close

Вход

Забыли?

вход по аккаунту

?

Effects of operating conditions on particle size in sonocrystallization.

код для вставкиСкачать
ASIA-PACIFIC JOURNAL OF CHEMICAL ENGINEERING
Asia-Pac. J. Chem. Eng. 2010; 5: 599–608
Published online 5 May 2010 in Wiley InterScience
(www.interscience.wiley.com) DOI:10.1002/apj.437
Special Theme Research Article
Effects of operating conditions on particle size
in sonocrystallization
Hussein Oubani,1 Ali Abbas,1 * Mourtada Srour1 and Jose A. Romagnoli2
1
2
School of Chemical and Biomolecular Engineering, University of Sydney, Sydney, New South Wales 2006, Australia
Chemical Engineering Department, Louisiana State University, Baton Rouge, LA 70803, USA
Received 2 November 2009; Revised 22 February 2010; Accepted 26 February 2010
ABSTRACT: This work presents systematic investigations on sonocrystallization to elucidate the effects of key variables
on sonocrystallization product properties. A novel continuous flow sonocrystallization apparatus was used to prepare
NaCl microparticles from a NaCl–ethanol–water antisolvent system. By implementing a full factorial experimental
design, we investigated the effects of ultrasonic power (75–225 W), antisolvent feed rate (0.5–6.5 l/h), system flow rate
(2.8–4.1 l/min) and sonication time (5–30 min) on product crystal size. Data from these experiments were regressed to
develop an empirical model that was found to be in agreement with experiments. The model identified the interaction
between sonication power and system flow to be rather significant. Model simulations found that particle size decreases
when antisolvent feed rate or ultrasonic power increases. This was found to be in contrast to increasing the system flow
which resulted in larger particle sizes. The regression model was subsequently used to determine optimal operating
conditions that minimize mean size, as smaller sizes are commonly required for pharmaceuticals such as for inhalation
particles. These optimal values were found to be as follows: antisolvent flow rate = 6.5 l/h, power ultrasound =
225 W, system flow = 2.8 l/min and sampling time = 15 min. The optimal mean size predicted at these conditions
was 28.6 ± 5.7 µm which is very close to the observed value of 27.6 µm. A high-speed camera was used to visualize
the ultrasonic irradiation in the sonoreactor and was crucial in explaining the significant interactive effect of sonication
power and system flow on crystal size.  2010 Curtin University of Technology and John Wiley & Sons, Ltd.
KEYWORDS: sonocrystallization; particle size; modeling; optimization; image analysis
INTRODUCTION
Ultrasound has been used in crystallization during the
last 70 years and has shown to have significant potential for modifying crystal product properties and yields.
Previous studies on sonocrystallization focused either
on the mechanism of this process or on the control of
different parameters involved to improve crystal properties. It has been generally observed that continuous
sonication leads to a reduction in crystal size and thus
generally accepted that high-intensity ultrasound (HIU)
promotes nucleation and inhibits crystal growth. Moreover, it provides better control over shape and size of
crystals which can be tailored through the manipulation
of the sonication time and supersaturation level.[1,2]
HIU was found to decrease the induction time,
defined as the time elapsed between the creation of
supersaturation and the appearance of crystals.[3 – 8]
Therefore, with HIU, nucleation can be induced at
*Correspondence to: Ali Abbas, School of Chemical and Biomolecular Engineering, University of Sydney, NSW 2006, Australia.
E-mail: ali.abbas@sydney.edu.au
 2010 Curtin University of Technology and John Wiley & Sons, Ltd.
lower levels of supersaturation and can reduce the
metastable zone width.[1,3] The use of ultrasound produces crystals with improved habit, allows control
over both size and size distribution of crystals through
adjusting ultrasound-related variables[7 – 10] and reduces
agglomeration.[7,9] Sonication is an effective method
that mixes reactants more uniformly,[7] increases the
diffusion coefficient which is the main reason for the
reduced induction time,[4,5] reduces the crystallization
time[10,11] as well as the amount of antisolvent required
in antisolvent crystallization and increases the final
yield.[10] The preparation of sodium chloride particles for inhalation via HIU was reported on two different occasions[12,13] where the manipulation of the
ultrasound power and working parameters (temperature, solution’s concentration, mixing rate and sonication time) were used to identify optimal conditions to
obtain particle sizes suitable for inhalation drug delivery. While previous works studied the batch sonocrystallization, fed-batch sonocrystallization involving controlled feeding of antisolvent into the system has not
600
H. OUBANI ET AL.
been studied. Adjunct to antisolvent feeding, continuous flow sonication, where the flow of the process is
passed through a sonication cell, has several attractive
features for control of crystal size. Although many studies investigated the effects of working parameters either
on the crystallization process or on the end product qualities, none has studied the effect of these parameters
on particle size in a systematic way. Thus, this study
sets out to investigate, via a novel experimental setup
and using surface response modeling approach, the main
effects and interactions of antisolvent feed rate, ultrasound power and the flow of system on the product
crystal size.
EXPERIMENTAL PROCEDURE
Materials
The salt used in our experiments is sodium chloride
powder (Astral Scientific, Sydney NSW Australia) and
the antisolvent used is high-grade ethanol (CSR Distilleries, Melbourne Victoria Australia). Isopropanol (CSR
Distilleries) was used as a dispersant solvent in the particle size analysis.
Experimental setup
Figure 1 shows a schematic diagram of the novel experimental setup used. The ultrasonic equipment consists of
a probe and a sonifier (model UP400S; Hielscher Ultrasound Technology, Germany). This ultrasonic transducer uses electric excitation to generate ultrasound
Asia-Pacific Journal of Chemical Engineering
which is transferred to the liquid medium via various
sonotrodes. The probe used is a flow horn that has a tip
diameter of 22 mm which was immersed in a 250-ml
standard jacketed flow cell (sonoreactor). The sonicator operates at a fixed frequency of 24 kHz and can
deliver a maximum power of 300 W with the sonotrode
used[14] . The sonifier is connected to a PC, enabling the
pre-adjustment of important control parameters such as
amplitude, pulse length and sonication time. Two pumps
were calibrated and used in these experiments: one to
adjust the flow of system (Grundfos 624U, Denmark)
with a maximum flow of 5.5 l/min and the other having
a maximum flow of 7.5 l/h to deliver the ethanol into
the flow (Grundfos DME). The starting temperature was
controlled at 10 ◦ C for all the experiments using a heating–cooling circulator (Julabo, Mettler-Toledo, USA).
This temperature control system allows the temperature
in the sonoreactor and the antisolvent feed vessel to be
monitored and regulated at the desired setpoint.
To conduct an experiment, sodium chloride solution
(30g NaCl/100 g water; volume 600 ml) was prepared
and charged to the jacketed reactor (2 l). Pump F was
turned on and adjusted to the desired system flow.
When the desired temperature (10 ◦ C) was reached,
the sonication process and delivery of antisolvent into
the flow were initiated simultaneously. Samples were
taken at 5-min increments directly from the flow.
Offline samples were vacuum filtered to remove the
solvents, and particle size was then measured using
laser light scattering technique (MastersizerS, Malvern
Instruments, UK). Small amounts of the filtered salt
were suspended in isopropanol in the MastersizerS
measurement cell. For each sample, three particle size
measurements were taken and average volume mean
(d43 ) values were reported along with volume-based
particle size distributions.
EMPIRICAL MODEL
Figure 1. Schematic diagram of the experimental setup: A,
sonoreactor; B, ultrasound generator; C, sonication probe;
D, process fluid jacketed vessel; E, antisolvent jacketed
vessel; F, process fluid pump; G, antisolvent pump; H,
heating/cooling circulator; I, temperature probe; J, computer
to control the ultrasound generator.
 2010 Curtin University of Technology and John Wiley & Sons, Ltd.
Four working variables (factors) were considered: feed
rate of antisolvent (R), ultrasonic power (P ), system
flow (F ) and the sampling time (t). These variables
are independent and are controlled in the experiments.
Modde package (Umetrics, Umea, Sweden) was used
to perform a full factorial experimental design (three
levels for each of the factors R, P and F ) leading
to 27 experiments. Throughout each experiment, three
samples were taken at different times, and thus 81
particle size measurements were obtained in total.
Table 1 summarizes the different operational conditions
along with sample times, the measured and predicted
crystal sizes.
In this response surface modeling, a partial least
square (PLS) model was fitted to the data aiming to
identify the relationship between the crystal size (L)
Asia-Pac. J. Chem. Eng. 2010; 5: 599–608
DOI: 10.1002/apj
Asia-Pacific Journal of Chemical Engineering
EFFECTS OF OPERATING CONDITIONS ON PARTICLE SIZE
Table 1. Experimental conditions along with observed and predicted crystal sizes.
Sample
ID
N1
N2
N3
N4
N5
N6
N7
N8
N9
N10
N11
N12
N13
N14
N15
N16
N17
N18
N19
N20
N21
N22
N23
N24
N25
N26
N27
N28
N29
N30
N31
N32
N33
N34
N35
N36
N37
N38
N39
N40
N41
N42
N43
N44
N45
N46
N47
N48
N49
N50
N51
N52
N53
N54
N55
N56
N57
N58
N59
Antisolvent flow
rate R (l/h)
System flow
F (l/min)
Nominal power
P (W)
Sampling time
t (min)
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
2.8
2.8
2.8
3.3
3.3
3.3
4.1
4.1
4.1
2.8
2.8
2.8
3.3
3.3
3.3
4.1
4.1
4.1
2.8
2.8
2.8
3.3
3.3
3.3
4.1
4.1
4.1
2.8
2.8
2.8
3.3
3.3
3.3
4.1
4.1
4.1
2.8
2.8
2.8
3.3
3.3
3.3
4.1
4.1
4.1
2.8
2.8
2.8
3.3
3.3
3.3
4.1
4.1
4.1
2.8
2.8
2.8
3.3
3.3
75
75
75
75
75
75
75
75
75
150
150
150
150
150
150
150
150
150
225
225
225
225
225
225
225
225
225
75
75
75
75
75
75
75
75
75
150
150
150
150
150
150
150
150
150
225
225
225
225
225
225
225
225
225
75
75
75
75
75
25
5
5
25
5
5
25
5
5
25
5
5
25
5
5
25
5
5
25
5
5
25
5
5
20
5
5
30
10
10
30
10
10
30
10
10
30
10
10
30
10
10
30
10
10
30
10
10
30
10
10
25
10
10
30
15
15
30
15
 2010 Curtin University of Technology and John Wiley & Sons, Ltd.
Observed size
L (µm)
69.9
69.0
63.2
62.5
63.6
58.9
54.2
61.4
61.5
49.1
35.8
56.0
55.7
60.5
56.7
59.8
59.0
56.3
61.1
33.1
35.9
59.7
55.2
48.3
57.6
51.8
60.8
79.2
70.2
53.6
67.4
29.5a
48.4
54.0
39.8a
62.6
64.9
31.4a
47.4
70.0
75.6a
27.1a
63.0
67.5
35.1a
73.8a
25.3
19.4
45.3
50.9
40.0
59.3
60.7
48.9
55.4a,b
70.8
54.5
55.4b
38.4a
Predicted size
τ (µm)
Confidence
interval (±)
73.6
66.5
57.7
66.8
63.7
57.6
60.0
60.8
57.6
60.4
53.3
44.5
60.1
57.0
50.9
59.8
60.6
57.3
47.2
40.1
31.3
53.3
50.2
44.2
60.8
60.3
57.1
72.2
65.2
56.3
65.5
62.3
56.3
58.7
59.5
56.3
59.0
52.0
43.1
58.7
55.6
49.6
58.4
59.2
56.0
45.9
38.8
29.9
52.0
48.9
42.8
59.5
59.0
55.7
72.2
65.2
56.3
65.5
62.3
5.6
5.1
5.2
3.6
4.2
3.4
5.5
5.7
5.6
4.6
4.1
4.0
2.8
3.6
2.7
4.2
4.2
4.2
5.6
5.4
5.2
3.4
4.1
3.3
5.5
4.9
5.1
5.7
4.4
5.4
3.6
3.2
3.6
5.5
4.9
5.5
4.6
3.1
4.1
2.9
2.3
2.8
4.3
3.2
4.3
5.6
4.6
5.2
3.5
3.1
3.5
5.2
4.3
5.3
5.7
4.4
5.4
3.6
3.2
Asia-Pac. J. Chem. Eng. 2010; 5: 599–608
DOI: 10.1002/apj
601
602
H. OUBANI ET AL.
Asia-Pacific Journal of Chemical Engineering
Table 1. (Continued).
Sample
ID
N60
N61
N62
N63
N64
N65
N66
N67
N68
N69
N70
N71
N72
N73
N74
N75
N76
N77
N78
N79
N80
N81
Antisolvent flow
rate R (l/h)
System flow
F (l/min)
Nominal power
P (W)
Sampling time
t (min)
Observed size
L (µm)
Predicted size
τ (µm)
Confidence
interval (±)
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
0.5
4
6.5
3.3
4.1
4.1
4.1
2.8
2.8
2.8
3.3
3.3
3.3
4.1
4.1
4.1
2.8
2.8
2.8
3.3
3.3
3.3
4.1
4.1
4.1
75
75
75
75
150
150
150
150
150
150
150
150
150
225
225
225
225
225
225
225
225
225
15
35
15
15
35
15
15
35
15
15
35
15
15
35
15
15
35
15
15
30
15
15
43.7
55.4b
38.0a
55.0
58.7
41.4
50.4
55.4b
69.1
32.7
55.4b
64.9
54.1
45.9
45.3
27.6
50.4
55.5
17.8a
63.2
52.2
51.3
56.3
57.4
58.2
54.9
57.7
50.7
41.8
57.4
54.3
48.2
57.1
57.9
54.7
44.5
37.5
28.6
50.7
47.5
41.5
58.2
57.6
54.4
3.6
5.8
4.5
5.9
5.1
2.7
4.8
3.7
1.7
3.7
4.9
2.8
4.9
6.0
4.2
5.7
4.2
2.7
4.3
5.4
4.1
5.9
a
Outliers
b
excluded from the model.
Missing values calculated by the model.
and the working variables. The resulting PLS model
after eliminating insignificant terms is as follows:
Model
term
L = β0 + β1 R + β2 F + β3 P + β4 t + β5 R × F
+ β6 F × P
Table 2. Calculated model coefficient values.
(1)
The coefficients βi of the regression model (Eqn 1)
are tabulated along with their values in Table 2. Their
p-values that determine the degree of significance of
each term are also listed in Table 2. The smaller the
p-value, the more significant is the corresponding
factor. As a result, the factors were found to be
decreasing in absolute significance in the following
order:
P , F × P , R, F , R × F , t
(2)
The observed vs predicted plot (Fig. 2) displays
the experimental runs within a 10% deviation of the
regression line. The correlation coefficient of the model
was determined to be R 2 = 0.63. The standard deviation
of the regression (30.6) is much larger than the standard
deviation of residuals (7.2) (Table 3). The F -value of
the regression (18.2) is much higher than the upper
critical value (2.3) of the F -distribution (F5,63 ) at a 5%
significance level.
 2010 Curtin University of Technology and John Wiley & Sons, Ltd.
Constant
R
F
P
t
R×F
F ×P
Coefficient
Coefficient
value
β0
β1
β2
β3
β4
β5
β6
54.6
−6.2
2.9
−5.6
−2.8
2.4
4.4
p-Value
Confidence
interval
(±)
0
0.000295
0.001478
2.08 × 10−8
0.090427
0.00905
4.26 × 10−6
1.7
3.2
1.7
1.7
3.2
1.8
1.8
R × F and F × P represent the interaction effects of the main
system flow with the antisolvent flow rate and the system flow with
ultrasonic power, respectively.
OPTIMIZATION
The regression model was subsequently used to determine the optimal values of different factors that minimize the crystal mean size. These were as follows:
antisolvent flow rate of 6.5 l/h, power ultrasound of
225 W, system flow of 2.8 l/min and sampling time
after 15 min. The optimal mean size predicted under
these optimal conditions is 28.6 ± 5.7 µm which is very
close to the observed value (27.6 µm). The prediction
plots shown in Fig. 3 display the predicted crystal mean
size against each of the factors while the other factors
are kept at their optimal values. These plots show that
Asia-Pac. J. Chem. Eng. 2010; 5: 599–608
DOI: 10.1002/apj
Asia-Pacific Journal of Chemical Engineering
EFFECTS OF OPERATING CONDITIONS ON PARTICLE SIZE
the mean size (Fig. 4). This corroborated with results
reported by others.[17,18] Two phenomena can explain
these effects: the first one is the intensified micromixing caused by increasing the ultrasonic power which
enhances the mass transfer between solvent and antisolvent and facilitates supersaturation generation, consequently resulting in smaller particle sizes and narrower CSDs. The second one is that nucleation is
promoted over growth at higher ultrasonic power[19]
which leads to birth of more nuclei and hence smaller
particles.
Effect of antisolvent flow rate on crystal size
and CSD
Figure 2. Observed vs predicted plot.
Table 3. ANOVA table.
DF
Total corrected
Regression
Residual
SS
69 8853
6 5609
63 3244
MS
Fp(variance) value Value
128
935
52
18.2
0
SD
11.3
30.6
7.2
DF, degree of freedom; SS, sum of square; MS, mean sum of square;
SD, standard deviation.
the antisolvent flow rate, the sonication time and the
power ultrasound are negatively correlated to the mean
size in contrast to the flow of system which is positively
correlated to the mean size.
MAIN EFFECTS AND INTERACTIONS
BETWEEN FACTORS
Investigations herein were focused on the effects
of intensity of power ultrasound (P ), rate flow of
antisolvent (R) and the system flow (F ) on the crystal size and crystal size distribution (CSD). The sonoprobe used in our study was immersed at the same
standard depth in the flow cell in all experiments,
being kept as a constant variable to eliminate the
influence of the immersion depth on the dissipation
of sonic waves as well as the flow pattern (mixing)
and the crystallization rate.[15,16] The sonicated volume
was also kept constant so that the influence of sonicated volume on the crystal size and CSD was also
eliminated.[9] We next discuss the effects of each of
these variables.
Effect of ultrasonic power on crystal size
and CSD
Our results showed that increasing the ultrasound
power resulted in narrowing the CSD and decreasing
 2010 Curtin University of Technology and John Wiley & Sons, Ltd.
In this study, the flow rate of antisolvent was controlled using a high precision pump and the antisolvent was delivered to the reactor for the whole duration of the experiment. Three flow rates of antisolvent were used: 0.5, 4 and 6.5 l/h. The volume of
water used to prepare the initial sodium chloride was
600 ml. The experimental results revealed that increasing the antisolvent feed rate resulted in smaller mean
sizes and narrower CSDs (Fig. 5). These results are
attributed to the enhancement, by the HIU, to the
mass transfer between solvent and antisolvent that lead
to high instantaneous supersaturation levels.[20] Such
supersaturation levels increase under higher feed rates
of antisolvent, in turn resulting in stronger nucleation,
and hence narrower CSDs and a smaller crystal size,
a finding similar to that reported by Holmback and
Rasmuson[21] and Doki et al .[22] Moreover, the continuous addition of antisolvent in the fed-batch crystallization kept the solution in the supersaturation zone
as the concentration of solute drops rapidly after primary nucleation, and thus continuous antisolvent addition provides the force for nucleation to keep occurring
which also leads to smaller crystal sizes and narrower
CSDs.
The antisolvent/solvent rate is a crucial factor in
determining the supersaturation rate and hence the crystal size. Smaller crystal sizes are expected at higher
antisolvent/solvent ratio. Table 4 shows a comparison between this study and previous studies involving the sonocrystallization of sodium chloride from a
water–ethanol system; it can be seen that the antisolvent/solvent ratio herein is equal to 3 in the highest
case. This is significantly less than ratios used in other
studies. Although in both cited studies[12,13] the minimum sizes achieved were smaller than the minimum
size achieved in this study, the differences in the antisolvent/solvent volume ratio are stark and represent
significant economic and operational advantages on the
part of this study. This suggests that sonocrystallization
system presented in this study is more efficient in terms
of antisolvent use and is more attractive for scale-up
Asia-Pac. J. Chem. Eng. 2010; 5: 599–608
DOI: 10.1002/apj
603
604
H. OUBANI ET AL.
Asia-Pacific Journal of Chemical Engineering
Figure 3. Prediction plots displaying the mean size against each factor while the others are
kept at their optimal values (, mean size; ž, lower confidential interval; , upper confidential
interval). The system flow values are 80, 100 and 120 which correspond, respectively, to 2.8, 3.3
and 4.1 l/min. The amplitude values are 25, 50 and 75% which correspond, respectively, to 75,
150 and 225 W nominal power.
Figure 4. Effect of ultrasound power on crystal size (a) and CSD (b). Antisolvent feed rate and system
flow were kept at their calculated optimal values.
 2010 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pac. J. Chem. Eng. 2010; 5: 599–608
DOI: 10.1002/apj
Asia-Pacific Journal of Chemical Engineering
EFFECTS OF OPERATING CONDITIONS ON PARTICLE SIZE
Figure 5. Effect of antisolvent feed rate on crystal size (a) and CSD (b). Power and system flow were kept
at their calculated optimal values.
Table 4.
Antisolvent/solvent volume ratio and
minimum sizes achieved in different studies.
Study
Tang et al .[12]
Abbas et al .[13]
This study
Volume ratio
(antisolvent/
solvent)
Mode of
antisolvent
addition
Minimum
size achieved
(µm)
50–100
67
0.5–3
Batch
Batch
Semi-batch
2
1
17.8
to commercial scales. This efficiency can be attributed
to the semi-batch addition mode of antisolvent through
the whole time of the crystallization process which
results in providing a driving force for nucleation to
keep occurring, and hence the supersaturation is mainly
consumed by nucleation rather than growth of existing
crystals.
Effect of system flow on crystal size and CSD
As far as the authors are aware, no previous work is
found in the literature that describes either a continuous
flow sonocrystallization system or the effect of the
system flow on the crystallization process or on end
product properties. In this study, it was found that
decreasing the system flow results in decreasing the
crystal size and narrowing the CSD (Fig. 6). This
effect can be due to coupling mixing with ultrasound,
observed elsewhere where it was found that under
higher stirring rates ultrasonic activity in the liquid
medium becomes hindered[12] In this study, this statement may be applicable to the macromixing effect produced by the system flow. In addition, the interactive
effect of the system flow with the power ultrasound
was found to be the second most significant factor
that influences the mean size according to the model
Figure 6. Effect of system flow on crystal size (a) and CSD (b). Power and antisolvent feed rate were kept
at their calculated optimal values.
 2010 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pac. J. Chem. Eng. 2010; 5: 599–608
DOI: 10.1002/apj
605
606
H. OUBANI ET AL.
The acoustic pressure field: (left) simulated
(Raman and Abbas, 2006) and (right) obtained by highspeed camera. This figure is available in colour online at
www.apjChemEng.com.
Figure 7.
presented above. To examine this effect further, a highspeed camera (FASTCAM-1024PCI, Photron, Japan)
was used to record the cavitation bubbles’ motion at
fixed power and different system flows. Initial visual
observations showed that previous modeling of ultrasonic pressure field by Raman and Abbas (2006)[23]
corroborated with the observed cavitation cloud zones
as compared in Fig. 7. Observations also showed that
higher system flows resulted in shorter and less intense
bubble clouds, which can be ascribed to the pressure
exerted by the flow in the opposite direction of the
ultrasonic pressure. The higher this pressure is, the
shorter is the bubble cloud and the lesser is the ultrasonic activity observed at the bottom of the vessel.
Moreover, it can be noticed that the bubble cloud is
skewed to the left (direction of the flow) at higher
flows which results in a less homogeneous bubble distribution, less cavitation activity and less intensified
micromixing which explain the high significance of
the interactive effect of power and system flow on
the mean size that was elucidated by the empirical
model.
Asia-Pacific Journal of Chemical Engineering
This result reveals the energy pattern distribution in the
cylindrical flow cell. Moreover, the compression and
rarefaction pressure fields can be distinguished in both
the simulated and the real image.
The energy distribution shown herein is mainly
affected by the applied ultrasonic power and the flow of
system in a constant geometry sonoreactor. A matrix of
nine images (Fig. 8) shows the images of bubble clouds
under conditions listed in Table 1. The observations
of Fig. 8 reveal that higher power ultrasound results
in longer bubble cloud, which is also corroborated
by Mandroyan et al .[24] This suggests that the energy
is more distributed throughout the sonoreactor which
leads to a more intense cavitation which is regarded
to be the dominant mechanism of ultrasound in affecting crystallization[1,25,26] . This is also attributed to the
higher levels of micromixing, generating higher supersaturation rates, more nuclei and consequently smaller
crystal sizes which can explain the dominant effect of
the ultrasonic power in deciding the crystal mean size
as per experimental observations and model predictions.
Interactive effect of power and system flow
on crystal size
The propagation of ultrasound into the sonoreactor
depends on the ultrasonic energy and on the geometry
of the sonoreactor. This implies a particular distribution
of ultrasonic energy due to the wave superposition
effect caused by wave’s reflection off the reactor
walls. In this study, a cylindrical sonoreactor and
ultrasonic probe (diameter 22 mm) were used, which
are the same as those used for the simulation of the
acoustic pressure field performed by Raman and Abbas
(2006)[23] and shown in Fig. 7(left). This simulation
was validated experimentally by visual comparison
against the high-speed image shown in Fig. 7(right).
 2010 Curtin University of Technology and John Wiley & Sons, Ltd.
Figure 8. High-speed images showing the bubble clouds
at different powers and system flows. Note that as
the power increases the bubble cloud length increases;
however, increasing the system flow results in decreasing
the bubble cloud length and shifting it to the left.
Asia-Pac. J. Chem. Eng. 2010; 5: 599–608
DOI: 10.1002/apj
Asia-Pacific Journal of Chemical Engineering
EFFECTS OF OPERATING CONDITIONS ON PARTICLE SIZE
explain the high significance of the interactive effect
of power and system flow on the mean size that was
elucidated by the empirical model.
The qualitative discussions around Fig. 8 could conceptually be extended to quantitative image analysis that
focuses on determination of the gray sonication intensity zones and the rest of the liquid. Such a concept
is introduced here via exemplification using three captured images. The image processing package Image J
(National Institute of Health, USA) is used for this
purpose. The sonication beam length and the ratio of
the area of sonication intensity zone (cone area) to
the area of the rest of liquid were measured at different power and fixed system flow while the deviation
angle of the sonication bubble cloud was measured at
different values of system flow and fixed power. The
methods used to measure these three parameters are
shown in Fig. 10 and the results are listed in Tables 5
and 6.
Figure 9. Micromixing effect of ultrasound
visualized using suspended PVC particles.
This figure is available in colour online at
www.apjChemEng.com.
Even though cavitation is believed to be the dominant
mechanism of the power ultrasound, the micromixing of
ultrasound can have a reinforcing effect especially on
mixing solvent and antisolvent molecules; to observe
this effect, a suspension of PVC particles in water was
subjected to ultrasound (Fig. 9). It can be seen that ultrasound is distributing the PVC particles evenly throughout the reactor which results in decreasing the segregation of zones[27] and enhancing the intrinsic mass
transfer.[28] Both these effects increase with increasing ultrasound power, and hence smaller crystal size
and narrower CSD can be expected at higher power
inputs.
The interactive effect of the system flow with the
power ultrasound was found to be the second important
factor that influences the mean size according to the
model discussed above. To examine this effect, bubbles’
motion and distribution were recorded at fixed power
and different system flows. The qualitative observations
of Fig. 8 reveal that higher system flows resulted
in shorter and less intense bubble clouds which can
be ascribed to the pressure exerted by the flow in
the direction opposite to the ultrasonic pressure. The
higher this flow pressure is, the shorter is the bubble
cloud and the lesser is the ultrasonic activity observed
at the bottom of the vessel. Moreover, it can be
noticed that the bubble cloud is shifted to the left
(direction of the flow) at higher flows which results in
a less homogeneous bubble distribution, less cavitation
activity and less intensified micromixing, which when
added to the shorter residence time at higher flows can
 2010 Curtin University of Technology and John Wiley & Sons, Ltd.
Figure 10. Methods used to measure (a) deviation angle,
(b) cone area and (c) beam length.
Table 5. Values of cone area ratio and beam length
under varying ultrasonic power and fixed system flow.
Power (W)
75
150
225
Cone area ratio
Length (pixels)
0.10
0.15
0.16
268
407
454
Table 6. Values of deviation angle of sonication bubble
cloud under varying system flow and fixed ultrasonic
power.
System flow (l/min)
2.8
3.3
4.1
Deviation angle
5.0
12.8
23.2
Asia-Pac. J. Chem. Eng. 2010; 5: 599–608
DOI: 10.1002/apj
607
608
H. OUBANI ET AL.
The results revealed that the beam length and
cone area ratio increase when the ultrasonic power is
increased which confirms the idea of better energy distribution at high ultrasonic powers. Similarly, the deviation angle and the system flow are positively correlated
which means that the higher the system flow is, the
more the bubble cloud is shifted to the left and the
more the system flow is hindering the ultrasonic effect
on the crystal size.
The use of high-speed imaging techniques and
image processing is recommended as more developed
high-speed photography techniques (>1 000 000 FPS)
become available; these techniques could provide
deeper insight into bubble implosion as well as the passage from cavitation to nucleation.
CONCLUSIONS
A novel continuous flow sonocrystallization apparatus
was used to prepare sodium chloride microparticles
from a NaCl–ethanol–water antisolvent system. Data
from experiments were used to develop an empirical model that was subsequently used to identify the
optimal operating conditions. By implementing a full
factorial experimental design, we systematically investigated the effects of ultrasonic power (75–225 W),
antisolvent feed rate (0.5–6.5 l/h) and system flow rate
(2.8–4.1 l/min) on product crystal size. It was found
that when the antisolvent feed rate or the ultrasonic
power increases, the particle size decreases. This is in
contrast to increasing the system flow which results in
larger particle sizes. Smaller particle sizes were prepared at optimal conditions calculated to be 6.5 l/h for
antisolvent flow rate, 225 W for power ultrasound and
2.8 l/min for the main system flow. The optimal mean
size predicted of 28.6 µm was found to be close to the
observed values (27.6 µm) under these optimal operational conditions. The novel process and approach presented in this article will play a significant role in largescale production of microparticles with tailored size
properties, such as inhalation drugs requiring smaller
particle sizes.
 2010 Curtin University of Technology and John Wiley & Sons, Ltd.
Asia-Pacific Journal of Chemical Engineering
REFERENCES
[1] L.J. McCausland, P.W. Cains, P.D. Martin. Chem. Eng. Prog.,
2001; 97(7), 56–61.
[2] L. McCausland, P. Cains. Chem. Ind. (London), 2003; 18,
15–17.
[3] N. Lyczko, F. Espitalier, O. Louisnard, J. Schwartzentruber.
Chem. Eng. J., 2002; 86(3), 233–241.
[4] Z. Guo, M. Zhang, H. Li, J. Wang, E. Kougoulos. J. Cryst.
Growth, 2005; 273(3-4), 555–563.
[5] Z. Guo, A.G. Jones, N. Li. Chem. Eng. Sci., 2006; 61(5),
1617–1626.
[6] Z. Guo, A.G. Jones, N. Li. J. Colloid Interface Sci., 2006;
297(1), 190–198.
[7] H. Li, H. Li, Z. Guo, Y. Liu. Ultrason. Sonochem., 2006;
13(4), 359–363.
[8] C. Virone, H.J.M. Kramer, G.M. van Rosmalen, A.H. Stoop,
T.W. Bakker. J. Cryst. Growth, 2006; 294(1), 9–15.
[9] H. Li, J. Wang, Y. Bao, Z. Guo, M. Zhang. J. Cryst. Growth,
2003; 247(1-2), 192–198.
[10] M.N. Patil, G.M. Gore, A.B. Pandit. Ultrason. Sonochem.,
2008; 15(3), 177–187.
[11] D. Qian, J.Z. Jiang, P.L. Hansen. Chem. Commun., 2003; (9),
1078–1079.
[12] P. Tang, H.K. Chan, E. Tam, N. de Gruyter, J. Chan. Ind. Eng.
Chem. Res., 2006; 45(12), 4188–4192.
[13] A. Abbas, M. Srour, P. Tang, H. Chiou, H.-K. Chan, J.A.
Romagnoli. Chem. Eng. Sci., 2007; 62(9), 2445–2453.
[14] Hielshcer Ultrasonics, 2006; UP400 lab manual.
[15] R.F. Contamine, A.M. Wilhelm, J. Berlan, H. Delmas. Ultrason. Sonochem., 1995; 2(1), S43–S47.
[16] I. Nishida. Ultrason. Sonochem., 2004; 11(6), 423–428.
[17] N. Amara, B. Ratsimba, A.M. Wilhelm, H. Delmas. Ultrason.
Sonochem., 2001; 8(3), 265–270.
[18] L.Y. Cao, C.B. Zhang, J.F. Huang. Mater. Lett., 2005; 59(1415), 1902–1906.
[19] E. Miyasaka, Y. Kato, M. Hagisawa, I. Hirasawa. J. Cryst.
Growth, 2006; 289(1), 324–330.
[20] R.S. Dhumal, S.V. Biradar, A.R. Paradkar, P. York. Pharm.
Res., 2008; 25(12), 2835–2844.
[21] X. Holmback, A.C. Rasmuson. J. Cryst. Growth, 1999; 199,
780–788.
[22] N. Doki, N. Kubota, M. Yokota, S. Kimura, S. Sasaki.
J. Chem. Eng. Jpn., 2002; 35(11), 1099–1104.
[23] V. Raman, A. Abbas, S.C. Joshi. COSMOL Users Conference,
2006; Bangalore, India, 17 Nov.
[24] R.V.A. Mandroyan, Y. Bailly, M.-L. Doche, J.-Y. Hihn.
Ultrason. Sonochem., 2009; 16, 88–96.
[25] G. Ruecroft, D. Hipkiss, T. Ly, N. Maxted, P.W. Cains. Org.
Process Res. Dev., 2005; 9(6), 923–932.
[26] K. Wohlgemuth, A. Kordylla, F. Ruether, G. Schembecker.
Chem. Eng. Sci., 2009; 64(19), 4155–4163.
[27] S.F.K. Okadap, Y. Nishimura, M. Matsubara. Chem. Eng.
Sci., 1971; 27, 529–535.
[28] L.C. Hagenson, L.K. Doraiswamy. Chem. Eng. Sci., 1998;
53(1), 131–148.
Asia-Pac. J. Chem. Eng. 2010; 5: 599–608
DOI: 10.1002/apj
Документ
Категория
Без категории
Просмотров
1
Размер файла
354 Кб
Теги
effect, sonocrystallization, operating, size, conditions, particles
1/--страниц
Пожаловаться на содержимое документа