European Journal of Pharmaceutical Sciences 69 (2015) 1925

Contents lists available at ScienceDirect

European Journal of Pharmaceutical Sciences

journal homepage: www.elsevier.com/locate/ejps

Mechanistic time scales in adhesive mixing investigated by dry particle

sizing
Duy Nguyen a, Anders Rasmuson a, Ingela Niklasson Bjrn b, Kyrre Thalberg b,
a
b

Department of Chemical Engineering, Chalmers University of Technology, SE-412 96 Gothenburg, Sweden

Astra Zeneca Pharmaceutical Development R&D Mlndal, SE-431 83 Mlndal, Sweden

a r t i c l e

i n f o

Article history:
Received 8 May 2014
Received in revised form 24 October 2014
Accepted 23 December 2014
Available online 7 January 2015
Keywords:
Adhesive mixing
Mixing dynamics
Dry particle sizing
Dry powders for inhalation
Dry particle coating
Time scales

a b s t r a c t
This study exploits the mechanisms governing blending of adhesive mixtures, i.e. random mixing, deagglomeration and adhesion, and their relative importance to achieve mixing homogeneity. To this
end, blending of micronized particles (nes) with carrier particles was carried out using a high shear
mixer. Dry particle sizing using laser diffraction coupled with a strong powder dispersion unit was
employed to measure the nes content in samples collected during mixing, and hence to assess blend
homogeneity. The method was also employed to evaluate the relative strength of the agglomerates present in the nes. Particle sizing using a non-destructive imaging technique was used to monitor changes in
particle size during blending. It could be shown that the de-agglomeration of the ne-particle agglomerates is the slowest mechanism and hence the rate-limiting step as regards achieving a homogeneous
adhesive mixture. Consequently, a longer mixing time is needed for blending of larger agglomerates.
Being fast, simple and reproducible, the laser diffraction technique was shown to be an efcient method
for measurement of ne particle content and homogeneity of a mixture, while the non-destructive image
analysis was able to give relevant information on the rate of de-agglomeration of the ne-particle
agglomerates as well as on the size of the resulting carrier-ne particle assemblies.
2015 Elsevier B.V. All rights reserved.

1. Introduction
The mixing in which ne particles are attached onto the surfaces of carrier particles has successfully been employed to produce homogeneous pharmaceutical formulations, both in the
case of very potent APIs and as regards dry powder formulations
for inhalations (Vikas Anand Saharan et al., 2008). The mixing process was rst described by the concept of ordered mixing where
mixing rate obeys rst-order kinetics (Hersey, 1975), i.e. same as
random mixing, but was later described as a dynamic equilibrium
process between adhesive and non-adhesive mixings using the
total mixing concept (Staniforth, 1981). The term adhesive mixing, was later recommended (Staniforth, 1987). Adhesive mixing
is applied to improve homogeneity of pharmaceutical dosage
forms as well as to alter the surface properties of several products
in a wide range of industrial applications (Jiang et al., 2006; Kale
et al., 2009; Swaminathan and Kildsig, 2002). It has been reported
that mixing homogeneity is a function of material properties such
as particle size (Sundell-Bredenberg and Nystrom, 2001), particle
shape (Wong and Pilpel, 1988), and surface properties (Buckton,
Corresponding author. Tel.: +46 70 9131964; fax: +46 317763727.
E-mail address: Kyrre.Thalberg@astrazeneca.com (K. Thalberg).
http://dx.doi.org/10.1016/j.ejps.2014.12.016
0928-0987/ 2015 Elsevier B.V. All rights reserved.

1997; Podczeck, 1998) and also depends on composition, addition

of other components (Jones and Price, 2006) and mixing time
(Bosquillon et al., 2001; Jones et al., 2010).
The mixing process itself is obviously a key unit operation in the
formulation of an adhesive mixture. Although the effects of material properties on product performance have been well documented, the mixing dynamics have not received much attention
(de Boer et al., 2012). Therefore, there is still a need to investigate
mixing mechanisms and dynamics, in order to provide better
understanding of the overall process. The aim of this work is thus
to estimate the time scale of the mechanisms governing the mixing
process as well as their relative importance as regards achieving a
homogeneous mixture.
Spherical mannitol pellets with a median diameter just above
200 lm and micronized lactose monohydrate with a median size
of 4 lm were selected for the experiments (details are given in Section 3.1). The latter component can be regarded as representing
drug particles, e.g. in a pharmaceutical dry powder formulation
for inhalation. Since such ne particles spontaneously forms
agglomerates during processing and handling, the study also deals
with the inuence of ne-particle agglomerates on the mixing
behaviour.

20

D. Nguyen et al. / European Journal of Pharmaceutical Sciences 69 (2015) 1925

In addition, the application of two different dry particle sizing

techniques is reported. In order to assess the content and content
uniformity of the ne particles in the blend, laser diffraction combined with a powerful powder dispersing unit was applied. By
means of a calibration procedure, the exact ne particle content
of a sample could be determined. The method was also used to
measure relative strength of the ne-particle agglomerates. As a
complement to this, a non-dispersing particle size method was
applied to monitor the disintegration of the ne-particle agglomerates and the formation of nes-carrier aggregates during mixing.

3. Material and methods

3.1. Material
D-Mannitol pellets (Nonpareil 108-200) supplied from Freund
(Japan) was used as model carrier, and have a narrow size distribution in the range of 150250 lm (D50 = 212 lm). Micronized lactose monohydrate from AstraZeneca was used as nes, thus
representing the drug component in a pharmaceutical adhesive
blend. The nes have a median size (D50) of 3.8 lm.

3.2. Sieving of nes

2. Mixing mechanisms
It has been documented that the mixing of ne particles with
carrier is achieved via a combination of random mixing and adhesive mixing (de Villiers, 1997). Four different mechanisms can be
distinguished in this process, as is illustrated in Fig. 1. The rst
one governs the distribution of assemblies, e.g. the mixing between
carrier particles and ne-particle agglomerates, and can be
described by the concept of random mixing developed by Lacey
(1943). The second mechanism is de-agglomeration of ne-particle
agglomerates which occurs as a consequence of mechanical collisions during blending (Ikegami et al., 2003; Kale et al., 2009). It
has been reported to occur rapidly during the random mixing stage
(de Villiers, 1997). Adhesion is the third mechanism implying the
binding of ne particle fragments onto the carrier surface by adhesion forces, e.g. van der Waals forces (Nystrm and Westerberg,
1986). The last mechanism deals with redistribution and exchange
of ne particles between carrier particles (Alonso et al., 1989; de
Boer et al., 2012, 2003). Finally, press-on forces acting during mixing may cause the compression of ne particles onto the carrier
surfaces (Podczeck, 1996).
For a given nes-carrier combination, the homogeneity of the
mixture and the properties of the nal blend are functions of the
above described mechanisms, which in turn depends on the type
of mixer and the process parameters applied. For instance, the dispersion performance of adhesive mixtures for inhalation was found
to be affected by the balance between de-agglomeration, re-distribution and compression mechanisms (Grasmeijer et al., 2013). The
present work focuses on studying the time scales of the mechanisms governing the high shear mixing process, in particular the
steps of random mixing, de-agglomeration and adhesion of nes.

Due to high cohesion, micronized particles often exist in the

form of self-agglomerates with a very broad size distribution. This
was also the case for the micronized lactose nes used here. To
investigate the effect of agglomerate size on mixing dynamics,
the material was gently sieved using sieves of mesh sizes;
0.71 mm, 1.0 mm, 1.4 mm, and 2.0 mm, to obtain ve size fractions
of nes-agglomerates.
3.3. High shear mixing
The mixing experiments were performed in a MiPro high shear
granulator (Procept, Belgium) installed with a 1.9 l vessel and a 3
bevelled-blades impeller, and was used without the chopper. A
moderate impeller speed of 500 rpm was used in all experiments.
A total of 260 g of material with 5% w/w of ne-particle agglomerates was loaded in each experiment. The agglomerates were placed
as a horizontal layer in the middle of the powder bed at start. The
mixer was stopped after 5, 15, 30, 45, 60, 75, 90 and 100 s to withdraw samples for analysis. At each time point, approximately 7 g
was removed for analysis. Experiments were carried out for all 5
size fractions of nes to evaluate the inuence of agglomerate size
on the mixing behaviour. Triplicate experiments were carried out
for the 1.42.0 mm fraction.
3.4. Sizing of primary particles
Particle sizing of primary particles was used for characterization of starting materials. Laser diffraction using Sympatec HELOS
equipped with the RODOS dispersing unit was employed. R5 and
R2 lenses were used for sizing of carrier and ne particles, respec-

Fig. 1. Schematic representation of the four mixing mechanisms.

21

D. Nguyen et al. / European Journal of Pharmaceutical Sciences 69 (2015) 1925

tively. A dispersion pressure of 3.0 bar was applied in all measurements. The size distribution did not change when further increasing the pressure, which indicates that fully de-aggregated primary
particles were detected. Each measurement was performed with a
sample weight of 100 mg to ensure sufcient statistical sampling.
3.5. Sizing of agglomerates
To access the size of the ne-particle agglomerates, care must be
taken not to disintegrate them during the analysis. To this end, the
Sympatec QicPic equipped with the GRADIS gravitational powder
feeder was employed. Both the original ne-particle agglomerates
and the nes-carrier assemblies formed during mixing points were
studied. Repeated measurements were carried out on each individual sample to ensure that the size distributions obtained were not
affected by the powder handling during measurement.
3.6. Assessment of nes content and blend uniformity
As the carrier and nes widely differ in particle size, the content
of ne particles in a blended sample can be assessed from the
bimodal particle size distribution, provided that agglomerates are
fully disintegrated during the test. Sympatec HELOS with RODOS
operated at 3.0 bar was found to be suitable for this. The R5 lens
was used because it is able to capture both the carrier and the ne
particles in the same measurement. A curve relating ne particle
content to the measured volume fraction of nes to actual ne particle content was constructed using samples with known ne particle content, prepared by weighing accurate amounts of carrier
and thereafter nes into test tubes, using a six digit analytical balance. For each concentration of nes, three samples were prepared
and analysed.
Blend uniformity was assessed by analysing 6 samples from the
top of the powder bed at each mixing time point. The sample size
was approximately 100 mg. The relative standard deviation (RSD)
of the nes content is used as the mixing index. The homogeneity
is regarded as acceptable if the RSD is less than 5% (Jiang et al., 2006).
3.7. Assessment of mechanical strength of agglomerates
Sympatec HELOS equipped with ASPIROS was operated at different pressures ranging from 0.1 bar to 1.0 bar, to assess the
nes-agglomerates. The relative mechanical strength of different
agglomerate size fractions was evaluated by comparison of the
fraction of particles less than 15 lm, obtained at different pressures., The cut off value of 15 lm was chosen to ensure that all ne
particles are taken into account. For assessment of the agglomerate
strength, the fraction <15 lm is used as a measure reecting how
easily the agglomerates are disintegrated. Three repeated measurements were carried out at all pressures for each sieved fraction.
4. Results and discussions
4.1. Size distribution of materials used
Table 1 summarizes the particle size characterization for the
primary carrier and lactose nes. The median diameter of the lac-

tose nes is well below 5 lm which indicates that they are suitable
to use as a model drug for inhalation.
The carrier particles have a very narrow size distribution which
indicates a good potential for obtaining homogeneous dosage
forms (Lai and Hersey, 1987; Nystrm et al., 1993). There was
almost no ne particles in the carrier as indicated by the very
low volume fraction <11 lm. Since both carrier and lactose nes
have narrow size distributions with a big gap in-between, any
cut-off value from 11 lm to 50 lm will serve to quantify the fraction of each material in a blend.
4.2. Measurement of nes content
The fraction of particles smaller than 11 lm was chosen to represent the amount of lactose nes, and was measured using Sympatec Helos/RODOS for a series of specially prepared samples, see
Section 3.6. The fraction of nes present in the carrier is then subtracted to construct a curve relating the measured nes content to
the loaded amount of ne lactose, Fig. 2.
The RSD of analytical weighing was less than 2%. The RSD for
the volume fractions obtained from Sympatec HELOS was around
6%, irrespective of the actual nes load, which reects the repeatability of the method. Considering that the R5 lens (measuring
range: 0.5875 lm) have rather low resolution in the size range
of the ne particles, an RSD value of 6% in the nes content is quite
good, and sufcient for assessment of content uniformity of
blended samples.
It is obvious from Fig. 2 that the relation between loaded and
measured volume fractions is far from unity. The line tted to
experimental data has the following equation:

y 1:67x 0:567 with R2 0:995

The content of nes measured by particle size distribution is hence
much higher than the actual ne content in the mixture. Different
densities of nes and carrier obviously play a role here. The difference in true density between lactose monohydrate (1.52 g/cm3) and
3
D-mannitol (1.49 g/cm ) is however small and cannot explain the
observed large deviation from unity. It is evident that the fraction
of ne particles in an adhesive mixture is heavily over-counted by
the Sympatec HELOS / RODOS, as has also been reported before
(Thalberg et al., 2012).
The overestimation behaviour can be regarded as the response
factor to nes content for the laser diffraction method. As such,
over-counting of nes should not invalidate the method. There
are still, however, uncertainties around the application of the
methodology. For instance, differences in structure and uniformity
between the samples prepared in small glass vials and the high
shear blended samples may lead to different response factors for
the two cases. Therefore, it is not possible to calculate the exact
content of nes in the blends using the curve in Fig. 2. In the context of the present work, however, exact values of the nes content
is not critical to the interpretation of the mixing mechanisms. Furthermore, the curve serves to demonstrate a linear relation
between measured and actual nes content as well as the reproducibility of the method, which is crucial in order to assess content
uniformity of the blends.

Table 1
Particle size distribution data for the materials used.
Size distribution (lm)

Fine lactose
Carrier

D10

D50

D90

Span

1.17
134

3.83
212

8.51
254

1.91
0.56

% volume lower than 5 lm

% volume lower than 11 lm

% volume lower than 50 lm

70
0.09

99
0.2

100
1.4

22

D. Nguyen et al. / European Journal of Pharmaceutical Sciences 69 (2015) 1925

4.4. Mixing dynamics

18

14
12
10
Unity line
Linear best t of the data

8
6
4
2
0
0

10

12

14

16

18

loaded nes fracon (%)

Fig. 2. Fines content measured using Sympatec Helos with RODOS versus loaded
nes content. Error bars represent standard deviations in weighing (x) and in
measured content (y).

4.3. Mechanical strength of ne-particle agglomerates

Fig. 3 shows data from the relative mechanical strength
method, as is described in Section 3.7. All agglomerate fractions
display virtually the same behaviour with respect to dispersion
pressure. As expected the dispersed ratio increases with a higher
dispersion pressure. At 1.0 bar, all lactose-nes agglomerates are
disintegrated into particles less than 15 lm.
The results indicate that all agglomerate size fractions behave
similarly under mechanical load, or in other words that all the
agglomerate fractions have the same mechanical strength.
Although the absolute mechanical strength cannot be obtained
by this method, the measurements demonstrate that the strength
is comparable, and hence that the size of the agglomerates is the
only difference between the sieved lactose fractions.
Recently, the agglomerate strength of a range of micronized
drugs commonly used in inhalation therapy was investigated using
an almost identical equipment and approach (Jaffari et al., 2013).
The notion of a critical primary pressure, CPP, was introduced
and dened as the pressure needed to ensure complete disintegration of the active pharmaceutical ingredient. CPP values obtained
ranged from 1.0 bar (salbutamol base) to 3.5 bar (salmeterol xinafoate). Adopting this approach, we may conclude that the CPP
for the lactose nes here used is P1.0 bar. It can be concluded that
micronized drugs predominantly consists of agglomerates, and
that the lactose nes here used as model drug is within the appropriate range regarding agglomerate strength.

4.4.1. Random mixing time scale

The random mixing mechanism outlined in Fig. 1 can be captured by studying the mixing of cohesionless particles, e.g. mixing
of carrier particles alone. The transient mixing of such particles in a
high shear mixer was recently investigated experimentally and by
EulerianEulerian multiphase simulations (Nguyen et al., 2014).
The mixer and the carrier material were the same as used in this
study. In the experiments, 10% of the mannitol carrier particles
were coloured (denoted the tracer) to enable high speed image
analysis of the mixing behaviour. Both experimentally and by simulation, it was shown that the time required to obtain a homogeneous carrier blend was of the same order as the time for ow
development, i.e. a couple of seconds. The evolution of homogeneity is illustrated in Fig. 4.
The mixing end point however also depends on the scale of
scrutiny, i.e. the sample size at which homogeneity is to be
achieved (Hogg, 2009). Cohesionless mixing is governed by two
mechanisms, convection and diffusion. The former proceeds rapidly to achieve homogeneity at the macro-level, i.e. long range
homogeneity, whereas it is a rather slow process to obtain short
range homogeneity, the micro-level. In the referred work, homogeneity was assessed for a sample size of 100 mg. Due to the large
number of particles in the sample, homogeneity at the micro-level
could not be studied.
The addition of 5% w/w of nes to the mixture is expected to
cause only minor changes to the ow pattern relative to the pure
carrier system. This was conrmed by comparing the impeller torques for the two cases. Therefore, the time needed to achieve the
random mixing step in this experiment can be expected to be similar to the random mixing of cohesionless particles, i.e. a few seconds only at the sample size of 100 mg.

4.4.2. De-agglomeration time scale

The de-agglomeration of the ne particle agglomerates (the second mechanism in Fig. 1) can be studied in terms of the evolution of
the particle size distribution of the mixture using a non-destructive
particle sizing method. In Fig. 5, distribution curves recorded at different time points are shown. At 5 s, the mixture displays a bimodal
distribution with large agglomerates in the range of 10002700 lm.
This range slightly exceeds the initial size distribution of the
sieved ne lactose, i.e. 14002000 lm, which indicates that not
only de-agglomeration but also agglomeration occurs during
mixing, as has also been reported (de Boer et al., 2004). As mixing
proceeds, the agglomerates are rapidly broken up; at 45 s the large
agglomerate peak is no longer seen in the size distribution curve.
100

100

0.16

80

0.12

60

0.08

40

0.04

20

Content

% lower than 15micrometer

0.2

60

40

20

d<0.71

0.71<d<1

1<d<1.4

1.4<d<2

experimental parcle content

80

experimental %RSD

d>2

%RSD

measured nes fracon (%)

16

0
0

0.5

1.5

2.5

3.5

Mixing me (s)

0
0

0.2

0.4

0.6

0.8

Dispersion pressure (bar)

Fig. 3. Dispersion curve for different agglomerate size fractions.

1.2

Fig. 4. Random mixing dynamics in the MiPro high shear mixer, impeller speed of
300 rpm, studied using EulerianEulerian multiphase simulation. Full line indicates
average concentration of tracer particles, dotted line indicates the RSD for 100 mg
samples.

23

D. Nguyen et al. / European Journal of Pharmaceutical Sciences 69 (2015) 1925

6
5s

Density distribuon

15s

30s

45s

60s

75s

90s

0.4

0.3

0.2

3
0.1

2
0
500

1000

1500

2000

2500

3000

0
0

500

1000

1500

2000

2500

3000

Parcle size (m)

Fig. 5. The evolution of the overall PSD of mixtures. Data refers to the sieved fraction 1.42.0 mm.

The de-agglomeration kinetics is further illustrated in Fig. 6, in

which different size fractions of agglomerates are plotted versus
mixing time. The size fractions are normalized relative to the corresponding values at the 5 s time point. The larger the agglomerates are, the faster is the de-agglomeration kinetics. For instance,
agglomerates larger than 1500 lm are reduced dramatically after
15 s, and are no longer present after 30 s. As for the size fraction
larger than 450 lm, 50 s of mixing is needed to reach a stable level.
During mixing, the de-agglomeration is accompanied by the
adhesion of agglomerate fragments onto the carrier surface, i.e.
the third mechanism in Fig. 1. It has been reported that the larger
daughter agglomerates remain as free assemblies whilst the ner
fragments become attached to the carrier surface (de Villiers,
1997). Consequently, the peak associated with the carrier will display an increase in size. Fig. 7 shows the evolution of the median
size of the blend. Due to the presence of the large agglomerates,
elevated values are obtained initially. The median size thereafter

1
0.9

0.7
0.6
d>1500

0.5
0.4

d>1050

240

d>625

239

d>450

238

D-50 (m)

Normalized fracon

0.8

drops rapidly, to a level of around 233 lm. This value is signicantly higher than the particle diameter of 212 lm of the pure carrier (measured by the same method) and indicates that an
adhesive mixture has been achieved.
The adhesion of ne particles to the carrier, mechanism 3 in
Fig. 1, can also be captured via Figs. 57. If the kinetics of this mechanism is slow, agglomerate fragments with a size smaller than the
carrier would be observed and a median particle size below that of
the carrier particles would be expected, due to the coexistence of
pure carrier particles and nes and/or small satellites of nes.
Apparently, this is not the case, and we may conclude that the adhesion of ne particles onto the carrier is a rapid process, relative to
the disintegration of the ne-particle agglomerates.
SEM imaging of the formulation after 100 s of mixing, Fig. 8, further conrms this picture and shows a structure of small clusters of
ne particles attached to the carrier. Apparently, a uniform and
continuous monolayer structure is not at hand. Anyhow, the structure observed in SEM seems to be consistent with the slight but
distinct increase in particle size of the mixture.
The levelling out of the overall particle size in Fig. 7 can be taken
as an indication that the mixing is ready. At even longer mixing
times, re-distribution of the ne particles between the carriers can
be expected, see Fig. 1. However, the size change during this stage
is not signicant and will furthermore not affect the homogeneity

0.3
0.2
0.1

Mean D-50 of mixture

237
236
235
234
233

0
0

10

20

30

40

50

60

70

80

90

Mixing me (s)

232
0

20

40

60

80

100

Mixing me (s)

Fig. 6. De-agglomeration rate for different agglomerate sizes. The fraction above
cut-off values are normalized to the corresponding fraction at the 5 s mixing time
point. Data refers to blending of agglomerate fraction 1.42.0 mm.

Fig. 7. Evolution of median particle size for the mixing of agglomerate fraction 1.4
2.0 mm.

24

D. Nguyen et al. / European Journal of Pharmaceutical Sciences 69 (2015) 1925

Fig. 8. SEM images of carrier (left) and nal blend (right) taken from a blend produced using the 1.42.0 mm size fraction.

of the mixture (see next section). Therefore, the mixing experiments

were nished after 100 s in this work.
4.4.3. Time scale to achieve homogeneity
Fig. 9 shows the evolution of the nes content and homogeneity
when mixing 5% ne lactose of the size fraction 1.42.0 mm. The
nes content was obtained by the analysis procedure described
in Section 3.6. At the very early mixing stage, ne-particle agglomerates were observed to oat on the top of powder bed, which
resulted in a very high ne content. Not surprisingly, the mixture
was strongly inhomogeneous, with an RSD of 24% at 5 s. The nes
content thereafter continuously decrease as agglomerates are broken up and distributed throughout the mixture. The progression of
mixing can be assessed by evaluating both the homogeneity and
the average nes content. After 60 s an RSD of 4.9% was reached,
i.e. a value just below the limit of 5% which is often set as a criterion for acceptable blend uniformity. At longer blending times the
uniformity is only slightly improved and the content remains at a
constant level.
The overall homogeneity achieved in the current experiments is
the combined result of all mechanisms, i.e. random mixing, deagglomeration and adhesion, acting together. By comparing the
time scale to achieve blend homogeneity with the time scale governing the different mechanisms, it is obvious that the mixing time
required to produce a uniform formulation is determined by the
de-agglomeration kinetics. In other words, once de-agglomeration
of the nes-agglomerates is completed, a homogeneous mixture is
achieved.
It is thus clear that by monitoring ne particle content and
blend homogeneity, the dynamics of the different mixing mechanisms can be followed. Laser diffraction using Sympatec Helos with

the RODOS dispersing unit was shown to be a simple and quick

way to monitor the evolution of the blend. In addition, a nondestructive particle sizing technique, here the Sympatec QicPic
with the GRADIS dispenser, was used to monitor the evolution of
the overall particle size of the mixture. Both the disappearance of
the large ne-particle agglomerates and the establishment of a stable endpoint level, inferring the presence of adhesive mixture
agglomerates, could be captured. The adhesive mixture structure
was further conrmed by SEM. Further changes to the structure
of the blend may certainly be the case at extended mixing times,
due to exchange and redistribution of ne particles and also due
to press-on forces which becomes more important when the mixing time is increased. Such changes may be important in certain
applications of adhesive mixtures, but was out of the scope for
the current study.
4.5. Inuence of the agglomerate size
Since adhesive mixing is seen to be dominated by the deagglomeration of ne-particle agglomerates, agglomerate size
and mechanical strength are obviously crucial factors affecting
the mixing kinetics. Fig. 10 illustrates the evolution of homogeneity for blends prepared with the various sieved agglomerate fractions. Mixing end-points are given in Table 2. It can be seen that
longer mixing times are required when mixing larger agglomerates. For instance, 2 min was required to meet the mixing endpoint for agglomerates larger than 2 mm, whilst less than 30 s is
sufcient for agglomerates smaller than 0.71 mm. Assuming a
homogeneous structure within the agglomerates, the energy consumption to break a unit volume of agglomerate would be identical
for agglomerates that have the same mechanical strength. There-

25
RSD d<0.71

%RSD 0.75<d<1

%RSD 1.4<d<2

RSD d>2mm

%RSD 1<d<1.4

25
average nes fracon

20

%RSD of nes concentraon

20

%RSD

Fines content and %RSD

30

15

15
10

10
5

0
0

20

40

60

80

100

120

Mixing me (s)

Fig. 9. Evolution of nes content for the mixing of agglomerate fraction 1.4
2.0 mm.

20

40

60

80

100

120

Mixing me (s)

Fig. 10. The evolution of homogeneity for mixing of different size fractions. The
dotted line represents the acceptance level.

25

D. Nguyen et al. / European Journal of Pharmaceutical Sciences 69 (2015) 1925

Table 2
Mixing end-points for different agglomerate fractions.
Size fractions
Mixing end-point (s)

<0.71 mm
24

0.711.0 mm
30

fore, the energy needed to break the agglomerates is a function of

size or in other words, larger agglomerates require more energy for
break-up. This is probably a main reason for the observed behaviour. Moreover, the mixing is also affected by segregation. During
the high shear mixing process, ne-particle agglomerates were
seen to be lifted up to the top of the powder bed due to differences
in density and size. The particle collision frequency and efciency
was thus reduced and the breakage process was decelerated. Due
to differences in size, the segregation intensity varied for different
agglomerate size fractions.
It is concluded that agglomerate size is an important factor
inuencing the mixing process. This nding further emphasizes
that the mixing dynamics is controlled by de-agglomeration of
the ne-particle agglomerates. Agglomerate size affects the breakage performance directly in terms of energy consumption and indirectly by reducing the particle collision rate due to segregation.
5. Conclusions
A quantitative study of the mixing dynamics of adhesive mixtures was carried out to evaluate the importance of different mixing mechanisms. In large, data were in agreement with the notion
of four different mechanisms, (i) random mixing, (ii) de-agglomeration, (iii) adhesion and (iv) exchange and redistribution. Time
scales for the rst three of these could be established. A main nding was that the de-agglomeration of ne-particle agglomerates is
the rate-limiting step as regards achieving satisfactory blend
homogeneity. The mixing time needed was affected by the size
of agglomerates, i.e. a longer mixing time is required for blending
of larger agglomerates.
This study thus provided in-depth understanding of the early
stages of adhesive mixing. The later stage mechanisms, i.e.
exchange, re-distribution and compression of the ne particles,
would also merit further investigation as these may affect key
properties of adhesive mixtures, such as the dispersibility of nes
in the case of adhesive mixtures for inhalation.
Dry particle sizing was found to be a rapid and reliable method
for assessment of mixing uniformity as well as the relative strength
of agglomerates. A non-destructive particle sizing technique, i.e.
the Sympatec QicPic with the GRADIS dispenser, was useful for
monitoring disappearance of large ne-particle agglomerates and
the establishment of a stable endpoint level, inferring the presence of adhesive mixture agglomerates. By combining the two
methods, in-depth information regarding mixing mechanisms
and dynamics could be obtained.
Acknowledgement
Financial support from POWTECH ITN (Grant 264722) is gratefully acknowledged.
References
Alonso, M., Satoh, M., Miyanami, K., 1989. Kinetics of nes transfer among carriers
in powder coating. Powder Technol. 59, 217224.
Bosquillon, C., Lombry, C., Preat, V., Vanbever, R., 2001. Inuence of formulation
excipients and physical characteristics of inhalation dry powders on their
aerosolization performance. J. Control. Release 70, 329339.

1.01.4 mm
35

1.42.0 mm
60

d > 2.0 mm
115

Buckton, G., 1997. Characterisation of small changes in the physical properties of

powders of signicance for dry powder inhaler formulations. Adv. Drug Deliv.
Rev. 26, 1727.
de Boer, A.H., Chan, H.K., Price, R., 2012. A critical view on lactose-based drug
formulation and device studies for dry powder inhalation: Which are relevant
and what interactions to expect? Adv. Drug Deliv. Rev. 64, 257274.
de Boer, A.H., Hagedoorn, P., Gjaltema, D., Goede, J., Kussendrager, K.D., Frijlink,
H.W., 2003. Air classier technology (ACT) in dry powder inhalation Part 2. The
effect of lactose carrier surface properties on the drug-to-carrier interaction in
adhesive mixtures for inhalation. Int. J. Pharm. 260, 201216.
de Boer, A.H., Hagedoorn, P., Gjaltema, D., Lambregts, D., Irngartinger, M., Frijlink,
H.W., 2004. The mode of drug particle detachment from carrier crystals in an air
classier-based inhaler. Pharm. Res. 21, 21672174.
de Villiers, M.M., 1997. Description of the kinetics of the deagglomeration of drug
particle agglomerates during powder mixing. Int. J. Pharm. 151, 16.
Grasmeijer, F.H.P., Frijlink, H.W., de Boer, H.A., 2013. Mixing time effects on the
dispersion performance of adhesive mixtures for inhalation. PLoS ONE, 8.
Hersey, J.A., 1975. Ordered mixing: a new concept in powder mixing practice.
Powder Technol. 11, 4144.
Hogg, R., 2009. Mixing and segregation in powders: evaluation, mechanisms and
processes. KONA Powder Part J. 27, 115.
Ikegami, K., Kawashima, Y., Takeuchi, H., Yamamoto, H., Mimura, K., Momose, D.-I.,
Ouchi, K., 2003. A new agglomerated KSR-592 b-form crystal system for dry
powder inhalation formulation to improve inhalation performance in vitro and
in vivo. J. Control. Release 88, 2333.
Jaffari, S., Forbes, B., Collins, E., Barlow, D.J., Martin, G.P., Murnane, D., 2013. Rapid
characterisation of the inherent dispersibility of respirable powders using dry
dispersion laser diffraction. Int. J. Pharm. 447, 124131.
Jiang, Y., Matsusaka, S., Masuda, H., Yokoyama, T., 2006. Evaluation of owability of
composite particles and powder mixtures by a vibrating capillary method. J.
Chem. Eng. Jpn. 39, 1421.
Jones, M., Price, R., 2006. The inuence of ne excipient particles on the
performance of carrier-based dry powder inhalation formulations. Pharm.
Res. 23, 16651674.
Jones, M.D., Santo, J.G., Yakub, B., Dennison, M., Master, H., Buckton, G., 2010. The
relationship between drug concentration, mixing time, blending order and
ternary dry powder inhalation performance. Int. J. Pharm. 391, 137147.
Kale, K., Hapgood, K., Stewart, P., 2009. Drug agglomeration and dissolution what
is the inuence of powder mixing? Eur. J. Pharm. Biopharm. 72, 156164.
Lacey, P.M.C., 1943. The mixing of solid particles. Trans. Inst. Chem. Eng. 21, 53.
Lai, F.K., Hersey, J.A., 1987. Simulated ordered powder mixture. Int. J. Pharm. 36,
157164.
Nguyen, D., Rasmuson, A., Niklasson Bjrn, I., Thalberg, K., 2014. CFD simulation of
transient particle mixing in a high shear mixer. Powder Technol. 258, 324330.
Nystrm, C., Westerberg, M., 1986. The use of ordered mixtures for improving the
dissolution rate of low solubility compounds&. J. Pharm. Pharmacol. 38, 161
165.
Nystrm, C., Westerberg, M., Nilsson, P., 1993. A pharmaceutical composition. EP
Patent 0,324,725.
Podczeck, F., 1996. Assessment of the mode of adherence and the deformation
characteristics of micronized particles adhering to various surfaces. Int. J.
Pharm. 145, 6576.
Podczeck, F., 1998. Adhesion forces in interactive powder mixtures of a micronized
drug and carrier particles of various particle size distributions. J. Adhes. Sci.
Technol. 12, 13231339.
Staniforth, J., 1981. Total mixing. Int. J. Pharm. Technol. Prod. Mfr 2, 712.
Staniforth, J.N., 1987. British Pharmaceutical Conference Science Award Lecture
1986: order out of chaos. J. Pharm. Pharmacol. 39, 329334.
Sundell-Bredenberg, S., Nystrom, C., 2001. The possibility of achieving an
interactive mixture with high dose homogeneity containing an extremely low
proportion of a micronised drug. Eur. J. Pharm. Sci.: Off. J. Eur. Fed. Pharm. Sci.
12, 285295.
Swaminathan, V., Kildsig, D.O., 2002. Effect of magnesium stearate on the content
uniformity of active ingredient in pharmaceutical powder mixtures. AAPS
PharmSciTech 3, 2731.
Thalberg, K., Berg, E., Fransson, M., 2012. Modeling dispersion of dry powders for
inhalation. The concepts of total nes, cohesive energy and interaction
parameters. Int. J. Pharm. 427, 224233.
Vikas Anand Saharan, V.K., Kataria, Mahesh, Kharb, Vandana, Choudhury, Pratim
Kumar, 2008. Ordered mixing: mechanism, process and applications in
pharmaceutical formulations. AJPS 3, 240259.
Wong, L.W., Pilpel, N., 1988. The Performance of a Small Y Gone Mixer. Drug Dev.
Ind. Pharm. 14, 609616.