Droplet and Particle Size Relationship and Shell Thickness

of Inhalable Lactose Particles during Spray Drying

RAN ALDERBORN,2 ULLA ELOFSSON1
JESSICA ELVERSSON,1,2 ANNA MILLQVIST-FUREBY,1 GO
1

YKI, Institute for Surface Chemistry, Box 5607, SE-114 86 Stockholm, Sweden

Department of Pharmacy, Uppsala University, Box 580, SE-751 23 Uppsala, Sweden

Received 7 November 2002; revised 11 November 2002; accepted 11 November 2002

ABSTRACT: To find means of controlling the size and density of particles intended for
inhalation the relationship between droplet and particle size during spray drying was
investigated. Lactose solutions were atomized with a two-fluid nozzle and dried in a
laboratory spray drier. The effects of nozzle orifice diameter, atomization airflow and feed
concentration on droplet and particle size were examined. Mass median diameter of both
droplets and particles were analyzed with laser diffraction. In addition, scanning electron
microscopy and transmission electron microscopy were used for studies of particle shape
and morphology. It was demonstrated that nozzle orifice diameter and airflow, but not
feed concentration controlled the droplet size during atomization. Increasing droplet size
increased particle size but the effect was also influenced by feed concentration. Particles
from solutions of a low concentration (1% w/w) were smaller than those from higher
concentrations (520% w/w). This may be partly explained by lower yields at higher feed
concentrations, but may also be related to differences in drying rate. Spray-dried lactose
solutions formed hollow particles, and it was suggested that the shell thickness of
the particles increased with increasing feed concentration. 2003 Wiley-Liss, Inc. and the
American Pharmaceutical Association J Pharm Sci 92:900910, 2003

Keywords: spray drying; droplet size; particle size; laser diffraction; shell thickness;
particle formation

INTRODUCTION
The pulmonary route is an attractive alternative
to oral and parenteral routes for systemic delivery
of protein and peptide drugs.1,2 The surface area
of the lungs (approximately 100 m2) is comparable
to the size of the gastrointestinal tract, but unlike
oral administration, pulmonary delivery allows
both exposure to drugs through a well-perfused
tissue and avoidance of the first-pass effect. In
addition, metabolic activity in the lungs is low.
However, the respiratory system in itself res-

Correspondence to: Jessica Elversson (Telephone: 46 8

790 96 68; Fax: 46 8 20 89 98;
E-mail: jessica.elversson@surfchem.kth.se)
Journal of Pharmaceutical Sciences, Vol. 92, 900910 (2003)
2003 Wiley-Liss, Inc. and the American Pharmaceutical Association

900

tricts the entrance of particulate matter by various means: for example, geometry of the airways,
morphology of the epithelial cells, and clearance
mechanisms of the lungs. As a result, inhalation
particles have to be aerodynamically optimized to
reach the thin (<0.2 mm) alveolar epithelium,3
where the absorption site of protein drugs, for
example, is assumed to exist. To enable delivery to
the alveolar tract particles for inhalation have
traditionally been made less than 5 mm. The
desired particle size is attained by milling of
freeze-dried or otherwise solidified material in
air-jet mills (i.e., micronization). Intense milling
can, however, cause unwanted change in the
physicochemical properties of the material, for
example, creation of amorphous regions at the
surface, which can affect humidity dependence
and stability,4 electrostatic charging, and cohesivity.5 In addition, chemical decomposition of

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

DROPLET AND PARTICLE SIZE RELATIONSHIP OF INHALABLE LACTOSE PARTICLES

thermolabile molecules has been observed during

micronization.6,7
To overcome the difficulties of preparing, handling, and administrating micronized powders for
inhalation, development concepts including formulation (e.g., ordered mixtures), manufacturing
(e.g., spheronization), and device design have been
realized. Over the last few years, Edwards and
coworkers8 have addressed the challenges of inhalation powders with an interesting approach. By
manipulation of particle size and density, aerosolization efficiency, and delivered dose of inhalation powders, whether produced by double- and
single-evaporation technique or spray drying, are
increased. First, an increased particle size results
in a decreased tendency to aggregate,9 and in
combination with a low mass, the aerodynamic
behavior is retained or even improved.8,10 The particles are still able to penetrate into the alveolar
region where dissolution followed by absorption
takes place. Second, phagocytic clearance can be
diminished, by making particles larger because
alveolar macrophages are unable to phagocyte
units larger than about 5 mm.11,12
For the preparation of porous protein particles
with a controlled particle size spray freeze drying
could be an alternative method.13 However, particle formation in for example liquid nitrogen would
be followed by a freeze-drying step, which makes
the process more time-consuming and complex.
Techniques using supercritical fluids have also
been used for the preparation of bio therapeutic
drugs of inhalable size.1416 Such particles are
generally of conventional densities, 11.5 g/cm3.
Spray drying could be a more attractive method for
preparing porous powders for inhalation. Particles
can be formed directly from solutions or emulsions.
The heat transfer to the particles is generally low
due to very short residence times in the drier
(approximately 1 s in a laboratory drier). Many
carbohydrate-containing particles are amorphous
by formation, which helps in stabilizing proteins
during dehydration.17,18 The particle size and
density are likely to be controlled by both processand formulation parameters. Therefore, spray
drying necessitates a consideration of several
important steps in addition to drying of droplets.19
These operations include preparation of a liquid
feed, atomization of the liquid into a spray and
separation of the dried particles from the drying
gas. In each of these steps the processing parameters will affect the products properties such as
particle size, morphology, and bulk density, as well
as the surface composition of particles.20 During

901

atomization the physical properties, such as

viscosity, surface tension, and density of the liquid
feed are likely to influence the break up of liquid,
and hence, the droplet size distribution of the
spray.21,22 Including proteins or polymers in the
solution may be expected to increase the complexity of viscosity effects due to non-Newtonian
behavior, at least at high concentrations. Theoretically, the particle size depends on the droplet size
and the dry matter content of the solution,23 at
least for the formation of nonporous spheres.
The relationship between droplet and particle
size during spray drying is a rarely investigated
area, although particle properties are highly
affected by the airliquid interface area (area-tovolume ratio) during atomization and drying.24
Atomizer performance and preparation of respirable particles by spray drying were investigated by
Dunbar et al.,25 but correlation between geometric
particle size and droplet size was not undertaken.
In the same study mass median aerodynamic
diameter was found to decrease with increasing
operating pressure of an air blast atomizer, but not
as a consequence of reduced droplet size. In a few
other investigations, the droplet size was calculated from particle size data assuming nonporous
particles,26,27 resulting in an overestimation of
droplet size because of the porous or agglomerated
particles being analyzed.28 In addition, both the
droplet size and the size distribution will change
during drying29 and drying kinetics for sprays are
more complex and less well studied than drying of
single droplets.19,30,31 Single-drop drying includes
both static32 and free-falling methods.33 The results from investigations of static drying (levitation) are difficult to apply to spray drying, due to
large differences in time scale, droplet size, and
flow conditions. Experimental methods for the
analysis of atomizer performance (droplet size)
and particle size are available and widely used.
However, the relationship between droplet and
particle size during spray drying deserves to be
investigated in more detail.
The present study was performed to evaluate
the possibility of controlling the size and density of
respirable particles, produced by spray drying.
The effects of atomization parameters (nozzle
orifice diameter and airflow through nozzle) and
feed concentration were investigated by comparative studies of droplet and particle size during
spray drying, using laser diffraction. In addition,
the variation of droplet size in different regions of
the spray and the morphology of the particles were
examined.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

902

ELVERSSON ET AL.

MATERIALS AND METHODS

Chemicals
a-Lactose monohydrate (in vitro grade) was purchased from Merck Eurolab (Stockholm, Sweden).
Rapeseed oil for particle analysis was supplied by
Karlshamns AB (Karlshamn, Sweden) and fractionated coconut oil (Miglyol1 812N) was obtained
from Condea Chemie GmbH (Witten, Germany).
Isopropanol (analytical grade) for TEM analysis
was purchased from Solveco Chemicals AB (Taby,
Sweden). Milli Q water was used throughout the
experiments.

Spray Drying
Spray drying was performed in a laboratory spray
drier built at the Institute for Surface Chemistry.
The drier operates in a cocurrent mode, with a
jacketed two-fluid nozzle and has a drying column
of 750 mm in length and 150 mm in diameter.
Compressed air from an in-house supply was
regulated to 2.4 bars by a pressure regulator
(Norgren, IMI Norgren Inc., St. Littleton, CO) and
used for atomization of the feed solution. The
standard conditions used in all experiments were:
an inlet temperature of 2008C, a liquid feed rate of
5 mL/min, a flow of drying air of approximately
0.8 m3/min, a jacket temperature of 258C and an
outlet temperature of 908C maintained by the
aspirator. The droplet size during spraying was
controlled by (a) a variable flow meter (Brooks
R-6-15-B/ sapphire, Brooks Inc., Chelmsford, MA)
that regulated the atomization airflow; and (b)
nozzle orifice diameter (1.5 and 2.0 mm). The
atomization airflow was varied between 20.6 and
31.9 L/min for the small orifice and between
28.9 and 35.7 L/min for the large orifice.
To ensure a representative collection of the
dried particles a membrane filter (Gore-Tex1
Membrane, W.L. Gore & Associates Scandinavia
AB, Molndal, Sweden) was used instead of a
cyclone (Figure 1). The cutoff of the filter was
>99.99% for particles of 0.100.15 mm. No particles were observed (by microscopy) either in the
filter housing or on the outside of the filter
membrane after drying. Shaking and tapping the
filter collected the particles and the recovery from
the filter was approximately 5080%. The batches
were stored over silica gel in a desiccator until
analysis. Between runs, the filter was soaked in
hot tap water for 1020 min, rinsed several times
and then dried in a drying cabinet for 2 h.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

Figure 1. Laboratory spray drier with filter for

particle separation.

Feed samples were prepared at concentrations

of 1, 5, 10, and 20% w/w of lactose in Milli Q water.
The solutions were stirred for at least 30 min
before use. The 20% w/w solutions were heated to
45508C to facilitate dissolution and cooled to
ambient temperature (21268C) before use.
Droplet Size Analysis
The distribution of droplet size was analyzed by
laser diffraction (Malvern Mastersizer X with
Helium-Neon laser, Malvern Instruments Ltd.,
Malvern, UK) and the setup is presented in
Figure 2. The instrument was adapted for spray
measurement with an adjustable holder for the
nozzle and a collection unit for the spray. An aspirator was mounted in the bottom of the collector

Figure 2. Setup for droplet size analysis by laser

diffraction.

DROPLET AND PARTICLE SIZE RELATIONSHIP OF INHALABLE LACTOSE PARTICLES

to prevent the aerosol from reentering the laser

area. The nozzle was placed 50 mm above the
laser beam in a vertical position. The distance between the central axis of the spray and the 300 mm
Fourier lens was 43 mm, which was within the
working range of the lens. The spray nozzles, the
air regulators and the peristaltic pump (Gilson
Minipuls II, Gilson Inc., Middleton, OH) were the
same as used in the spray drier. The nozzles were
calibrated with water before measurement to
give a predetermined droplet diameter. All experiments were performed for 1020 min (30
60 measurements) at ambient temperature (21
268C) and ambient relative humidity (3550%).
The size distribution was calculated with Mie
theory using the refractive indices of water
(n 1.33) and air (n 1.00).
By variation of the position of the nozzle during
analysis, droplet size in different locations of the
spray was examined (Figure 3). The distance
between the nozzle and (a) the lens (x-position) was
from 4070 mm, (b) the laser beam (y-position)
was from 2197 mm and (c) the perpendicular
distance from the laser axis (z-position) was
zero
12 mm.
Particle Size Analysis
A different laser diffraction instrument (Malvern
Mastersizer 2000, Malvern Instruments, Malvern, UK) was utilized for measurement of the
particle size distribution. Two sets of light sources
(Helium-Neon and Argon) gave a measuring size
range of 0.02600 mm. Fifty milligrams of powder,
withdrawn from the bulk with a spoon after

Figure 3. Experimental volume within the spray

cone. Upper corners a, dedge of spray; vertical axis
b,ccenter of spray; and bottom corner ein spray.

903

mixing and transferred to a 50-mL beaker, was

placed under vacuum (<15 mbar) before being
suspended in 20 mL of rapeseed oil, to facilitate
dispersion. Each suspension was stirred for 1 min
and ultrasonicated (Transonic T460/H ultrasonic
bath, Elma, Germany) for 5 min. No fragmentation
was observed under these conditions. Adequate
dispersion into primary particles was checked by
light-transmission microscopy (Axioplan, Carl
Zeiss Microscopy, Jena, Germany). Suspension
was added to a stirred sample cell to obtain an
obscuration value of 20%
1. A cycle of 5 measurements over 4 s started after one minute of
circulation. The analysis was repeated once with a
new sample from the same beaker. The sample
unit was rinsed twice with rapeseed oil between
measurements. The size distribution was calculated by Mie theory using the refractive indices of
lactose (n 1.5638, k 0.1) and rapeseed oil (n
1.471). The median diameter showed a relative
standard deviation of less than 3% (based on
seven samples withdrawn from the same powder
and analyzed at two different occasions), which
was considered acceptable to observe the effects of
process and formulation parameters.
A dispersant with a lower viscosity, in combination with a stronger ultrasonication was needed to
achieve complete deaggregation of particles from
solutions of 1% w/w lactose. Fifty milligrams of
particles were suspended in 40 mL of Miglyol
(n 1.448), stirred for 1 min, and ultrasonicated
for 5 min with a 13 mm standard probe (Vibracell
750, Sonics & Materials Inc., Newton, CT). Sampling and analysis was performed as described
above.
Particle size, shape, and surface morphology
were examined in an Environmental Scanning
Electron MicroscopeTM (ESEM), model XL 30 TMP
(W) (FEI Company, Hillsboro, ON), in high vacuum
mode. Acceleration voltage was typically 25 kV. A
sample was applied to a carbon tape covered Al
stub. The samples were coated with Au of 640 A
thickness (Sputter Coater SCD 050, Balzers union
AG, Balzers, Lichtenstein) prior to analysis.
ESEM was considered as a reference method to
laser diffraction because individual particles could
be identified.
Particle shell and interior was viewed by
Transmission Electron Microscopy (TEM), model
2000 FX (JEOL Inc., Akishima, Japan). Accelerating voltage was 200 kV and particles were dispersed in isopropanol and deposited onto carbon
coated copper grids, for analysis. To reduce the
influence of air moisture the sample was prepared
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

904

ELVERSSON ET AL.

in a custom-prepared dry box (addVise, Bromma,

Sweden).
Data Analysis
The volumetric or mass size distribution of droplets and particles was not perfectly log normal.
Therefore, mass median diameter, D (v, 0.5), was
reported instead of the geometric mean diameter.
To refer to RANGE and SPAN values, when describing the shape of the size distribution, median
diameter was used in favor of Sauter mean
diameter. RANGE, describing 80% of the droplet/particle volume, and SPAN, describing the
width of the distribution, are defined as:
RANGE Dv; 0:9-- Dv; 0:1
SPAN

RANGE
Dv; 0:5

The median diameter was reported as average

values of n measurements
standard deviation
(SD).
Median diameter of droplets was corrected for
loss of material in the spray drier tower by cutting
the high end of the cumulative droplet size distribution at the percentage corresponding to the
powder yield in each experiment and recalculating
the cumulative volumes in each size class, thereby
shifting D (v, 0.5) to lower sizes.

RESULTS
Variation of Droplet Size in Different
Locations of the Spray
When the position of the nozzle, relative to the
laser beam, was changed droplets in different
locations of the spray were analyzed (Figure 3). A
distance (x) to the lens of >50 mm increased the
spurious reading (beam steering) on the inner
detectors. Beam steering arises from density
gradients in the air causing abnormal refraction
of the laser beam and false registration of very
large droplets. A closer distance than 40 mm gave
considerable contamination of the lens. The droplet size was, however, unaffected by the vertical
position (y) of the nozzle, when analysis was
performed along the spray axis (z 0) (Table 1).
Varying the perpendicular position (z) of the
nozzle involved measurements in different regions
of the spray cone and droplet size increased as the
measurement position approached the edges of
the spray cone.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

Table 1. Droplet Size Distribution, Represented by

Volume Fractions 10%, D (v, 0.1), 50%, D (v, 0.5), and
90%, D (v, 0.9), at Different Locations in the Spray
Location
in Spray
a
b
c
d
e

D (v, 0.1)
(mm)

D (v, 0.5)
(mm)

D (v, 0.9)
(mm)

4.98
2.89
3.26
4.80
3.20

9.85
7.46
7.54
9.85
7.56

18.11
13.49
14.04
19.15
13.54

Droplet Size Control

The effects of nozzle orifice diameter and atomization airflow on droplet size were investigated by
laser diffraction. The median diameter of the
droplets could be varied between 6 and 20 mm by
means of nozzle orifice diameter and atomization
airflow (Figure 4). An increase of airflow decreased the droplet size in a linear manner while a
33% increase in orifice diameter of the nozzle
shifted the distribution to give droplets of approximately double the size.
Not only droplet size but also size distribution
was affected by primarily, orifice diameter, and
second, atomization airflow. RANGE was about
three times larger for the 2.0-mm nozzle compared
to the 1.5 mm nozzle (data not shown). Atomization
was considered more effective at higher airflows as
a more vaporous appearance of the spray was
observed visually: this was confirmed by SPAN
decreasing as the airflow increased (data not
shown).

Figure 4. Median droplet diameter as a function of

atomization airflow and nozzle orifice diameter, 1.5 mm
(open) and 2.0 mm (closed). Feed solutions: water (*),
n 60; lactose 10% w/w (&), n 30; and lactose 20% w/w
(~), n 1. Error bars: SD.

DROPLET AND PARTICLE SIZE RELATIONSHIP OF INHALABLE LACTOSE PARTICLES

905

The effect of feed concentration was demonstrated as droplet size of water and 10 and 20% w/w
lactose solutions, at constant atomization airflow
(28.9 L/min), were compared. The difference in
droplet size between the 10% w/w lactose concentration and pure water was negligible (Figure 4).
Droplet analysis of the 20% w/w solution was
interrupted by contamination of the lens and laserunit due to insufficient collection of the spray. Data
from a single measurement showed however no
evidential deviation from the 10% w/w solution.
Particle Size Control
By keeping the atomization airflow constant
(28.9 L/min), the effect of solids content in feed
solution on particle size was investigated. Droplets of two size ranges were included, small (1.5 mm
nozzle orifice diameter) and large (2.0 mm nozzle
orifice diameter), to observe how the size distribution changed during drying. The particle size was
analyzed with laser diffraction and ESEM.
Due to strong aggregation, the particles
prepared from 1% w/w lactose solutions were
measured according to a different protocol. Comparative measurements with a higher concentration of lactose showed a 20% reduction in average
median diameter when using the Miglyol method
instead of the rapeseed method. However, only a
minor decrease in the slope value between particle
and droplet diameters was observed between the
methods.
The relationship between solids content and
particle size was positive but leveled off at concentrations higher than 5% w/w (Figure 5). The
effect was most pronounced with the large nozzle
where particle size was nearly independent of
solids content at intermediate to high concentrations of lactose. In contrast, particles from 1% w/w
solutions were approximately half the size of
particles at other concentrations. The very small
particle size from 1% w/w solutions and the absence
of notable size difference between particles from
higher concentrations was confirmed by ESEM
(Figure 6).

Figure 5. Median diameter as a function of solids

content of feed solution. Experimental D (v, 0.5) (*) for
nozzle orifice 1.5 mm, experimental D (v, 0.5) (&) for
nozzle orifice 2.0 mm and predicted D (v, 0.5) (- - -) at
Ccrit 24%. SD < 0.07 mm, n 10. The atomization airflow was 28.9 L/min. Standard conditions were used for
spray drying.

droplet size (R > 0.790.99). Because particles

from 1% w/w lactose solutions were considerably
smaller than other particles, the slope (k) of the
relationship between particle and droplet diameters was lower than with higher lactose contents (e.g., k 0.16 compared to k 0.29 for the
median diameter). Particle distributions were,
throughout, narrower than droplet distributions:

Relationship between Droplet and Particle Size

One, 5, 10, and 20% w/w lactose solutions were
spray dried at various settings of atomization flow
and with two different nozzle orifice diameters
giving a wide range of droplet sizes. Figure 7
shows that there was a positive, almost linear,
relationship between the particle size and the

Figure 6. ESEM micrographs of spray-dried lactose

from solutions of (a) 1% w/w/, (b) 5% w/w, (c) 10% w/w,
and (d) 20% w/w solids content. The nozzle orifice
diameter was 2.0 mm and the atomization airflow was
28.9 L/min. Standard conditions were used for spray
drying.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

906

ELVERSSON ET AL.

Figure 7. Correlation between droplet and particle

size presented for the whole size distribution: (a)
D (v, 0.1), (b) D (v, 0.5), and (c) D (v, 0.9). Feed solutions:
lactose 1% w/w (*), 5% w/w (*), 10% w/w (&), and
20% w/w (~). R-value and slope (k) between particle and
droplet diameter is noted in the figures for clarity.

RANGE decreased approximately by a factor of

two during drying (not shown here).

DISCUSSION
Lactose, commonly used for pulmonary delivery
and for stabilization of proteins, was chosen as a
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

relevant model for the present investigation

where the relationship between droplet size and
particle size during spray drying was examined.
During atomization, surface tension and viscosity
of the feed solution are commonly thought to
affect the droplet size.21,22 Lactose has no surface
activity, and the viscosity of the 120% w/w
lactose solutions was between 0.910.95 mPas
(Ubbelohde viscometer, 258C), which is similar to
water (0.89 mPas). Accordingly, during atomization, lactose solutions could be assumed to break
up and disintegrate as water does. This was confirmed by droplet size measurements under
varying atomization airflow, nozzle orifice diameter, and solids content in feed (Figure 4).
However, at 20% feed solids content there were
problems with insufficient spray collection resulting in contamination of the lenses and distortion
of size distributions. To avoid such problems
droplet data of water were used instead of data
from the lactose solutions for comparison with
particle data.
The droplet size was substantially affected by
the atomization airflow: a higher flow resulted in
smaller droplets (Figure 4). The increased air
liquid surface area of the droplets is a result of
increasing relative motion between the surface
layer of the liquid and the air producing kinetic
energy for disintegration. Increasing the orifice
diameter from 1.5 to 2.0 mm decreased the air
velocity through the nozzle by 56%. The frictional
forces over the liquid surface thereby decreased
and droplet size increased. When the larger orifice
was used and with unchanged atomization airflow the median droplet size was approximately
2.5 times that of the smaller orifice.
Differences in nozzle cap tightness induced
small variations in droplet size due to changes in
adverse pressure. To investigate the consequences
of cap tightness on the subsequent particle size,
spray drying of lactose solution 10% w/w was
repeated three times (Table 2). The nozzle size was
1.5 mm and the atomization airflow was 28.9 L/
min. The nozzle cap was changed several times
between the first and the second runs but not
between the second and the third runs. Small
differences in median diameter, RANGE and
SPAN were noted, between the first and second
runs (Table 2). The results from the second and
third runs were, however, nearly identical
(Table 2). The overall variation in particle median
size induced by cap tightness is believed to be of
the same range as the variation between the first
and the second runs, (SD of
0.14).

DROPLET AND PARTICLE SIZE RELATIONSHIP OF INHALABLE LACTOSE PARTICLES

907

Table 2. Particle Size Distribution, Represented by Volume Fractions 10%, D (v, 0.1); 50%, D (v, 0.5); and 90%,
D (v, 0.9), of Lactose Solution (10% w/w) Spray Dried at 28.9 L/min Atomization Airflow with 1.5 mm Nozzle
Run No.

D (v, 0.1)a (mm)

D (v,0.5)b(mm)

D (v, 0.9)c (mm)

Range (mm)

Span ()

2.17
2.08
2.10

4.16
3.85
3.88

7.86
7.07
7.09

5.69
4.99
4.99

1.37
1.30
1.29

1
2
3
a

SD < 0.01 mm.

SD < 0.02 mm.
SD < 0.03 mm; n 10.

b
c

Only a moderate size difference in particles was

observed when lactose solutions of 5 to 20% w/w
were dried from droplets of equal size (Figure 5).
The particles from 1% w/w solutions were, however, significantly smaller than those from the
other concentrations. The difference in particle
size between 1% w/w and 5 to 20% w/w particles
was also clearly seen from the ESEM micrographs
(Figure 6). Furthermore, particles from the lowest
concentration also had a separate relationship
between droplet and particle sizes (Figure 7). A
lower slope value indicated that the size distribution of particles from 1% w/w solutions was narrower than the distributions of more concentrated
solutions. In a forthcoming study, the concentration interval of 15% w/w solids will be studied in
some detail for several carbohydrates.
Small differences in particle size implied that
particles from 5% w/w solutions had a lower density than particles from more concentrated solutions (10 and 20% w/w). This may be an effect of
loss of droplets/particles in the drying tower
because the loss increased as the feed concentration increased. Others have reported deficiency in
size difference between particles with different dry
matter contents: Masters19 stated that products of
film-forming materials show a significant reduction in mean size only when concentrations are
low. Concentrations lower than 5% w/w could
probably be considered as low: Fell and Newton27
investigated lactose solutions of 5 and 15% w/v and
Cassidy et al.,26 investigated lactose feed stocks of
10 to 50% w/v, and neither reported any size
differences between particles. In addition, the final
particle size was considered not to depend on the
initial droplet size.27 Considering the findings of
the present study, that the droplet size was independent of the lactose concentration of the feed,
then the droplet diameters calculated from particle data assuming solid particles, reported by Fell
and Newton27 are questionable. Calculation of
droplet size from particle size, applying material
balance, requires as well as accurate data on

primary particle size, experimental data on effective particle density. Density calculated from true
density of raw materials or determined by helium
pycnometry would considerably overestimate the
droplet size, as these values correspond only to
solid material.34 In the same study mercury porosimetry measurements of spray-dried emulsions
resulted in densities of 0.26 up to 0.39 g/cm3
compared to densities of 1.12 to 1.13 g/cm3 from
helium pycnometry.
The nonsolid structure of lactose particles was
observed from ESEM micrographs of broken
particles and TEM micrographs (Figure 8). Particles were smooth and spherical but with a hollow
interior. A hollow interior can arise from expansion of air in the droplets with a vapor-impervious
film or from air entrained in the liquid feed.35
Further, declining diffusion back into the drop and
capillary action on suspended solids will concentrate the dissolved material in a spherical shell.
Air incorporation during atomization is another
possible cause of vacuoles in particles.36 Lowdensity powders are more likely to be produced
from a two-fluid atomizer with internal mixing
(such as ours) than with external mixing, as the
possibility for air incorporation is increased.37
Contradictory information on whether an overor underpressure is formed in particles during
spray drying is found in the literature.38 In particles where the drying film is impervious to vapor,
an overpressure is created and blowholes or exploded particles may occur. Films formed by
hydrophilic materials, such as lactose, could, however, be permeable to water, and an underpressure
is more likely. The broken particles observed in
spray-dried lactose were always the largest of each
distribution. No blowholes were observed. Thus,
the particles probably fractured due to insufficient
shell thickness.
The hollow interior in combination with the
minor size differences between particles of various
solid contents gave the conclusion that the density
of particles from 520% w/w lactose solutions was
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

908

ELVERSSON ET AL.

Figure 8. (a) ESEM micrograph of spray-dried 20%

w/w lactose particle revealing its hollow interior. The
particle shell is smooth and compact with a defined
thickness; (b) TEM micrograph of 1% w/w lactose
particle illustrating a defined shell and a hollow interior.
Outer and inner shell surface are marked with arrows.

considerably lower than 1.47 g/cm3 (material density of amorphous lactose) and that the shell
thickness increased with concentration of the feed
solution.
It might be assumed that a critical concentration (Ccrit) is needed, in the outer layer of the
drying droplet, to establish a particle. At Ccrit, the
diffusion of solvent, over the drying surface, is
lower than the heat transfer; thus, the drying rate
declines (falling-rate period). Droplets from concentrated solutions will reach Ccrit earlier and the
particle size will be larger than for less concentrated solutions. Assuming the experimental
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

median diameter was equal to the droplet size at

Ccrit, then a particle with a size of 5 mm, prepared
from a 20% w/w lactose solution, would have a
lactose concentration of 56% w/w. Particle diameters calculated from Ccrit 56% resulted in
sizes corresponding well to experimental data for
the small nozzle orifice. However, for the large
orifice, calculated diameters were much larger
than experimental diameters. To explain this
deviation in some of the experiments, a review of
droplet and particle distribution data was undertaken. It was noted that in some cases, considerable loss of material (up to 53%), on the walls of the
drying tower and below the tower, occurred during
spray drying. The loss of material increased when
the large nozzle was used and as the solids content
of the feed increased. Ideally, a wider tower should
have been used to avoid droplets colliding with the
walls from the increase of the spray angle as result
of increased orifice diameter. The present study
showed that droplet size increased at the edges of
the spray (Figure 3); thus, it is reasonable to
believe that a population of mostly large droplets
was lost at the walls. This was confirmed by comparing the size of particles from walls and from the
collecting flask with particles collected from the
filter, by laser diffraction and ESEM (data not
shown). In addition, an increased solids content of
the feed might increase the tendency for droplets
to coalescence, on collision with each other or the
wall, due to higher impact force.
Considering the loss of large particles on drying,
Ccrit was recalculated from corrected droplet
diameters (see Data analysis). The corrected Ccrit
was 24% w/w, and the predicted values of particle
diameter corresponded well with the experimental
data on particle diameter (Figure 5), although a
crude method was used to correct the droplet size.
Moreover, it is interesting to note that Ccrit is
similar to the solubility of lactose (28% w/w) at the
wet bulb temperature corresponding to the operating conditions (approximately 458C). It is reasonable to assume that a minimal thickness of the
lactose film or shell is needed to form a mechanically strong particle, but it is unclear whether
the shell thickness is similar in the whole size
distribution of one sample. If particles are expanded during drying a distribution of shell thicknesses is more likely. According to this reasoning,
a median shell thickness was calculated from the
experimental and predicted median particle diameters (Table 3). The predicted shell thickness
increased from 0.07 to 0.30 mm as the solids content of the feed increased. The measured shell

DROPLET AND PARTICLE SIZE RELATIONSHIP OF INHALABLE LACTOSE PARTICLES

Table 3. Predicted Shell Thickness of Particles at a

Ccrit of 24%
Predicted Shell Thickness (mm)
Solids Content
(% w/w)
1
5
10
20

Orifice
1.5 mm

Orifice
2.0 mm

0.07
0.10
0.14
0.30

0.14
0.20
0.22
0.27

thickness for the 1% w/w particle in Figure 8b

corresponded well to the predicted values. It was
also observed that shell thickness varied within a
particle. However, from the ESEM micrograph of
the broken 20% particle (Figure 8a), it was
observed the relative shell thickness (shell thickness/ particle diameter) was considerably higher
than the predicted values. Observations of broken
particles indicated that relative shell thickness
varied over a wide range, suggesting that individual particles can differ substantially in density,
even though being prepared from the same feed
solution. It is well known that the drying conditions are different at different locations in the
spray.19 Thus, drying conditions experienced by
individual droplets vary considerably. Furthermore, the drying rate is influenced by the concentration of the droplet, the internal transport
conditions in the droplet, as well as interactions
between droplets. This indicates that it is likely
that droplets will reach their final size in a critical
concentration range, rather than at a defined
critical concentration.

CONCLUSIONS
As a result of the present study on the relationship between droplet and particle size during
spray drying, a mechanism describing the formation of hollow particles during spray drying
is proposed. The particle formation was complex
as drying proceeded in several stages. First, increasing droplet size during atomization increased
the particle size almost linearly. Second, increasing concentration of the feed also increased the
particle size but the effect was not linear. This
may partly be explained by lower yields at higher
feed concentrations but may also be related to
differences in drying rate. Nozzle orifice diameter
and atomization airflow, but not solids content of

909

the feed, controlled the size of droplets during

atomization. Third, a proposed critical concentration of the drying droplet determined both size
and density of particles and thus increasing the
feed concentration increased the particles shell
thickness. In view of this, it is likely that the
droplet size and concentration of the feed as well
as the choice of drying conditions, solvents, and
solutes control not only the size but also the
density of spray-dried particles, such as lactose.

ACKNOWLEDGMENTS
Nina Andersson at YKI, Institute for Surface
Chemistry, is gratefully acknowledged for the
TEM analysis. The authors also thank Dr. Hans
Karlsson and Dr. Marten Svensson at AstraZeneca R&D Lund, Sweden, for contributions to
the discussion regarding inhalation particles and
Dr. Paul Smith at YKI, Institute for Surface
Chemistry, for linguistic advice. This work was
supported financially by AstraZeneca R&D Lund,
Sweden.

REFERENCES
1. Patton JS, Platz RM. 1992. Pulmonary delivery of
peptides and proteins for systemic action. Adv Drug
Del Rev 8:179196.
2. Wall DA. 1995. Pulmonary absorption of peptides
and proteins. Drug Del 2:120.
3. Kerr JB. 1999. Atlas of functional histology.
London: Mosby.
4. Ward GH, Schultz RK. 1995. Process-induced
crystallinity changes in albuterol sulfate and its
effect on powder physical stability. Pharm Res 12:
773779.
5. Feeley JC, York P, Sumby BS, Dicks H. 1998.
Determination of surface properties and flow characteristics of salbutamol sulphate, before and after
micronisation. Int J Pharm 172:8996.
6. Dalby RN, Naini V, Byron PR. 1992. Droplet drying
and electrostatic collection, a novel alternative to
convential communation techniques. J Biopharm
Sci 3:9199.
7. Adjei AL, Gupta PK. 1997. Developmental pharmaceutics of inhaled peptides and proteins. In:
Adjei AL, Gupta PK, editors. Inhalation delivery of
therapeutic peptides and proteins. New York:
Marcel Dekker Inc., pp 735770.
8. Edwards DA, Hanes J, Caponetti G, Hrkach J, BenJebria A, Eskew ML, Mintzes J, Deaver D, Lotan N,
Langer R. 1997. Large porous particles for pulmonary drug delivery. Science 276:18681871.
JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

910

ELVERSSON ET AL.

9. French DL, Edwards DA, Niven RW. 1996. The

influence of formulation on emission, deaggregation and deposition of dry powders for inhalation. J
Aerosol Sci 27:769783.
10. Li WI, Perzl M, Heyder J, Langer R, Brain JD,
Englemeier KH, Niven RW, Edwards DA. 1996.
Aerodynamics and aerosol particle deaggregation
phenomena in model oral-pharyngeal cavities. J
Aerosol Sci 27:12691286.
11. Hardy JG, Chadwick TS. 2000. Sustained release
drug delivery to the lungs. Clin Pharmacokinet 39:
14.
12. Warheit DB, Hartsky MA. 1993. Role of alveolar
macrophage chemotaxis and phagocytosis in pulmonary clearance responses to inhaled particles:
Comparisons among rodent species. Microsc Res
Technol 26:412422.
13. Maa YF, Nguyen PA, Sweeney T, Shire SJ, Hsu CC.
1999. Protein inhalation powders: Spray drying vs
spray freeze drying. Pharm Res 16:249254.
14. Tom JW, Debenedetti PG. 1991. Formation of
bioerodible polymeric microspheres and microparticles by rapid expansion of supercritical solutions.
Biotechnol Prog 7:403411.
15. Hanna M, York P, Mazen H. 1995. Method and
apparatus for the formation of particles. University
of Bradford, GB, Pat. US5851453.
16. Moshashaee S, Bisrat M, Forbes RT, Nyqvist H,
York P. 2000. Supercritical fluid processing of
proteins I: Lysozyme precipitation from organic
solution. Eur J Pharm Sci 11:239245.
17. Allison SD, Chang B, Randolph TW, Carpenter JF.
1999. Hydrogen bonding between sugar and protein is responsible for inhibition of dehydrationinduced protein unfolding. Arch Biochem Biophys
365:289298.
18. Franks F, Hatley RHM, Mathias SF. 1991. Materials science and the production of shelf-stable
biologicals. BioPharm 4:38.
19. Masters K. 1991. Spray drying handbook, 5th ed.
Harlow: Longman Scientific & Technical.
20. Faldt P, Bergenstahl B. 1996. Changes in surface
composition of the spray-dried food powders due to
lactose chrystallization. Lebensm Wiss Technol 29:
438446.
21. Lefebvre AH. 1989. Atomization and sprays. New
York: Hemispere Publishing Corporation.
22. Filkova I, Cedlik P. 1984. Nozzle atomization in
spray drying. Adv Drying 3:181215.
23. Hickey AJ, Concessio NM, Van Oort MM, Platz
RM. 1994. Factors influencing the dispersion of dry
powders as aerosols. Pharm Technol August:5864.

JOURNAL OF PHARMACEUTICAL SCIENCES, VOL. 92, NO. 4, APRIL 2003

24. Maa YF, Nguyen PAT, Hsu SW. 1998. Spraydrying of air-liquid interface sensitive recombinant
human growth hormone. J Pharm Sci 87:152
159.
25. Dunbar CA, Concessio NM, Hickey AJ. 1998.
Evaluation of atomizer performance in production
of respirable spray-dried particles. Pharm Dev
Technol 3:433441.
26. Cassidy OE, Carter PA, Rowley G, Merrifield DR.
2000. Triboelectrification of spray-dried lactose
prepared from different feedstock concentrations.
J Pharm Pharmacol 52:1317.
27. Fell JT, Newton JM. 1971. The production and
properties of spray dried lactose: Part 2. Pharm
Acta Helv 46:425430.
28. van Mil PJJM, Hols G, Klok HJ. 1988. Spray drying
of concentrated milk: Relation between initial
droplet size and fine particle size. In: Bruin S,
editor. Preconcentration and drying of food materials. Amsterdam: Elsevier Science Publishers B.V.,
pp 193202.
29. Marshall WRJ. 1954. Atomization and spray drying. New York: American Institute of Chemical
Engineers.
30. Chawla JM. 1994. Effect of the droplet agglomeration on the design of spray dryer towers. Dry
Technol 12:13571365.
31. Perez-Correa JR, Faras F. 1995. Modelling and
control of a spray dryer: A simulation study. Food
Control 6:219227.
32. Hassan HM, Mumford CJ. 1993. Mechanisms of
drying of skin-forming materials. 3. Droplets of
natural products. Dry Technol 11:17651782.
33. Meerdink G. 1993. Drying of liquid food droplets:
Enzyme inactivation and multicomponent diffusion. PhD thesis, Agricultural University Wageningen, The Netherlands.
34. Christensen KL, Pedersen GP, Kristensen HG.
2001. Preparation of redispersible dry emulsions by
spray drying. Int J Pharm 212:187194.
35. Duffie JA, Marshall WR Jr. 1953. Factors influencing the properties of spray-dried materials, Part I.
Chem Eng Progr 49:417423.
36. Verhey JGP. 1972. Vacuole formation in spray
powder particles. I. Air incorporation and bubble
expansion. Neth Milk Diary J 26:186202.
37. Verhey JGP. 1972. Vacuole formation in spray
powder particles. II. Location and prevention of air
incorporation. Neth Milk Diary J 26:203224.
38. Crosby EJ, Marshall WR. 1958. Effects of drying
conditions on the properties of spray-dried particles. Chem Eng Progress 54:5663.