Powder Technology 206 (2011) 34–43

Contents lists available at ScienceDirect

Powder Technology
j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / p ow t e c

Wet granulation in laboratory-scale high shear mixers: Effect of chopper presence,

design and impeller speed
T.M. Chitu a,⁎, D. Oulahna a, M. Hemati b
a
RAPSODEE Research Centre, UMR EMAC-CNRS 2392, Ecole des Mines d'Albi-Carmaux, Campus Jarlard, 81013 Albi, France
b
Laboratoires de Génie Chimique UMR-CNRS 5503, ENSIACET, INPT, BP 1301, 5 Rue Paulin Talabot, 31106 Toulouse Cedex 1, France

a r t i c l e i n f o a b s t r a c t

Available online 25 July 2010 The effect of the main means of agitation in a high shear mixer has been investigated in this study.
Granulation runs have been performed on a fine cohesive microcrystalline cellulose powder (Avicel 105,
Keywords: d50 = 20 μm) often used as a pharmaceutical excipient in tablet formulations in two bowls of a Mi-Pro®
High shear mixer granulator laboratory high shear mixer with a capacity of 0.9 and 1.9 L, respectively. Torque curves recorded during
Torque curves granulation are found to allow good control of the process while increasing impeller speed is found to
Impeller speed
generally reduce granule size and the onset of breakage seems to occur for similar values of impeller tip
Chopper
Microcrystalline cellulose
speed. As a general rule, the chopper allows for better binder distribution in the Mi-Pro® and is found to be
necessary for successful granulation at low to moderate impeller speeds. For high impeller speeds in excess
of 4.4 m/s with or without a chopper, similar granule sizes and growth mechanisms are observed. Granule
roundness was found to increase with impeller speed up to a certain speed after which granule roundness
has been found to decrease with increasing impeller speed most probably because of increased breakage of
the granules. Dry granule strength has been found to increase with increasing impeller speed, presenting
only a slight decrease at the highest impeller speed studied.
© 2010 Elsevier B.V. All rights reserved.

1. Introduction chopper slightly decreases the mean granule size, Schaefer et al. [4] or
narrows the granule size distribution, Knight et al. [5]. Increasing the
Wet granulation using high shear mixers is a widespread technique impeller speed generally leads to a decrease in granule size and an
applied in fields such as pharmaceutical development and various increase in growth rate. As scale-up techniques usually imply finding the
branches of the chemical industry. Formulations are usually developed equivalence of impeller speeds between the scales following certain
at small scale and following validation are scaled-up to production units. rules, reviewed by Faure et al. [6], it seemed pertinent to investigate the
Most equipment manufacturers offer small-scale versions of their effect of impeller speed on the Mi-Pro high shear mixer.
industrial equipment in order to allow a smoother transition. In situ monitoring and identification of granule growth kinetics are
The Mi-Pro high shear granulator allows working at different scales, important aspects of the wet granulation process and transition
keeping similar ratios between bowl, impeller and chopper and can be between growth regimes can be influenced by changes in operating
seen as a scaled-down Collette-Graal high shear mixer. The instrumen- conditions as shown by Saleh et al. [7] and Benali et al. [8]. Describing
tation of the Mi-Pro allows the recording of torque curves during high shear mixer granulation is considered especially difficult as
granulation. As shown by Mackaplow et al. [1], some difficulties exist agglomeration (growth), breakage and attrition are often simultaneous.
related to using torque/power curves as a means to control granulation Breakage and attrition depend on the mechanical strength of the wet
given by their sensitivity to wall build-up and collapse; however, the agglomerate and the impact and shear forces developed by the main
relationship between torque/power curves and tensile strength means of agitation present in the mixer. It is reasonable to assume that
discussed by Betz et al. [2] shows the promising future of this technique, both chopper and impeller affect the consolidation of the granules while
which can relate material properties to operating conditions. the high tip speed of chopper also ensures breakage of lumps. The
The effect of the chopper depends largely on the equipment used, chosen powder for this study is a fine microcrystalline cellulose powder.
while some authors like Michaels et al. [3] find no effect of the chopper The granulation of this type of powders has been shown to go through a
on the granule size distribution, others have found that using the typical four-regime granulation: wetting, nucleation, growth and
overwetting by Benali et al. [8] while Bouwman et al. [9] have shown
that granules in the growth stage present a high deformability upon
impact and shear with pieces of the main granules breaking off and
⁎ Corresponding author. coalescing with other granules followed by a new rearrangement to a
E-mail address: Toma-Mihai.Chitu@enstimac.fr (T.M. Chitu). sphere.

0032-5910/$ – see front matter © 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.powtec.2010.07.016
 T.M. Chitu et al. / Powder Technology 206 (2011) 34–43 35

Table 1
Physicochemical characteristics of binder solutions, contact angle and microcrystalline cellulose Avicel 105.

Binder properties ρL (kg/m3) γL (mN/m) μL (mPa/s at 25 °C) θ (°)

Water 1000 72.2 1.0 69.0 ± 1.2

Powder properties D50 [μm] (volumetric) Span (D90-D10)/ D50 Bulk density [kg/m3] True density[kg/m3]

Avicel PH 105 20 1.93 244 1514

Avicel PH 101 60 2.1 291 1540

Torque curves have been related to different states of liquid In order to assess granule growth and morphological properties,
saturation [10]: pendular state when particles are held together by a representative samples of solid (15–30 g) were removed from the bed
liquid bridge at their contact points, capillary state that occurs when all and tray-dried at 40 °C overnight. Due to the small size of the mixer
the voids are filled with the liquid and the surface liquid is drawn back bowls and small mass of product handled (120 g MCC Avicel 105 for the
into the pores under capillary action and funicular state as a transition 1.9 L bowl), each sample was taken from a new batch with experiments
state between the pendular and capillary state where the voids are not being carried out twice. The dried granules were sieved and 15 size
fully saturated with liquid. Torque profiles allow the definition of these fractions were collected. Weight mean diameter was determined using
states as a function of the added binder where phase boundaries are the following equation:
marked at the intersection of drawn tangents to the curve with the
optimum liquid amount supposed to be found in the plateau phase
(growth zone) [11]. ∑ fi dpi
i
The aim of this study is to investigate the influence on the process Mean diameter : dpm = ð1Þ
∑ fi
of the main means of agitation in the high shear granulator (chopper i
and impeller) and the ability of torque curves to control it.

where fi is the particle mass fraction of size interval i, dpi is the mean
2. Materials and Methods diameter of size interval i (μm). For a better understanding of
agglomeration mechanisms, three characteristic granule classes were
2.1. Materials defined (Fig. 2): fine (corresponding to the initial powder with
granule diameters and dry agglomerates inferior to 200 μm),
Microcrystalline cellulose Avicel PH 105 (FMC Biopolymer, intermediate (weak granules with diameters between 200 and
Ireland) with a mean diameter of about 20 μm has been chosen as a 800 μm) and coarse agglomerate (strong, dense granules with
model product for this study. For comparison purposes, another diameters greater than 800 μm but smaller than 5 mm). Granules
microcrystalline cellulose grade, Avicel PH 101, with a mean diameter exceeding 5 mm are defined as lumps and are not taken into account
of 60 μm has also been used. Some relevant properties for these in the mean diameter equation. Fig. 2 shows the defined classes at 50%
materials are presented in Table 1 with ultra-pure water being used as liquid to solid ratio for an impeller speed of 800 rpm and chopper
a binder in all experiments. speed of 3000 rpm. This definition is considered to allow a good
description over a wide range of operating conditions; however,
instances may occur where the mean diameter value and granulation
2.2. Granulation equipment: the Mi-Pro® high shear mixer kinetics analysis can be influenced by these imposed boundaries. For
operating conditions leading to large agglomerates like low impeller
The Mi-Pro (Pro-C-Ept, Zelzate, Belgium) high shear mixer allows and chopper speeds for instance where a low number of high
granulating while recording real-time impeller torque values with diameter granules could be obtained or on the other hand for very
1 second intervals. This vertical axis high shear granulator can be high impeller speeds where granules would not grow above 800 μm
equipped with differently sized glass bowls with corresponding three- because of extended granule breakage leading to no coarse agglom-
bladed impellers and choppers. Granulation time was ~14 minutes, the erates class.
powder bed was homogenized by dry mixing for 2 minutes before the
water addition was done at a constant rate using a Dosimat 760 syringe
pump (Metrohm, Berchem, Belgium) through a 400-μm diameter
capillary tube. Granulation was studied at two scales in the 0.9 and 1.9 L
bowls (Fig. 1). Impellers have a similar design while the chopper is
slightly inclined in the 0.9 L bowl.

Fig. 2. Sieving results for a L/S ratio of 50% at an impeller speed of 800 rpm and a chopper
Fig. 1. Schematic representation of the investigated Mi-Pro bowls. speed of 3000 rpm for the granulation of MCC Avicel 105 with water.
36 T.M. Chitu et al. / Powder Technology 206 (2011) 34–43

Fig. 3. The Instron 5567 mechanical testing machine.

Kinetic results are expressed in function of the fraction added

liquid to initial dry powder mass named liquid to solid ratio:

L = S Ratio = Added Liquid Mass ½kg = Initial Powder Mass ½kg ð%Þ ð2Þ

Discussing kinetics as a function of added liquid will always

present a certain degree of relativity as the added liquid may or may
not correspond to the granule saturation, especially when poor liquid
distribution is involved. However, with products like microcrystalline
cellulose where liquid–solid interaction involves swelling [12] and
formation of a crystallite gel [13] during pelletisation, calculating
saturation is extremely difficult. Fig. 5. Torque curve (a) and characteristic fraction evolutions (b) obtained on the 1.9 L
When discussing the effects at a single scale, impeller speeds have Mi-Pro HSM for the granulation of MCC Avicel 105 with water.
been expressed in revolutions per minute; when comparing scales or
when expressing behavior common to multiple scales, the impeller diameter of the granulation run. Granule strength was assessed using
tip speed has been chosen as the common denominator. the following equation proposed by Adams et al. [14]:

2.3. Granule strength and roundness Fmax

σg = 4 ð3Þ
π⋅D2g
Granule strength was evaluated by single granule compression tests
using an Instron 5567 mechanical testing machine (Fig. 3). The Instron
testing machine allows recording a maximum force of 500 N, sufficient Where Fmax is the fracture force and Dg is the measured granule
to break all of the tested granules. When the granule is subjected to diameter.
compressive strain (displacement), a marked drop in the signal can be Imposed velocity of the steel probe was of 1 mm/min while the
observed when the granule first cracks (Fig. 4). The values recorded after tested granules diameter was in the 1000 to 1250 μm range.
this point on the load-displacement curve are of little practical use as Granule roundness was determined by an automated microscopy
they represent the compression of granule fragments and eventually and image analysis technique (PharmaVision 830, Malvern Instruments,
powder particles. Granules tested had a diameter close to the mean Inc., USA). Roundness is a measurement relating length to width and

Fig. 4. Characteristic load displacement curve and scanning electron microscope picture of the cracked granule.
 T.M. Chitu et al. / Powder Technology 206 (2011) 34–43 37

Fig. 6. Evolution of the lump fraction (granules with diameter above 5 mm) as a
function of impeller speed (a) and mean granule size evolution as a function of impeller
speed (b) in the Mi-Pro HSM V = 1.9 L for the granulation of MCC Avicel 105 with water.

presenting values in the range of 0 to 1. A perfect circle would present a

roundness of 1 while a needle-shaped particle would have a roundness
close to 0. The roundness is determined typically using the equation:
Fig. 8. Evolution of the characteristic size fractions with chopper rotating at 3000 rpm
4πAg for three impeller speeds in the Mi-Pro HSM V = 1.9 L for the granulation of MCC Avicel
Roundness = ð4Þ 105 with water.
Pg2

where Ag is the measured area and Pg is the perimeter of the granules. 1908 μm. Granules investigated for roundness were in the range of
The zoom objective was used at the 0.5 position at which it presents a 1000 to 1250 μm. Results can be expressed as a function of number or
resolution of 18 μm, a depth of field of 11 μm, a field of view of value distributions; however, in our experiments, both methods
4722 × 6166 μm and can analyze granules in the size range of 45 to presented very similar values as well as nearly identical trends.

Fig. 7. Evolution of the final granulated product in the Mi-Pro HSM V = 0.9 L without a chopper for three different speeds: (a) 300 rpm , (b) 650 rpm and (c) 950 rpm for the
granulation of MCC Avicel 105 with water.
38 T.M. Chitu et al. / Powder Technology 206 (2011) 34–43

most probably because of the very cohesive nature of the initial

powder rather than binder addition
• for values ranging from 20% to ~ 45%, nucleation occurs with fines
forming intermediate agglomerates in the 200 to 800 μm range
• for L/S ratios exceeding 50%, growth occurs by layering and
coalescence, first with the fine particles being agglomerated into
intermediate granules which achieve a maximum value at ~83% L/S
ratio and the apparition of the first coarse granules, secondly for
values exceeding 83% with agglomeration between intermediate
particles
• for L/S ratios above 100%, overwetting occurs, while some usable
granules could still be obtained up to 125% L/S ratio, the yield is low
and presents a high percentage of lumps. For values above 125%, L/S
ratio the wet mass evolves to highly cohesive paste which upon
Fig. 9. Mean granule size evolution on the Mi-Pro V = 1.9 L with and without chopper at
800 rpm for the granulation of MCC Avicel 105 with water. further addition for L/S ratios above 150% turns into a suspension.

While the torque curve is seperated into clear-cut segments as a

3. Results and discussion
function of its inflexion points (as defined by Leuenberger et al. [11],
granule growth mechanisms transcend these limits and it is not
3.1. Wet granulation of MCC Avicel 105 on the Mi-Pro HSM
uncommon to find nuclei for instance in the wetting stage as well as in
the growth stage. However, the SEM observations allowed us to define
Fig. 5a shows scanning electron microscope (SEM) pictures of
the dominating mechanism for the presented L/S ratio ranges.
granules obtained from microcrystalline cellulose and water on the
Mi-Pro high shear mixer as a function of liquid to solid ratio and the
3.2. Effect of impeller speed and chopper presence and design
corresponding torque curve. Fig. 5b shows the evolution of the three
characteristic classes, for an impeller speed of 800 rpm and a chopper
The influence of chopper on mean granule size is discussed in
speed of 3000 rpm allowing us to determine the optimum L/S ratio for
Fig. 6. Without a chopper, between 100 and 400 rpm lumps are found
the water–MCC Avicel 105 granulation:
to increase with impeller speed (Fig. 6a), leading to an inhomogenous
• for values of the S/L ratio between 0% and 20%, wetting occurs with product comprised of small fine undergranulated product and a large
only a very small percentage of fines forming into intermediates, percentage of lumps. This explains the high difference in mean

Fig. 10. Torque curve comparison for the Mi-Pro V = 1.9 L HSM with and without chopper at three different impeller speeds: 100, 400 and 800 rpm and comparison between speeds
with chopper active for the granulation of MCC Avicel 105 with water.
 T.M. Chitu et al. / Powder Technology 206 (2011) 34–43 39

granule size between 150 and 500 rpm as lumps (granules above call the resulting re-agglomeration as a secondary nucleation stage
5 mm) are not taken into account when calculating the mean granule because of the similar increase in torque signal with the nucleation
size (Fig. 6b). The single higher granule size value for granulation stage. Denser, smaller nuclei were observed to result from the broken-
without a chopper against granulation with chopper is obtained at up larger agglomerates. For the granulation run without chopper at
100 rpm and can be explained by our choice in defining lumps only for 100 rpm, the second nucleation stage and redistribution of powder
granules exceeding 5 mm as a mean granule size of 3.5 mm is still and binder does not take place and the end result corresponds to
high and indicative of less than perfect granulating conditions. From granules presenting a much larger mean granule size.
600 rpm onwards, while a small percentage of lumps may still exist, At 400 rpm, the drop in torque is characteristic for the granulation
the impeller speed alone becomes sufficiently high so as to break the of MCC Avicel 105 at moderate speeds in the Mi-Pro and is
lumps and reduce the percentage to a negligible value at 800 rpm and independent of chopper operation: when sufficient liquid is added,
beyond. It should also be noted that the high impeller speed also the powder sticks to the side walls exiting the impeller action causing
distributes the liquid binder better. Because the binder is added a sharp drop in impeller torque. This seems to be related to the
through a capillary and not sprayed, directing the liquid flow towards mechanical dispersion of the binder in the powder bed as well as the
the rotating chopper also insures a better distribution of the binder small particle size and high compressibility of the MCC Avicel 105.
over the powder bed. The same behavior showing that the lump Fig. 11 shows a torque comparison between MCC Avicel 105
fraction decreases with increasing impeller speed when granulating (D50 = 20 μm) and MCC Avicel 101 (D50 = 60 μm) and the reprodu-
without a chopper has also been observed on the 0.9 L Mi-Pro cibilty for MCCAvicel 101.
granulation bowl. Fig. 7 shows pictures of representative samples To better illustrate the difference in behavior between these two
taken from the granulator at three different speeds when granulating powders at 400 rpm, we performed a granulation run using water
without a chopper, showing diminishing amounts of lumps with coloured with a red tracer. Fig. 12 shows the contents of the bowl for
increasing impeller speed. MCC Avicel 105 and Avicel 101. Fig. 13 shows the bowl after the product
Setting the chopper at a high rotating speed and investigating the has been emptied from the bowl. As it can be observed in Fig. 12a, there
influence of impeller speed at 100, 400 and 800 rpm on the growth is a strong interaction between the MCC Avicel 105 powder and the glass
kinetics, we can observe that increasing impeller speed accelerates walls of the Mi-Pro while the same cannot be observed for the MCC
the consumption of fine particles and the formation of intermediates Avicel 101 where the product stays inside the bowl without sticking to
and coarse granules (Fig. 8). While for the lower impeller speeds, a the walls. When emptying the bowls, we can observe that for both
decrease in intermediate particle percentage is observed near the end powders, a certain amount of powder is being compressed to the glass
of the granulation runs for 800 rpm; the intermediate particles show wall in the space between impeller and wall. The red stains formed
the highest percentage and are still present at the end of the
granulation. This can be explained by the fact that at 800 rpm,
breakage occurs and with a mean granule size of 800 μm, the
intermediate fraction 200-800 μm does not contain low-density,
fragile granules, but rather high-density broken-down granules.
When looking at what happens with or without a chopper at
800 rpm, where the chopper would not be necessary in regards to
mean final granule size, we see however a significant difference in
terms of granule growth, with the chopper delaying the formation of
coarse granules (Fig. 9).
The effect of chopper presence on torque curves with the chopper
turning at 3000 rpm shows only small differences between operating
with or without chopper at medium and high impeller speeds
(Fig. 10). On the 400 rpm curve after the wetting stage (up to10% L/
S ratio) a sharp drop in torque can be observed at 30% L/S ratio. Where
torque curves are continuous (100 and 800 rpm), we observe an
evolution similar to that proposed by Leuenberger [11]. Also at 400
and 800 rpm, the chopper influence is not visible, the recorded torque
curve showing the same inflexion points independent of chopper use.
We expected the differences in torque curves for varying impeller
speeds to be relateable to granulation kinetics obtained by sieving of
the dry product. It would seem however that our sieving protocol
homogenizes the results between the studied impeller speeds being
perhaps too destructive with the agglomerates obtained at low
impeller speeds, which as we will show present a lower granule
strength.
For the low impeller speed of 100 rpm, we observed that the
chopper turning at high speed can favor granulation. Granulation at
100 rpm with the chopper active is characterised by an initial wetting
stage up to 10% L/S ratio followed by a first stage of nucleation.
However, while we use the same name for granulation mechanisms, it
should be stated that at 100 rpm, the nuclei formed in the first
nucleation stage are few but big, low-resistance agglomerates
resulting mainly from layering. This nucleation mechanism is a result
of poor binder dispersion at such low impeller speeds and is not
similar to the nucleation observed at 400 and 800 rpm. Once the Fig. 11. Comparison between torque curves at 400 rpm for the Mi-Pro V = 1.9 L HSM
agglomerates are large enough, they come in contact with the with chopper active for MCC Avicel 101 and 105 (a) and reproducibility for MCC Avicel
chopper and a redistribution of moisture and material occurs. We 101 (b).
40 T.M. Chitu et al. / Powder Technology 206 (2011) 34–43

Fig. 12. View from above of the full Mi-Pro 1.9 L HSM at 400 rpm with chopper active at 41% L/S ratio for the granulation of MCC Avicel 105 (a) and Avicel 101 (b).

where the coloured water has been added show a much greater local and binder distribution occurs only by breakage of lumps due to
overwetting for MCC Avicel 105 than for Avicel 101. shear forces inside the powder bed,
As noted by Hapgood et al. [15] for different grades of lactose, - and an intermediate regime between droplet- and shear-controlled
penetration times can vary greatly with mean particle size. They varied where some agglomeration does occur near the spray zone without
the mean surface diameter between 26 and 69 μm and mean weight complete caking or pooling.
diameters between 67 and 200 μm (roughly three times as is the case for
the difference between MCC Avicel 105 and Avicel 101). This phenom- The presumable difference in penetration time difference for our two
enon could account for the different behavior in the nucleation phase. MCC grades could explain taking the powder from the drop-controlled
Hapgood et al. [16] proposed a nucleation regime map describing a regime (or intermediate) to the mechanical dispersion controlled
dimensionless penetration time parameter as a function of a dimen- regime where good binder dispersion requires good mechanical mixing.
sionless spray flux. Given the binder “footprint” in Fig. 13b, it is reasonable to assume that
granulation for MCC Avicel 101 at 400 rpm takes place in the
tp intermediate regime while granulation for MCC Avicel 105 occurs in
τh = ð5Þ
tc the shear-controlled regime. In our case, the same binder flow rate is
imposed through the same binder addition method (same average drop
′
3V size), which leaves the velocity of the powder surface as the only free
Ψa = ð6Þ
2A′ dd variable in estimating ψa. We could not give an estimation for this
velocity as our high shear mixer lacked the instrumentation with a high-
where tp is the penetration time and tc the circulation time of the speed video camera and image analysis software or positron emission
powder in the nucleation zone, V′ is the volumetric spray rate, A′ the area particle tracking technology, two of the methods currently employed in
flux of powder traversing the spray zone and dd the droplet diameter. order to measure powder surface velocity. However, as shown by
While the penetration time is largely a function of formulation Hapgood et al. [16] for water pumped on a lactose bed, the values for ψa
properties, the adimensional spray flux ψa depends on the operating would be close to a value of unity (in a 25 L Fielder high shear mixer),
conditions. The regime describes three regimes of nucleation: which would make it borderline between the intermediate and the
shear-controlled regime.
- the drop-controlled regime where each individual drop complete- We compared the effect of chopper design (length) by mounting
ly wets the powder bed to form a single nuclei granule and nuclei the smaller and shorter chopper that equips the 0.9 L bowl on the 1.9 L
size distribution is controlled by the drop size distribution, bowl in order to further study the importance of the chopper for the
- shear-controlled regime (or mechanical dispersion regime) where Mi-Pro high shear mixers. The impeller speed chosen was of 650 rpm,
liquid pooling or caking occurs where the binder meets the bed considered sufficient to bring the powder mass in contact with both
 T.M. Chitu et al. / Powder Technology 206 (2011) 34–43 41

Fig. 13. Front and back view from above of the empty Mi-Pro 1.9 L bowls at 400 rpm with chopper active at 41% L/S ratio for the granulation of MCC Avicel 105 (a) and MCC Avicel 101 (b).

choppers. As can be observed in Fig. 14, there is a difference in granule Dry granule strength measurements (Fig. 16) show that increasing
size as well as amount of lumps. We obtained a mean granule size of the impeller speed leads to an increase in granule strength. Above
1200 μm with the smaller chopper and 950 μm with the regular one 800 rpm, however, a slight decrease in granule strength was also
(20% difference). observed which could be related to granule breakage becoming
dominant to such an extent that granule growth and consolidation are
affected while granule roundness decreases even more, making the
3.3. Granule roundness and strength and mixer design influence use of Eq. (3) for granule strength, in the absence of a spheronization
protocol at the end of granulation, questionable. It should also be
Using the granules in the 1000 to 1250 μm range, the roundness of mentioned that the granules obtained at 100 rpm in the studied
the obtained granules was studied. Table 2 and Fig. 15 show the granule size range (1000–1250 μm) show a lower reproducibility
evolution of roundness with increasing impeller speed. Up to 400 rpm, being very fragile.
the roundness increases while for values above 400 rpm, granules begin As we have shown when discussing chopper influence, mean
to decrease in roundness. granule size depends largely on impeller speed, with sizes decreasing

Fig. 14. (a) Choppers from the Mi-Pro HSM 1.9 L (left) and 0.9 L (right), (b) granules obtained in the 1.9 L bowl using the 0.9 L chopper, (c) granules obtained in the 1.9 L bowl using
the 1.9 L chopper.
42 T.M. Chitu et al. / Powder Technology 206 (2011) 34–43

Table 2
MCC Avicel 105 roundness as a function of impeller speed on the 1.9 L Mi-Pro bowl with
the chopper rotating at 3000 rpm.

Impeller speed [rpm] Roundness

D50 Span = (D90–D10)/D50

100 0.67 0.7

260 0.77 0.43
400 0.86 0.45
550 0.79 0.47
800 0.72 0.54
1050 0.58 0.48

upon increasing impeller speed. Comparing final granule size between

two scales of the Mi-Pro: 0.9 and 1.9 L (Fig. 17) with similar geometric
proportions in terms of bowl and impeller shape and expressing the Fig. 17. Evolution of mean granule size as function of impeller tip speed on two scales of
impeller speed as impeller tip speed, we found a greater granule size for the Mi-Pro HSM: 0.9 L and 1.9 L for the granulation of MCC Avicel 105 with ultra-pure
speeds lower than 1.5 m/s on the small scale (0.9 L). This seems water.
influenced by the same limit imposed for the definition of lumps where
a higher degree of agglomerates below the 5 mm boundary can have a
great impact on mean granule size on the smallest impeller speeds
tested while even slightly larger granules than 5 mm are not taken into
account. For speeds between 1.5 and 4.03 m/s, final granules are
consistently larger on the larger scale (1.9 L); however, the difference
does not seem dramatic. The evolution of slowly decreasing mean
granule sizes is common to both scales, for values above 4.03 m/s,
breakage occurs; and for values of 4.7 m/s and above, granule sizes
become identical. Our results are similar to those obtained by Ameye et
al. [17] who have shown that the impeller tip speed does not have a

Fig. 15. Roundness distributions by number as a function of impeller speed for the
granulation of MCC Avicel 105 with ultra-pure water on the 1.9 L Mi-Pro with the
chopper at 3000 rpm.

Fig. 18. Evolution of the characteristic size fractions for the same impeller tip speed in the
Fig. 16. Granule strength evolution as a function of impeller speed for the granulation of Mi-Pro HSM V= 1.9 L and V = 0.9 L with the choppers at 3000 rpm for the granulation of
MCC Avicel 105 with ultra-pure water on the 1.9 L Mi-Pro with the chopper at 3000 rpm. MCC Avicel 105 with ultra-pure water.
 T.M. Chitu et al. / Powder Technology 206 (2011) 34–43 43

dramatic effect on mean granule size between the 0.9 and 1.9 L bowls for dpi mean diameter of size interval i (m)
impeller speeds above 5 m/s. dpm mean diameters at time t (m)
In order to investigate differences between scales, we chose to look Dg is the measured granule diameter (m)
in more detail at the growth mechanisms at nearly identical impeller fi particle mass fraction of size interval i (% )(w/w)
tip speeds by analyzing the evolution of the three characteristic Fmax is the fracture force (N)
fractions (Fig. 18). We chose the speed of 400 rpm at the 1.9 L scale Pg is the perimeter of the granules (m)
and 460 rpm at the 0.9 L scale corresponding to impeller tip speeds of tp penetration time (s)
2.93 and 2.89 m/s, respectively. Fill ratio, binder flow per initial mass tc circulation time (s)
of dry powder, chopper speed, homogenization protocol, and V′ volumetric spray rate (m3/s)
granulation time have been kept constant between scales. The
wetting stage is similar occurring up until an S/L ratio of 30% is
reached. Afterwards, nucleation and growth start for the small scale,
Greek symbols
with coarse granules beginning to show at 30% L/S ratio, while for the
σg tensile strength (Pa)
larger scale, coarse granules begin to form only from 50% L/S ratio
ψa adimensional spray flux (−)
onwards. This behavior could be explained by the more confined
τh adimensional penetration time (−)
volume of the 0.9 L bowl causing an increase in contacts between
granules, which leads to an acceleration of growth mechanisms.

4. Conclusion References

[1] M.B. Mackaplow, L.A. Rosen, J.N. Michaels, Effect of primary particle size on
The effect of the main means of agitation in a high shear mixer has granule growth and endpoint determination in high-shear wet granulation,
been investigated in this study. We found that torque curves can allow Powder Technol. 108 (2000) 32–45.
[2] G. Betz, P.J. Burgin, H.L. Leuenberger, Power consumption profile analysis and
good control over the granulation process especially at high impeller tensile strength measurements during moist agglomeration, Int. J. Pharm. 252
speeds. At intermediate impeller speeds, torque curves show a low (2003) 11–25.
reproducibility especially because of the strong cohesive nature of the [3] J.N. Michaels, L. Farber, G.S. Wong, K. Hapgood, S.J. Heidel, J. Farabaugh, J.-H. Chou,
G.I. Tardos, Steady states in granulation of pharmaceutical powders with
employed grade of microcrystalline cellulose. We found that it is application to scale-up, Powder Technol. 189 (2009) 295–303.
important to study the effect of chopper presence at more than just [4] T. Schaefer, B. Taagegard, L.J. Thomsen, H.G. Kristensen, Melt pelletization in a
one speed as by increasing impeller speed the percentage of lumps high shear mixer. V. Effects of apparatus variables, Eur. J. Pharm. Sci. (1993)
133–141.
will increase for low to medium impeller speeds and decrease when
[5] P.C. Knight, An investigation of the kinetics of granulation using a high shear
impeller speed alone can break up the lumps for high impeller speeds. mixer, Powder Technol. 77 (1993) 159–169.
As a general rule, the chopper allows for better binder distribution in [6] A. Faure, P. York, R. Rowe, Process control and scale-up of pharmaceutical wet
granulation processes: a review, Eur. J. Pharm. Biopharm. 52 (2001) 269–277.
the Mi-Pro and is found to be necessary for successful granulation at
[7] K. Saleh, L. Vialatte, P. Guigon, Wet granulation in a batch high shear mixer, Chem.
low to moderate impeller speeds. For high impeller speeds in excess of Eng. Sci. 60 (2005) 3763–3775.
4.4 m/s with or without chopper, similar granule sizes and growth [8] M. Benali, V. Gerbaud, M. Hemati, Effect of operating conditions and physico–
mechanisms are observed. chemical properties on the wet granulation kinetics in high shear mixer, Powder
Technol. 190 (2009) 160–169.
Granule roundness was found to increase with impeller speed up [9] A.M. Bouwman, M.R. Visser, G.M.H. Meesters, H.W. Frijlink, The use of Stokes
to a certain speed, after which granule roundness was found to deformation number as a predictive tool for material exchange behaviour of granules
decrease with increasing impeller speed, most probably because of in the ‘equilibrium phase’ in high shear granulation, Int. J. Pharm. 318 (1–2) (2006)
78–85.
increased breakage of the granules. Dry granule strength has been [10] H.L. Leuenberger, M. Puchkov, E. Krausbacher, G. Betz, Manufacturing pharma-
found to increase with increasing impeller speed presenting only a ceutical granules: is the granulation end-point a myth, 3rd Int. Granulation
slight decrease at the highest impeller speed studied. Workshop, , 2007 Sheffield.
[11] H.L. Leuenberger, Granulation: new techniques, Pharm. Acta Helv. 57/3 (1982)
The effect of equipment geometry between the 0.9 and the 1.9 L 72–82.
Mi-pro high shear mixers has shown that at the lower scale, [12] P. Kleinebudde, Shrinking and swelling properties of pellets containing micro-
accelerated granule growth takes place while keeping a constant crystalline cellulose (MCC) and low substituted hydroxypropylcellulose (L-HPC).
I. Shrinking properties, Int. J. Pharm. 104 (1994) 209–219.
impeller tip speed can give good agreement in terms of size between [13] P. Kleinebudde, The crystallite-gel-model for microcrystalline cellulose in wet
the scales. granulation, extrusion and spheronization, Pharm. Res. 14 (1997) 804–809.
[14] M.J. Adams, M.A. Mullier, J.P.K. Seville, Agglomerate strength measurement using
a uniaxial confined compression test, Powder Technol. 78 (1994) 5–13.
List of symbols
[15] K. Hapgood, J.D. Litster, S.R. Biggs, T. Howes, Drop penetration into porous powder
A′ the area flux of powder traversing the spray zone (m2) beds, J. Colloid Interface Sci. 253 (2002) 353–366.
Ag is the measured granule area (m2 [16] K.P. Hapgood, J.D. Litster, R. Smith, Nucleation regime map for wet granulation,
C material constant (for uniform sphere C = 6) AIChE J. 49 (2) (2003) 350–361.
[17] D. Ameye, E. Keleb, C. Vervaet, J.P. Remon, E. Adams, D.L. Massart, Scaling-up of a
dd the droplet diameter lactose wet granulation process in Mi-Pro high shear mixers, Eur. J. Pharm. Sci. 17
di primary particle diameter (m) (2002) 247–251.