Influence of non-fat particulate network on fat bloom development in a model chocolate

Accepted Manuscript Influence of non-fat particulate network on fat bloom development in a model chocolate Huanhuan Zhao, Gokhan Bingol, Bryony J. James PII:

S0260-8774(18)30012-8

DOI:

10.1016/j.jfoodeng.2018.01.006

Reference:

JFOE 9139

To appear in:

Journal of Food Engineering

Received Date: 6 November 2017 Revised Date:

8 January 2018

Accepted Date: 9 January 2018

Please cite this article as: Zhao, H., Bingol, G., James, B.J., Influence of non-fat particulate network on fat bloom development in a model chocolate, Journal of Food Engineering (2018), doi: 10.1016/ j.jfoodeng.2018.01.006. 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 proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1

Influence of non-fat particulate network on fat bloom development in a

2

model chocolate

3

Huanhuan Zhao a, Gokhan Bingol a, and Bryony J. James a

4

a

Department of Chemical and Materials Engineering, University of Auckland, Auckland 1142, New Zealand

5

RI PT

6 Abstract: Particle size effect on fat bloom formation in chocolate was studied. Model chocolates with

8

68% (w/w) sand particles of varying particle size distributions (D90 of 18.7, 29.3, 41.4 and 51.1µm)

9

were produced and stored under cycling temperatures of 20°C (7 hours) and 29°C (17 hours). Fat

10

bloom formation was evaluated by the change of surface whiteness during 60 days’ storage, whilst the

11

associated changes of melting temperatures were determined using Differential Scanning Calorimetry.

12

The melting temperatures increased significantly (P<0.05) from 31.5 to 35.7°C with storage time due

13

to polymorphic transformation of fat; the increase in onset temperature indicated reduced liquid fat

14

content which decreased the rate of bloom formation from 9 days of storage. Image analysis showed

15

that smaller particle sizes had higher particle packing density, decreasing the radius of inter-particle

16

channels for fat movement. Therefore, the samples with smaller particle size were more resistant to fat

17

migration and had less changes of surface whiteness, thus bloom rate was slower.

M AN U

SC

7

18

EP

TE D

Keywords: Plain chocolate; Fat bloom; Particle size effect; Particle packing density

AC C

19

1

ACCEPTED MANUSCRIPT 20

1. Introduction

21 Plain (dark) chocolate is preferred by many consumers due to its flavour and, more recently, due to

23

the perceived health benefits of antioxidants. It has a shelf-life of more than one year when stored at a

24

temperature below 18°C (Cebula & Ziegleder, 1993). However, during storage fat bloom, a whitish

25

haze on the surface, may render the chocolate unacceptable for consumption due to the unappealing

26

appearance caused by the formation of flake- or needle-like fat crystals (≥5µm) (Lonchampt & Hartel,

27

2004; Rousseau, 2007). The process of bloom formation can be accelerated when chocolate or

28

chocolate products are stored under elevated and/or cycled temperatures (Ali, Selamat, Che Man, &

29

Suria, 2001; Ghosh, Ziegler, & Anantheswaran, 2002; Rousseau, 2007).

RI PT

22

There are many factors influencing bloom formation in chocolate products: 1) tempering methods

31

(De Graef et al., 2005; Lonchampt & Hartel, 2004; Svanberg, Ahrné, Lorén, & Windhab, 2011), 2)

32

composition (Liang & Hartel, 2004; Sonwai & Rousseau, 2010), 3) storage temperature (Ali et al.,

33

2001; Jin & Hartel, 2015) and 4) chocolate microstructure (Delbaere, Van de Walle, Depypere,

34

Gellynck, & Dewettinck, 2016; Rousseau & Sonwai, 2008). However, during commercial production

35

chocolate is normally properly tempered and is therefore relatively stable at optimum storage

36

temperature of 12-18°C (Afoakwa, Paterson, Fowler & Vieira, 2008a; Stauffer, 2017). The first two

37

factors can be well controlled during production and the third during storage. Particle size and

38

packing, as revealed by for example the D90 value of non-fat particles coupled with particle volume

39

ratio and morphology (Beckett, 2011; Delbaere et al., 2016), is harder to control and less well

40

understood. Therefore, a better understanding of the microstructure of chocolate, which is affected by

41

particle size distribution (PSD), plays a very significant role in control of fat bloom formation in plain

42

chocolate (Aguilera, 2005; Rousseau & Smith, 2008).

TE D

M AN U

SC

30

In the last two decades, many studies (Afoakwa, Paterson, Fowler & Vieira, 2009a; Aguilera, 2005;

44

Aguilera, Michel, & Mayor, 2004) have emphasised the importance of relating fat bloom

45

development with the particulate network of chocolate. However, almost all of these studies have

46

been carried out on the effect of PSD on bloom formation in filled chocolates, in which the plain or

47

model chocolate is in direct contact with fat-based substances, e.g. peanut butter (Choi, McCarthy,

48

McCarthy, & Kim, 2007; Dahlenborg, Millqvist-Fureby, & Bergenståhl, 2015a, 2015b); it has been

49

shown that the PSD of non-fat solids in chocolate significantly affects the migration rate of

50

triglycerides from the fat-based substances into the chocolate (Choi et al., 2007; Dahlenborg et al.,

51

2015a). However, for plain chocolate in isolation from fillings, thus far and to the best of our

52

knowledge, only two studies have been carried out to reveal the relationship between PSD and fat

53

bloom. Altimiras et al. (2007) showed that for sand-based chocolate, bloom formation rate was

54

inversely correlated to the mean particle size (D50) ranging from 5 to 120µm, and the authors reasoned

55

that when chocolate was partially melted Brownian displacement of small sand particles promoted

56

cocoa butter movement. However, on the other hand, for under-tempered dark chocolate with D90

AC C

EP

43

2

ACCEPTED MANUSCRIPT 57

value between 18 and 50µm, Afoakwa et al. (2009b) reported contrary findings regarding the effect of

58

particle size on fat bloom rate and they attributed the higher bloom rate in chocolate with larger

59

particle size to the larger inter-particle pathways for capillary migration. Although for filled chocolate products there is ample evidence from literature that the PSD affects

61

fat bloom formation, for plain chocolate, the number of studies in literature is scarce and furthermore

62

results are contradictory. Therefore, for the particle size range mainly used in the confectionery

63

industry a clear understanding of the particle size effect on bloom formation is crucial to develop

64

more bloom-resistant plain chocolates. The objective of this study was thus to investigate the

65

resistance offered by different particulate networks to bloom formation using a well-controlled model

66

system.

RI PT

60

SC

67 68 69

2. Materials and Methods

71

M AN U

70 2.1 Materials

Model chocolate was made of 32% (w/w) black sand particles (Bethells Beach, New Zealand) and

73

68% (w/w) refined cocoa butter (PureNature, Pure Ingredients Ltd, New Zealand). The black sand

74

was used to model the non-fat solid particles in real chocolate since it offered high visual contrast

75

with the fat phase, therefore enabling accurate image analysis. Black sand was firstly washed and

76

dried in a convection oven at 120°C for 24 hours, and then was ground in a ball mill at different

77

milling speeds and times to obtain sand particles of varying size distributions. In our preliminary

78

experiments, when using sand particles with D90 (90% smaller than the size) of 107.8µm for sample

79

preparation significant phase separation was observed prior to solidification, such that a cocoa butter

80

layer was separated from the mixture and migrated to the top surface, whereas sand particles settled at

81

the bottom. As a result, particles with smaller D90 values of 18.7, 29.3, 41.4 and 51.1µm were

82

prepared to minimise phase separation. The D90 was used to represent the size of sand particles in this

83

study since it is a key parameter to describe non-fat particles in chocolate industry (Afoakwa, 2010;

84

Beckett, 2011).

86

EP

AC C

85

TE D

72

2.2 Sample preparation

87

The sample (sand-based chocolate) making method was adapted from Altimiras et al.’s (2007) with

88

some minor modifications. Cocoa butter was held at 60°C in a water bath (WB-11, WiseBath, South

89

Korea) for 1 hour to completely melt it. Sand particles with the above-mentioned D90 values were

90

added and premixed by a mechanical stirrer (RW 20 digital, IKA-works Inc. NC, USA) for 30 min.

91

The mixture was then continuously mixed in a temper machine (Revolation 2, ChocoVision, USA) at

92

42.2°C for 30 min to ensure sufficient coating of sand particles with the cocoa butter. The mixture

93

was then cooled down in approximately 10 min to 24°C in a water bath at 21°C with continuous 3

ACCEPTED MANUSCRIPT 94

mixing, using a mechanical stirrer, to achieve a high viscosity prior to moulding. The mixture was

95

cast and solidified in shallow cylindrical moulds (Diameter = 45mm, Height = 12mm) in a fridge at

96

2°C for 15 hours; quick solidification at 2°C was used to avoid sedimentation of sand particles. After

97

solidification, all disc-shaped samples were stabilised at ambient temperature (20±0.5°C, relative

98

humidity <50%) for 6 hours. Each sample was then demoulded and placed mould side up since bloom

99

formation on the mould-side surface was used for DSC, VP-SEM and whiteness index measurements. Several discs of each type of model chocolates were run in parallel for different measurements.

RI PT

100 101 102

2.3 Quantification of particle size distribution

A laser diffractometer with a Hydro SM dispersion unit (Mastersizer 2000, Malvern Instruments

104

Ltd., Malvern, U.K.) was used. Approximately 0.1g of sand particles was dispersed in water until

105

achieving an obscuration of 20%. The refractive index of sand particles was 2.42. The mixing speed

106

was set to 2500 rpm and the dispersion was exposed to ultrasound treatment for 2 min before each

107

measurement. Particle volume diameter distribution and D90 values were obtained using the built-in

108

software (Mastersizer 2000, version 5.54).

109 110

2.4 Bloom induction

M AN U

SC

103

An elevated storage temperature within the temperature range of 18 to 30oC was recommended by

112

Lonchampt and Hartel (2004) to accelerate the bloom process without overheating the samples.

113

Therefore, after the initial measurements of DSC and surface whiteness, all disc samples were stored

114

in a dry incubator (Lab Armor Bead Bath 20L, Shel Lab, Australia) at 29±0.5°C for approximately 17

115

hours. Each day, the discs were transferred from the incubator to ambient temperature (20±0.5°C) for

116

7 hours to cycle the temperature (Jin & Hartel, 2015). Therefore, each day the discs experienced a

117

temperature cycle of 20°C for 7 hours and 29°C for 17 hours.

EP

118

TE D

111

2.5 Quantification of microstructure

120

2.5.1 Image acquisition

AC C

119

121

A piece of sample with a flat surface was cut from the mould side of sample disc using a scalpel.

122

Images were obtained using a variable pressure scanning electron microscope (VP-SEM) (Quanta 200

123

FEG SEM, FEI Company, Eindhoven, Holland). To avoid thermal damage, the sample was mounted

124

on a peltier-cooled stage with the temperature set to -5°C (James & Smith, 2009), and nitrous oxide

125

was used as an imaging gas with a pressure of 50.7 Pa. The distribution of sand particles in each

126

chocolate was imaged using a solid state backscatter detector and an accelerating voltage of 20 kV.

127 128

2.5.2 Image processing

129

ImageJ software (Schneider, Rasband, & Eliceiri, 2012) was employed to separate sand particles

130

from the VP-SEM images. Sand particles with areas smaller than 0.02µm2 were eliminated to remove 4

ACCEPTED MANUSCRIPT redundant pixels due to haze at particle edges. The cross-sectional area and the perimeter of individual

132

particles and the total number of particles were obtained. Thus, in a measured image area (A), the

133

total cross-sectional area of particles (Atotal), particle area fraction (AA) which is the ratio of Atotal to A,

134

the total cross-sectional perimeter of particles (Ptotal) and the number of particles per unit image area

135

(NA) were calculated to be able to accurately define the microstructure of the model chocolate.

136

Moreover, since a porous medium could be considered as an assembly of many tiny capillaries

137

(Chikao, 2006), the void space in between the sand particles, i.e. the space of fat matrix, could be

138

assumed as equal-sized capillaries, whose hydraulic radius (RH) was calculated using Eq.1 (Masoodi

139

& Pillai, 2012).

RH = 2

RI PT

131

A − Atotal Ptotal

141

SC

140

(1)

2.6 Colour analysis

To determine the colour changes on the surface of the chocolate a machine vision system,

143

developed by Luzuriaga et al. (1997), and the “two image” method, developed by Alçiçek and

144

Balaban (2012), were used. Two images were taken using the front- and back-lighting methods for

145

each measurement. The mould side surface of each disc was measured three times by horizontally

146

rotating around 30º between measurements as suggested by Lonchampt and Hartel (2006) to minimise

147

errors caused by diffraction of light due to uneven surface. The images were analysed using image

148

analysis software (LensEye, Gainesville, FL, USA). L* a* b* values were determined, and WI values

149

and the normalised values of WI were calculated using Eqs. 2 and 3, respectively.

TE D

M AN U

142

2 W I = 100 −  (100 − L * ) + a * 2 + b * 2   

0.5

 WI t − WI 0   × 100  WI 0 

Percentage normalised WI = 

152

where subscripts 0 and t denotes the initial and the WI value at time t, respectively.

EP

151

(3)

2.7 DSC measurements

AC C

150

(2)

153

The melting behaviour of model chocolate discs was analysed using Differential Scanning

154

Calorimetry (Shimadzu DSC-60, Columbia, MD, USA). Three samples, each about 5 mg, were cut

155

from the mould side surface of each disc, which was kept for a total of 50 days for measurements at

156

different time intervals. The sample was put in an aluminium pan and then was heated from 10 to

157

40°C at a rate of 2°C/min in a helium stream. The melting curve and the peak (Tpeak), onset (Tonset) and

158

final melting temperature (Tendset) values were obtained (TA-60WS Thermal Analyser).

159 160

2.8 Statistical analysis

161

The effect of storage time and sand particle size (D90) on melting temperatures were analysed by

162

two-way ANOVA at a significance level of 95% using SPSS (IBM SPSS Statistics for Windows, 5

ACCEPTED MANUSCRIPT 163

version 21.0, IBM Corp., Armonk, NY, USA). One-way ANOVA with a Tukey post-hoc test (α=0.05)

164

was used to examine the effect of particle size (D90) on AA, NA and RH values using the SPSS software.

165 166 167

3. Results and Discussion

168

3.1 Particle size distribution The D90 values of the four batches of sand particles used in this study, 18.7, 29.3, 41.4 and 51.1µm,

170

are in line with Afoakwa et al. (2009b) who used a particle size range of around 18 to 50µm (D90) for

171

non-fat particles in dark chocolate. As expected, and shown in Fig. 1, as the D90 value increased the

172

ratio of large particles and the range of the distribution increased. It is also seen that the PSD curves

173

of all sand particles were generally unimodal. However, Afoakwa et al. (2009b) not only observed

174

unimodal distributions but also bimodal and multimodal distributions. The modality of the distribution

175

gives a significant insight into the microstructure of chocolate, because the proportions of different

176

particle sizes could have a significant effect on fat bloom, since the percolation of fat through the

177

inter-particle channels, which are correlated with particle size, will be different. Therefore, when

178

making comparisons between different D90 values, working with unimodal curves could yield more

179

accurate results in terms of bloom formation.

M AN U

SC

RI PT

169

PSD of non-fat particles in the range of 0.3 to 100µm is commonly found in conventional

181

chocolates (Afoakwa, Paterson, Fowler, & Vieira, 2008b; Ziegler & Hogg, 2017). In our preliminary

182

experiment, the PSD of 107.8µm (D90) sand particles, wherein the percentage of particles larger than

183

100µm was 21.7%, caused significant phase separation. However, in this study the PSD of the

184

51.1µm (D90) sand particles contained 3.8% particles larger than 100µm (Fig. 1). Therefore, due to

185

the considerably smaller percentage of particles larger than 100µm, it is reasonable to assume that

186

phase separation caused by large particles will be negligible. Furthermore, it is clearly seen from Fig.

187

2 that the two histograms of sand particles on mould and non-mould sides of the 51.1µm samples

188

almost overlapped, which therefore indicates that a similar PSD existed on both sides of the chocolate

189

surface. As a result, it could be safely assumed that sand particles were homogenously distributed

190

within the samples, with no gravity driven separation, and the quantification of microstructure

191

represented all locations in the samples accurately.

193

EP

AC C

192

TE D

180

3.2 Effect of microstructure on bloom formation

194

Particle packing ability has been found to affect the resistance against liquid movement in a

195

particulate network (Chikao, 2006; Masoodi & Pillai, 2012) and is also an important factor

196

influencing fat migration in chocolate (Aguilera et al., 2004). Thus, using image processing, we have

197

characterised particle packing density (PPD) using parameters such as NA, AA and RH, which are

198

presented in Table 1. It is seen from Table 1 that as the D90 values increased, RH values increased

199

significantly (p<0.05) whereas the NA and AA values decreased significantly (p<0.05), except the 6

ACCEPTED MANUSCRIPT difference of AA values between 18.7 and 29.3µm, which suggested a decrease in PPD. The decrease

201

in PPD can also be visually observed from processed VP-SEM images, which illustrates sand particle

202

spatial distributions, as shown in Fig. 3 (B2, C2 and D2). This finding is consistent with Afoakwa et

203

al.’s (2009b) observations of the sugar network in molten dark chocolate, where chocolate containing

204

larger-sized sugar particles had a lower PPD. A lower PPD could increase the possibility of fat bloom

205

formation since there will be significantly larger void spaces, the significant (p<0.05) increase in RH,

206

between sand particles for the flow of liquid fat which will eventually result in a higher bloom rate.

RI PT

200

It should be noted that since the shape of sand particles was irregular, similar to that of sugar

208

crystals (grains of crystal-like shapes) in conventional chocolates, but different from cocoa particles

209

or milk powders which are porous and have more complex surface structures, the projected areas of

210

sand particles varied considerably based on orientation. However, since more than 13000 randomly

211

orientated sand particles were taken into account for the calculations, it can be expected that the

212

results presented in Table 1 are reliably representative.

214

M AN U

213

SC

207

3.3 The changes of melting temperatures during bloom formation

The Tonset, Tpeak and Tendset values, which represent different points of melting of the fat in model

216

chocolates, significantly (p<0.05) increased with the storage time, but did not significantly (p>0.05)

217

change with the particle size (D90); in addition, the effect of interaction of storage time and particle

218

size on the melting temperatures was not significant (p>0.05). This suggests that the melting

219

temperatures were primarily storage time-dependent, and thus Fig. 4 only presents DSC thermograms

220

of the 18.7µm (D90) samples during 50 days of storage.

TE D

215

The shift of Tpeak values to higher temperatures suggests the formation of relatively more stable fat

222

crystals. Cocoa butter has six polymorphic forms, I to VI, where I has the lowest and VI has the

223

highest melting point (Wille & Lutton, 1966). It was reported by Huyghebaert et al. (1971) that the

224

polymorphic form V and VI of cocoa butter have Tpeak values in the ranges of 31.3-33.2°C and 33.8-

225

36.0°C, respectively, and in our samples, the increase of Tpeak from 31.5 to 35.7oC indicates the

226

transformation of cocoa butter from form V to VI during the storage period. This type of

227

transformation along with the formation of storage bloom on well-tempered chocolates was also

228

reported by Lonchampt & Hartel (2004).

AC C

EP

221

229

The Tonset value gives information on amount of cocoa butter in the liquid state when the samples

230

were stored at 29°C. Hence, as the Tonset values increased during storage, the samples became more

231

heat resistant in terms of melting of the cocoa butter and consequently the ratio of liquid-solid fat

232

reduced. Since fat migration occurs in the liquid fat portion, a reduction in liquid-solid fat ratio can

233

considerably retard bloom formation, which can be seen in Fig. 5, after approximately 9 days of

234

storage the rate of increase in surface whiteness noticeably decreased.

235

7

ACCEPTED MANUSCRIPT 236

3.4 The changes of surface whiteness during bloom formation

237

Prior to storage, the samples had surface whiteness values (WI0) of approximately 14.5, 15.5, 15.9

238

and 17.0 for D90 of 18.7 28.3, 41.1 and 51.1µm, respectively, which is in line with the findings of

239

Afoakwa et al. (2009b). Therefore, in order to make concrete comparison of the changes of WI values

240

between different D90 values, we have used percentage normalised values of WI to eliminate the

241

initial minor differences of surface whiteness among the samples. The surface whiteness of all samples increased with time during storage as seen in Fig. 5, which

243

indicates the formation of fat bloom when liquid fat migrated and recrystallised onto surface. As we

244

have shown in Fig. 3 (A2-D2) and quantified using image processing, the existence of numerous

245

micro-channels filled with fat can be assumed where this assumption is in line with literature

246

(Aguilera & Mayor, 2004; Loisel, Lecq, Ponchel, Keller, & Ollivon, 1997). Since in our experiments

247

the samples were exposed to a temperature cycle of 20 to 29°C, above the Tonset temperature the cocoa

248

butter partially melted, and therefore the volume expanded; however, below the Tonset temperature it

249

recrystallised, and hence the volume contracted. Due to above-mentioned expansion and contraction

250

cycles a pressure gradient is formed which drives liquid fat from the internal volume to the surface

251

(Bricknell & Hartel, 1998). Therefore, the retardation of the change of surface whiteness after

252

approximately 9 days of cycling, shown in Fig. 5, could be explained by the shift of Tonset values to

253

higher temperatures during the whole period of storage since a higher Tonset value indicates the

254

presence of a lower liquid-solid fat ratio, which consequently lowers the rate of fat bloom formation.

M AN U

SC

RI PT

242

It can also be observed from Fig. 5 that during the first 9 days of storage, when the fat phase was

256

rich in the liquid state, as the D90 values increased the values for normalised surface WI also increased,

257

which indicated a faster rate of bloom formation. It is also seen from Fig 5 that after 9 days of storage

258

the slope of the curves for all D90 values significantly decreased and became almost flat after 21 days

259

of storage, which revealed that no further bloom was occurring. At the end of 21 days of storage the

260

samples with larger D90 values had higher normalised WI values, which could be due to the fact that

261

as the D90 values increased, the micro-channels, wherein liquid fat moved, had larger RH values as

262

presented in Table 1. This is in line with findings of Altimiras et al. (2007) who, by using Lucas-

263

Washburn equation, showed that under capillary effect, which could promote liquid fat movement in

264

plain chocolates (Aguilera & Mayor, 2004), migration of liquid cocoa butter was quicker in pathways

265

with larger diameter. Furthermore, it is also worth emphasising that the samples with larger D90 had

266

lower PPD (as shown in Table 1); therefore, the length of the tortuous pathways were shorter for fat

267

migration and thus offered less resistance to fat movement. As a result, a combination effect of larger

268

radius and lower tortuosity of fat migration pathways led to more rapid and more pronounced bloom

269

formation on the samples with larger D90 values.

AC C

EP

TE D

255

270

The findings of this study confirm and further clarify those of Afoakwa et al. (2009b) where

271

increasing particle size increased the rate of bloom formation. The apparent contradiction between the

272

results observed in this study and that of Altimiras et al (2007) is owing to the considerably large sand 8

ACCEPTED MANUSCRIPT 273

particles (e.g. D50: 120µm) those authors used. As significant phase separation was observed in our

274

preliminary experiment using sand particles with D90 of 107.8µm, sand particles with D50 of 120µm

275

could have led to gravity driven separation, which might have caused a non-homogeneous chocolate

276

system and therefore interfered with the particle size effect.

277 278 4. Conclusion

RI PT

279

The sand-based chocolate system was chosen to simplify the microstructure of multicomponent

281

real chocolate since the sand-based network was easy to be identified. Sand particles with D90 smaller

282

than 51.1µm were recommended to create relatively homogeneous chocolate systems. During storage,

283

the polymorphic transformation of cocoa butter increased the melting temperatures and thus led to the

284

reduction in liquid fat ratio which retarded bloom formation after 9 days of storage. The particle size

285

effect influenced the changes of surface whiteness but had limited impacts on cocoa butter melting

286

temperatures. Variations in particle size distributions affected the PPD of model chocolates and more

287

flocculated particulate networks were found in samples with smaller D90 values. It was also shown

288

that model chocolates with larger D90 values had more pronounced bloom formation because of lower

289

PPD. As the PPD reduced, the inter-particle channels became larger and less torturous, and as a result,

290

more liquid fat migrated to surface showing higher rate and extent of bloom.

M AN U

291

SC

280

Overall it is clear that microstructure of the non-fat solid particulate network contributes to bloom rate by altering the ease with which fat can migrate through inter-paticle channels.

293

microstructure of the solid network has no direct effect on the melting behaviour of the fat phase.

294 295

The

TE D

292

References

297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314

Afoakwa, E. O. (2010). Chocolate science and technology. York UK: Blackwell Publishing. Afoakwa, E. O., Paterson, A., Fowler, M. & Vieira, J. (2008a). Effects of tempering and fat crystallisation behaviour on microstructure, mechanical properties and appearance in dark chocolate systems. Journal of Food Engineering, 89(2), 128-136. Afoakwa, E. O., Paterson, A., Fowler, M., & Vieira, J. (2008b). Particle size distribution and compositional effects on textural properties and appearance of dark chocolates. Journal of Food Engineering, 87(2), 181-190. Afoakwa, E. O., Paterson, A., Fowler, M., & Vieira, J. (2009a). Microstructure and mechanical properties related to particle size distribution and composition in dark chocolate. International Journal of Food Science & Technology, 44(1), 111-119. Afoakwa, E. O., Paterson, A., Fowler, M., & Vieira, J. (2009b). Fat bloom development and structureappearance relationships during storage of under-tempered dark chocolates. Journal of Food Engineering, 91(4), 571-581. Aguilera, J. M. (2005). Why food microstructure? Journal of Food Engineering, 67(1–2), 3-11. Aguilera, J. M., Michel, M., & Mayor, G. (2004). Fat migration in chocolate: Diffusion or capillary flow in a particulate solid? A hypothesis paper. Journal of Food Science, 69(7), R167-R174. Alçiçek, Z., & Balaban, M. Ö. (2012). Development and application of “The Two Image” method for accurate object recognition and color analysis. Journal of Food Engineering, 111(1), 46-51.

AC C

EP

296

9

ACCEPTED MANUSCRIPT

EP

TE D

M AN U

SC

RI PT

Ali, A., Selamat, J., Che Man, Y. B., & Suria, A. M. (2001). Effect of storage temperature on texture, polymorphic structure, bloom formation and sensory attributes of filled dark chocolate. Food Chem, 72(4), 491-497. Altimiras, P., Pyle, L., & Bouchon, P. (2007). Structure–fat migration relationships during storage of cocoa butter model bars: Bloom development and possible mechanisms. Journal of Food Engineering, 80(2), 600-610. Beckett, S. T. (2011). Industrial chocolate manufacture and use: John Wiley & Sons. Bricknell, J., & Hartel, R. W. (1998). Relation of fat bloom in chocolate to polymorphic transition of cocoa butter. Journal of the American Oil Chemists' Society, 75(11), 1609-1615. Cebula, D. J., & Ziegleder, G. (1993). Studies of Bloom Formation Using X-Ray Diffraction from Chocolates after Long-Term Storage. Lipid / Fett, 95(9), 340-343. Chikao, K. (2006). Permeation (flow through porous medium). In H. Masuda, K. Higashitani & H. Yoshida (Eds.), Powder technology handbook (3 ed., pp. 317-324): CRC Press. Choi, Y. J., McCarthy, K. L., McCarthy, M. J., & Kim, M. H. (2007). Oil migration in chocolate. Applied Magnetic Resonance, 32(1-2), 205-220. Dahlenborg, H., Millqvist-Fureby, A., & Bergenståhl, B. (2015a). Effect of particle size in chocolate shell on oil migration and fat bloom development. Journal of Food Engineering, 146, 172181. Dahlenborg, H., Millqvist-Fureby, A., & Bergenståhl, B. (2015b). Effect of shell microstructure on oil migration and fat bloom development in model pralines. Food Structure, 5, 51-65. De Graef, V., Foubert, I., Agache, E., Bernaert, H., Landuyt, A., Vanrolleghem, P. A., & Dewettinck, K. (2005). Prediction of migration fat bloom on chocolate. European Journal of Lipid Science and Technology, 107(5), 297-306. Delbaere, C., Van de Walle, D., Depypere, F., Gellynck, X., & Dewettinck, K. (2016). Relationship between chocolate microstructure, oil migration, and fat bloom in filled chocolates. European Journal of Lipid Science and Technology, 118, 1800-1826. Ghosh, V., Ziegler, G. R., & Anantheswaran, R. C. (2002). Fat, moisture, and ethanol migration through chocolates and confectionary coatings. Critical Reviews in Food Science and Nutrition, 42(6), 583-626. Huyghebaert, A., & Hendrickx, H. (1971). Polymorphism of cocoa butter, shown by differential scanning calorimetry. Lebensm. Wiss. Technol, 4, 59–63. James, B. J., & Smith, B. G. (2009). Surface structure and composition of fresh and bloomed chocolate analysed using X-ray photoelectron spectroscopy, cryo-scanning electron microscopy and environmental scanning electron microscopy. LWT - Food Science and Technology, 42(5), 929-937. Jin, J., & Hartel, R. W. (2015). Accelerated fat bloom in chocolate model systems: Solid fat content and temperature fluctuation frequency. JAOCS, Journal of the American Oil Chemists' Society, 92(10), 1473-1481. Liang, B., & Hartel, R. W. (2004). Effects of milk powders in milk chocolate. Journal of Dairy Science, 87(1), 20-31. Loisel, C., Lecq, G., Ponchel, G., Keller, G., & Ollivon, M. (1997). Fat bloom and chocolate structure studied by mercury porosimetry. Journal of Food Science, 62(4), 781-788. Lonchampt, P., & Hartel, R. W. (2004). Fat bloom in chocolate and compound coatings. European Journal of Lipid Science and Technology, 106(4), 241-274. Lonchampt, P., & Hartel, R. W. (2006). Surface bloom on improperly tempered chocolate. European Journal of Lipid Science and Technology, 108(2), 159-168. Luzuriaga, D. A., Balaban, M. O., & Yeralan, S. (1997). Analysis of visual quality attributes of white shrimp by machine vision. Journal of Food Science, 62(1), 113-118. Masoodi, R., & Pillai, K. M. (2012). Single-phase flow (sharp-interface) models for wicking. In R. Masoodi & K. M. Pillai (Eds.), Wicking in porous materials: Traditional and modern modeling approaches (pp. 97-130). Boca Raton, FL: Taylor & Francis Group. Rousseau, D. (2007). The microstructure of chocolate. In D. J. McClements (Ed.), Understanding and Controlling the Microstructure of Complex Foods (pp. 648-690): Woodhead Publishing. Rousseau, D., & Smith, P. (2008). Microstructure of fat bloom development in plain and filled chocolate confections. Soft Matter, 4(8), 1706.

AC C

315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369

10

ACCEPTED MANUSCRIPT

RI PT

Rousseau, D., & Sonwai, S. (2008). Influence of the dispersed particulate in chocolate on cocoa butter microstructure and fat crystal growth during storage. Food Biophysics, 3(2), 273–278. Schneider, C. A., Rasband, W. S., & Eliceiri, K. W. (2012). NIH Image to ImageJ: 25 years of image analysis. Nat Meth, 9(7), 671-675. Sonwai, S., & Rousseau, D. (2010). Controlling fat bloom formation in chocolate – Impact of milk fat on microstructure and fat phase crystallisation. Food Chem, 119(1), 286-297. Stauffer, M. B. (2017). Quality control and shelf life Beckett's Industrial Chocolate Manufacture and Use (pp. 532-554): John Wiley & Sons, Ltd. Svanberg, L., Ahrné, L., Lorén, N., & Windhab, E. (2011). Effect of sugar, cocoa particles and lecithin on cocoa butter crystallisation in seeded and non-seeded chocolate model systems. Journal of Food Engineering, 104(1), 70-80. . Wille, R. L., & Lutton, E. S. (1966). Polymorphism of cocoa butter. Journal of the American Oil Chemists Society, 43(8), 491-496. Ziegler, G. R., & Hogg, R. (2017). Particle size reduction Beckett's Industrial Chocolate Manufacture and Use (pp. 216-240): John Wiley & Sons, Ltd.

SC

370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385

AC C

EP

TE D

M AN U

386

11

ACCEPTED MANUSCRIPT Fig. 1: Particle size distributions of sand particles with D90 of (a) 18.7µm, (b) 29.3µm, (c) 41.4µm and (d) 51.1µm. Fig. 2: Histograms of particle area distributions for individual sand particles on the mould- and nonmould-side surfaces of the sample with a D90 value of 51.1µm.

18.7µm, (B) 29.3µm, (C) 41.4µm and (D) 51.1µm.

RI PT

Fig. 3: Original (VP-SEM) and processed images of mould-side surfaces for samples with D90 of (A)

Fig. 4: DSC transitional thermograms of samples (D90: 18.7µm) during 50 days of storage. The Tpeak

SC

values were indicated.

Fig. 5: The percentage normalised values of Whiteness Index (WI) for samples with varying D90

AC C

EP

TE D

M AN U

values during 21 days of storage. Points are means ± standard error (n=3).

ACCEPTED MANUSCRIPT Table 1: Characteristic parameters of particulate network determined by the image analysis for samples with varying D90 values. Particle area fraction,

Hydraulic radius

NA (/1000µm2)

AA (-)

RH (µm)

18.7

199േ3 a

25.7േ0.5% a

1.21േ0.01 a

29.3

150േ2 b

26.0േ0.4% a

1.35േ0.03 b

41.4

142േ1 b

24.1േ0.3% b

51.1

116േ3 c

21.6േ0.5% c

RI PT

Particle number count

D90 (µm)

1.46േ0.01 c

1.86േ0.05 d

SC

Values are means ± standard error (n=3); Values in each row with different superscript letters are

AC C

EP

TE D

M AN U

significantly (p < 0.05) different.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

ACCEPTED MANUSCRIPT Highlights:

AC C

EP

TE D

M AN U

SC

RI PT

1. Smaller non-fat particle size resulted in higher particle packing density (PPD). 2. Higher PPD led to a lower rate and extent of bloom. 3. Chocolate with smaller particle size had less extensive bloom formation. 4. The increase of melting temperatures also resulted in a decreased rate of bloom.