Reconstitution Properties of Sucrose and Maltodextrins - American

Jan 3, 2017 - angle.3 It is an important driver of the reconstitution performance (e.g., poor ... the objective to get a wider range of particle sizes...
0 downloads 0 Views 2MB Size
Article pubs.acs.org/Langmuir

Reconstitution Properties of Sucrose and Maltodextrins Julien Dupas,*,† Vincent Girard,† and Laurent Forny‡ †

Nestlé Product Technology Center Orbe, 3 Rte de Chavornay, CH-1350 Orbe, Switzerland Nestlé Research Center, Rte du Jorat 57, CH-1000 Lausanne 26, Switzerland



ABSTRACT: Carbohydrates such as sucrose and maltodextrins are commonly used in dehydrated food beverages. However, these ingredients may have, in some cases, negative impacts on the reconstitution performance (e.g., lump formation), compromising key consumer’s expectations. In this study, we propose to discuss the performance of carbohydrates with regard to major physical steps of reconstitution (wetting, capillarity, dispersion, and dissolution). We show how particle size and water temperature drive the kinetics of dissolution of crystalline sucrose and propose descriptive equations. For amorphous maltodextrin, we quantify variations in wetting, capillarity, and dissolution performance as a function of important solid properties (moisture content, molecular weight, and particle size) as well as the liquid temperature. By doing so, we highlight the important role of the glass-transition temperature in relation to the moisture content of the powder. The comprehensive understanding provided by this work may be used to optimize product formulation in term of reconstitution performance.



INTRODUCTION

upon dissolution (i.e., higher local viscosity leads to a higher contact angle).7,8 Capillarity is the process of water penetration in between particles or inside the pores of particles themselves. Capillarity in nonsoluble cylinders has been vastly described in the literature.9,10 Those studies evidence the influence of the contact angle, liquid viscosity, and capillary size on the dynamics of penetration. If water penetration in cylinders requires a contact angle below 90°, then it was recently demonstrated that water penetration in a bed of spheres can happen only at a critical contact angle far below 90°.11 In addition, penetration in a porous network was shown to depend on the tortuosity,12−14 which is difficult to evaluate in food powders. However, when dealing with soluble polymer ingredients, classical equations fail and laws of capillarity must be revisited. Dispersion may be described by the breaking up of particle agglomerates and aggregates into smaller units. Different types of interactions (e.g., van der Waals, hydrophobic) can impact dispersion behavior. Liquid15 or solid bridges2 can delay dispersion performance. In the case of maltodextrins, previous studies show that poor dispersion may be responsible for lump formation.16 The energy input (e.g., stirring speed and hydrodynamic flow), composition, and particle size are known to impact the dispersibility of food powders.15−17 Dissolution describes how fast a material is solubilized in a liquid. The dissolution of crystalline ingredients can be

The reconstitution of dehydrated beverages is of upmost importance for the food industry. Products sold in retail shops or in dispensing systems (i.e., capsules or vending machines) must be convenient for consumers (i.e., reconstitution has to be fast and free of lumps upon consumption). To meet these expectations, powders must be optimized upon formulation and processing. Among all ingredients used in dehydrated food beverages, carbohydrates are considered to be challenging by most product developers. Indeed, crystalline sugar is sometimes found undissolved at the bottom of the cup whereas amorphous maltodextrin is frequently responsible for lump formation. Unfortunately, the hurdles driving such limited reconstitution performance, which makes product optimization challenging, are not yet well understood. In order to improve the understanding of reconstitution, it is common to decouple this process into four different physicochemical steps, namely, wetting, capillarity, dispersion, and dissolution.1,2 Wetting describes the affinity between the material constituting the powders and the liquid used for reconstitution (e.g., water or milk). It is characterized by the so-called contact angle.3 It is an important driver of the reconstitution performance (e.g., poor wetting combined with small particle size and density is responsible for floating powders).4 The wetting of soluble food polymers is complex and was poorly investigated until recently. It is driven by the water content under the spreading liquid.5,6 Therefore, it depends on the sorption isotherm and diffusion properties at a given spreading liquid velocity. Wetting is also influenced by viscous effects © 2017 American Chemical Society

Received: December 6, 2016 Revised: January 3, 2017 Published: January 3, 2017 988

DOI: 10.1021/acs.langmuir.6b04380 Langmuir 2017, 33, 988−995

Article

Langmuir

sealed aluminum bags at 4 °C to avoid major risks of physical and chemical degradation in between experiments. Eight samples were generated by storing maltodextrins DE12 and DE19 at different relative humidities (Table 2). Thin layers of powders (approximately 5 mm thick) were placed in trays and stored in incubators under controlled temperature and humidity for several weeks until reaching the desired glass-transition temperatures. Glasstransition temperatures Tg were measured by differential scanning calorimetry (DSC1, Mettler Toledo). Tg was determined from the onset temperature of inflection occurring in the heat flow curve. Water content on the wet basis, Wwb, was obtained by thermogravimetric analysis (TGA Discovery, TA Instruments) using a heating rate of 2 °C/min.28 Prior to water activity, aw, measurements (model Aqualab 4TEV), all powders with Tg greater than room temperature were heated for 2 h at Tg in small sealed aluminum bags to ensure the moisture spatial equilibration of particles, as recommended in a recent study.29 Samples were taken out of storage (4 °C) one night before assessing the reconstitution time in order for the powder to reach room temperature (i.e., 22 °C). The samples with Tg below 25 °C started to cake after approximately 1 day (i.e., after the reconstitution analysis had been completed). Measurement Procedures. All sucrose powders (Si) and equilibrated maltodextrin powders (M12i and M19i) were characterized in terms of reconstitution kinetics by means of a refractive index probe (Fiso, reference: FRI-NP-C5-F1-M2-R1-ST). The miniature probe had an acquisition frequency of 0.1 s−1. It was assembled on a 400 mL double-jacketed vessel connected to an external water bath in order to control the temperature (Figure 1). Magnetic stirring was performed using a bar (cylindrical 50 × 8 mm2) rotating at 500 rpm. The experiment consisted of discharging 20 g of powder into 300 mL of demineralized water (i.e., leading to the final concentration of 6.25 wt %) at either 5, 22, or 80 °C. The time t90 required to reach 90% of the final measured refractive index of the solution was used as a characteristic time of reconstitution. In order to evaluate the wetting properties of maltodextrin, experiments were conducted in the framework of a parallel study with maltodextrins DE2, DE6, and DE29. Thin layers of maltodextrin were prepared by spin-coating solutions onto bare silicon wafers (ACM France). The thickness e of the resulting layers was measured by ellipsometry and ranged from 100 nm to 1 mm depending on the initial solutions concentrations and spin-coating parameters. The average roughness was less than 1 nm as measured by interferometric profilometry. For instance, at 4000 rpm, the thickness e = 250 nm was obtained with a concentration of 10 wt % for maltodextrin DE2, 11.8 wt % for maltodextrin DE6, and 15 wt % for maltodextrin DE29 in relation to the higher viscosity of low-DE maltodextrins. Contact angle data were measured under controlled relative humidity30 using salt solutions (RH = 11, 43, 58, or 75%). Layers were pre-equilibrated at the same relative humidity before analysis. Droplets of water (3 μL) were deposited on layers. The spreading of the droplet was recorded by two video cameras, one from the side and one from the top. The side view allows the simultaneous determination of contact angle θ and instantaneous velocity U. The top view allows the quantification of hydration behavior as described in a previous study.5The results are not discussed here. The wetting of crystalline sucrose was evaluated using tablets made of icing sucrose produced at high pressure (280 MPa) by means of a hydraulic press. Indeed, the preparation of thin layers by spin-coating would have led to the formation of amorphous sucrose. The contact angle between water and sucrose was evaluated by depositing 6 μL droplets on those dense tablets. The intrinsic dissolution kinetics of maltodextrin was investigated using a single-particle approach. Therefore, cuboids of the eight maltodextrin powders M12i−M19i (new batches used with slightly different Tg values) were prepared by high-pressure compaction with the objective to get blocks of material with the smallest possible porosity (ideally dense). The obtained cuboids have a width W and length L of 13.8 mm (digital calliper measurement). Height H ranged from 10 to 16 mm depending on the mechanical properties of the powders, leading to varying porosities. Indeed, since compaction

described by the diffusion-layer model,18,19 where the limiting step is the diffusion of molecules through a film of liquid around the surface of the solid. Other models for dissolution can be found in the literature, such as the interfacial barrier model20 based on the activation energy and Danckwerts’ model21 based on solute absorption or the negative crystal growth model.22 The dissolution of amorphous polymers is a more complex process that has been studied in a few papers.23−27,34 De Gennes and Brochard23 described the process in three characteristic steps: swelling associated with a cooperative diffusion coefficient, the gel phase where the stretching of polymer chains limits swelling, and finally the nonFickian escape of molecules governed by the diffusion coefficient of reptation. Recent studies highlighted that the gel lifetime would rather be controlled by osmotic pressure and gel permeability.34 In this study, we propose to compare the reconstitution performance of different carbohydrate powders and to discuss the differences in view of major physical properties at stake in the main steps of the reconstitution process as described above.



EXPERIMENTAL SECTION

Materials. Two different kinds of materials were used to study the impact of the physical state: crystalline sucrose and amorphous maltodextrins. Three sucrose powders were selected for their differences in particle size: sugar white fine bulk (Schweizer Zucker AG, Switzerland), sugar EGII fine (Agrana, Austria), and sugar icing (Central Sugar Refinery, Malaysia). For ease of reading, sucrose powders are respectively named S1, S2, and S3. The particle size distribution of the powders was characterized by means of laser diffraction (Malvern Mastersizer 3000 fitted with an Aero dispersion module set at a dispersion pressure of 2 bar). The particle size distribution gives access to different parameters such as d10, d50, and d90, respectively, and diameters such as 10, 50, and 90% of the volume of particles are below these values. The span s = (d90 − d10)/d50 describes the width of the particle size distribution. Seven additional samples (S1a, S1b, S1c, S3a, S3b, S1S2, and S1S3) were generated by mixing or sieving the initial powders (S1, S2, and S3) with the objective to get a wider range of particle sizes d50 and spans s. Table 1 summarizes the preparation method and size characteristics of the different sucrose powders used in this study.

Table 1. Name, Preparation Method, d50, and Span s of the Sucrose Powders Used in This Study ref

preparation method

d50 [μm]

s

S1 S1a S1b S1c S2 S3 S3a S3b S1S2 S1S3

commercial powder (sugar white bulk) S1 > 800 μm 500 μm > S1 > 800 μm S1 < 500 μm commercial powder (sugar EGII fine) commercial powder (sugar icing) S3 > 100 μm S3 < 100 μm mix 50% S1/50% S2 mix 30% S1/70% S3

533 994 633 372 194 54 146 32 279 86

1.38 0.93 0.81 1.07 1.45 3.36 1.66 2.42 2.47 6.74

Five maltodextrin powders were selected for their differences in molecular weight (i.e., maltodextrin DE2, DE6, DE12, DE19, and DE29 (Roquette, France)). The DE (dextrose equivalence) describes the average fraction of glycoside bonds after starch hydrolysis. Therefore, while maltodextrin DE2 presents very long molecular chains (Mw ≈ 343 000 g/mol), and maltodextrin DE29 is made of a few monomers (Mw ≈ 2500 g/mol). All maltodextrins present a polydispersity in molecular weight of greater than 5. The same batches of powders were used throughout the experiments and were stored in 989

DOI: 10.1021/acs.langmuir.6b04380 Langmuir 2017, 33, 988−995

Article

Langmuir

Table 2. Name, Preparation Method (Storage Conditions and Thermal Treatment), d50, Tg, Wwb, and aw at 25°C for the Maltodextrin Powders Used in This Study ref

dextrose equivalent (DE)

M12a M12b M12c M12d M19a M19b M19c M19d

12 12 12 12 19 19 19 19

storage conditions

thermal treatment

d50 [μm]

Tg [°C]

Wwb [%]

aw [−]

2 h, 100 °C 2 h, 65 °C 2 h, 35 °C

75 ” ” ” 161 ” ” ”

98 72 35 14 69 63 28 17

4.9 6.9 9.8 12.4 5.9 6.2 9.0 10.8

0.15 0.36 0.61 0.73 0.30 0.35 0.59 0.68

7 days, 30 °C, 40% RH 4 days, 25 °C, 65% RH 4 days, 20 °C, 70% RH

2 h, 70 °C 2 h, 55 °C 2 h, 30 °C

7 days, 30 °C, 40% RH 4 days, 25 °C, 60% RH 17 days, 20 °C, 70% RH

Figure 2. Characteristic dissolution time t90 of sucrose powders in water as a function of particle size d90 for three different temperatures. Dashed lines are guides to the eye.

Figure 1. Experimental setup used to characterize reconstitution time t90 of carbohydrate powders and cuboids.

of particles, which are both claimed in different contexts to improve the reconstitution of bulk powders. Indeed, we easily demonstrate with this example that optimizing the reconstitution performance is actually more complex than generic rules and depends on the powder properties in combination with the reconstitution conditions (e.g., here, the temperature). In the case of amorphous maltodextrin, Figure 3 provides characteristic times of reconstitution t90 as a function of powder glass-transition temperatures for both maltodextrin DE12 and DE19 at three different water temperatures (5, 22, and 80 °C). The first observation that can be made is about the effect of the water temperature. As expected, the higher the temperature,

generates heat, powders with decreasing glass-transition temperatures had the tendency to stick to the mold, thus preventing us from controlling the final height and porosity of the cuboids. (Our experimental device operates only at ambient temperature.) Consequently, cuboids made of M12a and M19a had a porosity down to 5−7% (pressures around 600 MPa) whereas cuboids made of M12c and M19c had a porosity of between 15 and 35% (pressures from 100 to 200 MPa) and powders M12d and M19d could not be compacted properly. Four additional cuboids were prepared with maltodextrin DE19 to investigate the effect of porosity at constant Tg (69 °C) by varying the compaction pressure (100, 200, 400, and 600 MPa). All cuboids were dissolved in water under the same conditions as for powders (i.e., at three different temperatures (5, 22, or 80 °C), with a water-to-maltodextrin ratio of 15 and under intense stirring (500 rpm)). The refractive index probe was used to follow the kinetics of dissolution. Data curves were then fitted with a simple linear model, allowing the assessment of a constant dissolution speed vd as a function of time such as W(t) = max(0, W − 2vdt), L(t) = max(0, L − 2vdt), and H(t) = max(0, H − 2vdt) with dissolved matter proportional to the lost volume dV(t) = WLH − W(t) L(t) H(t).



RESULTS Reconstitution Experiment. The characteristic dissolution times t90 of crystalline sucrose powders demonstrate the important role of particle size (Figure 2). Whatever the temperature, the reconstitution time decreases when the particle size goes from 1600 μm to approximately 300 μm (represented here with a d90 value to account for the effect of large particles). However, an increase in t90 occurs below 300 μm at low temperature (5 and 22 °C). This trend was visually explained by lump formation in the presence of fine particles. Additionally, as expected, the kinetics of reconstitution were found to be faster at higher temperature. Overall, these results highlight the optimum to be made between antagonistic formulation strategies such as agglomeration and micronization

Figure 3. Characteristic reconstitution time t90 of maltodextrins versus glass-transition temperature Tg for three different water temperatures (5, 22, and 80 °C from left to right). Dashed blue and red lines are guides to the eye. Dashed gray lines indicate the water temperature. 990

DOI: 10.1021/acs.langmuir.6b04380 Langmuir 2017, 33, 988−995

Article

Langmuir

function of the glass-transition temperature Tg. It is observed that the dissolution speed logically increases with water temperature, ranging from 0.95 μm/s at 5 °C up to 36 μm/s at 80 °C. Also, cuboids made of maltodextrin DE19 are found to dissolve approximately 1.5 to 3 times faster than those made of DE12. To understand if the intrinsic dissolution rate is actually higher or if porosity is involved, complementary experiments performed with cuboids produced at constant Tg and variable compaction pressures were compared to cuboids produced with variable Tg (Figure 6).

the faster the reconstitution. For powder M19a, values go from t90 = 18.5s at 5 °C to t90 = 1.4s at 80 °C. For powder M12a, values go from t90 = 31.4s at 5 °C to t90 = 2.5s at 80 °C. Looking at the effect of the glass-transition temperature, it may be observed that characteristic reconstitution time t90 usually increases with increasing glass-transition temperatures Tg (i.e., lower water content). This is especially true at 5 °C. A similar trend is observed at 22 °C with a noticeable exception for powder M12d, which exhibits the lowest Tg, below the water temperature. At 80 °C, the influence of Tg is not obvious (i.e., within the experimental error). Contact Angle Measurements. The contact angle of crystalline sucrose was found to be 15.8 ± 0.3° at a contact line velocity of U = 0.01 mm/s. For maltodextrin, this value varies between approximately 30 and 3° depending on the water activity and molecular weight at the same contact line velocity, U = 0.01 mm/s (Figure 4). In more detail, it may be observed

Figure 6. Dissolution speed at 22 °C of maltodextrin cuboids as a function of porosity.

Figure 6 shows a linear correlation between porosity and the dissolution speed of maltodextrin DE19 at 22 °C (red hollow squares) (i.e., the higher the porosity, the higher the dissolution speed vd). Experimental points obtained at variable compaction pressure and constant Tg follow the same trend as those obtained at variable Tg and uncontrolled pressure. By extrapolating the intrinsic dissolution rate at 0% porosity, it is found that maltodextrin DE19 would dissolve at approximately 2.5 μm/s. In the case of maltodextrin DE12, this would lead to 1.6 μm/s (i.e., slower than for maltodextrin DE19).

Figure 4. Contact angle of water on 250-nm-thick maltodextrin layers equilibrated at various water activities at a contact line speed of U = 0.01 mm/s. Dashed lines are guides to the eye.

that the higher the molecular weight (i.e., low DE), the higher the contact angle θ. Additionally, the higher the initial water activity (i.e., the lower the Tg), the lower the contact angle θ. Dissolution Kinetics. The results of the dissolution speed measurements with maltodextrins cuboids at three different temperatures are presented in Figure 5, where vd obtained by fitting the experimental refractive index curves is plotted as a



DISCUSSION In order to explain the kinetics of reconstitution, a detailed examination of the mechanisms at stake (i.e., wetting, capillarity, dispersion, and dissolution) needs to be performed for each type of carbohydrate (crystalline sucrose and amorphous maltodextrin). Case of Crystalline Sucrose. In the case of crystalline sucrose, it is assumed that the wetting property does not vary from one powder to another (i.e., no water absorption in the crystalline structure, limited adsorption and capillary condensation at the relative humidity of the laboratory, and no amorphization upon short grinding). Consequently, the relatively low contact angle (15.8°) and intense stirring used for this study ensures fast sinking of dense sucrose particles in liquid water whatever the particle size. The dispersion of large particles is observed to be relatively easy, presumably because crystalline sucrose does not swell or become rubbery in the presence of water. However, visual observations and the kinetics of reconstitution show that dispersion starts to be difficult at small particle size, typically below 250−300 μm. This trend is probably explained by the fast viscosity buildup in between small particles, hindering rapid particle dispersion (more surface area for dissolution in a smaller volume for diffusion).

Figure 5. Dissolution speed of maltodextrin cuboids as a function of Tg for three different water temperatures (5, 22, and 80 °C from left to right). Dashed lines are guides to the eye. 991

DOI: 10.1021/acs.langmuir.6b04380 Langmuir 2017, 33, 988−995

Article

Langmuir Above the critical diameter of 250−300 μm, the reconstitution kinetics of crystalline sucrose appeared to be linearly correlated to the particle size. This suggests that the reconstitution kinetics are completely dominated by the dissolution mechanism. In order to validate this hypothesis, we propose to apply a simple fitting method to refractive index curves for samples with particle size d90 larger than 300 μm. Assuming constant dissolution speed vd at a given temperature, a spherical particle of radius Ri will see its size decreasing in time according to eq 1. ⎧ R (t = 0) ⎪ R i(t = 0) − vdt , t < i vd ⎪ R i(t ) = ⎨ R (t = 0) ⎪ 0, t ≥ i ⎪ vd ⎩

assumed that larger interparticle and intraparticle pores favor capillary penetration, leading to faster sinking and dispersion. In this study, we propose to focus our discussion on the effect of moisture, which is not as well covered in the literature. As explained in previous studies,5,6 the contact angle of hydrosoluble substrates such as maltodextrin is highly influenced by its water volume fraction ϕ under liquid for a given thickness and spreading liquid velocity. Therefore, the sorption and diffusion properties of the powders are key parameters to be considered. The results of wetting experiments are in agreement with expectations. Fast diffusion improves the wetting properties, especially when the glass transition is reached. Therefore, higher water activity and lower molecular weight decrease the value of θ. In addition, viscosity effects have an influence on wetting.7,8 More specifically, viscous materials such as low-DE maltodextrins significantly increase the viscosity in the wedge of the spreading liquid, leading to higher θ (with typically θ3 ∝ η for clean solid substrates, where η is the liquid viscosity). The observations made for the wetting behavior of maltodextrins could partially explain why the reconstitution time decreases when the Tg decreases (i.e., aw increases). Indeed, the higher contact angle measured at higher Tg could be responsible for significant differences in capillary penetration. Measuring or predicting capillary penetration within or in between soluble maltodextrin particles is a very challenging task that remains open. To make some progress, we propose to use numerical simulations. We consider the case of water penetrating the cylindrical channel of maltodextrin. Dissolution and swelling are taken into consideration. Modeling considerations are developed in the Appendix. Figure 8 shows the

(1)

Considering the complete particle size distribution by volume N(r), the volume V of dissolved material is directly proportional to the expression given by eq 2. V∝

∑ i

N (R i ) R i(t = 0)3

(R i(t = 0)3 − R i(t )3 )

(2)

Therefore, the value of vd can be obtained from refractive index data by minimizing the differences between the normalized experimental curve and the theoretical curve as illustrated in Figure 7 for sample S1. Applying this approach to

Figure 7. Refractive index curve obtained for sample S1 fitted with the dissolution model. The particle size distribution used for the fit is given in the inset. A dissolution speed of 3.2 μm/s is obtained in that case. Figure 8. Matlab simulation of the capillary penetration in channels of maltodextrin DE29 with different wall thickness (75 or 161 μm) and pore radius (8 or 18 μm). Distance of penetration of the liquid (lines) and liquid viscosity (dashed lines) as functions of time.

all samples, we obtain a rather constant value of vd at each temperature: 3.5 ± 0.5 μm/s at 5 °C, 7.8 ± 0.8 μm/s at 22 °C, and 44.5 ± 2.9 μm/s at 80 °C. It confirms that dissolution is the key driver under these reconstitution conditions for crystalline sucrose (low concentration and high shear flow). Regarding the effect of temperature, the dissolution kinetics of crystalline sucrose is limited by the diffusion mechanism of sucrose molecules from the crystal surface to the solution.31 The diffusion mechanism is often described by an Arrhenius type of equation (i.e., in exp(−Ea/RT)).29,32 We found that our vd data are well fit with this equation (with an activation energy Ea of 27.3 kJ/mol), which confirms the predominance of diffusion. This result is of the order of magnitude expected for sugars.33 Case of Amorphous Maltodextrin. In the case of amorphous maltodextrin powders, it is common practice to agglomerate small particles in order to improve the reconstitution performance in water.16 In this way, it is

results of our simulation (performed in Matlab). The distance of liquid penetration and the local liquid viscosity are represented as functions of time. We consider maltodextrin DE29 for our simulations because all input parameters are available from this study and previous work.5,6,30 The wall thickness e of the capillary is taken to be equal to d50 values with regard to maltodextrin DE12 and DE19 powders (i.e., 75 or 161 μm). Similarly, channel radius R0 is calculated by assuming tetrahedral holes in between spherical particles (i.e., radius ratio on the order of 0.22 leading to 8 and 18 μm, respectively). Maximum swelling is defined at 5% for this simulation, in agreement with previous work.30 992

DOI: 10.1021/acs.langmuir.6b04380 Langmuir 2017, 33, 988−995

Article

Langmuir It may be observed in Figure 8 that penetration stops because of the rapid increase in viscosity (i.e., the so-called viscosity buildup). The distance covered is much longer in the case of large particles (i.e., large pore radius). This confirms that it is more difficult for the assembly of small particles to imbibe water and thus be dispersed, leading to a longer reconstitution time and possibly lump formation (i.e., wet shell and dry core). In other words, it explains why agglomeration favors reconstitution by increasing the interparticle pore size as well as the intraparticle pore size, as can be observed by scanning electronic microscopy. Such a simulation could also be used, if input parameters are known, to explain why higher molecular weight is detrimental to reconstitution due to higher contact angles as well as higher local viscosity even though the intrinsic dissolution speed is lower. In the future, advanced simulations of capillary penetration in a polydisperse bed of particles would aid the further understanding of all limiting factors. Quantification of the dispersion behavior of powders is very difficult, independently of other mechanisms at stake. A previous study16 shows that lumps of maltodextrin could be formed by a bad dispersion. The main negative parameters are a low DE and a small particle size whereas a high liquid temperature and high stirring intensity are usually positive to overcome cohesive interactions. In our experiments performed at a relatively high stirring rate, dispersion was usually not considered to be a significant hurdle for reconstitution except for maltodextrin DE12 at low Tg and intermediate temperature (22 °C). In this case, bad dispersion is attributed to increasing cohesive forces between particles since the liquid temperature is above the glass-transition temperature but not yet high enough to promote fast dissolution and low viscosity as is the case at 80 °C. At 5 °C, the liquid temperature is below the initial Tg and particles might have time to disperse before becoming sticky. Nevertheless, water uptake is expected to be fast in the amorphous matrix, and it is possible that the dispersion mechanism always plays a non-negligible role, which remains difficult to estimate in our experimental setup. Complementary techniques such as the focused beam reflectance measurement (FBRM) would be required to measure online variations of particle size and thus further evaluate the contribution of this mechanism. Finally, the last mechanism to consider is of course the dissolution itself. On the basis of the cuboids dissolution experiments, it was not surprising to find that the intrinsic kinetic of dissolution vd varies with temperature, which certainly contributes to the variation of reconstitution time t90 observed at various temperatures (Figure 5). The activation energies were found to be 22.3 kJ/mol for maltodextrin DE12 (highest Tg) and 24.8 kJ/mol for maltodextrin DE19 (highest Tg) (i.e., slightly below sucrose but still of the same order of magnitude). It shows that sucrose and maltodextrin are close in terms of the sensitivity of dissolution kinetics to temperature change. Applying the same fitting protocol as the one used with sucrose powders but to the refractive index curves obtained with maltodextrin powders allows us to obtain approximate values of vd, which are summarized in Figure 9 for 5 and 22 °C. They are compared to the values measured with the cuboids. It is clear that measured dissolution speeds are larger in the case of powders, especially for maltodextrin DE19. It demonstrates that dissolution is not the only mechanism at stake for amorphous powders and that wetting, capillarity, and dispersion are certainly engaged to various extents depending on the powder properties and reconstitution conditions. For instance,

Figure 9. Dissolution speed of maltodextrin at 5 °C (left) and 22 °C (right). Comparison between data measured on powders (first series) and cuboids (second series) for both maltodextrin DE12 (blue) and DE19 (red).

maltodextrin DE19 probably sees its reconstitution favored by its larger particle size, which improves capillarity, as discussed previously. Indeed, it is assumed that the only mechanisms allowing reconstituting faster than intrinsic dissolution speed are imbibition in the porosity of particles and the breaking up of loose agglomerates into smaller units (rather unlikely here given the small size of particles). The nonspherical shape of particles and similar porosity may also contribute to the observed difference. Regarding the effect of moisture, dissolution speed measurements on cuboids with constant Tg and variable porosity (Figure 6) show that the observed effect of the glass-transition temperature in Figure 5 is likely to be an effect of porosity. Therefore, the intrinsic dissolution speed is not or only slightly impacted by the moisture content. This means that the effect of the glass-transition temperature observed on the powders’ reconstitution time may rather be an effect of contact angle variations (impacting wetting and capillarity) or stickiness (impacting dispersion) but not really the dissolution speed.



CONCLUSIONS This study shows that the reconstitution of carbohydrates involves several mechanisms, among which one mechanism can dominate the others depending on various physical and operational parameters (e.g., particle size, initial moisture content, and temperature). Typically, the reconstitution of large crystalline sucrose particles under intense stirring was found to be limited by the kinetics of dissolution. It is therefore logically influenced by temperature. But below approximately 250 μm, the dispersion becomes a limiting factor probably because of the rapid local viscosity buildup between the particles. In the case of amorphous maltodextrins, complex particle structure as well as water diffusion and viscosity buildup were found to be responsible for additional effects on wetting, capillarity, and dispersion mechanisms. For instance, wetting and therefore capillarity as well as the overall reconstitution inprove when the water activity increases (or the glasstransition temperature decreases). However, in specific situations such as high molecular weight and intermediate reconstitution temperature (i.e., 22 °C), the reconstitution 993

DOI: 10.1021/acs.langmuir.6b04380 Langmuir 2017, 33, 988−995

Article

Langmuir performance gets worse when the water activity increases because of the adverse effect on dispersion that is probably due to the cohesive interaction between rubbery particles. This study illustrates the complexity of the reconstitution process and explains why the optimization of product performance is so challenging, especially without deep scientific insight.

The capillary pressure is the motor for penetration leading to ∫ dz = 2γ Rcos θ The integration of eq 4 using eq 3 allows us dP dz



4R 0 ∫

0

U (r , z)2πr dr = a(z)R

(5)

0

∂ϕ ∂ϕ ⎞ ∂ ⎛ ⎜D(ϕ) ⎟ = ∂x ⎝ ∂x ⎠ ∂t

(6)

where D(ϕ) is the diffusion coefficient of water in the material. In this study, we used a function fitting the data obtained in the framework of another study.30 Solving eq 6 numerically in Mathematica allows us to calculate the average water volume fraction ϕ and thus the swelling S at all z coordinates of the channel according to the contact time with water: S=

(3)

ϕ 1−ϕ

(7)

The channel radius is therefore adjusted according to the values of S(z) and e. With regard to experimental observations (for instance, the hydration of thin layers) and the literature,23 we impose a maximal swelling Sm. Indeed, in practice the swelling saturates because of polymer chain stretching. As explained in the Discussion section, the value of 5% was chosen.

The fluid incompressibility imposes that for small variations of R along z, Q can be considered to be constant at a time t along the channel such as Q(z, t) ≈ Q(t). The pressure gradient is then dP = η(z)ΔU⃗ |z = 4η(z) a(z) dz

dz

πR 0



2

η(z) R(z , t )4

radius R0, the walls in contact with water dissolve at a speed vd such as R(z, t) = R0 + vdtcontact, where tcontact is the contact time. vd can be a function of liquid concentration φ, moisture ϕ, and speed U. On the basis of dissolution data obtained for maltodextrin DE29,30 the dissolution speed function presented in Table 3 was used. (The moisture effect was not considered). Dissolution then leads to a viscosity increase, which is also characterized by an experimental function from internal studies on maltodextrin DE29. The changes in liquid speed, material moisture, and liquid viscosity also lead to a modified contact angle that is given in Table 3. (Viscosity and moisture effects are neglected in this study.) A simplified swelling model can also be applied by taking into account diffusion into the material as a function of contact time with water using a 1D diffusion equation

walls will dissolve, leading to an increase in radius R, liquid concentration φ, and liquid viscosity η. Moreover, the walls could also swell as a result of hydration from behind the spreading liquid (i.e., immersed part) but also potentially ahead of the liquid by vapor evaporation and absorption (neglected here). Considering a speed profile of U(r, z)= a(z)[R2 − r2] (lubrication assumption, with a(z) as an arbitrary function), the flow Q at height z can be written as

∫0

L

From Q, it is therefore possible to compute the capillarity behavior in a soluble channel using an increment of time dt. At each step in the calculation, the displacement of water is Q Q dt Starting from a dz = 2 and the average speed is U = πR 2

Figure 10. Water penetration by capillarity in a soluble channel. The definition of parameters is at stake.

Q (r , z ) =

πγ cos θ

Q (t ) =

APPENDIX: CAPILLARITY SIMULATIONS We describe in this section the approach used to simulate capillary penetration in soluble channels of maltodextrin. Let us consider a channel of length L, radius R0, and wall thickness e (Figure 10). Upon capillary penetration along the z axis, the

R

0

to obtain an expression for Q:

(4)

Table 3. List of Functions Used in the Capillarity Simulations parameter

expression

function for maltodextrin DE29

viscosity

η(φ)

contact angle dissolution speed diffusion coefficient

θ(U, φ, ϕ) vd(U, φ, ϕ) D(ϕ)

0.001(0.87 + 60φ + 1.8e φ− 0.43/0.031) 3 + 6U0.42 or 80° if above 80° [17.04 + 31 atan(30(U − 0.08))][1 − 166φ] or 0.05 μm/s if below 0.05 μm/s ⎡ −15 ϕ 3.224 ⎤ 10 + 1.92110−14 2.7 (1 − ϕ)2 below ϕ = 0.6 or above 8.1 × 10−11 m2·s−1 ⎣⎢ ⎦⎥

maximal swelling wall thickness channel radius

Sm e R0

constant value (0.05) constant value (particle size) constant value (tetrahedral holes)

2

( )

994

DOI: 10.1021/acs.langmuir.6b04380 Langmuir 2017, 33, 988−995

Article

Langmuir



(19) Nernst, W. Theorie der Reaktionsgeschwindigkeit in heterogenen Systemen. Z. Phys. Chem. 1904, 47, 52−55. (20) Zdanovskii, A. B. The role of the interphase solution in the kinetics of the solution of salts. Zhur. Fiz. Khim. (USSR). 1946, 20, 869−880. (21) Danckwerts, P. V. Significance of liquid-film coefficients in gas absorption. Ind. Eng. Chem. 1951, 43, 1460−1467. (22) Koiranen, T.; Kilpio, T.; Nurmi, J.; Nordén, H. V. The modelling and simulation of dissolution of sucrose crystals. J. Cryst. Growth 1999, 198/199, 749−753. (23) de Gennes, P. G.; Brochart, F. Kinetics of polymer dissolution. Physico Chemical Hydrodynamics. 1983, 4, 313−322. (24) Ranade, V. V.; Mashelkar, R. A. Convective diffusion from a dissolving polymeric particle. AIChE J. 1995, 41, 666−676. (25) Parker, A.; Vigouroux, F.; Reed, W.F. Dissolution kinetics of polymer powders. AIChE J. 2000, 46, 1290−1299. (26) Miller-Chou, B. A.; Koenig, J. L. A review of polymer dissolution. Prog. Polym. Sci. 2003, 28, 1223−1270. (27) Wang, Q.; Ellis, P. R.; Ross-Murphy, S. B. Dissolution kinetics of water-soluble polymers: The guar gum paradigm. Carbohydr. Polym. 2008, 74, 519−526. (28) Vuataz, G.; Meunier, V.; Andrieux, J. C. TG−DTA approach for designing reference methods for moisture content determination in food powders. Food Chem. 2010, 122, 436−442. (29) Dupas-Langlet, M.; Dupas, J.; Samain, S.; Giardiello, M. I.; Meunier, V.; Forny, L. A new method to determine “equilibrated” water activity and establish robust sorption isotherm based on wise thermal treatment. J. Food Eng. 2016, 184, 53−62. (30) Dupas, J. Wetting of Soluble Polymers. Ph.D. Thesis, UPMC, Paris, 2012. (31) Schwartzberg, H. G., Hartel, R. W. Physical Chemistry of Foods; M. Dekker: New York, 1992. (32) Frias, J. M.; Oliveira, J. C.; Schittkowski, K. Modeling and parameter identification of a maltodextrin DE12 drying process in a convection oven. Applied Mathematical Modelling. 2001, 25, 449−462. (33) Beauchamp, G. Dissolution Kinetics of Solids: Application with Spherical Candy. J. Chem. Educ. 2001, 78, 523. (34) Valois, P.; Verneuil, E.; Lequeux, F.; Talini, L. Understanding the role of molar mass and stirring in polymer dissolution. Soft Matter 2016, 12, 8143−8154.

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Julien Dupas: 0000-0002-7715-4978 Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS This work was carried out at Nestlé PTC Orbe and ESPCI Paris. The authors acknowledge all people who contributed to the experimental results either by helping to perform the experiments or by feeding the discussions during data consolidation, especially F. Lequeux, E. Verneuil, L. Talini, M. Ramaioli, F. Perotti, C. Salzmann, and J. Kienle.



REFERENCES

(1) Hogekamp, S.; Schubert, H. Rehydration of Food Powders. Food Sci. Technol. Int. 2003, 9, 223−235. (2) Forny, L.; Marabi, A.; Palzer, S. Wetting, disintegration and dissolution of agglomerated water soluble powders. Powder Technol. 2011, 206, 72−78. (3) de Gennes, P. G.; Brochart, F.; Quéré, D. Gouttes, Bulles, Perles et Ondes; Springer: New York, 2004. (4) Dupas, J.; Forny, L.; Ramaioli, M. Powder wettability at a static air−water interface. J. Colloid Interface Sci. 2015, 448, 51−56. (5) Dupas, J.; Verneuil, E.; Ramaioli, M.; Forny, L.; Talini, L.; Lequeux, F. Dynamic wetting on a thin film of soluble polymer: effects of nonlinearities in the sorption isotherm. Langmuir 2013, 29, 12572− 8. (6) Dupas, J.; Verneuil, E.; Van Landeghem, M.; Bresson, B.; Forny, L.; Ramaioli, M.; Lequeux, F.; Talini, L. Glass transition accelerates the spreading of polar solvents on a soluble polymer. Phys. Rev. Lett. 2014, 112, 188302. (7) Voinov, O. V. Hydrodynamics of wetting. Fluid Dyn. 1976, 11, 714−721. (8) Cox, R. G. The dynamics of the spreading of liquids on a solid surface. Part 1. Viscous flow. J. Fluid Mech. 1986, 168, 169−194. (9) Washburn, E. W. The dynamics of capillary flow. Phys. Rev. 1921, 17, 273−283. (10) Zhmud, B. V.; Tiberg, F.; Hallstensson, K. The dynamics of capillary flow. Dynamics of capillary rise. J. Colloid Interface Sci. 2000, 228, 263−269. (11) Raux, P.; Cockenpot, H.; Ramaioli, M.; Quéré, D.; Clanet, C. Wicking in a powder. Langmuir 2013, 29, 3636−3644. (12) Pezron, I.; Bourgain, G.; Quéré, D. Imbibition in a fabric. J. Colloid Interface Sci. 1995, 173 (2), 319−327. (13) Benltoufa, S.; Fayala, F.; Ben Nasrallah, S. Capillary Rise in Macro and Micro Pores of Jersey Knitting Structure. J. Eng. Fibers Fabrics 2008, 3, 47−54. (14) Rouquerol, J.; et al. Liquid intrusion and alternative methods for the characterization of macroporous materials (IUPAC Technical Report). Pure Appl. Chem. 2011, 84 (1), 107−136. (15) Galet, L.; Oulahna, T. O.; Fages, J. The Wetting Behaviour and Dispersion Rate of Cocoa Powder. Food Bioprod. Process. 2004, 82, 298−303. (16) Mitchell, W. R.; Forny, L.; Althaus, T. O.; Niederreiter, G.; Palzer, S.; Hounslow, M.; Salman, A. D. Mapping the rate-limiting regimes of food powder reconstitution in a standard mixing vessel. Powder Technol. 2015, 270, 520−527. (17) Freudig, B.; Hogekamp, S.; Schubert, H. Dispersion of powders in liquids in a stirred vessel. Chem. Eng. Process. 1999, 38, 525−532. (18) Dokoumetzidis, A.; Macheras, P. A century of dissolution research: From Noyes and Whitney to the Biopharmaceutics Classification System. Int. J. Pharm. 2006, 321, 1−11. 995

DOI: 10.1021/acs.langmuir.6b04380 Langmuir 2017, 33, 988−995