 Higgins, J. E., Kosar‐Hashemi, B., Li, Z., Howitt, C. A., Larroque, O., Flanagan, B., Morell, M. K., & 425 Rahman, S. (2013). Characterization of starch phosphorylases in barley grains. Journal of the Science 426 of Food and Agriculture, 93(9), 2137-2145. 427 Iizuka, K., & Aishima, T. (1999). Starch Gelation Process Observed by FT‐IR/ATR Spectrometry with 428 Multivariate Data Analysis. Journal of food science, 64(4), 653-658. 429 John, R. P., Anisha, G., Nampoothiri, K. M., & Pandey, A. (2011). Micro and macroalgal biomass: a 430 renewable source for bioethanol. Bioresource Technology, 102(1), 186-193. 431 Knutson, C. A. (2000). Evaluation of variations in amylose-iodine absorbance spectra. Carbohydrate 432 Polymers, 42(1), 65-72. 433 Knutson, C. A. (1986). A simplified colorimetric procedure for determination of amylose in maize 434 starches. Cereal Chemistry, 63(2), 89-92. 435 Knutson, C. A. and M. J. Grove (1994). Rapid method for estimation of amylose in maize starches. 436 Cereal Chemistry, 71(5), 469-471.Liu, H., Lelievre, J., & Ayoung-Chee, W. (1991). A study of starch 437
Page 24 of 26
Accep
ted
Man
uscr
ipt
gelatinization using differential scanning calorimetry, X-ray, and birefringence measurements. 438 Carbohydrate Research, 210, 79-87. 439 Lopez-Rubio, A., Flanagan, B. M., Shrestha, A. K., Gidley, M. J., & Gilbert, E. P. (2008). Molecular 440 rearrangement of starch during in vitro digestion: toward a better understanding of enzyme 441 resistant starch formation in processed starches. Biomacromolecules, 9(7), 1951-1958. 442 Lopez‐Rubio, A., Flanagan, B. M., Gilbert, E. P., & Gidley, M. J. (2008). A novel approach for 443 calculating starch crystallinity and its correlation with double helix content: A combined XRD and 444 NMR study. Biopolymers, 89(9), 761-768. 445 Nara, S., Mori, A., & Komiya, T. (1978). Study on relative crystallinity of moist potato starch. 446 Starch‐Stärke, 30(4), 111-114. 447 Perry, P., & Donald, A. (2002). The effect of sugars on the gelatinisation of starch. Carbohydrate 448 Polymers, 49(2), 155-165. 449 Regina, A., Blazek, J., Gilbert, E., Flanagan, B. M., Gidley, M. J., Cavanagh, C., Ral, J.-P., Larroque, O., 450 Bird, A. R., & Li, Z. (2012). Differential effects of genetically distinct mechanisms of elevating amylose 451 on barley starch characteristics. Carbohydrate Polymers, 89(3), 979-991. 452 Rubens, P., Snauwaert, J., Heremans, K., & Stute, R. (1999). In situ observation of pressure-induced 453 gelation of starches studied with FTIR in the diamond anvil cell. Carbohydrate Polymers, 39(3), 231-454 235. 455 Sevenou, O., Hill, S., Farhat, I., & Mitchell, J. (2002). Organisation of the external region of the starch 456 granule as determined by infrared spectroscopy. International journal of biological macromolecules, 457 31(1), 79-85. 458 Shen, L., Worrell, E., & Patel, M. (2010). Present and future development in plastics from biomass. 459 Biofuels, Bioproducts and Biorefining, 4(1), 25-40. 460 Smits, A. L., Ruhnau, F. C., Vliegenthart, J. F., & van Soest, J. J. (1998). Ageing of starch based systems 461 as observed with FT-IR and solid state NMR spectroscopy. Starch-Starke, 50(11), 478-483. 462 Sterling, C. (1960). Crystallinity of potato starch. Starch/Stärke, 12, 182-185. 463 Tan, I., Flanagan, B. M., Halley, P. J., Whittaker, A. K., & Gidley, M. J. (2007). A method for estimating 464 the nature and relative proportions of amorphous, single, and double-helical components in starch 465 granules by 13C CP/MAS NMR. Biomacromolecules, 8(3), 885-891. 466 Tester, R. F., & Debon, S. J. (2000). Annealing of starch—a review. International journal of biological 467 macromolecules, 27(1), 1-12. 468 Van Soest, J. J., & Vliegenthart, J. F. (1997). Crystallinity in starch plastics: consequences for material 469 properties. Trends in biotechnology, 15(6), 208-213. 470 Waigh, T. A., Gidley, M. J., Komanshek, B. U., & Donald, A. M. (2000). The phase transformations in 471 starch during gelatinisation: a liquid crystalline approach. Carbohydrate Research, 328(2), 165-176. 472 Waigh, T. A., Kato, K. L., Donald, A. M., Gidley, M. J., Clarke, C. J., & Riekel, C. (2000). Side‐Chain 473 Liquid‐Crystalline Model for Starch. Starch‐Stärke, 52(12), 450-460. 474 Waigh, T. A., Perry, P., Riekel, C., Gidley, M. J., & Donald, A. M. (1998). Chiral side-chain liquid-475 crystalline polymeric properties of starch. Macromolecules, 31(22), 7980-7984. 476 Wang, T. L., Bogracheva, T. Y., & Hedley, C. L. (1998). Starch: as simple as A, B, C? Journal of 477 Experimental Botany, 49(320), 481-502. 478 Warren, F. J., Perston, B. B., Royall, P. G., Butterworth, P. J., & Ellis, P. R. (2013). Infrared 479 spectroscopy with heated attenuated total internal reflectance enabling precise measurement of 480 thermally induced transitions in complex biological polymers. Analytical chemistry, 85(8), 3999-4006. 481 Warren, F. J., Royall, P. G., Gaisford, S., Butterworth, P. J., & Ellis, P. R. (2011). Binding interactions of 482 α-amylase with starch granules: The influence of supramolecular structure and surface area. 483 Carbohydrate Polymers, 86(2), 1038-1047. 484 Wilson, R. H., Kalichevsky, M. T., Ring, S. G., & Belton, P. S. (1987). A Fourier-transform infrared study 485 of the gelation and retrogradation of waxy-maize starch. Carbohydrate Research, 166(1), 162-165. 486
Page 25 of 26
Accep
ted
Man
uscr
ipt
Zhang, B., Dhital, S., Flanagan, B. M., Luckman, P., Halley, P. J., & Gidley, M. J. (2015). Extrusion 487 induced low-order starch matrices: Enzymic hydrolysis and structure. Carbohydrate Polymers, 134, 488 485-496. 489 Zhang, B., Wang, K., Hasjim, J., Li, E., Flanagan, B. M., Gidley, M. J., & Dhital, S. (2014). Freeze-drying 490 changes the structure and digestibility of B-polymorphic starches. Journal of Agricultural and Food 491 Chemistry, 62(7), 1482-1491. 492
493
494
495
Page 26 of 26
Accep
ted
Man
uscr
ipt
Figure legends 496
Figure 1. Overlay SNV normalised FTIR-ATR spectra for all the starch samples analysed in 497
dry (A) and hydrated (B) states. The spectra have been zoomed to show the fingerprint region 498
(full spectra inset). 499
Figure 2. Principal component loadings for the PCA analysis of complete sample sets for dry 500
and hydrated starch FTIR-ATR spectra. A and B hydrated starch, PC1 and PC2 respectively; 501
C and D dry starch, PC1 and PC2 respectively. 502
Figure 3. Heat map plots of the SNV normalised FTIR-ATR spectra for the starch samples 503
used in this study. The spectra have been plotted relative to DSC enthalpy (y-axis) and 504
absorbance intensity is shown in colours, with “hot” colours (reds and yellows) indicating 505
high absorbance and “cool” colours (blues and greens) indicating low absorbance. A; 506
hydrated starch samples, B; dry starch samples. 507
Figure 4. Bi-plot showing the first two PC’s for a PLS model of the hydrated FTIR-ATR 508
spectra. Samples with a DSC enthalpy below 10 J/g have been marked with circles. 509