Journal of Food Engineering 253 (2019) 59–71 Contents lists available at ScienceDirect Journal of Food Engineering journal homepage: www.elsevier.com/locate/jfoodeng Applicability of the melanger for chocolate refining and Stephan mixer for T conching as small-scale alternative chocolate production techniques Michael Hinneha,b,c,∗, Davy Van de Wallea, Julie Haecka, Enoch Enorkplim Abotsia, Ann De Winnec, Arifin Dwi Saputroa,d, Kathy Messense, Jim Van Durmec, Emmanuel Ohene Afoakwab, Luc De Coomanc, Koen Dewettincka a Department of Food Technology, Safety and Health, Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, 9000, Ghent, Belgium bDepartment of Nutrition & Food Science, University of Ghana, P. O. Box LG 134, Legon, Accra, Ghana c Research Group Molecular Odor Chemistry, Department of Microbial and Molecular Systems (M2S), Research Cluster Food and Biotechnology, KU Leuven Technology Campus, 9000, Ghent, Belgium dDepartment of Agricultural Engineering, Faculty of Agricultural Technology, Universitas Gadjah Mada, Jl. Flora No. 1 Bulaksumur, 55281, Yogyakarta, Indonesia e Department of Biotechnology, Faculty of Bioscience Engineering, Ghent University, Valentin Vaerwyckweg, 1, 9000, Ghent, Belgium A R T I C L E I N F O A B S T R A C T Keywords: Conventional dark chocolate production methods that are applied in the industry consist of several steps, which Dark chocolate require equipment with big investment costs. There are however few cost-effective alternatives suggested for Small-scale processing small-scale production. Meanwhile, knowledge on these alternative equipment/techniques are insufficient to Melanger promote miniature production of high quality chocolates, either for research purposes in an industrial context, or Stephan mixer for producers in developing countries, where the cost of investing in conventional equipment seems to be an Refining Conching impediment. The aim of this study was two-fold; first, to assess the feasibility of utilizing the ECGC-12SLTA CocoaTown melanger as an alternative to the conventional 3-roll refiner at different settings and fat content. Thereafter, one optimal setting was selected for each equipment for further investigation on the impact of the refining on some quality attributes of the final dark chocolate (70% cocoa). Secondly, the Stephan mixer; being used to mimick a conching-like process, was assessed with respect to two processing factors; the dry conching temperature (60 °C, 80 °C) and the duration of vacuum pump connection (0, 30, 60min). The latter was to facilitate adequate moisture removal. The melanger proved to be a suitable alternative to the 3-roll refiner, provided that refining was carried out at moderate/high (ca. 40%) fat content, as is often the case for “high- percentage-cocoa” chocolates. Refining for 180min with the mini drum at 40% resulted in D (v,0.9) significantly (p < 0.05) lower than when the 3-roll refiner was used. Nonetheless, a comparative advantage of the latter would be its short throughput time (5–10min). Due to a resultant linear speed gradient of the chocolate mass due to the cylindrical roller stones, a more efficient refining was achieved with the mini drum than with the big drum. More so, refining in excess fat (40%) may have contributed to a more efficient coating of the newly created hydrophilic sugar surfaces, thus, limiting the possibility for moisture-induced agglomeration as may have been the case for the recipe with 27% fat. In spite of trivial difference in moisture content, chocolates manufactured following melanger and 3-roll refining showed significant (p < 0.05) differences in terms of particle size, flow parameters and color. For chocolates that were conched with the Stephan mixer, the vacuum duration had a significant (p < 0.05) impact on moisture content and D (v,0.9). Also, an impact of all factors and their interaction on the Casson yield values of the chocolates was observed. However, these factors proved to be less important in dictating the final viscosities of the chocolates. Among others, it is suggested that the influence of the high fat content of the chocolates may have played a more important role. Although all cho- colates exhibited less thixotropic behavior, a direct proportional relationship between the particle surface area and thixotropy was observed. Finally, the interaction effect of both factors also significantly (p < 0.05) influ- enced the color of the chocolates. ∗ Corresponding author.Department of Food Technology, Safety and Health, Faculty of Bioscience Engineering, Ghent University, Coupure Links 653, 9000, Ghent, Belgium. E-mail address: hinnehmichael@gmail.com (M. Hinneh). https://doi.org/10.1016/j.jfoodeng.2019.02.016 Received 29 August 2018; Received in revised form 24 January 2019; Accepted 15 February 2019 Available online 22 February 2019 0260-8774/ © 2019 Elsevier Ltd. All rights reserved. M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 Fig. 1. CocoaTown melangers (a) mini drum with cylindrical roller stones and (b) big drum with conical roller stones. 1. Introduction bottom of the vessel and a reverse-acting scraper arm fitted through the lib. The equipment is also fitted with steam and pressure nozzles which Chocolate is a complex suspension of solid particles (sugar, cocoa allow both temperature and pressure to be controlled in the vessel particles) dispersed in a continuous fat phase, mainly cocoa butter during processing (Stephan, 2018). By studying the rheological prop- (Afoakwa et al., 2008). The growing popularity of chocolate revolves erties of chocolates, Aidoo et al. (2014) provided some indication around its unique sensory and textural properties (Fowler, 2009; supporting the possibility of optimizing this equipment for use as an Torres-Moreno et al., 2015). However, the acclaimed health benefits alternative conching device. They concluded that optimum settings of attributed to chocolate in recent times have further promulgated its 65 °C for 10min dry conching followed by 50 °C for 15min wet con- demand among consumers (Steinberg et al., 2003; Latif, 2013). ching both at blade rotary speed of 443 g yielded similar flow properties The main conventional chocolate manufacturing process en- as the reference. However, information on some quality parameters compasses steps such as mixing, refining and conching. However, most such as moisture and particle size distribution (PSD), which have been classical processing equipment are designed to handle huge quantities demonstrated to have a huge impact on the flow behavior of chocolates, either in a continuous or batch process. Regardless of some attempts to were missing. Saputro et al. (2016b) also successfully applied a com- scale-down production, in practice, only a handful of equipment on the bination of the Stephan mixer and ball mill in an alternative production market have been recognized as suitable alternatives, considering the of dark chocolates sweetened with palm sap-based sugar. Here, they similitude of their final products to chocolates from a classical in- used the Stephan mixer for both mixing and liquefaction, whereas re- dustrial process in such areas as quality and flavor attributes. Common fining was carried in the ball mill. The first stage of the mixing process examples of these include the ball mill and the ELK'olino conche. was carried at 70 °C for 60min with blade speed of 750 rpm. During the Whereas the former is often applied for grinding nibs into liquor, and in last 10min of this process, a vacuum pump was activated in order to some cases, for alternative processing of chocolates/compounds, the facilitate moisture removal. Thereafter, a second mixing step was car- latter has served as an ideal conching equipment handling approxi- ried out at 50 °C for 30min with a blade speed of 1500 rpm. Next, the mately 5 kg per batch (Saputro et al., 2016a,b). The cost of investment chocolate mass was refined with the ball mill for 30min at 50 °C with associated with these equipment have made them unfavorable for maximum speed. Finally, liquefaction was performed at 50 °C for simple laboratory and even miniature industrial productions - where 15min with a blade speed of 1500 rpm using the Stephan mixer. only a small amount of chocolate needs to be produced from a limited The melanger, on the other hand, was initially developed for quantity of available beans. Notably, this also seems to be the challenge grinding, among others, dried seeds and nuts, as well as cocoa nibs. The for most small-scale bean-to-bar producers in various developing equipment can either be operated with a big drum (with a maximum countries, such as Ghana, Ivory Coast, Indonesia and India. It is of no capacity of ca. 3.6 kg) or mini drum (with a maximum capacity of ca. surprise that the idea of small-scale alternative processing has gained a 1 kg) (Fig. 1). The big drum is equipped with a set of conical granite lot of attention with some recent studies have been carried out on this roller stones, whereas, the mini drum operates with a set of cylindrical topic (Bolenz and Manske, 2013; Fistes et al., 2013; Pajin et al., 2013; granite roller stones that rotate at 135–140 rpm on a granite slab - Saputro et al., 2016a,b; Tan and Balasubramanian, 2017; Saputro et al., creating shearing forces - that crushes and reduces the particle sizes of 2018). For instance, the introduction of coarse conching using the ball cocoa solids during the grinding process (CocoaTown, 2017). This mill by Bolenz et al. (2014). It is apparent that the advantages asso- could result in particle sizes less than 20 μm after grinding/refining for ciated with alternative processing, such as its time and energy effi- about 8 h as indicated by Tan and Balasubramanian (2017). The in- ciency, compact nature (ability to combine several processing steps), tensity of the grinding/refining can be adjusted by regulating the ten- cost efficiency, as well as the need for less trained personnel may be sion between the roller stones and the slab. The minimal loss of sample some of the underlying factors driving its growing popularity and/or makes the melanger more efficient in terms of sample recovery. Besides, acceptance. its low investment cost, its compact configuration (takes up less than In this study, two types of such alternative processing equipment; 0.14m2 of space) and acclaimed suitability for use at different stages of the ECGC-12SLTA CocoaTown melanger (CocoaTown, Roswell, USA), the chocolate production process, may account for its growing popu- and the Stephan mixer (Stephan food service equipment GmbH, larity and demand among small-scale chocolate manufacturers in recent Hameln, Germany) were explored to evaluate their feasibility and times. The melanger however has some obvious flaws; chief of which is performance at different stages of the chocolate production process. the lack of a temperature control unit. Hence, operating this equipment The Stephan mixer is an all-purpose, robust system, equipped with a at room temperature in a conching-like process may limit the removal double jacket and a tightly fitted lid for all kinds of food processing of moisture and undesirable flavors. between 0.5 and 2.5 kg batch sizes. Here, mixing is achieved by means Apart from Tan and Balasubramanian (2017) who explored dif- of a set of rotating knives which is propelled by a shaft through the ferent analytical tools to measure particle size of chocolate refined with 60 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 the melanger, more comprehensive information including the different and big drums) was also carried out at 1.5, 3.0 and 4.5 h durations. possibilities by which the melanger along with other equipment could Since the melanger operates at room temperature (ca. 27 °C), it was be combined for small-scale chocolate production is scarce. The aim of necessary to monitor the temperature of the refining chocolate mass this study was therefore to assess the applicability of the melanger and during the entire process. Generally, temperature rose to approximately Stephan mixer for chocolate refining and conching respectively in 40 °C due to internal friction. However, at fat content of 27%, refining various small-scale alternative chocolate manufacturing techniques beyond 3 h with the mini drum resulted in a dramatic rise in tem- whilst comparing their outcome to those from conventional processes. perature (= 78 °C) of the chocolate mass. Meanwhile, the maximum temperature allowed for the operation of the melanger, according to the 2. Materials and methods manufacturer is 80 °C. For this reason, the process was discontinued beyond 3 h. For the same reason, refining with the big drum at 27% fat 2.1. Sample preparation (from bean to liquor) was not carried out in order to avoid the risk of damage through overheating. Following investigations by both PSD and microscopic Fully fermented and sun-dried cocoa beans (hybrid type Forastero) analyses, one optimal setting from each equipment was selected for were obtained from a farm in the Brong-Ahafo region of Ghana. further chocolate production. For the 3-roll refiner, the gap setting 2–1, Roasting was carried out at 135 °C for 35min in a MIWE Roll-in 1.0608- at 27% fat was selected, whereas, for the melanger, refining with the TL baking oven (Miwe, USA). After cooling down to room temperature, mini drum for 3 h at 40% fat was selected. Selection was done on the beans were deshelled using a Winn-15 winnower (Cacao Cucina, basis of maximum particle reduction in both equipment, and ad- U.S.A). Cocoa nibs (1.5 kg batch size) were first pre-broken using the ditionally, in the case of the melanger, for the shortest possible refining Stephan mixer at 45 °C. First, 8 min at 50% speed, then, 6min at 75% time. speed. Thereafter, nibs were then ground into liquor using ECGC- 12SLTA CocoaTown melanger (CocoaTown, Roswell, USA). For this, 2.2.1.2. Chocolate production. Two batches of 70% dark chocolate the big drum (1.5 kg batch size) was used for a duration of 150min in (total fat= 43%) consisting of 30% pre-broken sugar (Barry Callebaut order to achieve a final particle size (D90) of 26.5 μm based on the Belgium, Wieze, Belgium), 64.65% cocoa liquor, 5% cocoa butter finding from preliminary trials with both the mini and big drums (Puratos - Belcolade, Erembodegem, Belgium) and 0.35% soy lecithin (Appendix A). (Soya International Ltd, Cheshire, U.K.) were produced on a 5 kg scale (Figs. 3 and 4). Mixing was carried out using the VEMA BM 30/20 2.2. Experimental design planetary mixer (Machinery Verhoest NV/Vema Construct, Izegem, Belgium) for a duration of 20min at 45 °C. For the first batch, the mixed The study was conducted in two set-ups. The first set-up was focused ingredients (27% fat) was refined with the Exakt 80S 3-roll refiner on the stage of refining, where both the mini and big drums of the (Exakt Technologies, inc., USA) at gap setting 2–1, roller speed of melanger were studied in comparison to the 3-roll refiner for particle 400 rpm and temperature of 35 °C (referred to as choc 1). However, the size reduction. The outcome were evaluated on the basis of particle size second batch of mixed ingredients (40% or full fat) was refined using and microscopic imaging analyses. Thereafter, an optimal setting for the mini drum of the ECGC-12SLTA CocoaTown melanger (CocoaTown, the melanger was selected and further investigated in reference to the Roswell, USA) for a duration of 3 h (referred to as choc 2). In each case, 3-roll refiner for the impact of the different refining equipment on some the resulting refined chocolate mass was conched in a Bühler ELK'olino quality attributes of the final chocolate. The second set-up was also conche (Richard Frisse GmbH, Bad Salzuflen, Germany) in two phases. focused on the conching stage, where two processing parameters were The dry phase was carried out at 60 °C with 1200 rpm for 2 h explored using the Stephan mixer in a conching-like process. Hitherto, (clockwise) and 80 °C with 1200 rpm for 4 h (anti-clockwise). At the the impact of these processing parameters were also evaluated with liquid phase, calculated amounts of pre-conched cocoa liquor, cocoa respect to the same quality attributes of the end chocolates as in set-up butter and the soy lecithin were added, such that, the final fat content 1. Summary of the processes involved in both set-ups have been out- of the chocolate was 43%. Here, the process was carried out as follows; lined in Figs. 2–4. 45 °C with 2400 rpm for 15min (clockwise) and 15min (anti- clockwise). Specifically in the case of choc 1, pre-conching of part of 2.2.1. Experimental set-up 1 the cocoa liquor was necessary since the entire amount of cocoa liquor 2.2.1.1. Comparing melanger and 3-roll refiner for particle reduction during required to produce final chocolate consisting of 70% cocoa could not chocolate refining. Two different recipes were tested at different settings be included in the recipe before refining. This is because; for the of the ECGC-12SL CocoaTown melanger (CocoaTown, Roswell, USA) refining mass, a final fat content of 27% was required for optimal and the Exakt 80S 3-roll refiner (Exakt Technologies, inc., USA) (Fig. 2). processing by the 3-roll refiner. Hence, for this batch, a calculated Thereafter, particle size distribution (PSD) of samples before and after amount of cocoa liquor was previously dry-conched using the same dry refining were analyzed and compared. The first recipe had a total fat conching procedure, which was then subsequently added at the stage of content of 27% - being optimal for the 3-roll refiner, whereas the liquefaction in order to make up for this final concentration. However, second recipe, also referred to as “full fat”, had a total fat content of for choc 2, this was not the case because, here, since the melanger 40% as is recommended for the smooth running of the melanger. The refines at full fat, the entire amount of cocoa liquor required for the amounts of cocoa liquor and sugar needed were calculated on the production of 70% cocoa chocolate could be included at the mixing account of the previously determined fat content of the cocoa liquor. stage prior to refining. For the recipe 1 (27% fat), 900 g of sugar and 780 g of cocoa liquor were mixed for 20min in a VEMA BM 30/20 planetary mixer 2.2.2. Experimental set-up 2 (Machinery Verhoest NV/Vema Construct, Izegem, Belgium) at a Six 70% dark chocolates were produced with the same composition constant temperature of 45 °C. Whereas, for recipe 2 (40% fat), 900 g of ingredients just like choc 2. However, these were produced on a 1 kg sugar was mixed with 1939.5 g cocoa liquor using the same mixing scale due to the capacity of the Stephan mixer. Mixing of the sugar and equipment/process. cocoa liquor was carried out in a Hobart mixer (Troy, USA) for a The 3-roll refiner was operated at a constant temperature of 35 °C duration of 20min at 45 °C just like in the VEMA mixer. However, here, and roller speed of 400 rpm with two different gap settings 3 - 1 and 2 - the temperature of the mixing ingredients was maintained by means of 1. This means that the gap in between the first two rolls was adjusted to a heat gun (BOSCH, Germany). Thereafter, the mixed ingredients was either 3 or 2, while the gap in between the second and last roll was refined in the mini drum of the ECGC-12SLTA CocoaTown melanger always kept constant at 1. However, refining in the melanger (both mini (CocoaTown, Roswell, USA) as described in set-up 1. Next, the Stephan 61 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 Fig. 2. Chocolate refining using melanger and 3-roll refiner. Fig. 3. Outline of chocolate productions using conventional and alternative means. 62 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 Fig. 4. Conventional and alternative chocolate production steps. mixer (STEPHAN food service equipment GmbH, Hameln, Germany) (magnification=20× ). In the case of the former, 0.5 g of each molten was used for mimicking the conventional conching process. For dry sample was first diluted in 10ml isopropanol (VWR, Leuven, Belgium), conching, it was operated at blade speed of 1500 rpm for a duration of homogenized by shaking and then after, a representative drop was 60min. In order to facilitate moisture reduction and the loss of un- brought to the glass slide using a Pasteur pipette. A cover slip was then wanted volatiles during the process, a vacuum pump (KNF Neuberger, carefully placed on the sample. It was then mounted on the sample Inc., USA) was connected to the Stephan mixer as shown in Fig. 4. The holder (isothermal at 50 °C) for visualization. However, samples were pump was operated at – 0.7 bar. Here, a 2× 3 full factorial design was observed “as-is” under the polarized light. used, consisting of dry conching temperatures; 60 °C and 80 °C, and vacuum durations, 0, 30 and 60min. After the dry conching, the va- cuum pump was detached, the required amount of cocoa butter and 2.4. Scanning electron microscopy (SEM) lecithin were then added to the chocolates. Liquefaction was then carried out at 50 °C for 15min with blade speed of 1500 rpm according In other to enhance visualization, the samples were first partially to Saputro et al. (2016b). The resulting chocolates were hereby referred defatted by dissolving 0.5 g sample in 10ml isopropanol (VWR, Leuven, to as choc 3A, 3B, 3C, 3D, 3E, and 3F representing dry conching con- Belgium), filtered over Whatman No. 40 filter paper and dried in an ditions (temperature/vacuum duration); 60 °C/0min, 60 °C/30min, oven at 50 °C for 1 h. The surface morphology of the partially defatted 60 °C/60min, 80 °C/0min, 80 °C/30min, and 80 °C/60min respec- cocoa liquor and chocolate were then visualized using a JSM-7100 F tively. TTLS LV TFEG-SEM (Jeol Europe BV, Zaventem, Belgium) under high vacuum and at an accelerated voltage of 2 keV. Prior to electron beam 2.3. Light microscopy targeting, the samples were vitrified in liquid nitrogen and transferred to a PP3000T device (Quorum Technologies Ltd., East Sussex, UK) at Microstructures of cocoa liquor, the different refined chocolate −140 °C. Here, the samples were allowed to sublime for 15min at masses and finished chocolate were observed using a Leica DM2500 −70 °C in order to remove frost artifacts. Prior to the transfer from the microscope (Wetzlar, Germany) equipped with a temperature con- cryo-preparation room to the SEM chamber, a thin layer of a conductive trolled sample holder (Linkam Scientific Instrument Ltd, Surrey, UK). metal (Pt) was deposited on the samples. This sputter coating process Samples were observed under both normal and polarized light prevents charging of specimens with an electron beam. 63 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 Table 1 Particle size profiles of refined chocolates using melanger and 3-roll refiner. Sample Derived Diameter Distribution Percentiles Other D(4,3) (μm) D(3,2) (μm) D(v,0.1) (μm) D(v,0.5) (μm) D(v,0.9) (μm) SSA (m2/m³) Span FULL FAT Before refining (Full fat) 31.00 ± 2.76a 3.16 ± 0.05b 1.70 ± 0.04b 12.21 ± 0.39b 90.12 ± 10.57a 1.90 ± 0.03a 7.12 ± 0.69a Mini drum 1.5 h (Full fat) 15.62 ± 0.84de 2.95 ± 0.03de 1.59 ± 0.06ef 9.44 ± 0.43f 37.87 ± 3.39d 2.03 ± 0.02e 3.84 ± 0.18f Mini drum 3 h (Full fat) 10.68 ± 0.27g 2.59 ± 0.03g 1.34 ± 0.02h 7.80 ± 0.12g 23.86 ± 0.64g 2.32 ± 0.03c 2.89 ± 0.04i Mini drum 4.5 h (Full fat) 10.34 ± 0.12gv 2.59 ± 0.06g 1.39 ± 0.01h 7.57 ± 0.03g 22.89 ± 0.26g 2.32 ± 0.05c 2.84 ± 0.02i Big drum 1.5 h (Full fat) 18.58 ± 0.95c 3.08 ± 0.07cd 1.68 ± 0.05cde 10.16 ± 0.36ef 48.27 ± 3.51c 1.95 ± 0.05ef 4.58 ± 0.19d Big drum 3 h (Full fat) 15.65 ± 0.38de 2.78 ± 0.11ef 1.49 ± 0.08g 9.49 ± 0.24f 39.62 ± 0.99d 2.16 ± 0.09d 4.02 ± 0.07e Big drum 4.5 h (Full fat) 12.08 ± 0.89fg 2.77 ± 0.07f 1.52 ± 0.06fg 8.23 ± 0.29g 27.27 ± 2.42efg 2.17 ± 0.06d 3.12 ± 0.18h 3-roll refiner (3-1)(Full fat) 15.59 ± 0.05de 3.25 ± 0.05c 1.74 ± 0.02cd 10.63 ± 0.11de 36.80 ± 0.12d 1.85 ± 0.03f 3.30 ± 0.05h 3-roll re ner (2-1)(Full fat) 13.67 ± 0.24ef 3.11 ± 0.07cd 1.70 ± 0.02cdfi 10.00 ± 0.08ef 30.87 ± 0.74e 1.93 ± 0.04ef 2.92 ± 0.06i 27% FAT Before refining (27% fat) 32.36 ± 2.05a 3.77 ± 0.07a 2.14 ± 0.06a 14.88 ± 0.75a 89.75 ± 5.99a 1.60 ± 0.03b 5.87 ± 0.24b Mini drum 1.5 h (27% fat) 24.67 ± 3.67b 3.22 ± 0.17c 1.77 ± 0.12c 11.92 ± 1.84c 67.42 ± 8.97b 1.87 ± 0.10f 5.52 ± 0.11c Mini drum 3 h (27% fat) 24.31 ± 0.85b 3.15 ± 0.04c 1.68 ± 0.02cde 11.78 ± 0.48c 65.39 ± 2.02b 1.91 ± 0.02f 5.42 ± 0.06c Mini drum 4.5 h (27% fat) Nd Nd Nd Nd Nd Nd Nd Big drum 1.5 h (27% fat) Nd Nd Nd Nd Nd Nd Nd Big drum 3 h (27% fat) Nd Nd Nd Nd Nd Nd Nd Big drum 4.5 h (27% fat) Nd Nd Nd Nd Nd Nd Nd 3-roll refiner (3-1)(27% fat) 17.35 ± 0.23cd 3.12 ± 0.08cd 1.77 ± 0.01c 11.54 ± 0.23cd 41.97 ± 0.50d 1.93 ± 0.05ef 3.49 ± 0.03g 3-roll re f cdfiner (2-1)(27% fat) 12.83 ± 0.46 3.08 ± 0.24 1.67 ± 0.08de 9.52 ± 0.33f 28.97 ± 0.94ef 1.96 ± 0.15ef 2.87 ± 0.01i Different superscripts in a column indicate significant differences (p < 0.05). 2.5. Particle size distribution (PSD) stress at 5 s−1 for ramp up and ramp down measurements (Afoakwa, 2010). All analyses were done in triplicate. Particle size distribution (PSD) of refined masses and final choco- lates were determined with a Master Sizer S (Malvern Instruments Ltd., τ = τCA + ηCA . γ˙ (1) Worcestershire, UK), a laser diffraction particle size analyzer, equipped with a 300 RF lens and an active beam length of 2.4 to measure par- Casson model (τ=shear stress, τCA=Casson yield value, ηCA=Casson ticles in a range of 0.05–900 μm. First, 0.5 g of each molten sample was plastic viscosity, γ̇ =shear rate). diluted with 10ml of isopropanol (VWR, Leuven, Belgium) and placed in an oven at 50 °C for an hour. Then the suspension was further homogenized by shaking it with a vortex mixer. After this, a pasteur 2.8. Color pipette was used to inject a few drops of the suspension into the system, such that an obscuration between 20% and 30% was attained. The Color parameters of the molten chocolates were determined using a background was measured beforehand, at the same speed as the sample. Minolta Model CM-2500D spectrophotometer (Konica Minolta Sensing, Each sample was prepared in triplicate, and quintuplicate measure- Inc., Osaka, Japan). Only values of the SCE (Specular Component ments were taken per replicate. The percentiles; D (v, 0.1), D (v, 0.5) Excluded) were considered as these are claimed to be more correlated and D (v, 0.9) being the dimensions of which respectively 10%, 50% with observations of the human eye. The color was expressed in terms and 90% of the particles were smaller, Sauter mean diameter; D (3,2), of L* (lightness component), a* (green to red component) and b* (blue volume mean diameter; D (4,3), specific surface area (SSA) and the to yellow component). Here, five repeated measurements were taken span were the PSD parameters reported. Whereas the SSA was calcu- per replicate of a sample. lated based on the D (3,2), the span was calculated based on the parameters from the volume-weighed distribution. The latter provides an indication of the uniformity of the particles within the chocolate 2.9. Statistical analysis system (Ziegler and Hogg, 1999). Statistical analysis was performed with Minitab 18 (Minitab Inc, 2.6. Moisture content USA). For set-up 1, the different responses were subjected to Analysis of Variance (ANOVA) with a 5% significance level. Assumptions of nor- Moisture analysis of chocolates were carried out in triplicate ac- mality and equality of variance were tested prior to the analysis using cording to the AOAC (2005) method 931.04. Kolmogorov-Smirnov test and Modified Levene's test, respectively. Where assumptions were fulfilled, a post-hoc Tukey's test was used to 2.7. Flow behavior investigate significant differences between the different levels of the predictors. However, when assumption was not fulfilled, a non-para- The flow behavior of the chocolates were determined according to metric alternative, Welch was used along with Games Howell post-hoc Analytical method 46, stated by ICA (2000). An AR2000ex rheometer test. In set-up 2, a general linear model (GLM) was used to explore the was used, coupled with a concentric DIN cylinder geometry (cup and impact of factors (conching temperature and vacuum duration) and bob; stator inner radius of 15mm, rotor outer diameter of 14mm, cy- their interaction effect on the various responses. Here, assumptions of linder immersed height of 42mm and gap of 5920 μm). The tempera- normality of residuals, linearity of the covariate effect and constancy of ture of the rheometer was set at 40 °C. Chocolates were melted for 1 h at the variance were also verified. 52 °C before sampling 20 g for analysis. After, the data was fitted to the Casson model (Equation (1)) from which the Casson yield value (τCA) and Casson plastic viscosity (ηCA) were obtained. The value of thixo- tropy was also obtained by computing the difference between the shear 64 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 3. Results and discussion processes. From Fig. 5, a clear difference is seen. NLM revealed the micro- 3.1. Particle properties of refined chocolate mass using melanger and 3-roll structure of cocoa liquor being dominated by black circular spots re- refiner presenting solid cocoa particles with sizes range<20 μm. However, in the chocolate, additional sugar particles are represented by clear crys- 3.1.1. Particle size distribution tals with jagged irregular shapes and sharp edges which consist of ap- From Table 1, a clear downward trend in PSD with refining time proximate sizes ≤30 μm. This irregular breaking pattern is due to their was observed. Also, at the different fat contents, both mini and big brittle nature under mechanical stress during the process of refining drums proved to be effective in significantly (p < 0.05) reducing (Beckett, 2009). Similarly, this difference is also seen from PLM al- particle size with increasing refining time. Consequently, SSA also in- though under polarized light, only the presence of sugar crystals are creased. Meanwhile, for a given duration, particle size parameters were seen in the chocolate with a dense microstructure. Images from SEM slightly lower in the mini drum compared to the big drum. This may be were more detailed; consisting of a more loose packing of cocoa par- attributed to the difference in geometry of the roller stones (Fig. 1 and ticles of different sizes and shapes overlaid with some amount of fat B.1), as it dictates the linear speed gradient of product during the re- crystals due to partial defatting of the samples. Additionally, other el- fining process. Tan and Balasubramanian (2017), who reported similar lipsoidal components within the matrix which appeared to retain their findings, suggested that at constant rotational speed of the roller stones, shape in spite of the grinding and refining processes were observed. the linear speed of the product at the outer edge of the cylindrical roller According to Beckett (2009), milled cocoa particles appear as small stones in the mini drum is greater than that at the inner edge due to the platelets, but also included are cocoa starch granules. Of these, the difference in distance between the two edges of the roller stones to the starch granules are known to contribute roughly 7% of the weight of the center of the drum. This therefore creates a resultant linear speed liquor. Their sizes range from 2 to 12.5 μm, thus, due to their small gradient between the refining mass at the center and that at the walls of sizes, making it possible for them to retain their ellipsoidal shape even the rotating drum. It is suggested that this gradient consequently in- after milling or refining. The same was confirmed by Afoakwa (2016), troduces shearing forces which may have contributed to the crushing of who reported 6.1% starch granules in dried Forastero cocoa beans. more cocoa and sugar particles. Meanwhile, in the case of the big drum, Unlike the cocoa liquor, a large (≤30 μm), broken sugar particle with a which is equipped with conical roller stones, an opposite scenario is somewhat rectangular shape was additionally found in the SEM image observed where almost no or very limited gradient is created during the of the chocolate. Similar crystalline sugar particles have also been de- refining process, hence, less shearing forces are generated leading to scribed by Saputro et al. (2017) who studied sugar crystals with the relatively minimal particle reduction for the same duration of refining. SEM. Although the differences in the amount of samples used in both drums Indeed, as long as cocoa liquor has been sufficiently grinded, the may have also contributed to this observation, early work by Tan and sole purpose of refining is to reduce the size of sugar particles. As in- Balasubramanian (2017) during grinding of cocoa nibs proved that this dicated earlier, the initial liquor consisted of an average particle size; D was indeed insignificant, in comparison to the impact of the geometry (v,0.9)= 26.5 μm. This means that except for the samples refined be- of the roller stones. yond 3 h at full fat using the mini drum, the other processes were less Comparing the different settings of the 3-roll refiner, it was evident effective in further reducing the particle size of the cocoa solids during that the smaller the roller gap, the smaller the particle size due to a the process. Notwithstanding, NLM images (Figs. 6 and 7) of the various more effective crushing of the solid particles. According to Do et al. samples were also found to be in agreement with the earlier trends from (2007), optimum particle size of dark chocolate after refining should Table 1. be < 30 μm, as larger particles result in gritty mouth feel. Among other PSD parameters, the D (v, 0.9), being an estimation of the largest 3.2. Chocolate quality attributes as affected by different processing particle sizes, may be used to represent the fineness of the chocolate, equipment or techniques since this has been demonstrated to correlate well with human per- ception (Beckett, 2009). Considering the values of D (v, 0.9) from 3.2.1. Moisture content Table 1, for the full fat recipe, these were obtained after 3 h and 4.5 h of The moisture content of chocolate is known to be a key factor in- refining with the mini and big drums respectively. Additionally, at both fluencing its rheological and textural attributes. Comparing the cho- fat contents, the D (v,0.9) from setting (2-1) of the 3-roll refiner were colates from set-up 1, no significant (p > 0.05) difference between also sufficiently reduced. The differences in particle sizes of refined moisture contents of the final chocolates was found in spite of the re- chocolate masses from both melanger and 3-roll refiner could be due to fining technique used. This is probably due to the fact that both cho- the differences in both the equipment and the throughput time. colates were conched with the same equipment. However, in the case of Whereas the former was operated between 1.5 and 4.5 h, it only took set-up 2, a demonstration of the removal of moisture due to the con- 5–10min to refine the same amount of product with the latter. ching parameters was observed in spite of the initial moisture contents of the cocoa liquor and pre-broken sugar, being (1.99% ± 0.11) and 3.1.2. Microscopy (0.26% ± 0.01) respectively. Here, the moisture contents of the cho- Since dark chocolate consists of 65–75% suspended solid particles, colates ranged from 0.52% ± 0.03 in choc 3F to 1.03% ± 0.09 in the sizes of these particles will have a huge influence on the micro- choc 3A (Table 2). A significant (p < 0.05) decline in moisture content structure, and subsequently, other physical and flow properties of the was observed with increasing duration of the vacuum pump (Table 3). final chocolate (Afoakwa, 2010). The shapes and sizes of the particles This is because at fixed temperature, the decrease in pressure due to the that form the chocolate matrix originate from the ingredients; in this vacuum, may have resulted in a consequent decrease in the boiling case, the refined cocoa solids and sugar particles in the matrix. Hence, point of water, thereby facilitating its evaporation from the chocolate in order to understand the evolution of the microstructural changes matrix. As such, the longer the duration of the vacuum created by the during the refining process, three microscopic techniques were applied, pump, the higher the tendency for moisture removal, hence, the lower namely; Normal Light Microscopy (NLM), Polarized Light Microscopy the resulting moisture content of the chocolates. Nevertheless, the (PLM) and Scanning Electron Microscopy (SEM). First a deeper un- moisture content was trivial between the two conching temperatures derstanding of the microstructure was sought by comparing the initial for the same duration of vacuum pump connection. This is evident from cocoa liquor to that of the final chocolate, obviously, with the presence Table 3, where, only the vacuum duration showed significant of sugar particles being the difference between the two matrices. After (p < 0.05) impact on the moisture content. Also, there was no sig- this one technique was chosen to follow through the different refining nificant (p > 0.05) interaction effect. Interestingly, the moisture 65 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 Fig. 5. Images of NLM (top), PLM (middle) and Cryo-SEM (bottom) showing the microstructure of cocoa liquor (left) and chocolate (right). A: cocoa particles overlaid with fat crystals; B: cocoa starch granules; C: sugar particle. contents of all chocolates were within the range of 0.5–1.5% as re- Considering set-up 2, it can be observed from Table 3 that whereas commended by Afoakwa (2010) to be an acceptable range without dry conching temperature only had significant (p < 0.05) impact on D drastic effects on their flow properties. (4,3), D (v, 0.1) and D (v, 0.5), all particle parameters but D (v, 0.1) were significantly (p < 0.05) influenced by vacuum duration. Of these, only D (3,2), D (v, 0.1) and the SSA recorded significant (p < 0.05) 3.2.2. Particle size distribution interaction effects of the two factors. As stated earlier, the D (v, 0.9) Considering set-up 1 from Table 4, the chocolate refined with the generally shows a direct proportional relationship with other PSD melanger recorded significantly (p < 0.05) smaller particle parameters parameters, except for the SSA, in which case, the opposite is observed. than the chocolate refined with the conventional 3-roll refiner. This It is therefore highly essential in explaining the finesses of the choco- obviously implies a notable impact of the type of equipment/technique lates as perceived by consumers. The D (v, 0.9), decreased significantly in dictating the final particle size of the chocolate. Similar to an earlier (p < 0.05) with increasing vacuum pump duration for the two con- observation in Table 1 and Fig. 7, it is possible that the presence of ching temperatures (Table 4). In retrospect, a similar decreasing trend excess fat available within the matrix of choc 2 (refined at full fat) may in moisture content with increasing vacuum pump duration was in- have been responsible for coating the newly created sugar surfaces in itially observed (Table 2). Afoakwa (2010) explained that the presence the system, thereby, limiting the possibly of water-induced agglom- of moisture in chocolate promotes various interactions among hydro- eration through moisture reabsorption from either the ingredients or philic particles, such as aggregation of sugar particles resulting in lump the surrounding (Beckett, 2009; Saputro et al., 2016a,b). However, this formation with a consequential increase in particle sizes. This idea may was likely the case in choc 1 which was roll refined at 27% fat content. explain the possible effect of the moisture content on the decline in Additionally, as earlier suggested, the different durations employed for particle sizes as the duration of vacuum pump connection was in- the refining processes could have also contributed partially to this ob- creased. Hereby, the more moisture was removed by the action of the served difference. 66 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 Fig. 6. NLM images revealing sugar and cocoa particles during refining with melanger. vacuum pump, the lesser the tendency to promote agglomeration impact of the refining equipment/technique on yield value is exhibited, among the hydrophilic sugar particles. Interestingly, the aforemen- whereby choc 2 recorded a significantly (p < 0.05) higher yield value tioned trends are fairly reflected in the other particle size parameters as than choc 1. This is due to the smaller particle size with higher SSA in well. the former. The work of Bolenz and Manske (2013) provided insights on the linear relationship between the yield value and SSA of the choco- 3.2.3. Flow behavior late. They explained that a higher yield value was as a result of the The rheology of chocolate is one of the crucial quality attributes as it existence of smaller particle sizes with higher SSA, engaging in various gives an indication of its flow behavior under various conditions. particle-particle interactions. This results in a more rigid microstructure Molten chocolate is a non-Newtonian fluid whose flow properties; e.g. which is resistant to the initiation of flow. Unlike, in set-up 1, no clear yield stress and viscosity, are largely dependent on the composition and trend in yield value was found from the chocolates in set-up 2. The interactions between the constituents of the chocolate. Of these, the reasons for this could have been due to the combined effect of moisture importance of such factors as PSD, fat content, moisture, amount and and particle size distribution of the chocolates, both of which have type of emulsifier as influenced through various processing steps (re- opposite influence on yield value. Notwithstanding, choc 3D – 3F fining, conching, tempering) cannot be overemphasized (Vavreck, seemed to reveal a slight decreasing trend in yield value with increasing 2004; Schantz and Rohm, 2005). vacuum duration. With this trend being similar to that of the moisture The yield value according to Beckett (2009), corresponds to the content, we may suggest that in this instance, the impact of moisture on stress needed to initiate flow and it is mainly determined by the par- the matrix may have played a more important role in dictating the yield ticle-particle interactions in the microstructure of the chocolate. Dif- value of the chocolates. This is also evident from Table 3, where all ferent chocolates may have different yield values, depending on their factors and their interaction effect – both contributing to moisture re- specific composition and application. From set-up 1 (Table 2), an duction – also contributed significantly (p < 0.05) to the observed 67 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 Fig. 7. NLM images revealing sugar and cocoa particles during 3-roll refining. outcome. more important role in determining the viscosity of the chocolates. In The viscosity of the chocolate describes its resistance to flow once this study, the chocolates, were manufactured with a total fat content of the motion has been initiated. Contrary to the yield value, choc 2 with 43%. This suggests that the chocolate with smaller particles may be less lower particle size also recorded the lowest viscosity. According to likely to impede motion within the chocolate. More importantly, con- Chevalley (1999), the PSD is more pronounced on the yield value than trary to the 3-roll refiner, it appears that the ability of the melanger to the plastic viscosity of the chocolate. Hitherto, the free fat content, sufficiently coat most of the newly formed sugar surfaces with the extra consistency and arrangement packing arrangement of the particles play fat, may have played an additional role in viscosity reduction by way of a major role. Withal, Beckett (2009) recognized that unlike the yield keeping the hydrophilic sugar particles further apart. This may also be value, a phenomenon where the viscosity decreases amidst fine parti- responsible for the observed outcome of chocolates from set-up 2, cles is not surprising. He ascribed this to the packing arrangement of the where all chocolates had comparable viscosities. Here, in spite of the suspended solids and the increasing amounts of unbound fat in the slight differences in particle sizes (Table 4), the more efficient coating chocolate system making it possible for particles to slide over each of particles due to the same refining technique using the melanger and other with ease during motion. Whilst studying the impact of particle the additional impact of high fat content of the chocolate may have size on the Casson yield and viscosity of chocolate, it was observed that been chiefly responsible for the viscosities of these chocolates. Hitherto, a unimodal distribution resulted in a much lower effect on the flow the difference between the observed trend in set-up 1 as opposed to that parameters than in the case of a bimodal distribution. Moreover, of set-up 2, can be attributed to the different refining and conching whereas no effect was seen on the viscosity at fat content beyond 34%, equipment used. From Table 3, it was therefore not surprising both the impact on the yield value rather persisted to 45% fat content. Thus, factors and their interaction appeared to have no significant (p > 0.05) at high fat content, the excess amount of free fat in the system plays a impact on the Casson viscosities of the chocolates. Obviously, only 29% Table 2 Moisture content, flow parameters and color of chocolates produced through different processing means. Chocolates Moisture (%) Casson yield value (Pa) Casson viscosity (Pa.s) Thixotropy (Pa) L* a* b* Set-up 1 Choc 1 0.63 ± 0.11a 1.8 ± 0.0b 1.8 ± 0.0a 0.8 ± 0.0b 19.4 ± 0.4b 12.7 ± 0.2a 13.7 ± 0.7a Choc 2 0.62 ± 0.16a 3.4 ± 0.1a 1.3 ± 0.0b 1.0 ± 0.1a 22.5 ± 0.6a 10.6 ± 0.3b 9.4 ± 0.5b Set-up 2 Choc 3A 1.03 ± 0.09A 3.0 ± 0.1BC 1.4 ± 0.0A 0.9 ± 0.0B 18.9 ± 0.2B 13.2 ± 0.1B 14.7 ± 0.4C Choc 3B 0.82 ± 0.05B 2.7 ± 0.0D 1.4 ± 0.0A 0.9 ± 0.0B 19.6 ± 0.3A 13.6 ± 0.1A 16.1 ± 0.4A Choc 3C 0.62 ± 0.03CD 2.9 ± 0.0C 1.4 ± 0.0A 0.9 ± 0.0AB 18.8 ± 0.1B 13.2 ± 0.1B 15.4 ± 0.2B Choc 3D 1.01 ± 0.02A 3.3 ± 0.1A 1.4 ± 0.0A 0.7 ± 0.1C 18.9 ± 0.1B 13.6 ± 0.1A 16.1 ± 0.4A Choc 3E 0.74 ± 0.03BC 3.2 ± 0.1AB 1.4 ± 0.0A 0.9 ± 0.1AB 19.4 ± 0.1A 13.1 ± 0.0B 14.3 ± 0.2C Choc 3F 0.52 ± 0.03D 2.9 ± 0.1CD 1.4 ± 0.0A 1.0 ± 0.1A 19.0 ± 0.0B 13.3 ± 0.1B 15.4 ± 0.2B Different alphabets (lowercase: set-up 1 and uppercase: set-up 2) in each column indicate significant differences (p < 0.05). 68 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 Table 3 ANOVA summary showing F-values of chocolate quality attributes with varying dry conching temperature and vacuum pump duration using the Stephan mixer. Factor Moisture (%) PSD profile Flow parameters Color values D (4,3) D (3,2) D (v,0.1) D (v,0.5) D (v,0.9) SSA Span Yield Viscosity Thixotropy (Pa) L* a* b* (μm) (μm) (μm) (μm) (μm) (m2/m³) value (Pa) (Pa.s) T 8.65 19.11* 2.67 8.53* 24.52* 10.52 2.61 0.5 57.60* 1.00 0.00 0.00 0.11 1.47 V 138.44* 27.35* 4.48* 4.56 8.58* 29.13* 4.67* 10.72* 22.80* 1.00 31.20* 41.87* 4.14* 2.34 T×V 1.5 0.56 7.13* 4.16* 2.42 0.06 7.25* 1.45 25.20* 1.00 16.80* 3.64* 57.83* 68.45* R2 (%) 96.0 86.2 68.3 68.4 79.5 85.2 68.8 67.4 92.75 29.41 88.89 79.13 83.79 85.63 T = dry conching temperature (⁰C); V = vacuum pump duration (min); regression coefficients with (*) are significant at α = 0.05. of the variability could be explained, leaving 71% to other factors 4. Conclusions (suggestively, the impact of fat content and refining equipment) which were not considered in the model. The CocoaTown melanger could be considered as an ideal alter- Thixotropy is a unique property used to describe the efficiency of native to the 3-roll refiner at the refining step during small-scale cho- the processing equipment/technique to coat the surfaces of the sus- colate production, provided that a recipe with moderate to high fat pended particles in the system with the continuous fat phase. Generally, content (ca 40% fat) is targeted. Refining for 3 h with the mini drum at Table 2 revealed that chocolates with smaller particles (higher SSA) full fat resulted in D (v,0.9) significantly (p < 0.05) lower than when also exhibited high thixotropic behavior, although, statistically, there is the 3-roll refiner was used. Nonetheless, a comparative advantage of no clear trend in the case of the chocolates from set-up 2. It is however the latter would be its short throughput time. Whereas the former re- clear that the higher the extent to which larger particles are fragmented quired 3 h of refining time, the latter required about 5–10min of op- into smaller ones during the processing, the greater the need to coat eration time to refine the same amount of chocolate. Generally, a better these new surfaces with the continuous fat phase. It is expected that a refining was achieved with the mini drum than for the big drum. This is well-conched chocolate exhibits minimal thixotropic behavior (Servais due to the difference in linear speed gradient of the product during the et al., 2002; Afoakwa, 2010). Interestingly, in spite of the equipment/ refining process, owing to the difference in geometry of the roller technique applied, all chocolates proved to be less thixotropic with stones. Also, refining with a full fat recipe (40% fat) may have con- values≤ 1 Pa. tributed to a more efficient coating of the newly created sugar surfaces in the presence of excess fat in the system. In spite of trivial difference in moisture content, chocolates manufactured following melanger and 3.2.4. Color 3-roll refining showed significant (p < 0.05) differences in terms of The color is a key contributor to the appearance of the chocolate PSD, flow parameters and color. In this study where the Stephan mixer and influences the perception of consumers. In comparison to the values was employed on a small-scale for conching dark chocolate, both pro- in Table 2, the initial liquor (L* = 15.24 ± 0.10, a* = 12.58 ± 0.08, cessing factors; dry conching temperature and vacuum pump duration b* = 12.36 ± 0.07) appeared much darker (lowest L* value) than the proved to be significant in dictating the various quality attributes of the chocolates. Similarly, a* and b* components had lower values. This is end chocolates. Whereas the vacuum duration had a significant obviously due to the composition as no sugar has been added to the (p < 0.05) impact on moisture content and D (v,0.9), a consequential liquor. As sugar is added, there appears to be a “dilution” effect on the impact of all factors and their interaction on the Casson yield was ob- intensity of the color of the chocolates. From set-up 1, choc 2 (refined served. On the contrary, these factors proved to be less important in with the melanger), which recorded smaller particle size, also appeared dictating the final viscosities of the chocolates. Among others, it is lighter (higher L* value) with lower redness (a*) and yellowness (b*). suggested that the role of the high fat content of the chocolates may Afoakwa et al. (2008) explained that smaller particles promote a more have played a more important role. Although all chocolates exhibited dense packing that scatters more light, and therefore leads to a lighter less thixotropic behavior, a direct proportional relationship between appearance. Chocolates from set-up 2 - which were conched with the the particle SSA and thixotropy was observed. Finally, it is possible that Stephan mixer - showed no clear trend in terms of the color compo- the impact of the vacuum pump on the rigidity of the chocolate matrix nents. This may be indicative of the comparable color attributes of these through moisture removal during conching may have contributed to the chocolates. various color intensities of the chocolates. However, the interaction effect of both factors also proved significant (p < 0.05) for all color Table 4 Particle size profiles of chocolates produced through different processing means. Chocolates Derived Diameter Distribution Percentiles Other D(4,3) (μm) D(3,2) (μm) D(v,0.1) (μm) D(v,0.5) (μm) D(v,0.9) (μm) SSA (m2/m³) Span Set-up 1 Choc 1 12.38 ± 0.22a 2.83 ± 0.08a 1.42 ± 0.03a 8.53 ± 0.12a 28.00 ± 0.61a 2.12 ± 0.06b 3.11 ± 0.03a Choc 2 11.08 ± 0.05b 2.60 ± 0.07b 1.34 ± 0.01b 7.86 ± 0.10b 25.25 ± 0.09b 2.31 ± 0.06a 3.04 ± 0.05a Set-up 2 Choc 3A 12.69 ± 0.08A 2.68 ± 0.04AB 1.38 ± 0.03AB 8.42 ± 0.06AB 30.16 ± 0.33A 2.24 ± 0.03BC 3.42 ± 0.03A Choc 3B 12.50 ± 0.17A 2.70 ± 0.07A 1.41 ± 0.02A 8.49 ± 0.14A 29.36 ± 0.29AB 2.23 ± 0.06C 3.29 ± 0.03AB Choc 3C 11.93 ± 0.04BC 2.51 ± 0.01BC 1.31 ± 0.03AB 8.19 ± 0.13BC 27.98 ± 0.02C 2.39 ± 0.01AB 3.26 ± 0.05B Choc 3D 12.37 ± 0.26AB 2.65 ± 0.02ABC 1.36 ± 0.02AB 8.26 ± 0.02ABC 29.41 ± 0.87AB 2.27 ± 0.02ABC 3.40 ± 0.10AB Choc 3E 12.03 ± 0.04BC 2.50 ± 0.08C 1.29 ± 0.05B 8.07 ± 0.11C 28.53 ± 0.09BC 2.41 ± 0.08A 3.37 ± 0.06AB Choc 3F 11.66 ± 0.27C 2.59 ± 0.11ABC 1.31 ± 0.05B 7.99 ± 0.15C 27.33 ± 0.68C 2.32 ± 0.10ABC 3.26 ± 0.02B Different alphabets (lowercase: set-up 1 and uppercase: set-up 2) in each column indicate significant differences (p < 0.05). 69 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 components. Acknowledgements The authors would like to express their profound gratitude to the Declaration of interest Belgian Government through VLIR-UOS for funding this project (ICP PhD 2014-001). Also a big thanks to the staff of Cacaolab bvba for their The authors declare that they have no conflict of interest. technical assistance. Appendix A Fig. A1D (v,0.9) in function of grinding time during particle reduction in mini drum. Fig. A2. D (v,0.9) in function of grinding time during particle reduction in big drum. Fig. B1. Gradients of linear speed of cocoa product in (a) cylindrical and (b) conical roller stone melangers (According to Tan and Balasubramanian, 2017). 70 M. Hinneh, et al. Journal of Food Engineering 253 (2019) 59–71 References Crystallization and rheological properties of soya milk chocolate produced in a ball mill. J. Food Eng. 114 (1), 70–74. http://doi.org/10.1016/j.jfoodeng.2012.06.024. Saputro, A.D., Van de Walle, D., Aidoo, R.P., Mensah, M.A., Delbaere, C., De Clercq, N., Afoakwa, E.O., Paterson, A., Fowler, M., 2008. Effects of particle size distribution and Dewettinck, K., 2016a. Quality attributes of dark chocolates formulated with palm composition on rheological properties of dark chocolate. Eur. Food Res. Technol. 226, sap-based sugar as nutritious and natural alternative sweetener. European Food 1259–1268. https://doi.org/10.1007/s00217-007-0652-6. Research and Technology. http://doi.org/10.1007/s00217-016-2734-9. Afoakwa, E.O., 2010. Chocolate Science and Technology. Wiley-Blackwell Publishers, UK. Saputro, A.D., Van de Walle, D., Kadivar, S., Mensah, M.A., Van Durme, J., Dewettinck, Afoakwa, E.O., 2016. Chocolate Science and Technology, second ed. Wiley-Blackwell K., 2016b. Feasibility of a small-scale production system approach for palm sugar Publishers, UK. sweetened dark chocolate. Eur. Food Res. Technol. 243 (6), 955–967. Aidoo, R.P., De Clercq, N., Afoakwa, E.O., Dewettinck, K., 2014. Optimisation of pro- Saputro, A.D., Van de Walle, D., Kadivar, S., Sintang, M.D.B., Van der Meeren, P., cessing conditions and rheological properties using stephan mixer as conche in small- Dewettinck, K., 2017. Investigating the rheological, microstructural and textural scale chocolate processing. Int. J. Food Sci. Technol. 49 (3), 740–746. http://doi.org/ properties of chocolates sweetened with palm sap-based sugar by partial replace- 10.1111/ijfs.12360. ment. Eur. Food Res. Technol. 243, 1729–1738. https://doi.org/10.1007/s00217- AOAC, 2005. Official Method 931.04 Moisture in Cocoa Products, eighteenth ed. 017-2877-3. Association of Official Analytical Chemists, Washington, DC. Saputro, A.D., Van de Walle, D., Hinneh, M., Van Durme, J., Dewettinck, K., 2018. Aroma Beckett, S.T., 2009. Industrial Chocolate Manufacture and Use, fourth ed. Blackwell profile and appearance of dark chocolate formulated with palm sugar-sucrose blends. Publishing Ltd. http://doi.org/10.1002/9781444301588. Eur. Food Res. Technol. https://doi.org/10.1007/s00217-018-3043-2. Bolenz, S., Manske, A., 2013. Impact of fat content during grinding on particle size dis- Schantz, B., Rohm, H., 2005. Influence of lecithin-PGPR blends on the rheological tribution and flow properties of milk chocolate. Eur. Food Res. Technol. 236 (5), properties of chocolate. Food Sci. Technol. 38 (1), 41–45. http://doi.org/10.1016/j. 863–872. http://doi.org/10.1007/s00217-013-1944-7. lwt.2004.03.014. Bolenz, S., Manske, A., Langer, M., 2014. Improvement of process parameters and eva- Servais, C., Jones, R., Roberts, I., 2002. The influence of particle size distribution on the luation of milk chocolates made by the new coarse conching process. Eur. Food Res. processing of food. J. Food Eng. 51 (3), 201–208. Technol. 238 (5), 863–874. http://doi.org/10.1007/s00217-014-2165-4. Steinberg, F.M., Bearden, M.M., Keen, C.L., 2003. Cocoa and chocolate flavonoids: im- Chevalley, J., 1999. Chocolate flow properties. In: Beckett, S.T. (Ed.), Industrial Chocolate plications for cardiovascular health. J. Am. Diet Assoc. 103, 215–223. Manufacture and Use. Blackwell Science, Oxford. Stephan, 2018. Universal Machines, Standard. 15/01/2018 at 10:00am. http://stephan- CocoaTown, 2017. CocoaT Melanger–ECGC–12SLTA. 22/2/2017 at 12:20 pm. https:// machinery.com/ww-en/machines/universal-machines-standard/. cocoatown.com/product/ecgc-12slta/. Tan, J., Balasubramanian, B.M., 2017. Particle size measurements and scanning electron Do, T.A.L., Hargreaves, J.M., Wolf, B., Hort, J., Mitchell, J.R., 2007. Impact of particle microscopy (SEM) of cocoa particles refined/conched by conical and cylindrical size distribution on rheological and textural properties of chocolate models with roller stone melangers. J. Food Eng. 212, 146–153. reduced fat content. J. Food Sci. 72 (9), 541–552. http://doi.org/10.1111/j.1750- Torres-Moreno, M., Torrescasana, E., Salas-Salvadó, J., Blanch, C., 2015. Nutritional 3841.2007.00572. composition and fatty acids profile in cocoa beans and chocolates with different Fistes, A., Rakic, D., Pajin, B., Dokic, L., Nikolic, I., 2013. The effect of processing para- geographical origin and processing conditions. Food Chem. 166, 125–132. http://doi. meters on energy consumption of ball mill refiner for chocolate. Hem. Ind. 67 (5), org/10.1016/j.foodchem.2014.05.141. 747–751. Vavreck, A.N., 2004. Flow of molten milk chocolate from an efflux viscometer under Fowler, M.S., 2009. Cocoa beans: from tree to factory. In: Beckett, S.T. (Ed.), Industrial vibration at various frequencies and displacements. Int. J. Food Sci. Technol. 39 (4), Chocolate Manufacture and Use. Blackwell Publishing, Oxford, pp. 137–152. 465–468. ICA, 2000. Viscosity of Cocoa and Chocolate Products, Analytical Method 46. CAOBISCO, Ziegler, G., Hogg, R., 1999. Particle size reduction. In: Beckett, S.T. (Ed.), Industrial Bruxelles, Belgium. Chocolate Manufacture and Use. Chapman & Hall, New York, pp. 182–199. Latif, R., 2013. Chocolate/cocoa and health benefits: a review. J. Med. 71 (2), 63–68. Pajin, B., Dokić, L., Zarić, D., Šoronja-Simović, D., Lončarević, I., Nikolić, I., 2013. 71