Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
1
Enhanced creaming of milk fat globules in milk emulsions by the application of 1 ultrasound and detection by means of optical methods 2 3 P. Juliano1, * , A. Kutter1,2, L.J. Cheng1, P. Swiergon1, R. Mawson1, and M. A. 4
Augustin1 5
1CSIRO Food and Nutritional Sciences, 671 Sneydes Road, Werribee VIC, 3030, Australia 6
2Institute of Fluid Mechanics and Erlangen Graduate School in Advanced Optical Technologies, 7
University of Erlangen-Nuremberg, Cauerstrasse 4, 91058 Erlangen, Germany 8
Abstract 9
The effects of application of ultrasonic waves to recombined milk emulsions (3.5% fat, 10
7% total solids) and raw milk on fat destabilization and creaming were examined. 11
Coarse and fine recombined emulsions (D[4,3] = 2.7 µm and 9.3 µm) and raw milk 12
(D[4,3] = 4.9 µm) were subject to ultrasound for 5 min at 35°C and 400 kHz or 13
1.6 MHz (using a single transducer) or 400kHz (where the emulsion was sandwiched 14
between two transducers). Creaming, as confirmed by Turbiscan measurements, was 15
more evident in the coarse recombined emulsion and raw milk compared to that of the 16
recombined fine emulsion. Particle size analysis, supported by microscopic images, 17
showed that both flocculation and particle coalescence occurred upon ultrasound 18
treatment, the extent of which was dependent on the milk sample and ultrasound 19
conditions. These results imply that the ultrasound has potential to pre-dispose fat 20
particles in milk emulsions to creaming due to flocculation and coalescence, not only 21
in standing wave systems, but in systems with inhomogeneous sound distributions. 22
23
Keywords 24
Ultrasound, separation, milk, standing waves, milk fat globule, emulsion, coalescence 25
* Corresponding author. Tel.: +61 397313276; fax: + 397313201. E-mail address: [email protected] (P. Juliano).
2
1. Introduction 26
Ultrasonics has been proposed as a technique to separate particles under a standing 27
wave field. A number of publications concentrate on the removal of solid particles 28
(mainly polystyrene and polyamide beads) in batch (Tolt and Feke, 1993; Whitworth 29
et al., 1991) and continuous processes at laboratory scale (Johnson and Feke, 1995; 30
Pangu and Feke, 2004; Hawkes and Coakley, 2001; Mandralis and Feke, 1993; 31
Kapishnikov et al., 2006; Groeschl, 1998a; Groeschl, 1998b). In the food area, 32
ultrasound has been used to split canola oil emulsions by enhancing particle 33
flocculation and coalescence (Nii et al., 2009). 34
Particles in a standing pressure wave field experience the so called primary acoustic 35
radiation force, which can be expressed analytically for a compressible sphere 36
(Yosioka and Kawasima, 1955): 37
( )34sin 2
3ac acF R k E kxπ= − Φ (1) 38
where R denotes the particle radius, k the wave number of the ultrasound (=2π/λ, with 39
λ being the wavelength), Eac the average acoustic energy density, Φ the acoustic 40
contrast factor and x the distance from a nodal point of the standing wave. When 41
particles with a non-zero acoustic contrast factor (ACF) are subjected to a standing 42
ultrasonic wave the acoustic radiation force drives them either to the nodes or 43
antinodes for particles with positive and negative ACF values respectively. The 44
spatial distance between successive bands is therefore λ\2. The acoustic contrast 45
factor is given by 46
5 2
2M P M
M P P
ρ ρ βρ ρ β
−Φ = −+ (2) 47
3
where ρM, ρP and βM, βP are the density and the compressibility of the continuous 48
medium and the dispersed particles respectively. 49
In a pressure field, the container or chamber walls act as pressure antinodes as they 50
are the points where the pressure differences are built up. As can be derived from Eq, 51
2, fat particles in an aqueous environment show a negative contrast factor and 52
therefore have the tendency to move towards the walls. In addition, Eq. 2 shows that 53
even particles with zero density difference with respect to their surrounding medium 54
can be manipulated by ultrasound, as long as their compressibilities differ. After the 55
particles form bands of high concentration at the pressure nodes or antinodes the so 56
called “second acoustic force” or Bjerknes force (Apfel, 1988) drives them together, 57
leading to clusters. The second acoustic force originates from interactions of the 58
particles with the fluid, e.g. reflections of the ultrasound field at the particles’ surface. 59
A review dealing with the effects of ultrasound on proteins in dairy products was 60
recently published by Ashokkumar et al. (2010). However, the information on the use 61
of ultrasound for separation of fat in milk is scarce. A few references dealing with the 62
separation of milk fat globules are found in the field of biotechnology where fat 63
particles are separated from red blood cells using standing waves (Petersson et al., 64
2004; Petersson et al., 2005). Miles et al. (1995) determined the threshold levels for 65
the separation of micro-organisms from suspensions followed in a cuvette resonance 66
chamber set up and further suggests that the application of ultrasound standing waves 67
(1-3 MHz) can enhance not only bacteria but also cream separation. However, a 68
systematic evaluation of the effects of ultrasound standing waves on the creaming 69
properties of milk has not been done to date. The present contribution evaluates the 70
formation of milk fat globule banding, flocculation and coalescence, in coarse and 71
fine emulsions of recombined milks and in raw milk. 72
4
2. Materials and methods 73
Sample preparation 74
To study the effect of ultrasound on the creaming of milk fat globules three model 75
systems were used. These comprised of two models made out of recombined milk 76
emulsions (a coarse emulsion and a fine emulsion) and freshly collected raw milk. 77
The recombined milk emulsion was prepared by dispersion of skim milk powder 78
(SMP, Tatura Milk Industries Limited, Victoria, Australia) in water at 50ºC followed 79
by the addition of anhydrous milk fat (AMF) (Murray Goulburn Co-operative 80
Company Limited, Victoria Australia) to obtain 7% total solids emulsion with 3.5% 81
fat. The AMF was previously dyed by adding 0.01% oil-red-O colorant (BDH 82
Laboratory suppliers, Poole, England). An Ultra-turrax operated at 20,000 rpm for 83
3 min (ultra-turrax T25, IKA-Labortechnik, Staufen, Germany,) was utilised to form a 84
coarse recombined milk emulsion (hereafter termed “coarse emulsion”). To reduce 85
the particle size, the coarse emulsion was treated with a laboratory homogenizer after 86
two passes (FOSS NIRSystems Inc., Laurel, USA, Type D head, equivalent to a 87
treatment of approximately 140 bar). This homogenized emulsion of recombined milk 88
emulsion was hereafter termed a “fine emulsion”. All trials including raw milk used 89
fresh milk directly obtained from Bega Cheese (Victoria, Australia) on the day of the 90
experiment. 91
92
Resonance chamber configuration and sample treatment 93
The sonication of the samples (7 mL) was carried out in a Turbiscan tube 94
(Formulaction, Toulouse, France) which was sealed with a metal plate at the bottom, 95
acting as a sound transmitter into the fluid. Fig. 1 and Fig. 2 illustrate the 96
experimental set up for the one transducer plate (1) and two transducer plate (2) 97
5
configuration. Ultrasound frequencies used were 400 kHz and 1.6 MHz with the 98
single transducer set-up and only at 400 kHz with the two transducer plate set-up. 99
The sample was placed in a water bath with 35°C in order to minimize the 100
temperature increase due to the energy dissipation. In Fig. 2 the tube walls acted as a 101
sound transmitter as the set up had to be submerged in the water bath. Therefore, the 102
sound distribution in the tube in this case is complex and not predictable, as the glass 103
walls transmit and interfere with the external sound field of the transducers. In all 104
experiments the ultrasound treatment time was set to 5 min. All ultrasonication runs 105
were carried out as triplicates. 106
Two different ultrasound transducer systems were used with 400 kHz (Submersible 107
Transducers, Sonosys Ultraschallsysteme GmbH, Neuenbuerg, Germany) and 108
1.6 MHz (Nebulizer, APC International Inc., Mackayville, Pennsylvania, USA). To 109
estimate the power input into the liquid a slightly modified set up compared to Fig. 1 110
served as model system. A defined mass of water (10 g ± 0.1 g) was filled into the 111
tube, which was not submerged but in contact to air (metal cap still in the water bath). 112
By measuring the temperature increase after a defined period of time (120 s) the 113
power input into the liquid could be estimated. For transferring these results to the 114
dairy system this approach assumes that the differences in the physical properties of 115
water and milk are negligible or do not influence the power input. It is also assumed 116
that the heat conduction into the ambient through the glass walls is negligible 117
compared to the energy input through the dissipation of sound energy. The power 118
input values could be determined to be 1.6 W ± 0.2 W for the 400 kHz and 119
0.35 W ± 0.07 W for the 1.6 MHz unit, respectively. No acoustic streaming was 120
present at all times, which would lead to a remixing of the particles in the dairy 121
systems due to turbulent structures. 122
6
The sound intensity in the tube was estimated by using a hydrophone (TC 4038, 123
Reson A/S, Slangerup, Denmark). It was inserted into the set up at different positions 124
to get an approximation of the different sound levels throughout the tube. For the 125
described process parameters this sound pressure amplitude ranged around 350 to 126
500 kPa, which corresponds to an average acoustic energy density Eac of 10 to 25 J/m3. 127
128
Turbiscan measurements 129
In order to get an estimation of the creaming in the sample a Turbiscan MA 2000 130
(Formulaction, Toulouse, France) was employed, which scans the whole sample in the 131
tube. The Turbiscan measures backscattered and transmitted light as a function of the 132
axial tube coordinate and time. Due to the increase of backscattered light at the layers 133
with a higher number of dispersed particles, it is possible to measure high cream 134
concentration regions. 135
Scans were performed over a 10 minute period right after removing the samples from 136
the ultrasonic set up. For each model system a non-sonicated control was treated 137
exactly in the same way as the sonicated samples. They were filled into the Turbiscan 138
tube, put into the water bath and after 5 min they were placed into the Turbiscan and 139
measured for further 10 min. 140
141
Creaming and recovery rates 142
To compare the different creaming extents the ratio of the area of the peak 143
corresponding to the separated cream phase was calculated with respect to the total 144
backscattering measured in the sample (Eq. 3). Fig. 3 illustrates the computation of 145
the “creaming extent” (CE). The cream phase was defined as the curve area, starting 146
where the backscattering value is higher than 1.5 % in comparison to the baseline 147
7
value representing the bulk liquid. The 1.5% value represents the measurement error 148
of backscattering determined for the Turbiscan equipment. For comparison between 149
different measurements (e.g. initial particle concentration) the area under the cream 150
phase curve was related to the area under the whole curve. Analytically this can be 151
expressed as 152
( )
( )
creamphase0
0
creaming extent =
L
L
BS z dz
BS z dz
∫
∫ (3) 153
where z represents the axial position in the tube (origin at the bottom), L the free 154
surface and BS(z) the backscattering at position z. 155
This approach assumes that the “backscattering potential” and hence number of 156
particles, which equals concentration, is shifted when a cream phase is formed. To 157
assure that this procedure yields valid results, the so called “recovery rate” (RR) was 158
computed with the following equation 159
( )
( )
sample0
control0
recovery rate=
L
L
BS z dz
BS z dz
∫
∫ (4) 160
where BS(z)sample stands for the backscattering of the sample and BS(z)control for the 161
corresponding non-ultrasonic treated control at the same time. If this value is close to 162
unity this means that the overall backscattering of the sample and the control match. 163
For instance, assuming no change in particle sizes, a recovery rate of one can be 164
attributed to the fact that all particles are moved to another position in the tube but do 165
not flocculate or coalesce. It also guarantees that no significant non-linear effects are 166
8
introduced due to high particle concentrations and the backscattering is directly 167
proportional to the square root of the concentration. 168
169
Particle sizing and microscopy 170
All particle size measurements were conducted with a Galai CIS-1 (Galai Production 171
LTD, Haemek, Israel), which provides particle size ranges allowing the detection of 172
changes in size. All measurements were repeated twice per sample treatment. Samples 173
were adjusted to a detectable dilution range with distilled water. To differentiate 174
between flocculation (which is a reversible phenomenon in which the integrity of the 175
interfacial layer of the initial fat droplet is maintained) and coalescence (where there 176
disruption of the interfacial layer and the formation of a larger oil droplet) the particle 177
size increase for the same set of described experiments was conducted a second time. 178
After sonication the cream phase was remixed by shaking the Turbiscan tube gently to 179
break up particle clusters and determine if particle coalescence occurred. After the 180
Turbiscan measurements microscopy images (400x lens,Olympus BH-2 light 181
microscope, soft imaging system Color View IIIu, Japan) and particle size 182
distributions of the top cream layer (1 ml extracted from the surface with a pipette) 183
were recorded to observe potential changes in the globule structure and concentration. 184
185
Statistics 186
Changes in particle size were evaluated by using the General Linear Model in the 187
Minitab Statistical package to a 95% level of significance (Minitab ® Release 14.11, 188
2003, Coventry, UK). The Tukey method was applied at 95% confidence level for 189
pairwise comparisons. 190
9
3. Results and Discussion 191
192 3.1 Particle size of initial emulsions 193 194 195
Fig. 4 shows the number distribution of all samples, where a wider distribution can be 196
found in the case of raw milk. The calculated volume mean diameters D[4,3] for the 197
samples were 4.9 µm for raw milk, 2.7 µm for the fine emulsion and 9.3 µm for the 198
coarse emulsion. 199
200 3.2 Effect of ultrasonication 201 202
3.2.1 Creaming behavior 203
The creaming behavior expressed as the creaming extent from Turbiscan readings as 204
well as particle size results are presented after sonication runs. The extent of creaming 205
was dependent on the milk sample used, the frequency applied and the transducer 206
geometry. Differences in creaming were obtained with the use of a single transducer 207
at the two frequencies (400kHz or 1.6 MHz) and also at fixed frequency (400 KHz) 208
when different transducer set-ups were used (single or two transducer geometry). 209
210
3.2.1.1. Single transducer configuration at various frequencies 211
By applying one transducer and a reflector (Fig. 1) and adjusting the distance between 212
them a planar standing wave could be achieved. Under these conditions the force on a 213
particle is maximized. This effect is shown as an example in Fig. 5 for the coarse 214
emulsion of recombined milk. It can be seen that the ultrasound leads to the formation 215
of bands with higher fat concentration. After switching off the ultrasound field 216
floccules start rising immediately and re-disperse at the same time. If the sonicated 217
sample is stirred all floccules completely re-disperse back into the solution. This re-218
10
dispersing suggests that the fat globules clustered in bands made up of discrete single 219
droplets. Miles et al. (1995) have identified the threshold amplitudes to form bands in 220
suspensions of latex microspheres at particle size ranges between 1 and 5 µm at 1 221
MHz and higher frequencies. The minimum average sound pressure determined for 222
our systems was 350 kPa, which exceeds the threshold values reported for banding of 223
all particles sizes tested at 1 MHz. 224
The enhanced creaming extent caused by ultrasonics in the coarse emulsion, in 225
comparison to the untreated control, could be observed with the naked eye. The top 226
layer exhibited a more intense red color which is due to the higher concentration of fat 227
particles. In these experiments with the Turbiscan tubes, the Turbiscan readings (Fig. 228
6) reveal that hardly any natural creaming could be observed for the coarse emulsion 229
of recombined milk without ultrasound after 10 min. The results for the single 230
transducer for the two frequencies are summarized in Table 1 and Table 2. Comparing 231
the results for the least stable coarse emulsion at 400 kHz and 1.6 MHz it becomes 232
obvious that the higher frequency leads to a two-fold increase in creaming extent. The 233
increase with higher frequency was also observed with the raw milk. In this case a 234
weak level of creaming could be measured at 400 kHz after 10 min, in contrast to 235
1.6 MHz. The more prominent separation observed in the coarse emulsion could be 236
due to the presence of larger particles in the 9 to 15 µm range, which were not present 237
in milk. 238
The fine emulsion showed no creaming, regardless of the frequency and settling time 239
during the Turbiscan trials. The smaller particle sizes of the fine emulsion compared 240
to the coarse emulsion and the raw milk contributed to the resistance for the fine 241
emulsion to creaming. Table 1 and Table 2 reveal that all the observed separations 242
occurred during the sonication for the coarse emulsion at 400 KHz and 1.6 MHz and 243
11
for the raw milk at 1.6 MHz. Thus, these experiments clearly reveal that ultrasound 244
can be used to accumulate and separate milk fat in its native environment. 245
246
3.1.2. Particle size changes 247
During particle flocculation the probability of coalescence rises due to the increased 248
local concentration. Both processes, pure flocculation and coalescence, differ from the 249
point of view of their physics. Both phenomena result in an increase in particle size. 250
In the case of flocculation where two or more fat droplets aggregate while each 251
maintaining their interfacial layer, there is an apparent hydrodynamic increase. In 252
flocculation, the clustered fat globules may be re-dispersed. In the case of coalescence, 253
there is disruption of the original interfacial layer of droplets, leading to an 254
irreversible increase in particle size of the oil droplet. Both flocculation and 255
coalescence result in enhanced creaming. This has been shown in previous studies 256
which mention these mechanisms operating for the ultrasonic enhanced separation of 257
algae and vegetable oil from the continuous medium (Bosma et al., 2003; Nii et al., 258
2009). 259
Fig. 7 shows that no significant particle size (P>0.05) increase was detected for the 260
coarse emulsion in the 400 KHz single transducer set up. The same was found for raw 261
milk and the fine emulsion for both single transducer geometries at 400 kHz and 1.6 262
MHz. However, the coarse emulsion showed that the number of particle sizes in the 2-263
3 µm range increased when using the 1.6 MHz unit. 264
These observations imply that the primary and secondary acoustic forces drive the 265
particles together so they form floccules or clusters, which redisperse once ultrasonic 266
radiation is turned off. This was verified by remixing the fat phase and finding no 267
increased particle size. Extending the sonication time from 5 min to 30 min (data not 268
12
shown) did not result in further coalescence in all samples for the single transducers at 269
400 kHz and 1.6 MHz. 270
271
3.2. Two-transducer configuration at 400 KHz 272
3.2.1. Creaming behavior 273
Major differences were found when comparing the coarse emulsion after sonication in 274
the one and two-transducer set ups, which were clearly observable with the naked eye. 275
Firstly, there were large clusters forming at the tube walls in the two-transducer 276
resonance chamber (Fig. 8). The clusters formed in the two transducer geometry were 277
visually larger in size than those obtained in the one transducer geometry forming a 278
band. After switching off the ultrasound the clusters destabilized, rising to the top of 279
the tube, and redispersed; but not as quickly as in the case of the single transducer. 280
The destabilized fat clusters obtained upon sonication in the two transducer geometry 281
could still be measured minutes later with the Turbiscan (which was not the case in 282
the single transducer geometry). Where sonication was applied with two transducers, 283
the fat particles rose to the top forming a more distinct cream layer, which means that 284
more particles flocculated or coalesced, and as a consequence more fat was 285
accumulated. This effect can be noticed in Fig. 9 for the coarse emulsion and in Fig. 286
10 for the raw milk. Both the coarse emulsion and raw milk samples showed higher 287
creaming extents in contrast to the fine emulsion where no creaming was observed 288
(see Table 3). 289
In addition, the treatment with the two transducer configuration results in very high 290
concentrations of milk fat in the top (i.e., creamed) layer of raw milk after sonication 291
(Fig. 11d and Fig. 12d). These images illustrate that sonication with 1.6 MHz (Fig. 292
11b and Fig. 12b) drives bigger particles to the top right after ultrasonication, which 293
13
could also be confirmed with particle size analysis (data not shown). It is worth noting 294
that the raw milk was not influenced by the 400 kHz sonication in the single as it was 295
in the two transducer set up. The microscopy images in Fig. 11 and Fig. 12 show 296
excellent agreement with the creaming extents in tables 1 to 3. 297
Although the sound distribution in the tube is not predictable with the two transducer 298
geometry, the sound intensity seems to be higher and the particles are driven together 299
with a bigger force, leading to these characteristic floccules which tend to cluster at 300
the tube walls. The fact that these floccules rose slower than in the single transducer 301
geometry (Fig. 9 and Fig. 10) implies that the fat particles clustered more strongly to 302
the walls due to the higher intensity of the sound field. These higher forces may lead 303
to changes in the particle size, and possibly interfacial membranes, which is addressed 304
in the next section. 305
The recovery rate can be used to demonstrate the ability of the clustered fat droplets to 306
re-disperse after treatment in the two transducer set up at 400 kHz. To account for the 307
peaks appearing in the Turbiscan readings as a result of the large floccules distributed 308
along the tube walls in the two transducer set up (Fig. 9 and Fig. 10), the recovery rate 309
(Eq. 4) is calculated by using the integral of the curve obtained after 10 min in the 310
denominator. Instead of using the reading for the control sample, the 10 min reading 311
is used as a baseline for comparison between recovery rates. The single peak 10 min 312
reading mainly occurs when all floccules have risen to the top of the tube and 313
therefore allows the evaluation of the extent of re-dispersion of large floccules in 314
relative to this point in time. 315
Thus, all recovery rates measured through time in the tube show values higher than 316
one as a result of the additional peaks encountered by the Turbiscan optic sensor. 317
While the formed clusters disassembled into floccules or particulates rising towards 318
14
the top cream layer the recovery rates decrease with time (Fig. 13), which indicated 319
that some globules forming the clusters partially re-dispersed. 320
321
3.2.2. Particle size changes 322
A comparison of particle sizes of the unsonicated milk (control) with milk sonicated 323
in the two transducer set up shows no significant increase in particle sizes in raw milk 324
and in the fine emulsion. For the coarse emulsion, however, a significant increase can 325
be observed for particles in the size range greater than 6 µm (P>0.05), which is 326
demonstrated by Fig. 7. This means that the ultrasonic forces in the coarse emulsion 327
were enough as to promote fat destabilization probably by means of particle 328
coalescence in addition to flocculation. 329
On the other hand, even though flocculation was achieved, the raw milk matrix has 330
shown to require greater forces to achieve particle coalescence. This may be due to 331
differences in the interfacial properties of the different fat globules and the presence 332
of larger sized droplets in the 9 to 15 µm range in the coarse emulsion, which were 333
not present in milk. In addition, in fresh raw milk, the fat droplet is surrounded by the 334
natural milk fat globule membrane (MFGM), which is rich in phospholipids, whereas 335
in recombined emulsions, proteins form the interfacial layer. 336
Only a few literature references deal with the phenomenon of coalescence in standing 337
ultrasound fields, as solid particles are often used as dispersed phase, which usually 338
do not exhibit coalescence. In particular some work has shown instantaneous 339
coalescence of particles in vegetable oil emulsions in standing ultrasonic wave fields 340
(Pangu and Feke, 2007; Pangu and Feke, 2009). 341
342
15
4. Conclusion 343
This contribution shows for the first time that the application of ultrasound at 344
frequencies lower than 1 MHz leads to enhanced creaming in a coarse recombined 345
milk emulsion formulated with skim milk powder and milk fat and in raw milk. This 346
was supported by microscopy imaging as well as Turbiscan readings, which showed 347
higher particle concentration in the top cream layer as a result of ultrasound. Particle 348
flocculation and clustering was detected in both the coarse emulsion and raw milk, 349
where in the coarse emulsion particle coalescence was observed in the two transducer 350
system at 400 kHz. However, no coalescence was observed in milk in the two 351
transducer set up. These findings suggest that a non standing wave system with two 352
transducers can enhance particle coalescence. 353
Separation of milk fat in dairy streams can benefit from the increased coalescence 354
phenomena by pre-treating milk with high frequency ultrasound. Further research is 355
required to determine the kinetics of flocculation, coalescence and creaming of milk 356
fat globules in high frequency ultrasonic systems. Parameters from kinetic models 357
will enable the design of ultrasonic based unit operations applicable to the dairy 358
industry. 359
360
Acknowledgements 361
The authors gratefully acknowledge the support by the Erlangen Graduate School in 362
Advanced Optical Technologies (SAOT) by the German National Science Foundation 363
(DFG) in the framework of the excellence initiative. The authors would also like to 364
acknowledge Dr. Kai Knoerzer and Ms. Elodie Cotte for their contribution in this 365
project. 366
367
16
References 368
Apfel, R.E. 1988. Acoustically induced square law forces and some speculations 369 about gravitation. American Journal of Physics 56:726-29. 370
Ashokkumar, M., Bhaskaracharya, R., Kentish, S., Lee, J., Palmer, M., and Zisu, B. 371 2010. The ultrasonic processing of dairy products - An overview. Dairy 372 Sci.Technol. 90:147-68. 373
Bosma, R., van Spronsen, W.A., Tramper, J., and Wijffels, R.H. 2003. Ultrasound, a 374 new separation technique to harvest microalgae. Journal of Applied Phycology 375 15:143-53. 376
Groeschl, M. 1998a. Ultrasonic Separation of Suspended Particles - Part I: 377 Fundamentals. Acta Acustica united with Acustica 84:432-47. 378
Groeschl, M. 1998b. Ultrasonic Separation of Suspended Particles - Part II: Design 379 and Operation of Separation Devices. Acta Acustica united with Acustica 380 84:632-42. 381
Hawkes, J.J. and Coakley, W.T. 2001. Force field particle filter, combining ultrasound 382 standing waves and laminar flow. Sensors and Actuators B-Chemical 75:213-383 22. 384
Johnson, D.A. and Feke, D.L. 1995. Methodology for fractionating suspended 385 particles using ultrasonic standing wave and divided flow fields. Separations 386 Technology 5:251-58. 387
Kapishnikov, S., Kantsler, V., and Steinberg V. 2006. Continuous particle size 388 separation and size sorting using ultrasound in a microchannel. Journal of 389 Statistical Mechanics: Theory and Experiment 2-15. 390
Mandralis, Z.I. and Feke, D.L. 1993. Continuous suspension fractionation using 391 acoustic and divided-flow fields. Chemical Engineering Science 48:3897-905. 392
Miles, C.A., Morley, M.J., Hudson, W.R., and Mackey, B.M. 1995. Principles of 393 separating micro-organisms from suspensions using ultrasound. Journal of 394 Applied Bacteriology 78:47-54. 395
Nii, S., Kikumoto, S., and Tokuyama, H. 2009. Quantitative approach to ultrasonic 396 emulsion separation. Ultrasonics Sonochemistry 16:145-49. 397
Pangu, G.D. and Feke, D.L. 2004. Acoustically aided separation of oil droplets from 398 aqueous emulsions. Chemical Engineering Science 59:3183-93. 399
Pangu, G.D. and Feke, D.L. 2007. Droplet transport and coalescence kinetics in 400 emulsions subjected to acoustic fields. Ultrasonics 46:289-302. 401
Pangu, G.D. and Feke, D.L. 2009. Kinetics of ultrasonically induced coalescence 402 within oil/water emulsions: Modeling and experimental studies. Chemical 403 Engineering Science 64:1445-54. 404
17
Petersson, F., Nilsson, A., Holm, C., Jonsson, H., and Laurell, T. 2005. Continuous 405 separation of lipid particles from erythrocytes by means of laminar flow and 406 acoustic standing wave forces. Lab on A Chip 5:20-22. 407
408
Tolt, T.L. and Feke, D.L. 1993. Separation of dispersed phases from liquids in 409 acoustically driven chambers. Chemical Engineering Science 48:527-40. 410
Whitworth, G., Grundy, M.A., and Coakley, W.T. 1991. Transport and Harvesting of 411 Suspended Particles Using Modulated Ultrasound. Ultrasonics 29:439-44. 412
Yosioka, K. and Kawasima, Y. 1955. Acoustic Radiation Pressure On A 413 Compressible Sphere. Acustica 5:167-73. 414
415 416
417
18
Figures 418 419
420 Fig. 1. Experimental set up for the one transducer geometry (400 kHz and 1.6 MHz) 421 422 423
424
Fig. 2. Experimental set up for the two transducer geometry. In this set up sound is 425 also transferred into the fluid through the tube walls (only 400 kHz) 426 427
19
428
429
Fig. 3. Example for the computation of the creaming extent. This figure shows the 430
backscattering of the coarse emulsion control (no ultrasonic treatment) in comparison 431
with the sonicated sample (measurement 1 min after switching of ultrasound). The 432
total area under the graphs is equal for all times 433
434
20
435
436
0
10
20
30
40
50
60
70
80
90
100
0-1 1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-20 20-30
particle size [µm]
cum
ulat
ive
num
ber
dist
ribut
ion
[%]
raw milkcoarse emulsion fine emulsion
437
Fig. 4. Cumulative number distribution for the raw milk, coarse emulsion and fine 438 emulsion. The coarse emulsion shows more similarities regarding the particle sizes 439 and can therefore be used as a milk model, with the advantage of using dye to 440 visualize the particle concentration 441
442
21
443
444
Fig. 5. a) Dyed coarse emulsion under sonication (400 kHz) in an open vessel (no 445
reflector). The distinct bands can clearly be observed. b) After switching of the 446
ultrasound field the formed floccules start to rise immediately. The formation of the 447
streaks leads to the conclusion that the floccules comprise of single particles 448
22
449
Fig. 6. Example of the Turbiscan reading for the control samples. Here the results for 450
the coarse emulsion are used, as it represents the most unstable of all three dispersions. 451
In neither of the samples any natural creaming could be observed. Only the range 452
between 8 mm and 63 mm where the backscattering shows the constant value of 53 % 453
can be considered to be inside the tube. The region below 8 mm marks the metal base 454
and the strong decay of backscattering above 63 mm the beginning of the free surface 455
of the sample. This holds true for all following Turbiscan readings 456
457
23
458
Fig. 7. Particle size distribution after the sonication of the coarse emulsion at 400 kHz 459
(one and two transducer geometry) and 1.6 MHz at 35°C and gentle remixing of the 460
cream and the “skimmed” phase. Different letters represent significant differences per 461
treatment (P<0.05) 462
463
464
24
465
466
Fig. 8. Coarse emulsion after sonication in the sandwich two transducer geometry. 467
Large particle clusters were formed at the tube wall showing a greater lag time for re-468
dispersion than bands observed in a single transducer set up. 469
470
25
471
Fig. 9. Turbiscan reading after the sonication of the coarse emulsion in the one and 472
two transducer geometry. For the two transducers the formed clusters persist longer 473
than for the single transducer 474
475
26
476
Fig. 10. Turbiscan reading after the sonication of raw milk in the one and two 477
transducer geometry. Like in the coarse emulsion the particle clusters persist longer 478
than for the single transducer 479
480
27
481
Fig. 11. Microscopy image of the top layer 10 min after sonication for the different 482
frequencies for the coarse emulsion. The particle concentration correlates very well 483
with the computed creaming extents in the tables 1 to 3 484
485
28
486
Fig. 12. Microscopy image of the top layer 10 min after sonication for the different 487
frequencies for the raw milk. The particle concentration correlates very well with the 488
computed creaming extents in the tables 1 to 3 489
490
29
491
Fig. 13. Example for the recovery rate as a function of time for raw milk in the two 492
transducer geometry. The recovery rate is greater than one, showing the higher 493
integral backscattering compared to the steady state 10 min after sonication. The 494
decrease can be assigned to reversible re-dispersion of particles out of the floccules. 495
496
30
Tables 497
498
Table 1 499 Creaming extents (CE) and recovery rates (RR) for the experiments with the single 500 transducer geometry at a frequency of 400 kHz for the coarse emulsion (coarse), fine 501 emulsion (fine) and raw milk 502 Time
[min]
CE
coarse
RR
coarse
CE
fine
RR
fine
CE
raw milk
RR
raw milk
0* 0.086 99% 0 100% 0 100%
1 0.100 99% 0 100% 0 100%
2 0.108 99% 0 100% 0 100%
5 0.104 99% 0 100% 0 100%
10 0.104 99% 0 100% 0.0185 100%
*1 min after sonication 503
504 Table 2 505 Creaming extents (CE) and recovery rates (RR) for the experiments with the single 506 transducer geometry at 1.6 MHz for the coarse emulsion (coarse), fine emulsion (fine) 507 and raw milk 508
Time
[min]
CE
coarse
RR
coarse
CE
fine
RR
fine
CE
raw milk
RR
raw milk
0* 0.265 99% 0 100% 0.116 99%
3 0.236 98% 0 100% 0.119 99%
5 0.237 98% 0 100% 0.119 99%
7 0.239 98% 0 100% 0.119 99%
10 0.238 99% 0 100% 0.119 99%
*1 min after sonication 509
510 Table 3 511 Comparison between the creaming extents (CE) and recovery rates (RR) for the 512 experiments with the one and two transducer geometry at 400 kHz for the coarse 513 emulsion (coarse), fine emulsion (fine) and raw milk (10 min after sonication, as only 514 then all particle clusters had risen to the top) 515
CE
coarse
RR
coarse
CE
fine
RR
fine
CE
raw milk
RR
raw milk
one transducer
0.104 99% 0 100% 0.0185 100%
two transducer
0.551 91% 0 100% 0.643 85%
516