- Email: [email protected]
The microstructure and component distribution in spray-dried emulsion particles
Accepted Manuscript Title: The microstructure and component distribution in spray-dried emulsion particles Author: Marta Munoz-Ibanez Marine Nuzzo Christelle Turchiuli Bj¨orn Bergenst˚ahl Elisabeth Dumoulin Anna Millqvist-Fureby PII: DOI: Reference:
S2213-3291(16)30009-0 http://dx.doi.org/doi:10.1016/j.foostr.2016.05.001 FOOSTR 47
To appear in: Received date: Revised date: Accepted date:
4-1-2016 10-5-2016 13-5-2016
Please cite this article as: Munoz-Ibanez, Marta., Nuzzo, Marine., Turchiuli, Christelle., Bergenst˚ahl, Bj¨orn., Dumoulin, Elisabeth., & Millqvist-Fureby, Anna., The microstructure and component distribution in spray-dried emulsion particles.Food Structure http://dx.doi.org/10.1016/j.foostr.2016.05.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The microstructure and component distribution in spray-dried emulsion particles Marta Munoz-Ibaneza, Marine Nuzzob,c, Christelle Turchiulia,d, Dumoulina, Anna Millqvist-Furebyb*
Björn Bergenståhlc, Elisabeth
a
UMR Ingénierie Procédés Aliments, AgroParisTech, Inra, Université Paris-Saclay, 1 avenue des olympiades, 91300 Massy, France b
SP Technical Research Institute of Sweden, Box 5607, SE-114 86 Stockholm, Sweden
c
Lund University, Department of Food Technology, Engineering and Nutrition, P.O. Box 124, SE-221 00 Lund, Sweden d
Université Paris-Sud, IUT Orsay, Plateau de Moulon, F- 91400 Orsay, France
∗ Corresponding author. Tel.: +46 (0)10 516 60 46 E-mail address: [email protected] (A. MillqvistFureby).
1
Highlights
Confocal Raman microscopy is a powerful tool for internal mapping of particles.
Phase segregation occurs even during short drying time of particles.
Particle morphology of a given formulation is governed by emulsion processing, atomization and drying rate.
2
Abstract Microencapsulation by spray drying of oil-in-water (o/w) emulsions provides a means to encapsulate functional lipophilic ingredients. The active ingredient is dispersed in continuous solid phase providing protection. However, the encapsulation efficiency depends on the microstructure and morphology of the dry particles influenced by several mechanisms occurring during processing such as oil droplet breakup during atomization, ingredient diffusivity, interfacial adsorption of surface active ingredients, and drying kinetics. In this work, sunflower oil (model for lipophilic compounds) was encapsulated in solid particles composed of acacia gum and maltodextrin DE12. Three powders with different initial emulsion size (e.g. about 0.1 and 1 μm) and atomized under high and low shear rate were analyzed for the morphology and distribution of oil droplets and matrix constituents within the solid particle (20 to 100 μm). The microscopic (optical, SEM, LVSEM, confocal Raman), spectroscopic (XPS) and analytical (solvent extraction) techniques used were either qualitative or quantitative. Their combination made it possible to determine both the composition at the surface and inside the particle. The surface differs from the bulk in composition, confirming the constituent segregation during spray drying, and depended on the initial emulsion size and atomization conditions that must be controlled for an efficient encapsulation. Especially, the use of confocal Raman microscopy is promising for the study of processstructureproperties relationship. Keywords: Spray-dried emulsions; particle microstructure; Confocal Raman Microscopy, phase segregation.
3
1 Introduction Spray-drying is a common technology to encapsulate fats and oils in solid matrices to create a powder product with improved shelf stability and new functionality compared to a liquid emulsion. The functional properties of the powder, such as wetting, flowability, oxidation stability, etc. are strongly linked to the structure of the particles on different levels and the oil encapsulation efficiency (Kim, Chen, & Pearce, 2002; Nijdam & Langrish, 2006). Encapsulation of oil in powders is an area that has been researched over many years, and the encapsulation efficiency and stability of the encapsulated oil are two important parameters of practical relevance. There are several different aspects that influence the properties and microstructure of the powder, including formulation and processing conditions. The emulsion droplet size as well as particle size influence the oil encapsulation and surface oil (Jafari, Assadpoor, Bhandari, & He, 2008; Sarkar, Arfsten, Golay, Acquistapace, & Heinrich, 2016; Soottitantawat, et al., 2005; Soottitantawat, Yoshii, Furuta, Ohkawara, & Linko, 2003). However, the optimum emulsion droplet size to minimize surface oil depended on the oil type, emulsion processing technology and particle size. The increase in surface oil at larger oil droplet size has also been observed by several other authors (Jafari, et al., 2008; Millqvist Fureby, 2003; Soottitantawat, et al., 2005; Soottitantawat, et al., 2003). The instability of emulsion droplets during spray-drying was explained by the fast dehydration, leading to close contact between droplets, resulting in coalescence and weak stabilising surface layers on the droplets (Taneja, Ye, Jones, Archer, & Singh, 2013). This may contribute to the common explanation of high surface oil based on rupture of oil droplets at the particle surface (Fäldt & Bergenståhl, 1995; Kim, Chen, & Pearce, 2009; Nijdam, et al., 2006). The processing conditions during spray-drying were also linked to the oil encapsulation efficiency, where effects of shear have been put forward by Risch and Reineccius (1988) and Soottitantawat et al. (2005). This is further supported by Munoz-Ibanez, Azagoh, Dubey, Dumoulin, and Turchiuli (2015), who showed that the droplet breakup during the atomisation step of spray drying is influenced by the shear exerted on the droplets. This leads to decrease in droplet size, but it is not clear whether it is the droplet size per se, or other effects that cause the increase in surface fat for large initial emulsion droplet sizes. To extend the understanding of the complex mechanisms leading to oil encapsulation and the final microstructure of spray-dried particles different analytical tools for detailed characterisation are necessary. The effects of composition and processing for spray-dried dairy-like emulsions (Kim, et al., 2002; Kim, et al., 2009; Millqvist-Fureby, Elofsson, & Bergenståhl, 2001; Murrieta-Pazos, Gaiani, Galet, & Scher, 2012; Wu, et al., 2014) as well as other food protein and biopolymer formulations (e.g., Fäldt, Bergenståhl, & Carlsson, 1993; Fäldt & Bergenståhl, 1994; Fäldt, et al., 1995;; Nuzzo, Millqvist-Fureby, Sloth, & Bergenståhl, 2015) on the surface composition have been studied rather extensively. These studies show that the surface active components are over-represented at the dry particle surface, since these adsorb to the surface of the drying drop. The surface oil coverage, as determined by XPS in a surface layer of approximately 2-5 nm, can still be very high due to spreading of oil originating from ruptured oil droplets (Chew, Fu, Gengenbach, Chen, & Selomulya, 2015; Keogh et al., 2001; Millqvist-Fureby et al., 2001; Nikolova et al., 2015). The morphology of the particles is influenced by several parameters, such as the solids content in the feed, the drying rate (Walton & Mumford, 1999), and the composition and viscoelastic properties of the adsorbed surface film of the drying drop (Elversson & Millqvist-Fureby, 2006; Nuzzo et al., 2015a).
4
The composition of the internal structure on the other hand still remains largely theoretical. Drusch et al (2008) have studied how the emulsion droplets are located in spray-dried emulsion powders using confocal laser scanning microscopy (CLSM), and found that powders with higher oil content presented higher levels of surface oil and oil droplets close to the surface compared to powders with less oil content. CSLM has been used to provide information about particle internal structure in different types of milk powder (Auty, Twomey, Guinee, & Mulvihill, 2001). In this study the whole milk powder showed very low levels of surface fat, and an even distribution of oil droplets throughout the particles, while a high fat cream powder showed some surface fat and larger oil droplets. Recently, Sarkar et al. (2016) used CSLM to visualise the close packing of oil droplets in spray-dried emulsions with ultra-high oil content and cross-linked whey protein as the emulsifier and matrix material. However, CLSM requires fluorescent labelling of the components to be visualized, and is thus limited to the determination of components of sufficiently different properties, i.e. oil, carbohydrate and protein, and does not distinguish between components of similar dyeing properties, such as different polysaccharides. Confocal Raman imaging, on the other hand, allows for internal particle investigations with the same level of resolution as CLSM, but no labelling is required as the full Raman spectrum in each pixel of the image is used to construct an image of the localisation of the different molecular components. Recently, Nuzzo et al (2015b) showed phase segregation in spraydried powder particles and in individually dried particles composed of maltodextrin and hydroxypropyl methyl cellulose using confocal Raman microscopy. The same technique was also used to investigate the internal structure of single particles composed of lactose and BSA, HPMC or Poloxamer 188 (Nuzzo et al, 2015b). In these investigations, the more surface active macromolecules were located at the surface, and a layer with a thickness in the micrometre range enriched in this component was found at the surface. The aim of this paper is to elucidate how oil droplet breakup during atomization influences the localization of ingredients and the oil encapsulation in spray-dried powders composed of sunflower oil, maltodextrin and acacia gum. We address the distribution of dry emulsion components within the particles by means of confocal Raman microscopy coupled with traditional microscopic (optical, Scanning Electron Microscopy, Low Vacuum, Scanning Electron Microscopy), spectroscopic (X-ray photoelectron spectroscopy) and analytical (solvent extraction) techniques.
2
Materials and methods
2.1 Materials Maltodextrin DE12 (Glucidex ®, Roquette, France) was used as matrix material in combination with acacia gum (Instantgum AA, Nexira, France), also bringing emulsifying and film forming properties. Food grade sunflower oil (60 g/100 g poly-unsaturated fatty acids, 29 g/100 g mono-unsaturated fatty acids, 11 g/100 g saturated fatty acids) was used as a model for lipophilic compounds. Ultra-purified water (Elga option 3, U.S) was used as solvent. 2.2
Methods
2.2.1 Preparation of emulsion The preparation at pilot scale was based on a previous study (Turchiuli, Lemarié, Cuvelier, & Dumoulin, 2013). All the emulsions were prepared with the same concentration and quantity (2500g). They consisted of oil (4%w/w), acacia gum (14.4%w/w), maltodextrin (21.6%w/w), and water (60%w/w) leading to a total dry matter of 40 %w/w where 10%w/w of the dry matter corresponded to oil and the remaining 90%w/w to maltodextrin and acacia gum with a weight ratio of 3/2. 5
Aqueous phases were prepared by first dissolving the acacia gum, and second, the maltodextrin powder, in water at 40°C using a high speed blender with 3-blade axial flow impeller (Eurostar Ika, Labortechnik). The large oil droplet size coarse emulsions (about 1µm) were prepared by dispersing the oil into the aqueous solution by rotor-stator homogenization (AXR, Silverson Machines Ltd, UK) at 3500 rpm for 20 minutes. To obtain the small oil droplet size emulsion (average droplet size about 0.1 µm), the coarse emulsion was further emulsified using a high pressure homogenizer (Rannie Slow 22-50, APV, UK) at 300 bar with re-circulation for 4 min (flow rate 7.7 kg·min-1, ~12 passes). 2.2.2 Spray drying The emulsions were spray dried in a pilot dryer (Niro Minor, GEA Process Engineering, France). The dimensions of the cylindro-conical drying chamber were H 1.1 m x L 0.8 m. The initial liquid emulsion was atomized with a rotary atomizer with a diameter of 5 cm, 24 rectangular vanes of 3.5 x 6 mm and operated by compressed air. In order to produce powders with and without change in oil droplet size distribution during atomization, coarse emulsions were atomized at two different rotational speeds chosen from a previous study (Munoz-Ibanez et al., 2015): a conventional speed used in spray drying (i.e. 33200 rpm) and found to lead to monomodal oil droplet breakup and a low rotational speed (i.e. 3270 rpm)., corresponding to a capillary number value below the critical value obtained (i.e. 0.7), in order to preserve the emulsion microstructure. For the small oil droplet size emulsion, atomization was performed solely at conventional speed (i.e. 33200 rpm) since no condition leading to oil droplet breakup has been found in this case. For the three trials, the liquid feed inlet temperature was 20°C with a flow rate of 33 g·min-1. The inlet air temperature was set at 220°C (at the outlet of the heater) with a flow rate of 100 kg·h-1. The range of air temperature measured at the chamber outlet is shown in Table 1.
Table 1. Conditions for the production of the spray dried powders analysed. Name
d50,initial (µm)
Emulsion viscosity (Pa·s)
Atomizer speed (rpm)
Ca value
Tinlet (°C)
Toutlet (°C)
Status of oil droplet after spraying Breakup
d50, reconstituted (µm)
1µm-B
1.34
0.35
33200
2.1
220
129125
Yes
0.89
0.1µmNoB
0.13
0.30
33200
0.2
220
129125
No
0.13
1µmNoB
1.94
0.35
3270
0.1
220
134
No
1.94
2.2.3 Characterization of emulsions Oil droplet particle size distribution The size distribution of oil droplets in emulsions was measured by laser light diffraction after dispersion in purified water (Mastersizer 2000, Malvern, UK). The refractive index used was 1.33 for 6
water and 1.475 for sunflower oil droplets (measured). The absorption value was set to 0.001. The oil droplet size distribution was calculated applying a modified Mie-scattering model. 2.2.4 Characterization of spray-dried particles Powder particle size distribution Particle size distribution in powders was measured by laser light diffraction (Mastersizer 2000, Malvern, UK) using the Scirroco 2000 dry dispersion unit. About 10 g of spray dried emulsions were dispersed in a compressed air stream (P= 4 bar). The instrument calculates the particle size by a modified Mie-scattering model. For 1µm-NoB, the quantity of powder recovered after spray drying was not enough for the analysis and the particle size was estimated by observation of SEM micrographs of the powder particles (Error! Reference source not found.). Surface oil extraction Surface oil extraction was carried out according to Niro Analytical Method N° A 10a (GEA Niro, 2005). 50 mL of petroleum ether (boiling point 40-60°C, analytical grade, Fisher-Scientific) were added into a 250 mL Erlenmeyer flask containing 10 g ± 0.1 g of powder. The flask was agitated for 15 min in a shaking device (Grant OSL 200) at 200 rpm, and filtered by gravity. Filtration parts were wetted before filtration and rinsed with petroleum ether at the end to minimise losses. The filtrate was poured into two aluminium dishes (approximately 30 mL each), left under the fume extractor until the petroleum ether had evaporated (1h 30 min approximately), and then dried in a drying oven at 105°C for 1h. The aluminium dishes containing the extracted oil were cooled down in a desiccator for 10 min and weighed. Surface oil is expressed in % of total oil. 1µm-B and 0.1µm-NoB were analysed in duplicate, 1µm-NoB, only once due to limited amount of powder. Optical microscopy Spray dried emulsion particles were dispersed in silicon oil prior observation under optical microscope (BX51, Olympus, France) in order to reveal their internal morphology.
Scanning Electron Microscopy (SEM) Powders were mounted on aluminium stubs with double-sided sticky carbon tape. Samples were sputter coated in gold. The samples were analysed in a Scanning Electron Microscope (FEI Quanta 200). The SEM images where obtained under high vacuum (< 6.10 -4 Pa ) using 20 kV accelerating voltage with a secondary electron detector. Low-vacuum Scanning Electron Microscopy (LVSEM) The analyses were performed in an Environmental Scanning Electron Microscope from FEI–XL 30 Series operating in low vacuum mode. The LVSEM images were taken under low vacuum (13-133Pa). The samples (uncoated) were analysed with the Large Field Detector (LFD) or Gas Backscattered Secondary Electrons (CBS) detectors. At the accelerating voltage used (i.e. 20kV), the approximate depth scan of the LFD and GSE detectors is 1 nm and 500 nm respectively. Confocal Raman Microscopy
7
The distribution of the components in the spray dried particles was analysed with the confocal Raman microscopy technique. The measurements were performed with a WITec alpha300 system which has a lateral resolution of 250 nm and a vertical resolution of 500 nm. The depth scan range for one image is 25x25μm and 100x100 pixels using a 532nm laser for excitation. The integration time per Raman spectrum is 100 ms. One full Raman spectrum is collected in each pixel. Reference spectra for the individual components are used to allow calculation of the distribution of these components based on a least squares algorithm (Nuzzo et al, 2015b). X-ray photoelectron spectroscopy Particles surface chemistry of pure components (maltodextrin, acacia gum,) and on powder samples was characterized using X-ray Photoelectron Spectroscopy (AXIS UltraDLD x-ray photoelectron spectrometer, Kratos Analytical, UK). Sunflower oil spectrum was taken from our database. The samples, placed in high vacuum (4·10-7 Pa), were irradiated with well-defined X-ray radiation (monochromatic Al x-ray source) resulting in the emission of photoelectrons from the outermost surface layers. The position and area of the peaks in the spectrum obtained enabled quantitative determination of the atomic composition of the sample surface analysed (an area of 700 x 300 µm). The depth of analysis is approximately 2-5 nm, with a signal that decays exponentially with the depth. Two measurements were taken for each sample. The average standard deviation in the data for the atomic composition was 0.3 atom%. The surface composition in terms of molecular components was estimated using the patch model of Fäldt et al (1993).
3
Results
3.1 Oil droplet size distribution The oil droplet size distribution was analysed before and after spray drying (Table 2). Prior to spray drying the emulsions were all monomodal with a median volume diameter d50 of 1.34 µm and 1.94 µm for the 1µm-B and 1µm-NoB samples, respectively, and of 0.13 µm for the 0.1µm-NoB sample. After reconstitution of the dried emulsions with water to the original concentration the emulsion droplet size distributions were still monomodal but decreased for 1m-B, whilst for powders 0.1 µm-NoB and 1µmNoB the emulsion droplet size distribution remained close to that of the original emulsions. The reduction in emulsion droplet size is related to the break-up of droplets due to combination of high shear force during atomization and large original droplet size. The low shear force applied for 1µmNoB preserves the droplet size, and in 0.1 µm-NoB the droplets are too small to be affected by the high shear force. Table 2. Powder particle median diameter d50 and d10 and d90 (volume distribution) measured by Laser light diffraction or 1Estimated from SEM micrographs. Powder sample
d10 (μm)
d50 (μm)
d90 (μm)
1µm-B 0.1µm-NoB 1µm-NoB
16 6
37 19 ~801
85 42
3.2 Characterization of spray-dried particles The three powders obtained were characterized by different quantitative (XPS, oil extraction) and qualitative (confocal Raman microscopy, LVSEM) techniques to provide detailed and complementary information regarding the distribution of the different components. When imaging techniques have 8
been used for the characterization, only representative images are shown here. Typically, several images (at least three) were obtained for each sample and technique to ascertain that the structure is well represented in the selected images. Two of the powders were prepared in duplicate (1-µm-B and 0.1µm-NoB), and averages of values measured or images that are representative of both samples are shown. 3.2.1 Particle morphology The particle morphology was studied by light microscopy and scanning electron microscopy using different detectors. The optical microscopy of particles dispersed in silicon oil shows the overall morphology and the presence of vacuoles in the particles. Depending on processing conditions, the particle size and morphology was affected (Figure 1) so that 1µm-B displayed irregular hollow particles with a rather narrow particle size distribution. 0.1µm-NoB on the other hand showed a bimodal size distribution with small solid, irregular particles and large spherical hollow particles. 1µm-NoB particles were very homogeneous in size, and solid with irregular shape with few exceptions. The shell thickness in the observed hollow particles was between 3 and 10 μm. 1µm-B
0.1µm-NoB
1µm-NoB
H S S
H 100µm
100µm
100µm
Fig. 1. Light microscopy images of spray-dried emulsion to illustrate internal morphology: Solid (S) or Hollow (H). Magnification of the objective was 20x, and the scale bar is 100 µm. Samples dispersed in silicon oil. More detailed studies of the morphology were conducted using SEM in different modes. The gold sputtered samples in Figure 2 confirm the results in Figure 1. 1µm-B (Fig. 2a, top row) showed mainly smooth particles with few wrinkles, ridges or dimples; 0.1µm-NoB displayed two different types of particles, as observed with light microscopy: small wrinkled particles and larger smooth particles (Fig. 2b top row) and for 1µm-NoB, particles were larger and wrinkled. The surface fat was removed by solvent extraction, to display some small round features (pock marks) on powder 1µm-B (Fig. 2a, bottom row), also observed elsewhere (Jones, et al., 2013). Particles in 0.1µm-NoB and 1µm-NoB powders presented oil-free surfaces very similar to the non-extracted counterparts.
9
Fig. 2. Scanning electron microscope micrographs of spray dryed emulsions. Top row : Overview of the powder particles showing their topography. Bottom row: Particle surface after free oil extraction. (a) 1µm-B, (b) 0.1µm-NoB, and (c) 1µm-NoB.
The internal distribution of the oil droplets was observed in sectioned particles by SEM and close to the surface in intact particles by LV-SEM (Fig. 3). The sectioned particles (Fig. 3 top row) show that the oil droplets, observed as holes, are distributed throughout the thickness of the shell, but not in the close vicinity of the surface. The shell thickness corresponds to the thickness observed by light microscopy in Figure 1 (e.g. 3 to 10 µm). The LV-SEM images of particles without gold sputtering and using the CBS detector show the material just below the surface of the particles (Fig. 3 middle row). The dark spots correspond to material of lower electron density, i.e., the oil droplets. Contrast between maltodextrin and acacia gum was not observed, which is to be expected due to the similarity of the materials. Powder 1µm-B displays many oil droplets close to the surface, while the frequency of oil droplets in the other two powders is lower, indicating that the oil droplets are located some distance away from the particle surface. The same images using the LFD detector (Fig. 3 bottom row) show only the surface topography, and powder 1µm-B clearly shows protrusions in the surface corresponding to the location of oil droplets, while the other two samples show smooth surfaces.
10
Fig. 3. Localisation of oil droplets in the powder particles. SEM images of cross sections (top row). LV-SEM images using gaseous backscattered secondary electrons (CBS) (middle row) and large field detector (LFD) (bottom row). (a) 1µm-B, (b) 0.1µm-NoB and (c) 1µm-NoB.
3.2.2 Internal distribution of components Confocal Raman microscopy, a vibrational spectroscopy technique providing chemical imaging, was used to locate the different components in the interior of the particles. From the spectra of reference components (Fig. 4a), the composition at a given particle position is identified (Fig. 4b). Fig. 5 shows representative images of the internal mapping obtained for the three spray-dried powder types. The analysed powder particles were hollow with a shell thickness of about 3-5 µm at the studied locations, which corresponds to observations by optical microscopy (Fig 1) and SEM (Fig 3).
11
(a)
(b) + sunflower oil
Maltodextrin Sunflower oil
maltodextrin acacia gum
Fig. 4. Raman spectra for reference components (a), and example of spectra at a given position in a spray-dried particle (b). The arrows point out distinctive peaks in the reference spectra which can also be observed in the sample spectrum. The oil droplets were in general evenly distributed across the powder particles (solid shell) as can be observed in Fig. 5 where oil appears in yellow on the depth scanned images of bottom row. Powder 1µm-B displayed oil droplets with different sizes as a result of the droplet breakup during spray drying. Some droplets seem to reach the surface or be located close to it. Droplets in powder 0.1µmNoB were too small to be distinguished from the matrix as the droplets size is at or below the image resolution (250 nm), but their presence was detected throughout the solid matrix. In 1µm-NoB particles oil droplets close to the surface were located below a well-defined layer of acacia gum (in red). The size and location of the oil droplets corresponded to the location of oil droplets with SEM (Fig. 2), and to the emulsion droplet sizes determined for the reconstituted emulsions in Table 1. Acacia gum was overrepresented at the outer surface of the particles, and seemed to form a layer at the outer surface with a thickness of at least 0.5 µm (Fig. 5). Maltodextrin (in blue) and acacia gum tended to phase segregate during drying, observed as the distinct acacia gum layer and weak acacia gum peaks in the maltodextrin rich layer. The surface layer of acacia gum increased in thickness as the powder particle size increased: 1µm-NoB >1µm-B> 0.1µm-NoB. Small domains of acacia gum (red) can be observed in a maltodextrin matrix (blue), and correlation of the location of these with the location of oil droplets suggests that acacia gum is associated with the oil droplets. The maltodextrin concentration increased when approaching the particle core, as indicated by the relative peak sizes in the spectra.
12
0.1µm-NoB
1µm-B
1µm-NoB
Fig. 5. Internal mapping by confocal Raman microscopy of spray-dried emulsions. (a), (d) 1µm-B, (b), (e) 0.1µm-NoB, and (c), (f) 1µm-NoB. Top row: Depth scanned images showing an overlay image of maltodextrin and acacia gum. Bottom row: Depth scanned images showing overlay image of all the components. Red : acacia gum, blue: maltodextrin, yellow : sunflower oil. 3.2.3 Oil encapsulation The oil encapsulation was determined by surface oil extraction and XPS. By solvent extraction, the oil on the surface or connected to the surface via pores or cracks is removed. In XPS, the atomic composition of the surface layer (about 2-5 nm thickness) is detected, and these data are used to provide the surface coverage of different molecular components using a patch model (Fäldt, et al., 1993) and the atomic concentration of reference pure components. The atomic concentrations of O, C and N in the pure components used in this work (maltodextrin DE12, acacia gum and sunflower oil) are given in Table 3. Table 3. Atomic concentration of reference pure components. The data are averages of two measurements. The average standard deviation in the measurements is 0.3 atom%.
Maltodextrin DE12 Acacia gum Sunflower oil
O (%) 44.1 40.4 9.6
C (%) 55.9 57.2 90.4
N (%) 0 2.4 0
The data for surface oil extraction and estimations of the surface molecular composition based on XPS data are shown in Table 4. Surface oil by solvent extraction was low in all cases (< 4 of total oil in powder), and lowest in the powders which did not experience droplet breakup (0.1µm-NoB, 1µmNoB). XPS analysis showed an overrepresentation of the oil on the particle surface in all powders, which indicates that the oil spreads over the surface of the particles in a thin layer. Surface oil in 1µmNoB was, however, higher than expected, which is attributed to the mechanical breakup of still sticky 13
particles after spray-drying. By mechanical breaking it is reasonable to assume that some of the encapsulated oil was released and accounted for by extracted oil (Table 4) and the oil surface coverage by XPS analysis (Table 4). The 1µm-NoB particles, taking longer time to dry (larger particles), showed how the only matrix component on the surface was acacia gum, indicating that acacia gum adsorbs to the drying drop surface. Giving sufficient time for adsorption, acacia gum will cover the surface. Table 4 Surface oil extracted by petrol ether and molecular component surface coverage analysed by XPS. Extraction Surface oil (% of total oil)
Oil surface Acacia coverage gum (%) surface coverage Powder sample (%) 1µm-B 3.4±0.2 67 12 0.1µm-NoB 0. 4±0.0 30 21 1µm-NoB 1.11 72 29 1 Only one repetition of the analysis for this sample.
XPS Maltodextrin Maltodextrin surface / acacia gum coverage (in bulk: 1.5) (%) 21 49 0
1.75 2.33 0
4 Discussion The combination of the results presented for the different methods together provide a more comprehensive insight in the microstructure of spray dried emulsions. The methods are complementary in that they give information on different scales and from different areas of the particles. A comparison of depth of analysis and resolution of these techniques is shown in Fig. 6. The oil extraction provides the amount of fat accessible from the surface and through cracks and pores in the particles, and is thus not a purely surface sensitive technique. XPS and oil extraction provide quantitative information, while the imaging techniques are mainly qualitative.
Fig. 6 Comparison of depth of analyses and resolution of XPS, confocal Raman microscopy and, LVSEM characterization techniques. The surface analysed by XPS is indicated by a darker outline on one of the powder particles. The particle morphology was found to be related to the atomization, more specifically the drop size and correspondingly the drying time. The larger drops in 1µm-NoB dried more slowly, and dense, wrinkled particles were produced. The drops produced at higher atomization speed were smaller, thus 14
dried faster and a larger fraction of the drops locked their size at an earlier stage of drying, leading to vacuole formation. The surface will be spherical or wrinkled depending on the viscoelastic properties of the air/liquid surface film during shrinkage, with additional influence of the drying rate. The equilibrium surface tension of the feed solution using protein solutions has been proposed to describe the morphology of spray-dried particles (Xu, Howes, Adhikari, & Bhandari, 2012). However, it was recently shown that it is rather the surface rheological properties than the surface tension (dynamic or equilibrium) that determines morphology at given solids content and drying conditions (Nuzzo et al, 2015a). The dented morphology is quite typical of acacia gum containing particles dried at low solids content. Here the surface dilatational modulus was about 70 mN/m (obtained by oscillating pendant drop method, data not shown), which is a high level and in accordance with data of Sanchez, Renard, Robert, Schmitt, and Lefebvre (2002). However, the high evaporation rates and high solids content in the spray-drying feed will lead to early crust formation and thus the particles can still be spherical as is the case in most of the particles in powder 1µm-B and 0.1µm-NoB. Accordingly, the particle morphology is determined by the different parallel non-equilibrium processes in the drying drop and at its surface. The internal distribution of the components has hitherto been poorly explored, with a small number of studies using confocal laser scanning microscopy indicating how the protein and fat are distributed in solid particles (Auty et al., 2001; Drusch & Berg, 2008; Taneja et al., 2013). However, to our knowledge, there are no studies on how different matrix materials with similar properties are distributed. Here, we have shown that the distribution of matrix materials as well as of oil droplets depend on the drying conditions and processing prior to drying (emulsion droplet size). The confocal Raman imaging shows that acacia gum and maltodextrin phase segregates from an acacia gum rich surface layer in 1µm-NoB, i.e., the most slowly drying sample. A surface enrichment of acacia gum is observed in the other powders too, but to a lesser extent, which is attributed to faster drying. This corroborated by the XPS results, which show higer ratio of maltodextrin/ acacia gum at the surface for these samples (1µm-B and 0.1 µm-NB) (Table 4). This is in turn explained by the surface active character of acacia gum, which leads to its adsorption at the air-water interface of the drying drops in the spray-dryer. Such behaviour of surface active components has been investigated in several studies (Elversson et al., 2006; Fäldt & Bergenståhl., 1994; Jones et al., 2013; Keogh et al., 2001; MillqvistFureby et al., 2001; Millqvist Fureby, 2003; Murrieta-Pazos et al., 2012; Nijdam et al., 2006; Nuzzo et al., 2015a). The surface enrichment of acacia gum then directs the phase segregation so that a surface layer enriched in this component is created. Similar behaviour has been observed for phase segregation in confined droplets (Wassén, et al., 2013). The oil droplets were shown by several techniques to be located close to the surface of the particles when droplet break-up occurred during atomisation. This was also associated with higher levels of free fat, while surface fat appeared more related to the oil droplet size. In the present case (1µm-B), the droplet break-up results in an increase of the oil droplet surface area by a factor of 1.5. It can be speculated that when an oil droplet breaks up during atomization there is insufficient time for the macromolecular surfactant to adsorb to the new interface before drying is completed, resulting in incomplete surface occupation at the oil/water interface which in turn leads to unstable droplets. These more fragile droplets would then more easily rupture at the particle surface and result in higher levels of surface oil, as detected by oil extraction. A further aspect that could result in higher surface fat is that larger droplets are less stable than smaller ones, resulting in more frequent rupture of the oil droplets in 1µm-B and 1µm-NoB than in 0.1µm-NoB, and also substantially higher oil release per ruptured oil droplet.
15
XPS and oil extraction provide information on different aspects of the surface oil. XPS quantifies the coverage at the outermost layer of the particles, a property that is relevant for powder particle interactions, initial behaviour in wetting, and dispersion. Oil extraction, on the other hand, quantifies the accessible oil, including oil accessible through cracks and pores. It seems clear that oil droplet breakup in 1µm-B during atomization brought the droplets close to the surface as LV-SEM and Raman imaging suggests ( Fig. 3, Fig. 5), and the surface was covered by fat to a high level (Table 4). A protecting layer (>0.5 µm) of acacia gum was detected on the surface of the powders where no oil droplet breakup occurred as indicated by LV-SEM images, and confocal Raman microscopy (Fig. 3, Fig. 5), and this is further supported by the XPS data (Table 4) although the oil coverage of 1µm-NoB was high. The main part of the oil in spray dried particles is present as droplets in the matrix, and at the surface the oil can be present as a thin (even monomolecular) layer as well as in the form of patches. The low surface energy of the oil enables efficient spreading of the liquid oil on the particle surface. Oil could spread on the particle surface during manufacture, handling and storage (Fäldt & Bergenståhl., 1995; Fäldt & Sjöholm, 1996), and it has been shown that the composition of the hydrophilic matrix remains constant while the oil coverage varies if the composition is kept constant but not the processing conditions (Keogh et al., 2001; Millqvist-Fureby et al., 2001). The theoretical thickness of oil spread in the particle surface was calculated by combining both techniques (Table 4) and the particle size data (Error! Reference source not found., Fig. 1). For a hollow particle representative of powder 1µm-B with a diameter of 50 µm, surface to core thickness of 5 µm, and 67% of oil surface coverage, the average oil layer thickness is 23 nm. The corresponding oil layer thickness for 1µm-NoB is 7 nm, and for 0.1µm-NoB it is 6 nm. Raman microscopy showed no evidence of an oil-rich surface layer within the resolution limits of the technique, thus this layer is expected to be below 200 nm. These calculated oil layer thicknesses indicate that the oil is present in patches on the surface, since the layer thickness exceeds the depth of analysis of XPS and both acacia gum and maltodextrin are detected. Moreover, some of the extracted oil may be derived from the interior of the particles through cracks, pores and interconnected oil droplets, as well as broken particles.
5 Conclusions Even in the short timescales of drying there is time for the surface active ingredient to adsorb at the drop surface and for the two polymeric matrix ingredients to phase segregate as shown by analysis by confocal Raman microscopy. This is of interest when surface modification of the particle is required, and is also related to the encapsulation capacity of the formulation. Oil droplet breakup during atomization was found to have an influence on the surface oil content, determined by XPS and oil extraction. Droplet elongation and breakup brought the droplets close to or on the emulsion drop surface. In emulsions which, due to their small droplet size or gentle atomization conditions, preserved their original size distribution, the oil droplets in the particles were located below the surface active component layer. The surface analytical techniques reveal the distribution of components at different depths into the particles. XPS provides the composition of the outer surface which is important for e.g. wetting properties. Confocal Raman microscopy gives good opportunity to characterize the microstructure as it provides structural information through the whole particle depth. However, due to the resolution limitations, observation of submicron oil droplets was not possible, and the results showed here were only qualitative. It would therefore be interesting to explore the quantitative possibilities of confocal
16
Raman microscopy in order to obtain the relative composition at any region of the particle, although this is an ambitious endeavour. In summary, the techniques studied in this work provided complementary information on the spraydried emulsions, improving our understanding of the mechanisms behind events occurring during spray drying, such as droplet breakup, molecular adsorption and diffusion, and morphology build-up.
6 Acknowledgements The authors acknowledge the European Commission 7th Framework Program through the Marie Curie Actions, Initial Training Network ―Powtech‖ (FP7-PEOPLE-2010-ITN 264722) for supporting this research. Thanks also to Giana Perre and Nathalie Ruscassier at Laboratoire de Génie des Procédés et Matériaux, at Ecole Centrale Paris for performing the SEM imaging. Thanks to Johan Andersson and Mikael Sundin at SP Chemistry, Materials and Surfaces for the LVSEM imaging, and XPS analysis, respectively. Further, the authors acknowledge the Troëdsson Foundation for the Confocal Raman Microscope at SP Chemistry, Materials and Surfaces.
Industrial relevance text The study presented in this paper is of clear industrial relevance as it highlights how the processing parameters include oil encapsulation efficiency as well as the external and internal structure on nanoand micro-levels. These structural details have a strong influence on the behaviour of the powder when used in different applications. In particular, the novel revelation of the phase segregation occurring in spray-dried powders may in future developments be used to enhance encapsulation efficiency if various encapsulation applications in foods as well as in other areas.
17
References
Auty, M. A. E., Twomey, M., Guinee, T. P., & Mulvihill, D. M. (2001). Development and application of confocal scanning laser microscopy methods for studying the distribution of fat and protein in selected dairy products. Journal of Dairy Research, 68(03), 417-427. Chew, J. H., Fu, N., Gengenbach, T., Chen, X. D., & Selomulya, C. (2015). The compositional effects of high solids model emulsions on drying behaviour and particle formation processes. Journal of Food Engineering, 157(0), 33-40. Drusch, S., & Berg, S. (2008). Extractable oil in microcapsules prepared by spray-drying: Localisation, determination and impact on oxidative stability. Food Chemistry, 109(1), 17-24. Elversson, J., & Millqvist-Fureby, A. (2006). In situ coating—An approach for particle modification and encapsulation of proteins during spray-drying. International journal of pharmaceutics, 323(1), 52-63. Fäldt, P., Bergenståhl, B., & Carlsson, G. (1993). The surface coverage of fat on food powders analyzed by ESCA (electron spectroscopy for chemical analysis). Food structure, 12. Fäldt, P., & Bergenståhl, B. (1994). The surface composition of spray-dried protein—lactose powders. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 90(2), 183-190. Fäldt, P., & Bergenståhl, B. (1995). Fat encapsulation in spray-dried food powders. Journal of the American Oil Chemists’ Society, 72(2), 171-176. Fäldt, P., & Sjöholm, I. (1996). Characterization of spray-dried whole milk. Milchwissenschaft, 51(2), 88-92. Jafari, S. M., Assadpoor, E., Bhandari, B., & He, Y. (2008). Nano-particle encapsulation of fish oil by spray drying. Food Research International, 41(2), 172-183. Jones, J. R., Prime, D., Leaper, M. C., Richardson, D. J., Rielly, C. D., & Stapley, A. G. F. (2013). Effect of processing variables and bulk composition on the surface composition of spray dried powders of a model food system. Journal of Food Engineering, 118(1), 19-30. Keogh, M., O'Kennedy, B., Kelly, J., Auty, M., Kelly, P., Fureby, A., & Haahr, A. M. (2001). Stability to Oxidation of Spray‐Dried Fish Oil Powder Microencapsulated Using Milk Ingredients. Journal of Food Science, 66(2), 217-224. Kim, E. H.-J., Chen, X. D., & Pearce, D. (2002). Surface characterization of four industrial spraydried dairy powders in relation to chemical composition, structure and wetting property. Colloids and Surfaces B: Biointerfaces, 26(3), 197-212. Kim, E. H.-J., Chen, X. D., & Pearce, D. (2009). Surface composition of industrial spray-dried milk powders. 2. Effects of spray drying conditions on the surface composition. Journal of Food Engineering, 94(2), 169-181. Millqvist-Fureby, A., Elofsson, U., & Bergenståhl, B. (2001). Surface composition of spray-dried milk protein-stabilised emulsions in relation to pre-heat treatment of proteins. Colloids and Surfaces B: Biointerfaces, 21(1), 47-58. Millqvist Fureby, A. (2003). Characterisation of spray-dried emulsions with mixed fat phases. Colloids and Surfaces B: Biointerfaces, 31, 65-79. Munoz-Ibanez, M., Azagoh, C., Dubey, B. N., Dumoulin, E., & Turchiuli, C. (2015). Changes in oilin-water emulsion size distribution during the atomization step in spray-drying encapsulation. Journal of Food Engineering(0). Murrieta-Pazos, I., Gaiani, C., Galet, L., & Scher, J. (2012). Composition gradient from surface to core in dairy powders: agglomeration effect. Food Hydrocolloids, 26(1), 149-158. Nijdam, J. J., & Langrish, T. A. G. (2006). The effect of surface composition on the functional properties of milk powders. Journal of Food Engineering, 77(4), 919-925. Nikolova, Y., Petit, J., Sanders, C., Gianfrancesco, A., Scher, J., & Gaiani, C. (2015). Toward a better determination of dairy powders surface composition through XPS matrices development. Colloids and Surfaces B: Biointerfaces, 125(0), 12-20. Nuzzo, M., Millqvist-Fureby, A., Sloth, J., & Bergenståhl, B. (2015a). Surface composition and morphology of particles dried individually and by spray drying. Drying Technology, 33(6), 757-767. Nuzzo, M., Sloth, J., Brandner, B., Bergenståhl, B., & Millqvist-Fureby, A. (2015b). Confocal Raman microscopy for mapping phase segregation in individually dried particles composed of lactose 18
and macromolecules. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 481(0), 229-236. Risch, S. J., & Reineccius, G. A. (1988). Spray-Dried Orange Oil. Flavor Encapsulation, 370(370), 67-77. Sanchez, C., Renard, D., Robert, P., Schmitt, C., & Lefebvre, J. (2002). Structure and rheological properties of acacia gum dispersions. Food Hydrocolloids, 16(3), 257-267. Sarkar, A., Arfsten, J., Golay, P.-A., Acquistapace, S., & Heinrich, E. (2016). Microstructure and long-term stability of spray dried emulsions with ultra-high oil content. Food Hydrocolloids, 52, 857-867. Soottitantawat, A., Bigeard, F., Yoshii, H., Furuta, T., Ohkawara, M., & Linko, P. (2005). Influence of emulsion and powder size on the stability of encapsulated d-limonene by spray drying. Innovative Food Science & Emerging Technologies, 6(1), 107-114. Soottitantawat, A., Yoshii, H., Furuta, T., Ohkawara, M., & Linko, P. (2003). Microencapsulation by Spray Drying: Influence of Emulsion Size on the Retention of Volatile Compounds. Journal of Food Science, 68(7), 2256–2262. Taneja, A., Ye, A., Jones, J. R., Archer, R., & Singh, H. (2013). Behaviour of oil droplets during spray drying of milk-protein-stabilised oil-in-water emulsions. International Dairy Journal, 28(1), 15-23. Turchiuli, C., Lemarié, N., Cuvelier, M.-E., & Dumoulin, E. (2013). Production of fine emulsions at pilot scale for oil compounds encapsulation. Journal of Food Engineering, 115(4), 452-458. Walton, D., & Mumford, C. (1999). The morphology of spray-dried particles: the effect of process variables upon the morphology of spray-dried particles. Chemical Engineering Research and Design, 77(5), 442-460. Wassén, S., Lorén, N., van Bemmel, K., Schuster, E., Rondeau, E., & Hermansson, A.-M. (2013). Effects of confinement on phase separation kinetics and final morphology of whey protein isolate-gellan gum mixtures. Soft Matter, 9(9), 2738-2749. Wu, W. D., Liu, W., Gengenbach, T., Woo, M. W., Selomulya, C., Chen, X. D., & Weeks, M. (2014). Towards spray drying of high solids dairy liquid: Effects of feed solid content on particle structure and functionality. Journal of Food Engineering, 123(0), 130-135. Xu, Y. Y., Howes, T., Adhikari, B., & Bhandari, B. (2012). Investigation of relationship between surface tension of feed solution containing various proteins and surface composition and morphology of powder particles. Drying Technology, 30(14), 1548-1562.
19
Figure captions Fig. 1. Light microscopy images of spray-dried emulsion to illustrate internal morphology: Solid (S) or Hollow (H). Magnification of the objective was 20x, and the scale bar is 100 µm. Samples dispersed in silicon oil. Fig. 2. Scanning electron microscope micrographs of spray dryed emulsions. Top row : Overview of the powder particles showing their topography. Bottom row: Particle surface after free oil extraction. (a) 1µm-B, (b) 0.1µm-NoB, and (c) 1µm-NoB. Fig. 3. Localisation of oil droplets in the powder particles. SEM images of cross sections (top row). LV-SEM images using gaseous backscattered secondary electrons (CBS) (middle row) and large field detector (LFD) (bottom row). (a) 1µm-B, (b) 0.1µm-NoB and (c) 1µm-NoB. Fig. 4. Raman spectra for reference components (a), and example of spectra at a given position in a spray-dried particle (b). The arrows point out distinctive peaks in the reference spectra which can also be observed in the sample spectrum. Fig. 5. Internal mapping by confocal Raman microscopy of spray-dried emulsions. (a), (d) 1µm-B, (b), (e) 0.1µm-NoB, and (c), (f) 1µm-NoB. Top row: Depth scanned images showing an overlay image of maltodextrin and acacia gum. Bottom row: Depth scanned images showing overlay image of all the components. Red : acacia gum, blue: maltodextrin, yellow : sunflower oil. Fig. 6 Comparison of depth of analyses and resolution of XPS, confocal Raman microscopy and, LVSEM characterization techniques. The surface analysed by XPS is indicated by a darker outline on one of the powder particles.
20
Table 5. Conditions for the production of the spray dried powders analyzed. Name
d50,initial (µm)
Emulsion viscosity (Pa.s)
Atomizer speed (rpm)
Ca value
Tinlet (°C)
1µm-B
1.34
0.35
33200
2.1
220
0.1µmNoB 1µmNoB
0.13
0.30
33200
0.2
220
1.94
0.35
3270
0.1
220
Toutlet (°C)
129125 129125 134
Status of oil droplet after spraying Breakup
d50, reconstituted (µm)
Yes
0.89
No
0.13
No
1.94
21
Table 6. Powder particle median diameter d50 and d10 and d90 measured by Laser light diffraction or 1 Estimated from SEM micrographs. Powder sample
d10 (μm)
d50 (μm)
d90 (μm)
1µm-B 0.1µm-NoB 1µm-NoB
16 6
37 19 ~801
85 42
22
Table 7. Atomic concentration of reference pure components. The data are averages of two measurements. The average standard deviation in the measurements is 0.3 atom%.
Maltodextrin DE12 Acacia gum Sunflower oil
O (%) 44.1 40.4 9.6
C (%) 55.9 57.2 90.4
N (%) 0 2.4 0
23
Table 8 Surface oil extracted by petrol ether and molecular component surface coverage analyzed by XPS.
Extraction
XPS
Surface oil Acacia gum Maltodextrin Oil surface Powder (% of total surface surface coverage coverage (%) sample oil) coverage (%) (%) 1µm-B 3.4±0.2 67 12 21 0.1µm-NoB 0. 4±0.0 30 21 49 1µm-NoB1 1.11 72 29 0 1 Only one repetition of the analysis for this sample was possible.
Maltodextrin / acacia gum (in bulk: 1.5 ) 1.75 2.33 0
24
Overview
150 µm
100 µm
150 µm
Particle surface after free oil extraction
(a)
50 µm
20 µm
20 µm (b)
(c)
2 µm
(a)
2 µm
6 µm
10 µm
20 µm
50 µm
10 µm
20 µm
50 µm
(b)
(c)
(a)
(b)
(c)
(d)
(e)
(f)
XPS Particle surface 1nm 2-5 nm 10nm Average, image or 100nm spot 1μm 300 µm 10μm 100μm
Confocal Raman microscopy 25 x 25 μm image
LV-SEM Back scattering electrons 100nm Material contrast Scattering probability =f(Atomic number)
Resolution: 200nm
Secondary electrons 1nm Surface sensitivity
Resolution: <5nm
Surface oil extraction
XPS, 2-5 nm
Confocal Raman microscopy
vacuole
LV-SEM (500 nm)
SEM (1 nm) ̴ 50 µm
S2213-3291(16)30009-0 http://dx.doi.org/doi:10.1016/j.foostr.2016.05.001 FOOSTR 47
To appear in: Received date: Revised date: Accepted date:
4-1-2016 10-5-2016 13-5-2016
Please cite this article as: Munoz-Ibanez, Marta., Nuzzo, Marine., Turchiuli, Christelle., Bergenst˚ahl, Bj¨orn., Dumoulin, Elisabeth., & Millqvist-Fureby, Anna., The microstructure and component distribution in spray-dried emulsion particles.Food Structure http://dx.doi.org/10.1016/j.foostr.2016.05.001 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
The microstructure and component distribution in spray-dried emulsion particles Marta Munoz-Ibaneza, Marine Nuzzob,c, Christelle Turchiulia,d, Dumoulina, Anna Millqvist-Furebyb*
Björn Bergenståhlc, Elisabeth
a
UMR Ingénierie Procédés Aliments, AgroParisTech, Inra, Université Paris-Saclay, 1 avenue des olympiades, 91300 Massy, France b
SP Technical Research Institute of Sweden, Box 5607, SE-114 86 Stockholm, Sweden
c
Lund University, Department of Food Technology, Engineering and Nutrition, P.O. Box 124, SE-221 00 Lund, Sweden d
Université Paris-Sud, IUT Orsay, Plateau de Moulon, F- 91400 Orsay, France
∗ Corresponding author. Tel.: +46 (0)10 516 60 46 E-mail address: [email protected] (A. MillqvistFureby).
1
Highlights
Confocal Raman microscopy is a powerful tool for internal mapping of particles.
Phase segregation occurs even during short drying time of particles.
Particle morphology of a given formulation is governed by emulsion processing, atomization and drying rate.
2
Abstract Microencapsulation by spray drying of oil-in-water (o/w) emulsions provides a means to encapsulate functional lipophilic ingredients. The active ingredient is dispersed in continuous solid phase providing protection. However, the encapsulation efficiency depends on the microstructure and morphology of the dry particles influenced by several mechanisms occurring during processing such as oil droplet breakup during atomization, ingredient diffusivity, interfacial adsorption of surface active ingredients, and drying kinetics. In this work, sunflower oil (model for lipophilic compounds) was encapsulated in solid particles composed of acacia gum and maltodextrin DE12. Three powders with different initial emulsion size (e.g. about 0.1 and 1 μm) and atomized under high and low shear rate were analyzed for the morphology and distribution of oil droplets and matrix constituents within the solid particle (20 to 100 μm). The microscopic (optical, SEM, LVSEM, confocal Raman), spectroscopic (XPS) and analytical (solvent extraction) techniques used were either qualitative or quantitative. Their combination made it possible to determine both the composition at the surface and inside the particle. The surface differs from the bulk in composition, confirming the constituent segregation during spray drying, and depended on the initial emulsion size and atomization conditions that must be controlled for an efficient encapsulation. Especially, the use of confocal Raman microscopy is promising for the study of processstructureproperties relationship. Keywords: Spray-dried emulsions; particle microstructure; Confocal Raman Microscopy, phase segregation.
3
1 Introduction Spray-drying is a common technology to encapsulate fats and oils in solid matrices to create a powder product with improved shelf stability and new functionality compared to a liquid emulsion. The functional properties of the powder, such as wetting, flowability, oxidation stability, etc. are strongly linked to the structure of the particles on different levels and the oil encapsulation efficiency (Kim, Chen, & Pearce, 2002; Nijdam & Langrish, 2006). Encapsulation of oil in powders is an area that has been researched over many years, and the encapsulation efficiency and stability of the encapsulated oil are two important parameters of practical relevance. There are several different aspects that influence the properties and microstructure of the powder, including formulation and processing conditions. The emulsion droplet size as well as particle size influence the oil encapsulation and surface oil (Jafari, Assadpoor, Bhandari, & He, 2008; Sarkar, Arfsten, Golay, Acquistapace, & Heinrich, 2016; Soottitantawat, et al., 2005; Soottitantawat, Yoshii, Furuta, Ohkawara, & Linko, 2003). However, the optimum emulsion droplet size to minimize surface oil depended on the oil type, emulsion processing technology and particle size. The increase in surface oil at larger oil droplet size has also been observed by several other authors (Jafari, et al., 2008; Millqvist Fureby, 2003; Soottitantawat, et al., 2005; Soottitantawat, et al., 2003). The instability of emulsion droplets during spray-drying was explained by the fast dehydration, leading to close contact between droplets, resulting in coalescence and weak stabilising surface layers on the droplets (Taneja, Ye, Jones, Archer, & Singh, 2013). This may contribute to the common explanation of high surface oil based on rupture of oil droplets at the particle surface (Fäldt & Bergenståhl, 1995; Kim, Chen, & Pearce, 2009; Nijdam, et al., 2006). The processing conditions during spray-drying were also linked to the oil encapsulation efficiency, where effects of shear have been put forward by Risch and Reineccius (1988) and Soottitantawat et al. (2005). This is further supported by Munoz-Ibanez, Azagoh, Dubey, Dumoulin, and Turchiuli (2015), who showed that the droplet breakup during the atomisation step of spray drying is influenced by the shear exerted on the droplets. This leads to decrease in droplet size, but it is not clear whether it is the droplet size per se, or other effects that cause the increase in surface fat for large initial emulsion droplet sizes. To extend the understanding of the complex mechanisms leading to oil encapsulation and the final microstructure of spray-dried particles different analytical tools for detailed characterisation are necessary. The effects of composition and processing for spray-dried dairy-like emulsions (Kim, et al., 2002; Kim, et al., 2009; Millqvist-Fureby, Elofsson, & Bergenståhl, 2001; Murrieta-Pazos, Gaiani, Galet, & Scher, 2012; Wu, et al., 2014) as well as other food protein and biopolymer formulations (e.g., Fäldt, Bergenståhl, & Carlsson, 1993; Fäldt & Bergenståhl, 1994; Fäldt, et al., 1995;; Nuzzo, Millqvist-Fureby, Sloth, & Bergenståhl, 2015) on the surface composition have been studied rather extensively. These studies show that the surface active components are over-represented at the dry particle surface, since these adsorb to the surface of the drying drop. The surface oil coverage, as determined by XPS in a surface layer of approximately 2-5 nm, can still be very high due to spreading of oil originating from ruptured oil droplets (Chew, Fu, Gengenbach, Chen, & Selomulya, 2015; Keogh et al., 2001; Millqvist-Fureby et al., 2001; Nikolova et al., 2015). The morphology of the particles is influenced by several parameters, such as the solids content in the feed, the drying rate (Walton & Mumford, 1999), and the composition and viscoelastic properties of the adsorbed surface film of the drying drop (Elversson & Millqvist-Fureby, 2006; Nuzzo et al., 2015a).
4
The composition of the internal structure on the other hand still remains largely theoretical. Drusch et al (2008) have studied how the emulsion droplets are located in spray-dried emulsion powders using confocal laser scanning microscopy (CLSM), and found that powders with higher oil content presented higher levels of surface oil and oil droplets close to the surface compared to powders with less oil content. CSLM has been used to provide information about particle internal structure in different types of milk powder (Auty, Twomey, Guinee, & Mulvihill, 2001). In this study the whole milk powder showed very low levels of surface fat, and an even distribution of oil droplets throughout the particles, while a high fat cream powder showed some surface fat and larger oil droplets. Recently, Sarkar et al. (2016) used CSLM to visualise the close packing of oil droplets in spray-dried emulsions with ultra-high oil content and cross-linked whey protein as the emulsifier and matrix material. However, CLSM requires fluorescent labelling of the components to be visualized, and is thus limited to the determination of components of sufficiently different properties, i.e. oil, carbohydrate and protein, and does not distinguish between components of similar dyeing properties, such as different polysaccharides. Confocal Raman imaging, on the other hand, allows for internal particle investigations with the same level of resolution as CLSM, but no labelling is required as the full Raman spectrum in each pixel of the image is used to construct an image of the localisation of the different molecular components. Recently, Nuzzo et al (2015b) showed phase segregation in spraydried powder particles and in individually dried particles composed of maltodextrin and hydroxypropyl methyl cellulose using confocal Raman microscopy. The same technique was also used to investigate the internal structure of single particles composed of lactose and BSA, HPMC or Poloxamer 188 (Nuzzo et al, 2015b). In these investigations, the more surface active macromolecules were located at the surface, and a layer with a thickness in the micrometre range enriched in this component was found at the surface. The aim of this paper is to elucidate how oil droplet breakup during atomization influences the localization of ingredients and the oil encapsulation in spray-dried powders composed of sunflower oil, maltodextrin and acacia gum. We address the distribution of dry emulsion components within the particles by means of confocal Raman microscopy coupled with traditional microscopic (optical, Scanning Electron Microscopy, Low Vacuum, Scanning Electron Microscopy), spectroscopic (X-ray photoelectron spectroscopy) and analytical (solvent extraction) techniques.
2
Materials and methods
2.1 Materials Maltodextrin DE12 (Glucidex ®, Roquette, France) was used as matrix material in combination with acacia gum (Instantgum AA, Nexira, France), also bringing emulsifying and film forming properties. Food grade sunflower oil (60 g/100 g poly-unsaturated fatty acids, 29 g/100 g mono-unsaturated fatty acids, 11 g/100 g saturated fatty acids) was used as a model for lipophilic compounds. Ultra-purified water (Elga option 3, U.S) was used as solvent. 2.2
Methods
2.2.1 Preparation of emulsion The preparation at pilot scale was based on a previous study (Turchiuli, Lemarié, Cuvelier, & Dumoulin, 2013). All the emulsions were prepared with the same concentration and quantity (2500g). They consisted of oil (4%w/w), acacia gum (14.4%w/w), maltodextrin (21.6%w/w), and water (60%w/w) leading to a total dry matter of 40 %w/w where 10%w/w of the dry matter corresponded to oil and the remaining 90%w/w to maltodextrin and acacia gum with a weight ratio of 3/2. 5
Aqueous phases were prepared by first dissolving the acacia gum, and second, the maltodextrin powder, in water at 40°C using a high speed blender with 3-blade axial flow impeller (Eurostar Ika, Labortechnik). The large oil droplet size coarse emulsions (about 1µm) were prepared by dispersing the oil into the aqueous solution by rotor-stator homogenization (AXR, Silverson Machines Ltd, UK) at 3500 rpm for 20 minutes. To obtain the small oil droplet size emulsion (average droplet size about 0.1 µm), the coarse emulsion was further emulsified using a high pressure homogenizer (Rannie Slow 22-50, APV, UK) at 300 bar with re-circulation for 4 min (flow rate 7.7 kg·min-1, ~12 passes). 2.2.2 Spray drying The emulsions were spray dried in a pilot dryer (Niro Minor, GEA Process Engineering, France). The dimensions of the cylindro-conical drying chamber were H 1.1 m x L 0.8 m. The initial liquid emulsion was atomized with a rotary atomizer with a diameter of 5 cm, 24 rectangular vanes of 3.5 x 6 mm and operated by compressed air. In order to produce powders with and without change in oil droplet size distribution during atomization, coarse emulsions were atomized at two different rotational speeds chosen from a previous study (Munoz-Ibanez et al., 2015): a conventional speed used in spray drying (i.e. 33200 rpm) and found to lead to monomodal oil droplet breakup and a low rotational speed (i.e. 3270 rpm)., corresponding to a capillary number value below the critical value obtained (i.e. 0.7), in order to preserve the emulsion microstructure. For the small oil droplet size emulsion, atomization was performed solely at conventional speed (i.e. 33200 rpm) since no condition leading to oil droplet breakup has been found in this case. For the three trials, the liquid feed inlet temperature was 20°C with a flow rate of 33 g·min-1. The inlet air temperature was set at 220°C (at the outlet of the heater) with a flow rate of 100 kg·h-1. The range of air temperature measured at the chamber outlet is shown in Table 1.
Table 1. Conditions for the production of the spray dried powders analysed. Name
d50,initial (µm)
Emulsion viscosity (Pa·s)
Atomizer speed (rpm)
Ca value
Tinlet (°C)
Toutlet (°C)
Status of oil droplet after spraying Breakup
d50, reconstituted (µm)
1µm-B
1.34
0.35
33200
2.1
220
129125
Yes
0.89
0.1µmNoB
0.13
0.30
33200
0.2
220
129125
No
0.13
1µmNoB
1.94
0.35
3270
0.1
220
134
No
1.94
2.2.3 Characterization of emulsions Oil droplet particle size distribution The size distribution of oil droplets in emulsions was measured by laser light diffraction after dispersion in purified water (Mastersizer 2000, Malvern, UK). The refractive index used was 1.33 for 6
water and 1.475 for sunflower oil droplets (measured). The absorption value was set to 0.001. The oil droplet size distribution was calculated applying a modified Mie-scattering model. 2.2.4 Characterization of spray-dried particles Powder particle size distribution Particle size distribution in powders was measured by laser light diffraction (Mastersizer 2000, Malvern, UK) using the Scirroco 2000 dry dispersion unit. About 10 g of spray dried emulsions were dispersed in a compressed air stream (P= 4 bar). The instrument calculates the particle size by a modified Mie-scattering model. For 1µm-NoB, the quantity of powder recovered after spray drying was not enough for the analysis and the particle size was estimated by observation of SEM micrographs of the powder particles (Error! Reference source not found.). Surface oil extraction Surface oil extraction was carried out according to Niro Analytical Method N° A 10a (GEA Niro, 2005). 50 mL of petroleum ether (boiling point 40-60°C, analytical grade, Fisher-Scientific) were added into a 250 mL Erlenmeyer flask containing 10 g ± 0.1 g of powder. The flask was agitated for 15 min in a shaking device (Grant OSL 200) at 200 rpm, and filtered by gravity. Filtration parts were wetted before filtration and rinsed with petroleum ether at the end to minimise losses. The filtrate was poured into two aluminium dishes (approximately 30 mL each), left under the fume extractor until the petroleum ether had evaporated (1h 30 min approximately), and then dried in a drying oven at 105°C for 1h. The aluminium dishes containing the extracted oil were cooled down in a desiccator for 10 min and weighed. Surface oil is expressed in % of total oil. 1µm-B and 0.1µm-NoB were analysed in duplicate, 1µm-NoB, only once due to limited amount of powder. Optical microscopy Spray dried emulsion particles were dispersed in silicon oil prior observation under optical microscope (BX51, Olympus, France) in order to reveal their internal morphology.
Scanning Electron Microscopy (SEM) Powders were mounted on aluminium stubs with double-sided sticky carbon tape. Samples were sputter coated in gold. The samples were analysed in a Scanning Electron Microscope (FEI Quanta 200). The SEM images where obtained under high vacuum (< 6.10 -4 Pa ) using 20 kV accelerating voltage with a secondary electron detector. Low-vacuum Scanning Electron Microscopy (LVSEM) The analyses were performed in an Environmental Scanning Electron Microscope from FEI–XL 30 Series operating in low vacuum mode. The LVSEM images were taken under low vacuum (13-133Pa). The samples (uncoated) were analysed with the Large Field Detector (LFD) or Gas Backscattered Secondary Electrons (CBS) detectors. At the accelerating voltage used (i.e. 20kV), the approximate depth scan of the LFD and GSE detectors is 1 nm and 500 nm respectively. Confocal Raman Microscopy
7
The distribution of the components in the spray dried particles was analysed with the confocal Raman microscopy technique. The measurements were performed with a WITec alpha300 system which has a lateral resolution of 250 nm and a vertical resolution of 500 nm. The depth scan range for one image is 25x25μm and 100x100 pixels using a 532nm laser for excitation. The integration time per Raman spectrum is 100 ms. One full Raman spectrum is collected in each pixel. Reference spectra for the individual components are used to allow calculation of the distribution of these components based on a least squares algorithm (Nuzzo et al, 2015b). X-ray photoelectron spectroscopy Particles surface chemistry of pure components (maltodextrin, acacia gum,) and on powder samples was characterized using X-ray Photoelectron Spectroscopy (AXIS UltraDLD x-ray photoelectron spectrometer, Kratos Analytical, UK). Sunflower oil spectrum was taken from our database. The samples, placed in high vacuum (4·10-7 Pa), were irradiated with well-defined X-ray radiation (monochromatic Al x-ray source) resulting in the emission of photoelectrons from the outermost surface layers. The position and area of the peaks in the spectrum obtained enabled quantitative determination of the atomic composition of the sample surface analysed (an area of 700 x 300 µm). The depth of analysis is approximately 2-5 nm, with a signal that decays exponentially with the depth. Two measurements were taken for each sample. The average standard deviation in the data for the atomic composition was 0.3 atom%. The surface composition in terms of molecular components was estimated using the patch model of Fäldt et al (1993).
3
Results
3.1 Oil droplet size distribution The oil droplet size distribution was analysed before and after spray drying (Table 2). Prior to spray drying the emulsions were all monomodal with a median volume diameter d50 of 1.34 µm and 1.94 µm for the 1µm-B and 1µm-NoB samples, respectively, and of 0.13 µm for the 0.1µm-NoB sample. After reconstitution of the dried emulsions with water to the original concentration the emulsion droplet size distributions were still monomodal but decreased for 1m-B, whilst for powders 0.1 µm-NoB and 1µmNoB the emulsion droplet size distribution remained close to that of the original emulsions. The reduction in emulsion droplet size is related to the break-up of droplets due to combination of high shear force during atomization and large original droplet size. The low shear force applied for 1µmNoB preserves the droplet size, and in 0.1 µm-NoB the droplets are too small to be affected by the high shear force. Table 2. Powder particle median diameter d50 and d10 and d90 (volume distribution) measured by Laser light diffraction or 1Estimated from SEM micrographs. Powder sample
d10 (μm)
d50 (μm)
d90 (μm)
1µm-B 0.1µm-NoB 1µm-NoB
16 6
37 19 ~801
85 42
3.2 Characterization of spray-dried particles The three powders obtained were characterized by different quantitative (XPS, oil extraction) and qualitative (confocal Raman microscopy, LVSEM) techniques to provide detailed and complementary information regarding the distribution of the different components. When imaging techniques have 8
been used for the characterization, only representative images are shown here. Typically, several images (at least three) were obtained for each sample and technique to ascertain that the structure is well represented in the selected images. Two of the powders were prepared in duplicate (1-µm-B and 0.1µm-NoB), and averages of values measured or images that are representative of both samples are shown. 3.2.1 Particle morphology The particle morphology was studied by light microscopy and scanning electron microscopy using different detectors. The optical microscopy of particles dispersed in silicon oil shows the overall morphology and the presence of vacuoles in the particles. Depending on processing conditions, the particle size and morphology was affected (Figure 1) so that 1µm-B displayed irregular hollow particles with a rather narrow particle size distribution. 0.1µm-NoB on the other hand showed a bimodal size distribution with small solid, irregular particles and large spherical hollow particles. 1µm-NoB particles were very homogeneous in size, and solid with irregular shape with few exceptions. The shell thickness in the observed hollow particles was between 3 and 10 μm. 1µm-B
0.1µm-NoB
1µm-NoB
H S S
H 100µm
100µm
100µm
Fig. 1. Light microscopy images of spray-dried emulsion to illustrate internal morphology: Solid (S) or Hollow (H). Magnification of the objective was 20x, and the scale bar is 100 µm. Samples dispersed in silicon oil. More detailed studies of the morphology were conducted using SEM in different modes. The gold sputtered samples in Figure 2 confirm the results in Figure 1. 1µm-B (Fig. 2a, top row) showed mainly smooth particles with few wrinkles, ridges or dimples; 0.1µm-NoB displayed two different types of particles, as observed with light microscopy: small wrinkled particles and larger smooth particles (Fig. 2b top row) and for 1µm-NoB, particles were larger and wrinkled. The surface fat was removed by solvent extraction, to display some small round features (pock marks) on powder 1µm-B (Fig. 2a, bottom row), also observed elsewhere (Jones, et al., 2013). Particles in 0.1µm-NoB and 1µm-NoB powders presented oil-free surfaces very similar to the non-extracted counterparts.
9
Fig. 2. Scanning electron microscope micrographs of spray dryed emulsions. Top row : Overview of the powder particles showing their topography. Bottom row: Particle surface after free oil extraction. (a) 1µm-B, (b) 0.1µm-NoB, and (c) 1µm-NoB.
The internal distribution of the oil droplets was observed in sectioned particles by SEM and close to the surface in intact particles by LV-SEM (Fig. 3). The sectioned particles (Fig. 3 top row) show that the oil droplets, observed as holes, are distributed throughout the thickness of the shell, but not in the close vicinity of the surface. The shell thickness corresponds to the thickness observed by light microscopy in Figure 1 (e.g. 3 to 10 µm). The LV-SEM images of particles without gold sputtering and using the CBS detector show the material just below the surface of the particles (Fig. 3 middle row). The dark spots correspond to material of lower electron density, i.e., the oil droplets. Contrast between maltodextrin and acacia gum was not observed, which is to be expected due to the similarity of the materials. Powder 1µm-B displays many oil droplets close to the surface, while the frequency of oil droplets in the other two powders is lower, indicating that the oil droplets are located some distance away from the particle surface. The same images using the LFD detector (Fig. 3 bottom row) show only the surface topography, and powder 1µm-B clearly shows protrusions in the surface corresponding to the location of oil droplets, while the other two samples show smooth surfaces.
10
Fig. 3. Localisation of oil droplets in the powder particles. SEM images of cross sections (top row). LV-SEM images using gaseous backscattered secondary electrons (CBS) (middle row) and large field detector (LFD) (bottom row). (a) 1µm-B, (b) 0.1µm-NoB and (c) 1µm-NoB.
3.2.2 Internal distribution of components Confocal Raman microscopy, a vibrational spectroscopy technique providing chemical imaging, was used to locate the different components in the interior of the particles. From the spectra of reference components (Fig. 4a), the composition at a given particle position is identified (Fig. 4b). Fig. 5 shows representative images of the internal mapping obtained for the three spray-dried powder types. The analysed powder particles were hollow with a shell thickness of about 3-5 µm at the studied locations, which corresponds to observations by optical microscopy (Fig 1) and SEM (Fig 3).
11
(a)
(b) + sunflower oil
Maltodextrin Sunflower oil
maltodextrin acacia gum
Fig. 4. Raman spectra for reference components (a), and example of spectra at a given position in a spray-dried particle (b). The arrows point out distinctive peaks in the reference spectra which can also be observed in the sample spectrum. The oil droplets were in general evenly distributed across the powder particles (solid shell) as can be observed in Fig. 5 where oil appears in yellow on the depth scanned images of bottom row. Powder 1µm-B displayed oil droplets with different sizes as a result of the droplet breakup during spray drying. Some droplets seem to reach the surface or be located close to it. Droplets in powder 0.1µmNoB were too small to be distinguished from the matrix as the droplets size is at or below the image resolution (250 nm), but their presence was detected throughout the solid matrix. In 1µm-NoB particles oil droplets close to the surface were located below a well-defined layer of acacia gum (in red). The size and location of the oil droplets corresponded to the location of oil droplets with SEM (Fig. 2), and to the emulsion droplet sizes determined for the reconstituted emulsions in Table 1. Acacia gum was overrepresented at the outer surface of the particles, and seemed to form a layer at the outer surface with a thickness of at least 0.5 µm (Fig. 5). Maltodextrin (in blue) and acacia gum tended to phase segregate during drying, observed as the distinct acacia gum layer and weak acacia gum peaks in the maltodextrin rich layer. The surface layer of acacia gum increased in thickness as the powder particle size increased: 1µm-NoB >1µm-B> 0.1µm-NoB. Small domains of acacia gum (red) can be observed in a maltodextrin matrix (blue), and correlation of the location of these with the location of oil droplets suggests that acacia gum is associated with the oil droplets. The maltodextrin concentration increased when approaching the particle core, as indicated by the relative peak sizes in the spectra.
12
0.1µm-NoB
1µm-B
1µm-NoB
Fig. 5. Internal mapping by confocal Raman microscopy of spray-dried emulsions. (a), (d) 1µm-B, (b), (e) 0.1µm-NoB, and (c), (f) 1µm-NoB. Top row: Depth scanned images showing an overlay image of maltodextrin and acacia gum. Bottom row: Depth scanned images showing overlay image of all the components. Red : acacia gum, blue: maltodextrin, yellow : sunflower oil. 3.2.3 Oil encapsulation The oil encapsulation was determined by surface oil extraction and XPS. By solvent extraction, the oil on the surface or connected to the surface via pores or cracks is removed. In XPS, the atomic composition of the surface layer (about 2-5 nm thickness) is detected, and these data are used to provide the surface coverage of different molecular components using a patch model (Fäldt, et al., 1993) and the atomic concentration of reference pure components. The atomic concentrations of O, C and N in the pure components used in this work (maltodextrin DE12, acacia gum and sunflower oil) are given in Table 3. Table 3. Atomic concentration of reference pure components. The data are averages of two measurements. The average standard deviation in the measurements is 0.3 atom%.
Maltodextrin DE12 Acacia gum Sunflower oil
O (%) 44.1 40.4 9.6
C (%) 55.9 57.2 90.4
N (%) 0 2.4 0
The data for surface oil extraction and estimations of the surface molecular composition based on XPS data are shown in Table 4. Surface oil by solvent extraction was low in all cases (< 4 of total oil in powder), and lowest in the powders which did not experience droplet breakup (0.1µm-NoB, 1µmNoB). XPS analysis showed an overrepresentation of the oil on the particle surface in all powders, which indicates that the oil spreads over the surface of the particles in a thin layer. Surface oil in 1µmNoB was, however, higher than expected, which is attributed to the mechanical breakup of still sticky 13
particles after spray-drying. By mechanical breaking it is reasonable to assume that some of the encapsulated oil was released and accounted for by extracted oil (Table 4) and the oil surface coverage by XPS analysis (Table 4). The 1µm-NoB particles, taking longer time to dry (larger particles), showed how the only matrix component on the surface was acacia gum, indicating that acacia gum adsorbs to the drying drop surface. Giving sufficient time for adsorption, acacia gum will cover the surface. Table 4 Surface oil extracted by petrol ether and molecular component surface coverage analysed by XPS. Extraction Surface oil (% of total oil)
Oil surface Acacia coverage gum (%) surface coverage Powder sample (%) 1µm-B 3.4±0.2 67 12 0.1µm-NoB 0. 4±0.0 30 21 1µm-NoB 1.11 72 29 1 Only one repetition of the analysis for this sample.
XPS Maltodextrin Maltodextrin surface / acacia gum coverage (in bulk: 1.5) (%) 21 49 0
1.75 2.33 0
4 Discussion The combination of the results presented for the different methods together provide a more comprehensive insight in the microstructure of spray dried emulsions. The methods are complementary in that they give information on different scales and from different areas of the particles. A comparison of depth of analysis and resolution of these techniques is shown in Fig. 6. The oil extraction provides the amount of fat accessible from the surface and through cracks and pores in the particles, and is thus not a purely surface sensitive technique. XPS and oil extraction provide quantitative information, while the imaging techniques are mainly qualitative.
Fig. 6 Comparison of depth of analyses and resolution of XPS, confocal Raman microscopy and, LVSEM characterization techniques. The surface analysed by XPS is indicated by a darker outline on one of the powder particles. The particle morphology was found to be related to the atomization, more specifically the drop size and correspondingly the drying time. The larger drops in 1µm-NoB dried more slowly, and dense, wrinkled particles were produced. The drops produced at higher atomization speed were smaller, thus 14
dried faster and a larger fraction of the drops locked their size at an earlier stage of drying, leading to vacuole formation. The surface will be spherical or wrinkled depending on the viscoelastic properties of the air/liquid surface film during shrinkage, with additional influence of the drying rate. The equilibrium surface tension of the feed solution using protein solutions has been proposed to describe the morphology of spray-dried particles (Xu, Howes, Adhikari, & Bhandari, 2012). However, it was recently shown that it is rather the surface rheological properties than the surface tension (dynamic or equilibrium) that determines morphology at given solids content and drying conditions (Nuzzo et al, 2015a). The dented morphology is quite typical of acacia gum containing particles dried at low solids content. Here the surface dilatational modulus was about 70 mN/m (obtained by oscillating pendant drop method, data not shown), which is a high level and in accordance with data of Sanchez, Renard, Robert, Schmitt, and Lefebvre (2002). However, the high evaporation rates and high solids content in the spray-drying feed will lead to early crust formation and thus the particles can still be spherical as is the case in most of the particles in powder 1µm-B and 0.1µm-NoB. Accordingly, the particle morphology is determined by the different parallel non-equilibrium processes in the drying drop and at its surface. The internal distribution of the components has hitherto been poorly explored, with a small number of studies using confocal laser scanning microscopy indicating how the protein and fat are distributed in solid particles (Auty et al., 2001; Drusch & Berg, 2008; Taneja et al., 2013). However, to our knowledge, there are no studies on how different matrix materials with similar properties are distributed. Here, we have shown that the distribution of matrix materials as well as of oil droplets depend on the drying conditions and processing prior to drying (emulsion droplet size). The confocal Raman imaging shows that acacia gum and maltodextrin phase segregates from an acacia gum rich surface layer in 1µm-NoB, i.e., the most slowly drying sample. A surface enrichment of acacia gum is observed in the other powders too, but to a lesser extent, which is attributed to faster drying. This corroborated by the XPS results, which show higer ratio of maltodextrin/ acacia gum at the surface for these samples (1µm-B and 0.1 µm-NB) (Table 4). This is in turn explained by the surface active character of acacia gum, which leads to its adsorption at the air-water interface of the drying drops in the spray-dryer. Such behaviour of surface active components has been investigated in several studies (Elversson et al., 2006; Fäldt & Bergenståhl., 1994; Jones et al., 2013; Keogh et al., 2001; MillqvistFureby et al., 2001; Millqvist Fureby, 2003; Murrieta-Pazos et al., 2012; Nijdam et al., 2006; Nuzzo et al., 2015a). The surface enrichment of acacia gum then directs the phase segregation so that a surface layer enriched in this component is created. Similar behaviour has been observed for phase segregation in confined droplets (Wassén, et al., 2013). The oil droplets were shown by several techniques to be located close to the surface of the particles when droplet break-up occurred during atomisation. This was also associated with higher levels of free fat, while surface fat appeared more related to the oil droplet size. In the present case (1µm-B), the droplet break-up results in an increase of the oil droplet surface area by a factor of 1.5. It can be speculated that when an oil droplet breaks up during atomization there is insufficient time for the macromolecular surfactant to adsorb to the new interface before drying is completed, resulting in incomplete surface occupation at the oil/water interface which in turn leads to unstable droplets. These more fragile droplets would then more easily rupture at the particle surface and result in higher levels of surface oil, as detected by oil extraction. A further aspect that could result in higher surface fat is that larger droplets are less stable than smaller ones, resulting in more frequent rupture of the oil droplets in 1µm-B and 1µm-NoB than in 0.1µm-NoB, and also substantially higher oil release per ruptured oil droplet.
15
XPS and oil extraction provide information on different aspects of the surface oil. XPS quantifies the coverage at the outermost layer of the particles, a property that is relevant for powder particle interactions, initial behaviour in wetting, and dispersion. Oil extraction, on the other hand, quantifies the accessible oil, including oil accessible through cracks and pores. It seems clear that oil droplet breakup in 1µm-B during atomization brought the droplets close to the surface as LV-SEM and Raman imaging suggests ( Fig. 3, Fig. 5), and the surface was covered by fat to a high level (Table 4). A protecting layer (>0.5 µm) of acacia gum was detected on the surface of the powders where no oil droplet breakup occurred as indicated by LV-SEM images, and confocal Raman microscopy (Fig. 3, Fig. 5), and this is further supported by the XPS data (Table 4) although the oil coverage of 1µm-NoB was high. The main part of the oil in spray dried particles is present as droplets in the matrix, and at the surface the oil can be present as a thin (even monomolecular) layer as well as in the form of patches. The low surface energy of the oil enables efficient spreading of the liquid oil on the particle surface. Oil could spread on the particle surface during manufacture, handling and storage (Fäldt & Bergenståhl., 1995; Fäldt & Sjöholm, 1996), and it has been shown that the composition of the hydrophilic matrix remains constant while the oil coverage varies if the composition is kept constant but not the processing conditions (Keogh et al., 2001; Millqvist-Fureby et al., 2001). The theoretical thickness of oil spread in the particle surface was calculated by combining both techniques (Table 4) and the particle size data (Error! Reference source not found., Fig. 1). For a hollow particle representative of powder 1µm-B with a diameter of 50 µm, surface to core thickness of 5 µm, and 67% of oil surface coverage, the average oil layer thickness is 23 nm. The corresponding oil layer thickness for 1µm-NoB is 7 nm, and for 0.1µm-NoB it is 6 nm. Raman microscopy showed no evidence of an oil-rich surface layer within the resolution limits of the technique, thus this layer is expected to be below 200 nm. These calculated oil layer thicknesses indicate that the oil is present in patches on the surface, since the layer thickness exceeds the depth of analysis of XPS and both acacia gum and maltodextrin are detected. Moreover, some of the extracted oil may be derived from the interior of the particles through cracks, pores and interconnected oil droplets, as well as broken particles.
5 Conclusions Even in the short timescales of drying there is time for the surface active ingredient to adsorb at the drop surface and for the two polymeric matrix ingredients to phase segregate as shown by analysis by confocal Raman microscopy. This is of interest when surface modification of the particle is required, and is also related to the encapsulation capacity of the formulation. Oil droplet breakup during atomization was found to have an influence on the surface oil content, determined by XPS and oil extraction. Droplet elongation and breakup brought the droplets close to or on the emulsion drop surface. In emulsions which, due to their small droplet size or gentle atomization conditions, preserved their original size distribution, the oil droplets in the particles were located below the surface active component layer. The surface analytical techniques reveal the distribution of components at different depths into the particles. XPS provides the composition of the outer surface which is important for e.g. wetting properties. Confocal Raman microscopy gives good opportunity to characterize the microstructure as it provides structural information through the whole particle depth. However, due to the resolution limitations, observation of submicron oil droplets was not possible, and the results showed here were only qualitative. It would therefore be interesting to explore the quantitative possibilities of confocal
16
Raman microscopy in order to obtain the relative composition at any region of the particle, although this is an ambitious endeavour. In summary, the techniques studied in this work provided complementary information on the spraydried emulsions, improving our understanding of the mechanisms behind events occurring during spray drying, such as droplet breakup, molecular adsorption and diffusion, and morphology build-up.
6 Acknowledgements The authors acknowledge the European Commission 7th Framework Program through the Marie Curie Actions, Initial Training Network ―Powtech‖ (FP7-PEOPLE-2010-ITN 264722) for supporting this research. Thanks also to Giana Perre and Nathalie Ruscassier at Laboratoire de Génie des Procédés et Matériaux, at Ecole Centrale Paris for performing the SEM imaging. Thanks to Johan Andersson and Mikael Sundin at SP Chemistry, Materials and Surfaces for the LVSEM imaging, and XPS analysis, respectively. Further, the authors acknowledge the Troëdsson Foundation for the Confocal Raman Microscope at SP Chemistry, Materials and Surfaces.
Industrial relevance text The study presented in this paper is of clear industrial relevance as it highlights how the processing parameters include oil encapsulation efficiency as well as the external and internal structure on nanoand micro-levels. These structural details have a strong influence on the behaviour of the powder when used in different applications. In particular, the novel revelation of the phase segregation occurring in spray-dried powders may in future developments be used to enhance encapsulation efficiency if various encapsulation applications in foods as well as in other areas.
17
References
Auty, M. A. E., Twomey, M., Guinee, T. P., & Mulvihill, D. M. (2001). Development and application of confocal scanning laser microscopy methods for studying the distribution of fat and protein in selected dairy products. Journal of Dairy Research, 68(03), 417-427. Chew, J. H., Fu, N., Gengenbach, T., Chen, X. D., & Selomulya, C. (2015). The compositional effects of high solids model emulsions on drying behaviour and particle formation processes. Journal of Food Engineering, 157(0), 33-40. Drusch, S., & Berg, S. (2008). Extractable oil in microcapsules prepared by spray-drying: Localisation, determination and impact on oxidative stability. Food Chemistry, 109(1), 17-24. Elversson, J., & Millqvist-Fureby, A. (2006). In situ coating—An approach for particle modification and encapsulation of proteins during spray-drying. International journal of pharmaceutics, 323(1), 52-63. Fäldt, P., Bergenståhl, B., & Carlsson, G. (1993). The surface coverage of fat on food powders analyzed by ESCA (electron spectroscopy for chemical analysis). Food structure, 12. Fäldt, P., & Bergenståhl, B. (1994). The surface composition of spray-dried protein—lactose powders. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 90(2), 183-190. Fäldt, P., & Bergenståhl, B. (1995). Fat encapsulation in spray-dried food powders. Journal of the American Oil Chemists’ Society, 72(2), 171-176. Fäldt, P., & Sjöholm, I. (1996). Characterization of spray-dried whole milk. Milchwissenschaft, 51(2), 88-92. Jafari, S. M., Assadpoor, E., Bhandari, B., & He, Y. (2008). Nano-particle encapsulation of fish oil by spray drying. Food Research International, 41(2), 172-183. Jones, J. R., Prime, D., Leaper, M. C., Richardson, D. J., Rielly, C. D., & Stapley, A. G. F. (2013). Effect of processing variables and bulk composition on the surface composition of spray dried powders of a model food system. Journal of Food Engineering, 118(1), 19-30. Keogh, M., O'Kennedy, B., Kelly, J., Auty, M., Kelly, P., Fureby, A., & Haahr, A. M. (2001). Stability to Oxidation of Spray‐Dried Fish Oil Powder Microencapsulated Using Milk Ingredients. Journal of Food Science, 66(2), 217-224. Kim, E. H.-J., Chen, X. D., & Pearce, D. (2002). Surface characterization of four industrial spraydried dairy powders in relation to chemical composition, structure and wetting property. Colloids and Surfaces B: Biointerfaces, 26(3), 197-212. Kim, E. H.-J., Chen, X. D., & Pearce, D. (2009). Surface composition of industrial spray-dried milk powders. 2. Effects of spray drying conditions on the surface composition. Journal of Food Engineering, 94(2), 169-181. Millqvist-Fureby, A., Elofsson, U., & Bergenståhl, B. (2001). Surface composition of spray-dried milk protein-stabilised emulsions in relation to pre-heat treatment of proteins. Colloids and Surfaces B: Biointerfaces, 21(1), 47-58. Millqvist Fureby, A. (2003). Characterisation of spray-dried emulsions with mixed fat phases. Colloids and Surfaces B: Biointerfaces, 31, 65-79. Munoz-Ibanez, M., Azagoh, C., Dubey, B. N., Dumoulin, E., & Turchiuli, C. (2015). Changes in oilin-water emulsion size distribution during the atomization step in spray-drying encapsulation. Journal of Food Engineering(0). Murrieta-Pazos, I., Gaiani, C., Galet, L., & Scher, J. (2012). Composition gradient from surface to core in dairy powders: agglomeration effect. Food Hydrocolloids, 26(1), 149-158. Nijdam, J. J., & Langrish, T. A. G. (2006). The effect of surface composition on the functional properties of milk powders. Journal of Food Engineering, 77(4), 919-925. Nikolova, Y., Petit, J., Sanders, C., Gianfrancesco, A., Scher, J., & Gaiani, C. (2015). Toward a better determination of dairy powders surface composition through XPS matrices development. Colloids and Surfaces B: Biointerfaces, 125(0), 12-20. Nuzzo, M., Millqvist-Fureby, A., Sloth, J., & Bergenståhl, B. (2015a). Surface composition and morphology of particles dried individually and by spray drying. Drying Technology, 33(6), 757-767. Nuzzo, M., Sloth, J., Brandner, B., Bergenståhl, B., & Millqvist-Fureby, A. (2015b). Confocal Raman microscopy for mapping phase segregation in individually dried particles composed of lactose 18
and macromolecules. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 481(0), 229-236. Risch, S. J., & Reineccius, G. A. (1988). Spray-Dried Orange Oil. Flavor Encapsulation, 370(370), 67-77. Sanchez, C., Renard, D., Robert, P., Schmitt, C., & Lefebvre, J. (2002). Structure and rheological properties of acacia gum dispersions. Food Hydrocolloids, 16(3), 257-267. Sarkar, A., Arfsten, J., Golay, P.-A., Acquistapace, S., & Heinrich, E. (2016). Microstructure and long-term stability of spray dried emulsions with ultra-high oil content. Food Hydrocolloids, 52, 857-867. Soottitantawat, A., Bigeard, F., Yoshii, H., Furuta, T., Ohkawara, M., & Linko, P. (2005). Influence of emulsion and powder size on the stability of encapsulated d-limonene by spray drying. Innovative Food Science & Emerging Technologies, 6(1), 107-114. Soottitantawat, A., Yoshii, H., Furuta, T., Ohkawara, M., & Linko, P. (2003). Microencapsulation by Spray Drying: Influence of Emulsion Size on the Retention of Volatile Compounds. Journal of Food Science, 68(7), 2256–2262. Taneja, A., Ye, A., Jones, J. R., Archer, R., & Singh, H. (2013). Behaviour of oil droplets during spray drying of milk-protein-stabilised oil-in-water emulsions. International Dairy Journal, 28(1), 15-23. Turchiuli, C., Lemarié, N., Cuvelier, M.-E., & Dumoulin, E. (2013). Production of fine emulsions at pilot scale for oil compounds encapsulation. Journal of Food Engineering, 115(4), 452-458. Walton, D., & Mumford, C. (1999). The morphology of spray-dried particles: the effect of process variables upon the morphology of spray-dried particles. Chemical Engineering Research and Design, 77(5), 442-460. Wassén, S., Lorén, N., van Bemmel, K., Schuster, E., Rondeau, E., & Hermansson, A.-M. (2013). Effects of confinement on phase separation kinetics and final morphology of whey protein isolate-gellan gum mixtures. Soft Matter, 9(9), 2738-2749. Wu, W. D., Liu, W., Gengenbach, T., Woo, M. W., Selomulya, C., Chen, X. D., & Weeks, M. (2014). Towards spray drying of high solids dairy liquid: Effects of feed solid content on particle structure and functionality. Journal of Food Engineering, 123(0), 130-135. Xu, Y. Y., Howes, T., Adhikari, B., & Bhandari, B. (2012). Investigation of relationship between surface tension of feed solution containing various proteins and surface composition and morphology of powder particles. Drying Technology, 30(14), 1548-1562.
19
Figure captions Fig. 1. Light microscopy images of spray-dried emulsion to illustrate internal morphology: Solid (S) or Hollow (H). Magnification of the objective was 20x, and the scale bar is 100 µm. Samples dispersed in silicon oil. Fig. 2. Scanning electron microscope micrographs of spray dryed emulsions. Top row : Overview of the powder particles showing their topography. Bottom row: Particle surface after free oil extraction. (a) 1µm-B, (b) 0.1µm-NoB, and (c) 1µm-NoB. Fig. 3. Localisation of oil droplets in the powder particles. SEM images of cross sections (top row). LV-SEM images using gaseous backscattered secondary electrons (CBS) (middle row) and large field detector (LFD) (bottom row). (a) 1µm-B, (b) 0.1µm-NoB and (c) 1µm-NoB. Fig. 4. Raman spectra for reference components (a), and example of spectra at a given position in a spray-dried particle (b). The arrows point out distinctive peaks in the reference spectra which can also be observed in the sample spectrum. Fig. 5. Internal mapping by confocal Raman microscopy of spray-dried emulsions. (a), (d) 1µm-B, (b), (e) 0.1µm-NoB, and (c), (f) 1µm-NoB. Top row: Depth scanned images showing an overlay image of maltodextrin and acacia gum. Bottom row: Depth scanned images showing overlay image of all the components. Red : acacia gum, blue: maltodextrin, yellow : sunflower oil. Fig. 6 Comparison of depth of analyses and resolution of XPS, confocal Raman microscopy and, LVSEM characterization techniques. The surface analysed by XPS is indicated by a darker outline on one of the powder particles.
20
Table 5. Conditions for the production of the spray dried powders analyzed. Name
d50,initial (µm)
Emulsion viscosity (Pa.s)
Atomizer speed (rpm)
Ca value
Tinlet (°C)
1µm-B
1.34
0.35
33200
2.1
220
0.1µmNoB 1µmNoB
0.13
0.30
33200
0.2
220
1.94
0.35
3270
0.1
220
Toutlet (°C)
129125 129125 134
Status of oil droplet after spraying Breakup
d50, reconstituted (µm)
Yes
0.89
No
0.13
No
1.94
21
Table 6. Powder particle median diameter d50 and d10 and d90 measured by Laser light diffraction or 1 Estimated from SEM micrographs. Powder sample
d10 (μm)
d50 (μm)
d90 (μm)
1µm-B 0.1µm-NoB 1µm-NoB
16 6
37 19 ~801
85 42
22
Table 7. Atomic concentration of reference pure components. The data are averages of two measurements. The average standard deviation in the measurements is 0.3 atom%.
Maltodextrin DE12 Acacia gum Sunflower oil
O (%) 44.1 40.4 9.6
C (%) 55.9 57.2 90.4
N (%) 0 2.4 0
23
Table 8 Surface oil extracted by petrol ether and molecular component surface coverage analyzed by XPS.
Extraction
XPS
Surface oil Acacia gum Maltodextrin Oil surface Powder (% of total surface surface coverage coverage (%) sample oil) coverage (%) (%) 1µm-B 3.4±0.2 67 12 21 0.1µm-NoB 0. 4±0.0 30 21 49 1µm-NoB1 1.11 72 29 0 1 Only one repetition of the analysis for this sample was possible.
Maltodextrin / acacia gum (in bulk: 1.5 ) 1.75 2.33 0
24
Overview
150 µm
100 µm
150 µm
Particle surface after free oil extraction
(a)
50 µm
20 µm
20 µm (b)
(c)
2 µm
(a)
2 µm
6 µm
10 µm
20 µm
50 µm
10 µm
20 µm
50 µm
(b)
(c)
(a)
(b)
(c)
(d)
(e)
(f)
XPS Particle surface 1nm 2-5 nm 10nm Average, image or 100nm spot 1μm 300 µm 10μm 100μm
Confocal Raman microscopy 25 x 25 μm image
LV-SEM Back scattering electrons 100nm Material contrast Scattering probability =f(Atomic number)
Resolution: 200nm
Secondary electrons 1nm Surface sensitivity
Resolution: <5nm
Surface oil extraction
XPS, 2-5 nm
Confocal Raman microscopy
vacuole
LV-SEM (500 nm)
SEM (1 nm) ̴ 50 µm