Food Research International 42 (2009) 200–209 Contents lists available at ScienceDirect Food Research International journal homepage: www.elsevier.com/locate/foodres Influence of tempering and fat crystallization behaviours on microstructural and melting properties in dark chocolate systems Emmanuel Ohene Afoakwa a,b,*, Alistair Paterson a, Mark Fowler b, Joselio Vieira b a b Centre for Food Quality, SIPBS, University of Strathclyde, Royal College Building, 204 George Street, Glasgow G1 1XW, Scotland, UK Nestlé Product Technology Centre York, P.O. Box 204, Haxby Road, York YO91 1XY, England, UK a r t i c l e i n f o Article history: Received 3 June 2008 Accepted 19 October 2008 Keywords: Differential scanning calorimetry Tempering Fat crystallization Melting properties Particle size distribution Scanning electron microscopy Chocolate a b s t r a c t Particle size distribution (PSD) and temper influences on dark chocolate fat crystallization were studied using differential scanning calorimetry (DSC) and microscopy to establish relationships with their melting properties and microstructure. Variations in PSD had no influence on crystallinity of products at all temper regimes. Particle size (PS) increases had limited effects on Tonset, Tpeak, and DHmelt independent of temper regime but significant decreases in Tend and Tindex were noted. Contrary, varying temper regime influenced the crystallinity and melting properties (Tend, Tindex and DHmelt) of products. Under-tempered chocolate showed widened crystal size distribution (CSD) with significant changes in Tend, Tindex and DHmelt of products. Over-tempering caused moderate increases in CSD and melting properties, with significant effect on Tend, Tindex and DHmelt but no changes were noted in Tonset, Tpeak of products. Fat–sugar melting profiles showed similar levels in all products independent of temper regime, suggesting fat and sugar components are present in similar amounts in under-, over- and optimally-tempered products. Micrographs revealed clear crystalline network structure and well defined inter-crystal networks among tempered and over-tempered samples. Under-tempered products showed re-arrangement and re-crystallization of unstable fat crystals to smaller numbers of larger agglomerates with formation of solid bridges between the crystalline network structures. Attainment of optimal temper regime during precrystallization of dark chocolate is necessary for the achievement of premium quality products and avoidance of defects in structure and melting character. Ó 2008 Elsevier Ltd. All rights reserved. 1. Introduction Chocolate is a multi-component system consisting of cocoa powder and sucrose (total solids of 65–75%) in a continuous fat phase, primarily composed of cocoa butter. During chocolate manufacture, tempering – a technique of controlled pre-crystallization is used to induce a more thermodynamically stable polymorphic form of cocoa butter to effect good product snap, contraction, gloss and shelf-life characteristics. Pre-crystallization of a small proportion of triglycerides forms nuclei for the remaining lipid to set in an appropriate form, through nucleation that is highly dependent on the shear-temperature–time parameter combinations. Important physical and functional characteristics (i.e., texture, snap and gloss) of chocolate products are dictated by the crystal network formed by its constituent lipid during crystallization (Afoakwa, Paterson, & Fowler, 2007; Campos, Narine, & Marangoni, 2002; Marangoni * Corresponding author. Address: Centre for Food Quality, SIPBS, University of Strathclyde, Royal College Building, 204 George Street, Glasgow G1 1XW, Scotland, UK. Tel.: +44 (0) 7984288727. E-mail addresses: [email protected], e.afoakwa@strath. ac.uk (E.O. Afoakwa). 0963-9969/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved. doi:10.1016/j.foodres.2008.10.007 & Narine, 2002). In industrial chocolate manufacture, tempering is crucial, influencing quality characteristics such as colour, hardness, handling, finish and shelf-life characteristics (Afoakwa, Paterson, Fowler, & Vieira, 2008a; Altimiras, Pyle, & Bouchon, 2007; Debaste, Kegelaers, Liégeois, Ben Amor & Halloin, 2008; Herrera & Hartel, 2000; Nelson, 1999; Pérez-Martínez, AlvarezSalas, Charo-Alonso, Dibildox-Alvarado & Toro-Vazquez, 2007; Toro-Vazquez, Pérez-Martinez, Dibildox-Alvarado, Charo-Alonso & Reyes-Hernandez, 2004). Polymorphism – the existence of two or more distinct crystalline forms of the same substance – is a critical concept in the study of fat crystal structure. Cocoa butter crystallizes in different forms as a function of processing condition, time and temperature of storage, as follows: six polymorphic forms (I–VI), have been identified, the principals being a, b and b’. The unstable forms I (m.p. 16– 18 °C) and II (22–24 °C) transform slowly into III (24–26 °C) and IV (26–28 °C), all possessing less stable polymorphs but upon optimal tempering sets in a more stable form V (32–34 °C) polymorph (Beckett, 2000; Loisel, Lecq, Keller, & Ollivon, 1998; Timms, 1984; Willie & Lutton, 1966). Form V, a b polymorph, most desirable and found in optimal-tempered chocolate, gives glossy appearance, good snap, contraction and resistance to bloom (Beckett, 1999). E.O. Afoakwa et al. / Food Research International 42 (2009) 200–209 Form VI, the most stable, is difficult to generate but formed on lengthy chocolate storage, accompanied by fat bloom, melts at 34–36 °C, and has crystals experienced as large and gritty on the tongue. All these polymorphic forms could be formed directly from melted TAGs, or via melt-mediated or solid-state monotropic phase transformations (Marangoni, 2002). Phase transitions in cocoa butter polymorphs from less to more stable are irreversible and dependent on temperature and time. Polymorphism in relation to solid continuous phases of cocoa butter has a large impact on chocolate quality, dictating their structural properties (Schenk & Peschar, 2004). Structural factors such as microstructural elements and microstructure characteristics can provide quantitative information about the mechanical properties of the network, and therefore information about the sensory hardness of the network (Narine & Marangoni, 1999a). Polymorphic changes can be observed as overall contraction of chocolate, appearance, or undesirable fat bloom formation dependent on relative stabilities of crystal forms and temperature. Fat crystallization is a complex process influenced by processing conditions that determines chocolate microstructure and physical properties (Bell, Gordon, Jirasubkunakorn, & Smith, 2007; Campos et al., 2002; Himawan, Starov, & Stapley, 2006; Pérez-Martínez et al., 2007). Hartel (2001) noted that the control of crystallization is critical for texture, melting properties and other quality characteristics. Melting profiles of chocolates have been studied using pulsed nuclear magnetic resonance (pNMR) and differential scanning calorimetry (DSC) (Smith, Cain, & Talbot, 2007; Tabouret, 1987; Walter & Cornillon,2001, 2002). As information on cocoa butter isothermal phase behaviour during chocolate manufacture is important for optimizing production processes that maintain product quality (Foubert, Dewettinck, & Vanrolleghem, 2003; Marangoni & McGauley, 2003), this work would provide vital information on the effects of changes in processing conditions on the microstructure and melting properties of fat systems during industrial chocolate manufacture. Additionally, it would provide valuable quality control indicators to ensure the structure and melting properties of fat networks being produced on a production line during dark chocolate manufacture are consistent and would yield the desired mechanical and sensory qualities during post-processing handling, supply-chain management and consumption. The objective of this study was to investigate the influences of tempering and fat crystallization behaviours on melting properties and crystallized network microstructure in dark chocolates varying in particle size distribution. 2. Materials and methods 2.1. Materials Cocoa liquor of Central West African Origin was obtained from Cargill Cocoa Processing Company (York, UK); sucrose (pure extra fine granulated) from British Sugar Company (Peterborough, UK); pure prime pressed cocoa butter and soy lecithin from ADM Cocoa Limited (Koog aan de Zaan, Netherlands) and Unitechem Company Ltd. (Tianjin, China), respectively. The recipe, formulation and production of samples were as described previously (Afoakwa, Paterson, & Fowler, 2008). Chocolates were formulated with total fat of 35% (w/w) from cocoa liquor and cocoa butter. Experimental samples (5 kg batch for each formulation) were produced by mixing sucrose and cocoa liquor in a Crypto Peerless Mixer (Model K175, Crypto Peerless Ltd, Birmingham, UK) at low speed for 2 min and then at high for 3 min, then using a 3-roll refiner (Model SDX 600, Buhler Ltd., CH-9240 Uzwil, Switzerland) to a specified particle size (D90 [90% finer than this size]: 18 ± 1 lm, 25 ± 1 lm, 35 ± 1 lm and 50 ± 1 lm) conducting 201 particle size analysis, during refining, to ensure D90 values. The refined chocolates were melted at 50–55 °C for 24 h and the chocolate mass conched in a Lipp Conche (Model IMC-E10, Boveristr 40-42, D-68309, Mannhein, Germany) at low speed for 3.5 h at 60 °C. Lecithin and cocoa butter were added and mixed then conched at high speed for 30 min to effect adequate mixing and liquefaction. Samples were kept in sealed plastic containers under ambient temperature (20–22 °C) and moisture and fat contents determined using Karl Fischer and Soxhlet methods (ICA, 1988, 1990). 2.2. Determination of particle size distribution A MasterSizerÒ Laser Diffraction Particle Size Analyzer equipped with MS 15 Sample Presentation Unit (refractive index 1.590) (Malvern Instrument Ltd., Malvern, England) was used. About 0.2 g of refined dark chocolate was dispersed in vegetable oil (Refractive index 1.450) at ambient temperature (20 ± 2 °C) until an obscuration of 0.2 was obtained. The sample was placed under ultrasonic dispersion for 2 min to ensure particles were independently dispersed and thereafter maintained by stirring during the measurement. Size distribution was quantified as the relative volume of particles in size bands presented as size distribution curves (Malvern MasterSizerÒ Micro Software v 2.19). PSD parameters obtained included specific surface area, largest particle size (D90), mean particle volume (D50), smallest particle size (D10) and Sauter mean diameter (D[3,2]). 2.3. Tempering experiment Samples were incubated at 50 °C for 4 h for melting and tempered using Aasted Mikrovert laboratory continuous three-stage tempering unit (Model AMK 10, Aasted Mikroverk A/S, Farum, Denmark). Chocolate was pumped through the multi-stage units and a worm screw drove the product through the heat exchangers. Sensors located at specific points in the equipment measured the temperature of both the chocolate and the coolant fluid at each stage. Based on our earlier work modelling temperature controls to study tempering behaviour (Afoakwa, Paterson, Fowler, & Vieira, 2008b), the temperature of each of the coolant fluids (Zones 1:2:3) were thus set as 26:24:32 °C, 21:19:32 °C and 18:16:32 °C, respectively, for attaining the under-tempered, optimally-tempered and over-tempered regimes. The degree of pre-crystallization was measured using a computerized tempermeter (Exotherm 7400, Systech Analytics, Neuchâtel, Switzerland) and a built-in algorithm provided the tempering curves and temper readings in chocolate temper index (Slope), corresponding to optimal temper (Slope 0), under-temper (Slope 1.0) and over-temper regimes (Slope 1.0). The principle of this method has been described by Nelson (1999). Chocolate from the three regimes were moulded using plastic moulds: 80 mm length; 20 mm breadth; and 8 mm height. The final products were allowed to cool in a refrigerator (10 ± 2 °C) for 2 h before de-moulding onto plastic trays and conditioned at 20 ± 2 °C for 14 days before analysis. Triplicate measurements were taken for each product composition and the mean values were recorded. 2.4. Determination of melting properties Differential scanning calorimeter (DSC Series 7, Perkin Elmer Pyris, Norwalk, CT, USA) equipped with a thermal analysis data station was calibrated using indium and octadecane at a scan rate of 5 °C/min using an aluminium pan as reference. Samples (5 mg) were loaded into 40 ll capacity pans with holes, which were sealed with lids using a sample press. Pans were heated at 5 °C/min from 15–55 °C in a N2 stream. Onset temperature (Tonset), peak temperature (Tpeak), end temperature (Tend) and enthalpy of melting 202 E.O. Afoakwa et al. / Food Research International 42 (2009) 200–209 (DHmelt) were calculated automatically by the software. Melting index (Tindex) was computed as (Tend  Tonset), and measures duration of melting as described by Vasanthan and Bhatty (1996) and Afoakwa, Paterson, Fowler, and Vieira (2008c). Each sample was analysed in triplicate and mean values and standard deviations reported. Thermal behaviour of fat and sugar components in samples from the different temper regimes were analysed using DSC. Pans containing 5 mg were heated at 10 °C/min from 15 to 200 °C in a N2 stream and melting profiles of the fat and sugar calculated by the software. To calculate the DHmelt sugar/DHmelt fat ratio, the melting enthalpy of the sugar was divided by the melting enthalpy of the fat peak, technique reported to provide information on the possible structural changes in the fat and/or sugar components in bloomed chocolates (Lonchampt & Hartel, 2006). Triplicate analyses were conducted and mean value and standard deviation reported. 3. Results and discussion 3.1. Particle size distribution The volume histograms (Fig. 1) show narrow (18 lm PS) and wide (25 lm PS) bimodal and narrow (35 lm PS), and wide (50 lm PS) multimodal size distributions. This PSD range 18– 50 lm using D90 values (>90% finer) covers optimum minimum and maximum sizes ranging, respectively, from very fine to coarse particles as detected by micrometer screw gauge (Ziegler & Hogg, 1999), and have direct effects on rheology, texture and sensory character of chocolates in manufacture (Afoakwa, Paterson, Fowler, & Vieira, 2008d; Beckett, 2000). Data from the PSD (reported previously – Afoakwa, Paterson, & Fowler, 2008) showed variations in specific surface area, mean particle volume D(v,50), Sauter mean (D[3,2]) and mean particle diameter (D[4,3]) with increasing D90 particle sizes. Increasing D90 from 18 to 50 lm led to significant reduction in specific surface area, with increases in Sauter mean and mean particle diameter. This explains that the largest PS (D90) is directly proportional to the D10, D50, Sauter mean diameter (D[3,2]) and mean particle diameter (D[4,3]), and inversely proportional to specific particle surface area. As size increases, particles become more spherical, leading to broadening of PSD with consequential increases (from narrow to wide) in their size and number of modal distributions and reduction in solid loading. Reduction in specific surface area with increasing particle sizes of component PSD have been reported (Ziegler & Hogg, 1999). Beckett (1999) concluded that the largest particle size and solids specific surface area are the two key parameters in chocolate manufacture. The former determines coarseness and textural character, the latter desirable flow properties. Specific surface area is inversely correlated with a component of PSD (Afoakwa, Paterson, & Fowler, 2008; Beckett, 1999; Sokmen & Gunes, 2006; Ziegler & Hogg, 1999). 2.5. Scanning electron microscopy Microstructural studies were carried out on optimally-, underand over-tempered chocolates after 14 days in storage using a 1200 EX JEM scanning electron microscopy (SEM; Joel Ltd., Akishima, Japan). Sectioned samples (20  20 mm) were lyophilized (Heto Model DW3, Allerød, Denmark), then transferred and separately placed on grids with the help of double-sided tape, sputter-coated with gold (2 min, 2 mbar). Microstructures were observed at 5 kV and 9.75  105 Torr vacuum taking 12 micrographs for each section (500, 1500 and 2500) showing typical micrographs for each temper regime. 2.6. Experimental design and statistical analysis Two experimental variables (temper regime and PSD) were varied keeping refiner temperature and pressure, conching time and temperature constant. A 3  4 factorial experimental design was used comprising; temper regime: optimal-, under- and over-temper, and PSD (D90): 18, 25, 35 and 50 lm. Statgraphics Plus 4.1 (STCC, Inc, Rockville, USA) examined melting properties (Tonset, Tend, Tindex, Tpeak, DHmelt) using two-way ANOVA and multiple range tests to determine effects and interactions of the factors. Tukey multiple comparisons (95% significance) determined differences between levels. Mean values of triplicate experiments were reported. 3.2. Fat crystallization behaviours during tempering Three different temper regimes (under-, over- and optimal tempering) characterized (Fig. 2) with unique crystallization behaviours were studied. In optimal tempering, the temperature of the chocolate dropped rapidly during cooling until reaching thermodynamic equilibrium. At this point, the crystallization heat released is balanced by an equal amount of cooling energy rending a rather flat time–temperature curve (with a zero slope). The equilibration 10 Volume (%) 8 6 4 d c b 2 a 0 0.1 1.0 10.0 100.0 1000.0 Particle Diameter (µm) Fig. 1. Particle size distribution of dark chocolate with D90 of (a) 18 lm; (b) 25 lm; (c) 35 lm and (d) 50 lm. 203 E.O. Afoakwa et al. / Food Research International 42 (2009) 200–209 31 Optimally-tempered 29 Over-tempered Under-tempered 27 Temperature (°C) 25 23 21 19 17 15 1 61 121 181 241 301 361 421 481 Time (sec) Fig. 2. Pre-crystallization curves of the different temper regimes (18 lm PS). temperature attained promotes formation of stable fat crystals, which subsequently undergo further growth and maturity during cooling and storage to effect shelf stability of the product. The temperature of the chocolate then dropped further when the liquid cocoa butter was transformed into solid crystals resulting in solidification of the products (Fig. 2). Foubert et al. (2003) noted that fat crystallization process can also be followed by means of viscosity changes as function of time. Before crystallization starts, the melt shows Newtonian behaviour. With the formation and growth of crystals, the viscosity increases almost linearly with the amount of crystals in the suspension until it reaches a thermodynamic equilibrium (Breitschuh & Windhab, 1998). This technique has also been used by Loisel et al. (1998), Toro-Vazquez et al. (2000), Chen, Lai, Ghazali, and Chong (2002) to follow the isothermal crystallization of refined palm oil, chocolate and palm stearin in sesame oil, respectively. Under-tempering (insufficient tempering) was caused by the relatively higher temperatures released between the multi-stage heat exchangers during tempering. The process causes development of more crystallization heat within the product during solidification, effecting quick cooling, as more liquid fat is transformed quickly into solid form. The distinct increase in temperature observed at the beginning of the crystallization, declined again after reaching a maximum point where most of the stable crystals formed were re-melted prior to cooling, resulting in the formation of very few stable fat crystals (Fig. 2). Previous studies revealed that un-tempering – another insufficient temper regime, produced no stable fat crystals as the heat exchange system generated higher crystallization heat during cooling, resulting in consistent cooling of the completely melted product with no inflexion point for stable fat crystal formation (Afoakwa et al., 2008a). Beckett (2000) explained that the crystallization processes in both un-tempered and under-tempered chocolates lead to the formation of unstable ðb01 Þ polymorph, which later transforms into more stable Form VI (b1) polymorph during storage. Additionally, it was noted (Afoakwa et al., 2008a) that un-tempering and under-tempering regimes exhibit different crystallization behaviours but results in similar unstable fat crystal nucleation and growth, with similar associated storage polymorphic transformations and defects in textural properties and appearance of products. Over-tempering occurred when relatively lower temperatures were exchanged between the multi-stage heat exchangers of the tempering equipment. This causes significant part of the liquid fat to solidify within the continuous phase of the chocolate, transforming the product into solid form when less liquid fat was available for pumping it through the multiple coolant regions of the 204 E.O. Afoakwa et al. / Food Research International 42 (2009) 200–209 temperer. The process effects very slow cooling as very little crystallization heat is released during the process, rendering a rather flat and slow cooling curve causing the chocolate to solidify very quickly (Fig. 2). In over-tempering, the crystallization heat released is balanced by an equal amount of cooling energy causing nucleation of stable fat crystals (b2) to effect shelf stability of the product. However, the period of equilibrium is very short relative to that of optimal tempering, and this is suspected to affect the crystal size, mass (number), strengths and adequacy of the fat crystals formed and thus possible defects in structure and product shelf stability (Hartel, 2001; Lonchampt & Hartel, 2006). As a substantial part of the phase transition (from liquid to solid) took place before the chocolate reached the mould, less contraction occurred in the mould, leading to de-moulding problems with defects in final product texture and appearance (gloss and colour) and consequential effects on shelf-life of products (Afoakwa et al., 2008a; Lonchampt & Hartel, 2004). 3.3. Effect of temper regime and PSD on melting properties 3.3.1. Effects of temper regime Fig. 3 shows typical DSC thermograms used for evaluating the melting properties of dark chocolates manufactured from the optimally-, over- and under-tempered regimes. Similar to previous reports (Afoakwa et al., 2008c; Walter & Cornillon, 2002), all the samples exhibited similar distinct single endothermic transitions between 15 and 55 °C, the range expected for chocolate melting profiles. McFarlane (1999) explained that peak onset corresponds to the temperature at which a specific crystal form starts to melt; peak maximum, that at which melting rate is greatest; and end of melting, completion of liquefaction – all these information are related to the crystal type. Peak height, position and resolution are dependent on sample composition and crystalline state distribution. Data from the DSC (Fig. 3) showed that differences in temper regime produced changes in crystallinity and melting properties, observed in the differences in their peak widths. These suggest that variations in crystallization behaviour in dark chocolates during tempering influence the degree of crystallinity of their derived products. Under-tempered (bloomed) chocolates showed the greatest peak width, followed by the over-tempered samples having slightly wider crystal formation than the optimally-tempered products with resultant variation in their melting profiles (Fig. 3). Hartel (2001) concluded that arrangement and distribution of crystallinity (total and specific characteristics of crystalline material) in foods play key roles in final product quality. Number of crystals and range of sizes, shapes, and polymorphic stability, as well as arrangements in network structures dictates mechanical and rheological properties. Knowledge and control of crystallinity during chocolate manufacture can be important for optimizing processing conditions. Data from the DSC on Tonset, Tend, Tpeak, DHmelt and Tindex in relation to temper regime (Table 1) analysed by ANOVA and multiple comparison tests showed significant (P < 0.05) differences for Tonset and Tpeak differing in temper regime (Table 2) and highly significant differences (P < 0.001) among Tend, Tindex and DHmelt (Table 2). The differences in temper yielded mean Tend values of 33.0, 33.7 and 35.9 °C, respectively, for the optimally, over- and under-tempered chocolates. There was a significant (P < 0.05) inverse relationship between Tend and PSD (Table 2). Such observations suggest that under-tempered chocolate completed melting at higher temperatures than optimally and over-tempered products. As Tend values provides representation of the melting point of the product and can be used as a measure of polymorphic status (Afoakwa, Paterson, & Fowler, 2007; Beckett, 2000; Marangoni & McGauley, 2003), the changing Tend values the samples revealed that the crystallites in optimally and over-tempered were in ßV polymorph while that of under-tempered in ßVI. Similarly, under-tempered (bloomed) chocolate had higher Tindex values of 8.8, 8.7, 8.5 and 8.2 °C inversely related to PS from 18 to 50 lm, while the optimal and overtempered products had Tindex ranges of 7.1–6.0 °C, and from 7.6 to 6.6 °C, respectively, suggesting that the under-tempered chocolate took longer to melt than the optimally and over-tempered products. Values for Tindex has been reported to provide information 4.5 Optimally-tempered Normalized Heat Flow (W/g) 4.3 Over-tempered Under-tempered 4.1 3.9 3.7 3.5 3.3 15 17.433 21.183 24.933 28.683 32.433 36.183 39.933 43.683 47.433 51.183 54.933 Temperature (°C) Fig. 3. Typical DSC thermograms showing fat melting profile of tempered, over-tempered and under-tempered (bloomed) dark chocolates at 18 lm PS. 205 E.O. Afoakwa et al. / Food Research International 42 (2009) 200–209 Table 1 Effect of temper regime and particle size distribution on melting properties. Temper regime Particle size (D90) (lm) Melting properties Tonset (°C) Tend (°C) Tindex (°C) Tpeak (°C) DHmelt (J/g) Tempered 18 25 35 50 26.5 ± 0.4 26.4 ± 0.3 26.6 ± 0.2 26.5 ± 0.4 33.6 ± 0.3 33.3 ± 0.4 32.7 ± 0.3 32.5 ± 0.4 7.1 ± 0.2 6.7 ± 0.4 6.1 ± 0.2 6.0 ± 0.4 31.9 ± 0.1 31.7 ± 0.2 31.7 ± 0.1 31.8 ± 0.2 37.73 ± 0.65 37.56 ± 0.92 36.87 ± 0.58 36.76 ± 0.72 Over-tempered 18 25 35 50 26.6 ± 0.2 26.5 ± 0.4 26.7 ± 0.2 26.6 ± 0.3 34.2 ± 0.3 33.8 ± 0.4 33.5 ± 0.2 33.2 ± 0.4 7.6 ± 0.2 7.3 ± 0.4 6.8 ± 0.2 6.6 ± 0.4 32.6 ± 0.2 32.7 ± 0.1 32.5 ± 0.2 32.6 ± 0.2 41.26 ± 0.61 40.42 ± 0.88 40.47 ± 0.57 40.36 ± 0.52 Under-tempered 18 25 35 50 27.4 ± 0.2 27.3 ± 0.4 27.2 ± 0.2 27.4 ± 0.3 36.2 ± 0.3 36.0 ± 0.4 35.7 ± 0.3 35.6 ± 0.4 8.8 ± 0.2 8.7 ± 0.4 8.5 ± 0.2 8.2 ± 0.4 33.8 ± 0.2 33.7 ± 0.1 33.6 ± 0.2 33.6 ± 0.1 44.45 ± 0.88 44.10 ± 0.51 43.87 ± 0.86 43.80 ± 0.58 Means ± standard deviation from triplicate analysis. Table 2 ANOVA summary of F-values of melting properties. Process variables Tonset (°C) Tend (°C) Tpeak (°C) Tindex (°C) DHmelt (J/g) A: particle size (D90) B: temper regime AB 1.78 198.75*** 1.18 12.17* 261.19*** 2.45* 3.74 22.57** 7.26*** 34.73** 1107.80*** 160.33*** 6.96 462.78*** 3.67** * ** *** Significant F-ratios at P 6 0.05. Significant F-ratios at P 6 0.01. Significant F-ratios at P 6 0.001. on the duration of melting as measured by the DSC, with possible impacts on melting behaviour during consumption (Afoakwa et al., 2008c). Multiple comparison tests showed that the over-tempered samples took longer to melt than the optimally-tempered. It is predicted that these would have likely impact on their behaviour during consumption, attributable to the relative strengths of their mechanical properties (hardness and stickiness). Similarly, undertempered chocolate had higher DHmelt values at all PS than the optimally- and over-tempered products (Table 1), with significant (P < 0.05) interactions with PS. The observed differences in melting behaviour of under-, over- and optimally-tempered chocolates have been attributed to their reported variations in structure and mechanical properties (Afoakwa et al., 2008a). Under-tempered chocolates were reported to possess the greatest hardness (texture), attributable to the re-crystallisation process undergone by the fat resulting in intense hardening of products. Further multiple comparison test revealed that over-tempered chocolates showed higher Tindex and DHmelt than the optimally-tempered, a significant finding for process quality control. This could also be attributed to the over-tempered products possessing relatively higher mechanical properties than their corresponding optimally-tempered products, probably due to the reported differences in their fat crystallization behaviour resulting different fat crystal network structure. The extent of fat crystal network organization during tempering has been reported to influence the mechanical properties of their derived confectionery products (Campos et al., 2002; Marangoni & Narine, 2002). 3.3.2. Effects of particle size distribution In chocolate, PSD influences rheological, melting and microstructural properties as well as texture in tempered products (Afoakwa et al., 2008d; Afoakwa, Paterson, Fowler, & Vieira, 2008e; Beckett, 2008; Do, Hargreaves, Wolf, Hort, & Mitchell, 2007). In this study, peak shapes and sizes were similar with all values for PSD suggesting only little or no differences in crystallinity between them. Examination of important DSC parameters (Table 1) – Tonset, Tend, Tpeak, DHmelt and Tindex – suggested increasing PS from 18 to 50 lm yielded no significant (P = 0.2782) changes in Tonset, for any temper regime (Table 2), with values in the ranges of 26.5–26.6 °C, 26.5–26.7 °C and 27.2–27.4 °C, respectively, for the optimally, over- and under-tempered chocolate. Similar observations were made for Tpeak (Table 1) – ranging from 31.7 to 31.9 °C, 32.1 to 32.3 °C and 33.6 to 33.8 °C for the optimally, overand under-tempered chocolates. Similarly, DHmelt in products with increasing PS from 18 to 50 lm decreased marginally from 37.73 to 36.76, 41.26 to 40.36 and 44.45 to 43.80 J/g in the optimally, over- and under-tempered products, respectively, (Table 1), with insignificant differences (P > 0.05) in PS found at all temper regimes. The lack of significant relationship between PS and DHmelt, implies that enthalpy of melting was similar for chocolates at all PS independent of temper regime. In contrast, varying PS had significant effects on Tend and Tindex with general inverse relationships between particle size and Tend, and Tindex, at all temper regime (Table 1). Optimally-tempered products with smaller PS (18 lm) had Tend value of 33.6 °C, and decreased consistently to 32.5 °C in the 50 lm samples, representing a difference of 0.9 °C. Similar marginal but significant (P < 0.05) decreasing trends in Tend with increasing PS were observed with both the over-tempered and under-tempered (bloomed) samples (Table 2). These findings suggested that dark chocolates with larger PS (50 lm) require slightly lower temperatures to complete melting than the smaller PS (18 lm) products at all temper regimes. Chocolates with smaller PSD (D90, 18 lm) have been found to contain higher particle-to-particle strengths with resultant increases in hardness (texture) than their corresponding larger PSD (D90, 50 lm) (Afoakwa et al., 2008b, 2008c; Do et al., 2007). Similar inverse relationships were observed between Tindex and PS at all temper regimes. The data (Table 1) showed that increasing PS for 18–50 lm in the optimally-tempered products caused significant (P 6 0.05) reductions in Tindex from 7.1 to 6.0 °C, respectively, and similar trends were noted with the over-tempered and 206 E.O. Afoakwa et al. / Food Research International 42 (2009) 200–209 under-tempered (bloomed) products. ANOVA showed significant (P < 0.001) influence of PS on Tend and Tindex with significant interactions with temper regime (Table 2). Multiple comparison test revealed significant differences (P = 0.001) between Tend of products containing 18 lm, 35 lm and 50 lm, suggesting that chocolates with finer particles (18 lm) would take relatively longer time to melt than their corresponding products with larger particles (35 lm and 50 lm) independent of temper regime, attributable to the relative strengths of the inter-particle aggregations and flocculation in the different PS products (Afoakwa et al., 2008d; Marangoni & McGauley, 2003; Narine & Marangoni, 1999a, 1999b). Do et al. (2007) noted that quantitative decreases in particle aggregation and structure in chocolate influence melting behaviour suggesting that in its crystallized state, structures with larger PS are less interconnected, providing less resistance to breakage and melting. This could be important for predicting oral melting behaviour with impacts on temporal components of flavour release and oral epithelial sensations. mechanical and melting properties. The DSC thermograms (Fig. 4) showed differences in fat melting profile, resulting from the widened peak width in the under-tempered (bloomed) sample; but no differences were noted in the sugar melting profiles, suggesting that blooming of under-tempered dark chocolate is associated with structural transformations in the fat component alone while the crystalline network of the sugar remains unchanged. The DSC data on fat and sugar melting properties (Tonset, Tend, Tpeak, DHfat, DHsugar and DHsugar/DHfat) related to temper regime (Table 3) were similar to the trends for fat (Table 1). Fat melting profiles suggested the ßV polymorph in both optimally and over-tempered chocolates with Tend of 32.3 °C and 32.9 °C, respectively, and a more stable ßVI polymorph in under-tempered sample with Tend of 35.8 °C, showing significant (P < 0.001) influences (Table 4) on Tonset, Tpeak, DHfat in chocolates. Beckett (2008) explained that the melting points of ßV and ßVI polymorphs of cocoa butter are in the ranges of 32– 34 °C and 34–36 °C, respectively. Contrary, the results of the sugar melting properties (Table 3) showed only marginal differences in all the melting properties with varying temper regime. ANOVA showed no significant differences (P > 0.05) in all the studied melting properties (Tonset, Tend, Tpeak, DHsugar) on chocolates from the three temper regimes (Table 4), suggesting that no structural change in sugar were found in products from the three temper regimes. Similarly, the ratios of sugar to fat melting enthalpies in products from optimal, over- and under-tempered samples were 1.25, 1.24 and 1.17, respectively, (Table 3), with no significant different (P = 6.853) among them 3.4. Thermal behaviours and ratio of sugar/fat melting enthalpies in products Thermal behaviours and ratio of sugar/fat melting enthalpies in chocolates differing in temper regime were studied using DSC to provide information on differences in structure. Marangoni & McGauley (2003) explained that the structure of fat in a food product is an important property that strongly influences its perceived B A 3.9 Optimally-tempered Over-tempered Normalized Heat Flow (W/g) 3.4 Under-tempered 2.9 2.4 1.9 1.4 15 25 39.666 54.333 69 83.666 98.333 113 127.666 142.333 157 171.666 186.333 201 Temperature (°C) Fig. 4. Typical DSC thermograms showing (A) fat and (B) sugar melting profiles of tempered, over-tempered and under-tempered (bloomed) dark chocolates at 18 lm PS. Table 3 Thermal properties of fat and sugar components in dark chocolates from different temper regimes. Temper regime Tempered Over-tempered Under-tempered Fat melting properties Sugar melting properties Sugar/fat relations Tonset (°C) Tend (°C) Tpeak (°C) DHmelt (J/g) Tonset (°C) Tend (°C) Tpeak (°C) DHmelt (J/g) DHsugar/DHfat 26.2 ± 0.24 26.4 ± 0.18 27.3 ± 0.53 32.3 ± 0.44 32.9 ± 0.28 35.8 ± 0.19 30.8 ± 1.04 31.4 ± 0.83 33.5 ± 0.71 37.60 ± 0.66 39.58 ± 0.42 42.07 ± 0.73 179.43 ± 0.43 178.85 ± 0.18 178.21 ± 0.47 191.98 ± 0.39 191.34 ± 0.83 190.82 ± 0.50 188.83 ± 0.52 187.37 ± 0.74 186.87 ± 0.58 50.64 ± 0.64 49.13 ± 0.47 49.16 ± 0.76 1.25 ± 0.21 1.24 ± 0.34 1.17 ± 0.28 Means ± standard deviation from triplicate analysis. 207 E.O. Afoakwa et al. / Food Research International 42 (2009) 200–209 Table 4 ANOVA summary of F-values of fat and sugar thermal properties. Process variable Temper regime * ** Fat melting properties Sugar melting properties Sugar/fat relations Tonset (°C) Tend (°C) Tpeak (°C) DHmelt (J/g) Tonset (°C) Tend (°C) Tpeak (°C) DHmelt (J/g) DHsugar/DHfat 12.41* 42.83** 3.86* 32.89** 2.07 1.52 2.54 3.28 1.95 Significant F-ratios at P 6 0.05. Significant F-ratios at P 6 0.001. (Table 4). The lower DHsugar/DHfat ratio noted in the under-tempered sample resulted from the higher DHfat as a result of re-crystallization of fat (Hartel, 2001; Lonchampt & Hartel, 2004). These findings support our earlier report that fat and sugar components are present in similar quantities in both bloomed and optimallytempered dark chocolates, but contrast with the report of Lonchampt and Hartel (2006) that the melting peak of fat in bloomed chocolate was almost non-existence with DHfat being tenfold smaller than that obtained for optimally-tempered chocolate, concluding that the whitish spots in bloomed chocolates were mainly composed of sugar crystals and cocoa powder and nearly devoid of fat. Kinta and Hatta (2005) also reported the presence of fat components in bloomed dark chocolate, suggesting mechanisms of bloom development in bloomed chocolate involves phase separation associated with the growth of xenomorphic fat crystals. distributions of network mass among optimally-, over- and under-tempered samples, becoming well defined with increasing magnification from (i) 800 (ii) 1500 to (iii) 2500 (Fig. 5). Effect of PS on the microstructure was not presented because we realized from our preliminary studies that there were no clear differences in crystal structure among the varying PS (18–50 lm) at the same temper regime. Microscopy of optimally-tempered chocolate showed an even spatial distribution of dense crystalline network with well defined inter-crystal connections within the structure (Fig. 5a). The micrographs of over-tempered chocolate also showed spatial distribution of a dense mass of smaller crystals within a network structure containing mixtures of both well- and ill-defined crystal-to-crystal connections (Fig. 5b). These large numbers 3.5. Effect of temper regime on scanning electron microstructure Microstructural examination using scanning electron microscopy after 14 days of conditioning showed clear variations in crystalline network structure, inter-crystal connections and spatial Fig. 5. Scanning electron micrographs showing crystalline network microstructures at magnifications of (i) 800 (ii) 1500 (c) 2500 for (a) tempered, (b) overtempered and (c) under-tempered (bloomed) dark chocolates at 18 lm PS. C shows some of the well defined crystal structures; iC shows some of the ill-defined crystal structures; i shows some of the inter-crystal connections. The arrows indicate some of the pores, cracks and crevices; B shows some of solid bridges; L shows some of the large (crystal) lumps on the crystal structure. Fig. 5 (continued) 208 E.O. Afoakwa et al. / Food Research International 42 (2009) 200–209 properties and shelf-life (Campos et al., 2002; Hartel, 2001; Pérez-Martínez et al., 2007). Parameters such as cooling rate and thermal history (i.e., crystallization temperature and tempering) influence kinetics and ultimate physical properties of the crystallized fat systems. 4. Conclusion Fig. 5 (continued) of smaller crystals noted within the crystalline networks in the over-tempered samples is suspected to result from early nucleation and growth of seed crystals due to the slow cooling (Fig. 2), leading to the formation of sub-micron primary crystallites from the melt, with the resulting fat crystal network stabilized by van der Waals forces, possibly with steric and Coulombic forces (deMan, 1999; Marangoni & Narine, 2002; Tang & Marangoni, 2008). Under-tempered (bloomed) chocolates showed dissolution, rearrangement and re-crystallization of the numerous crystals noted in the over- and optimally-tempered products to a smaller number of larger (lumps) fat crystals (Ostwald ripening). This results in polymorphic transformation, nucleation and growth of new large crystals in a more stable polymorphic form, inducing formation of solid bridges with weak and less inter-crystal connections within the crystalline structures (Fig. 5c). Hartel (2001) suggested this phenomenon is brought about by thermodynamic differences in equilibrium between large and small crystals within a network structure leading to re-crystallization of unstable fat polymorphs. This hypothesis suggests that differences in crystallization behaviour during tempering leads to formation of different microstructural organizations of crystal network structure with associated physical changes in chocolates. Characterizing the nature of crystals in confectionery is an important step in quantifying the physical and sensory properties, as the resulting three-dimensional fat crystal network along with the phase behaviour and structural arrangements impact on mechanical, rheological, and melting Fat crystallization behaviour during tempering of dark chocolates plays a key role in defining their ultimate structure and melting properties. Variations in PSD had no influence on crystallinity of chocolates whether optimally, over- or under-tempered. Particle size had a limited but significant direct relationship with certain melting parameters – Tonset, Tpeak, and DHmelt – independent of temper but significant inverse relationship with others – Tend and Tindex. Contrariwise, varying temper influenced crystallinity and chocolate melting properties (Tend, Tindex and DHmelt). Under-tempering of chocolate resulted in widened crystal size distribution with significant changes in Tend, Tindex and DHmelt. Over-tempering caused moderate increases in crystal size distribution, with significant effects on Tend, Tindex and DHmelt but no changes were noted in Tonset, or Tpeak. Fat-sugar melting profiles were similar in all chocolates independent of PS and temper regime. Electron micrographs showed an even spatial distribution of numerous stable b-polymorph crystals in a network with well defined inter-crystal connections in optimally-tempered chocolate. With over-tempered chocolate there were large numbers of very small crystals in network with mixtures of well- and ill-defined crystal–crystal networks resulting from formation of a stable bpolymorph with early nucleation: the outcome was growth of seed crystals from the melt into sub-micron primary crystallites and a fat crystal network stabilized by van der Waals forces. Under-tempering resulted in dissolution of a large number of small crystals, re-arrangement and re-crystallization into a small number of larger (lumps) fat crystals (Ostwald ripening). In this process there was polymorphic transformation, nucleation and growth of new large crystals in a more stable polymorphic form with formation of solid bridges with weak and fewer inter-crystal connections within the chocolate structure. Thus, attainment of optimal temper regime during tempering (pre-crystallization) of dark chocolate is necessary for the achievement of premium quality products and avoidance of defects in structure and melting character. Acknowledgements This study was co-funded by the Government of Ghana and Nestlé Product Technology Centre (York, UK). The sponsors are gratefully acknowledged for the Research Support. The authors thank Drs. Steve Beckett, John Rasburn and Angel Manez (Nestlé PTC, York) for expert technical discussions. References Afoakwa, E. O., Paterson, A., & Fowler, M. (2007). Factor influencing rheological and textural qualities in chocolate – A review. Trends in Food Science and Technology, 18(6), 290–298. Afoakwa, E. O., Paterson, A., & Fowler, M. (2008). Effects of particle size distribution and composition on rheological properties of dark chocolate. European Food Research and Technology, 226, 1259–1268. doi:10.1007/s00217-007-0652-6. 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. doi:10.1016/j.jfoodeng.2008.04.021. Afoakwa, E. O., Paterson, A., Fowler, M., & Vieira, J. (2008b). Modelling tempering behaviour of dark chocolates from varying particle size distribution and fat content using response surface methodology. Innovative Food Science and Emerging Technologies, 9(4), 527–533. doi:10.1016/j.ifset.2008.02.002. Afoakwa, E. O., Paterson, A., Fowler, M., & Vieira, J. (2008c). Characterization of melting properties in dark chocolate from varying particle size distribution and E.O. Afoakwa et al. / Food Research International 42 (2009) 200–209 composition using differential scanning calorimetry. Food Research International, 41(7), 751–757. doi:10.1016/j.foodres.2008.05.009. Afoakwa, E. O., Paterson, A., Fowler, M., & Vieira, J. (2008d). Particle size distribution and compositional effects on textural properties and appearance of dark chocolates. Journal of Food Engineering, 87(2), 181–190. doi:10.1016/ j.jfoodeng.2007.11.025. Afoakwa, E. O., Paterson, A., Fowler, M., & Vieira, J. (2008e). Microstructural and mechanical properties relating to particle size distribution and composition in dark chocolate. International Journal of Food Science and Technology. doi:10.1111/j.1365-2621.2007.01677.x. 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, 600–610. Beckett, S. T. (1999). Industrial chocolate manufacture and use (3rd ed.). Oxford: Blackwell Science. pp. 153–181, 201–230, 405–428, 460–465. Beckett, S. T. (2000). The science of chocolate. London: Royal Society of Chemistry. Beckett, S. T. (2008). The science of chocolate (2nd ed.). London: Royal Society of Chemistry. Bell, A., Gordon, M. H., Jirasubkunakorn, W., & Smith, K. W. (2007). Effects of composition on fat rheology and crystallisation. Food Chemistry, 101, 799–805. Breitschuh, B., & Windhab, E. J. (1998). Parameters influencing cocrystallization and polymorphism in milk fat. Journal of American Oil Chemists Society, 75, 897–904. Campos, R., Narine, S. S., & Marangoni, A. G. (2002). Effect of cooling rate on the structure and mechanical properties of milk fat and lard. Food Research International, 35, 971–981. Chen, C. W., Lai, O. M., Ghazali, H. M., & Chong, C. L. (2002). Isothermal crystallization kinetics of refined palm oil. Journal of the American Oil Chemists’ Society, 79, 403–410. Debaste, F., Kegelaers, Y., Liégeois, S., Ben Amor, H., & Halloin, V. (2008). Contribution to the modelling of chocolate tempering process. Journal of Food Engineering, 88, 568–575. deMan, J. M. (1999). Relationship among chemical, physical, and textural properties of fats. In N. Widlak (Ed.). Physical properties of fats, oils and emulsions (pp. 79–95). Champaign, IL, USA: AOCS Press. Do, T.-A. L., Hargreaves, J. M., Wolf, B., Hort, J., & Mitchell, J. R. (2007). Impact of particle size distribution on rheological and textural properties of chocolate models with reduced fat content. Journal of Food Science, 72(9), E541–E552. Foubert, I., Dewettinck, K., & Vanrolleghem, P. A. (2003). Modelling of the crystallization kinetics of fats. Trends in Food Science & Technology, 14, 79–92. Hartel, R. W. (2001). Crystallization in food. Gaithersburg, USA: Aspen Publishers Inc.. Herrera, M. L., & Hartel, R. W. (2000). Effect of processing conditions on the crystallization kinetics of milk fat model systems. Journal of the American Oil Chemists Society, 77, 1177–1187. Himawan, C., Starov, V. M., & Stapley, A. G. F. (2006). Thermodynamic and kinetic aspects of fat crystallization. Advances in Colloid and Interface Science, 122, 3–33. ICA (1988). Determination of moisture content of cocoa and chocolate products. Analytical Method 26. Bruxelles, Belgium: CAOBISCO. ICA (1990). Determination of fat content of cocoa and chocolate products. Analytical Method 37. Brussels, Belgium: CAOBISCO. Kinta, Y., & Hatta, T. (2005). Composition and structure of fat bloom in untempered chocolate. Journal of Food Science, 70, S450–452. Loisel, C., Lecq, G., Keller, G., & Ollivon, M. (1998). Dynamic crystallization of dark chocolate as affected by temperature and lipid additives. Journal of Food Science, 63, 73–79. Lonchampt, P., & Hartel, R. W. (2004). Fat bloom in chocolate and compound coatings. European Journal of Lipid Science & Technology, 106, 241–274. doi:10.1002/ejlt.2004008938. 209 Lonchampt, P., & Hartel, R. W. (2006). Surface bloom on improperly tempered chocolate. European Journal of Lipid Science & Technology, 108, 159–168. doi:10.1002/ejlt.200500260. Marangoni, A. G. (2002). Special issue of FRI—crystallization, structure and functionality of fats. Food Research International, 35(10), 907–908. Marangoni, A. G., & McGauley, S. E. (2003). Relationship between crystallization behavior and structure in cocoa butter. Crystal Growth & Design, 3(1), 95–108. Marangoni, A. G., & Narine, S. S. (2002). Identifying key structural indicators of mechanical strength in networks of fat crystals. Food Research International, 35, 957–969. McFarlane, I. (1999). Instrumentation. In S. T. Beckett (Ed.), Industrial chocolate manufacture and use (pp. 347–376). New York: Chapman & Hall. Narine, S. S., & Marangoni, A. G. (1999a). Relating structure of fat crystal networks to mechanical properties: A review. Food Research International, 32, 227–248. Narine, S. S., & Marangoni, A. G. (1999b). Fractal nature of fat crystal networks. Physical Reviews E, 59, 1908–1920. Nelson, R. B. (1999). Tempering. In S. T. Beckett (Ed.), Industrial chocolate manufacture and use (pp. 347–376). New York: Chapman & Hall. Pérez-Martínez, D., Alvarez-Salas, C., Charó-Alonso, M., Dibildox-Alvarado, E., & Toro-Vazquez, J. F. (2007). The cooling rate effect on the microstructure and rheological properties of blends of cocoa butter with vegetable oils. Food Research International, 40, 47–62. Schenk, H., & Peschar, R. (2004). Understanding the structure of chocolate. Radiation Physics and Chemistry, 71, 829–835. Smith, K. W., Cain, F. W., & Talbot, G. (2007). Effect of nut oil migration on polymorphic transformation in a model system. Food Chemistry, 102, 656–663. Sokmen, A., & Gunes, G. (2006). Influence of some bulk sweeteners on rheological properties of chocolate. LWT – Food Science & Technology, 39, 1053–1058. Tabouret, T. (1987). Detection of fat migration in a confectionery product. International Journal of Food Science and Technology, 22, 163–167. Tang, D., & Marangoni, A. G. (2008). Modified fractal model and rheological properties of colloidal networks. Journal of Colloid and Interface Science, 318, 202–209. Timms, R. E. (1984). Phase behaviour of fats and their mixtures. Progress in Lipid Research, 23, 1–38. Toro-Vazquez, J. F., Briceno-Montelongo, M., Dibildox-Alvarado, E., Charo-Alonso, M., & Reyes-Hernandez, J. (2000). Crystallization kinetics of palm stearin in blends with sesame seed oil. Journal of the American Oil Chemists’ Society, 77, 297–310. Toro-Vazquez, J. F., Pérez-Martínez, D., Dibildox-Alvarado, E., Charó-Alonso, M., & Reyes-Hernández, J. (2004). Rheometry and polymorphism of cocoa butter during crystallization under static and stirring conditions. Journal of the American Oil Chemists Society, 81, 195–203. Vasanthan, T., & Bhatty, R. S. (1996). Physicochemical properties of small and large granule starches of waxy, regular, and high amylase barleys. Cereal Chemistry, 73, 199–207. Walter, P., & Cornillon, P. (2001). Influence of thermal conditions and presence of additives on fat bloom in chocolate. Journal of the American Oil Chemists Society, 78, 927–932. Walter, P., & Cornillon, P. (2002). Lipid migration in two-phase chocolate systems investigated by NMR and DSC. Food Research International, 35, 761–767. Willie, R. L., & Lutton, E. S. (1966). Polymorphism of cocoa butter. Journal of the American Oil Chemists Society, 43, 491–496. Ziegler, G., & Hogg, R. (1999). Particle size reduction. In S. T. Beckett (Ed.), Industrial chocolate manufacture and use (pp. 182–199). New York: Chapman & Hall.