Complete Thesis

University of Groningen
Improving protein stabilization by spray drying

Grasmeijer, Niels
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from
it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date:
2016
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):

Grasmeijer, N. (2016). Improving protein stabilization by spray drying: formulation and process
development. [Thesis fully internal (DIV), University of Groningen]. University of Groningen.
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the
author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
The publication may also be distributed here under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license.
More information can be found on the University of Groningen website: https://www.rug.nl/library/open-access/self-archiving-pure/taverne-
amendment.
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately
and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the
number of authors shown on this cover page is limited to 10 maximum.
Download date: 05-11-2023

Improving protein stabilization by spray drying
Formulation and process development
Niels Grasmeijer
Paranimfen
Floris Grasmeijer
Maryse Grasmeijer
The research presented in this PhD thesis was performed at the Department of
Pharmaceutical Technology and Biopharmacy of the University of Groningen and falls
within the research program of the Groningen University Institute for Drug Exploration
(GUIDE).
Cover: Scanning Electron Microscope (SEM) image of inulin particles prepared

by spray drying, taken by Anko Eissens.
ISBN: 978-90-367-8519-8 (printed version)

ISBN: 978-90-367-8518-1 (electronic version)
Printing: Ridderprint BV, The Netherlands

Printing of this thesis was supported by the University of Groningen
© Copyright 2015, Niels Grasmeijer. All rights reserved. No part of this thesis may be
reproduced in any form or by any means without prior permission of the author.
Improving protein stabilization by
spray drying
Formulation and process development
Proefschrift
ter verkrijging van de graad van doctor aan de

Rijksuniversiteit Groningen
op gezag van de
rector magnificus prof. dr. E. Sterken
en volgens besluit van het College voor Promoties.
De openbare verdediging zal plaatsvinden op
maandag 11 januari 2016 om 11:00 uur
door
Niels Grasmeijer
geboren op 22 juni 1984

te Groningen
Promotor
Prof. dr. H.W. Frijlink
Copromotor
Dr. W.L.J. Hinrichs
Beoordelingscommissie
Prof. dr. F. Picchioni
Prof. dr. ir. W.E. Hennink
Prof. dr. G. van den Mooter
Table of contents
Chapter 1 Introduction 3
Chapter 2 Unraveling protein stabilization mechanisms: vitrification 13

and water replacement in a glass transition temperature
controlled system
Chapter 3 Identifying critical process steps to protein stability during 35

spray drying using an ultrasonic or a two-fluid nozzle
Chapter 4 A user-friendly model for spray drying to aid pharmaceutical 57

product development
Chapter 5 An adaptable model for growth and/or shrinkage of droplets 83

in the respiratory tract during inhalation of aqueous particles
Chapter 6 Model to predict inhomogeneous protein-sugar distribution 113

in powders prepared by spray drying
Chapter 7 General discussion and perspectives 135
Appendix A Summary 141
Appendix B Samenvatting 147
Appendix C Model discussed in chapter 5 153
Appendix D Model discussed in chapter 6 163
Appendix E Biography and list of publications 179
Appendix F Dankwoord 183
1
2 | Chapter 1
Chapter 1
Introduction
Introduction | 3
Biopharmaceuticals, such as therapeutic proteins, peptides, and vaccines, have been
around for quite some time, with the first recombinant biopharmaceutical being
accepted in 1982. Some of the most well-known examples are insulin for diabetes
patients and influenza vaccines against the flu. Especially the latter has received a lot
of attention around the time of the swine flu H1N1 outbreak around 2009. Despite a
slowdown in the increase of the number of approved applications of
biopharmaceuticals in the last few years, they remain an important group of
therapeutics for diseases that are otherwise untreatable [1].
Most biopharmaceuticals are administered via injection or infusion using an
aqueous solution. This solution can either be formulated directly as an aqueous
solution, or reconstituted from a dried formulation. Despite efforts to stabilize the
biopharmaceuticals in aqueous solutions using excipients, these solutions still have a
very limited shelf life and are only suitable for immediate use unless stored under
refrigerated conditions. This often poses a problem in developing countries where
refrigerated transport and storage, the so called ‘cold chain’, is too expensive, is prone
to failures, wasting product (even in western countries) and makes stockpiling (e.g. of
vaccines for pandemics) more expensive as well. Furthermore, for reconstitution from
a dried formulation, drying biopharmaceuticals often results in poor storage stability.
An optimal solution to this problem would be to formulate a dry powder in
which the biopharmaceutical is stable and retains its activity for prolonged periods
under ambient storage conditions. Not only would this solve the aforementioned
problems, but also have the added advantage of being able to keep a larger stockpile.
This makes preparing larger batches of rarely used materials a lot easier and more cost
effective. Moreover, the development of stable dry powder products would allow for
the use of alternative dosage forms and routes of administration. There has, for
example, been an increasing interest in inhalation therapies using dry powder inhalers,
which could then be used for local or systemic administration of biopharmaceuticals.
Naturally, powder suitable for reconstitution for injection/infusion therapies will also
benefit from the development of stable dry powder products containing
biopharmaceuticals. However, the challenge is not so much to obtain a dry powder; a
simple case of e.g. evaporation or sublimation by spray or freeze drying, respectively,
would accomplish this. However, maintaining the bioactivity of the biopharmaceutical
during drying and subsequent storage is unfortunately a completely different story.
There are many factors that may cause a biopharmaceutical to lose its activity
during drying and subsequent storage. In this thesis we are mostly referring to
proteins, since this is the major building block of many biopharmaceuticals. In general,
the activity of proteins depends on their 3-dimensional structure, which allows them
to fit onto other molecules to perform some sort of function. A classic example is that
of a (non-therapeutic) enzyme such as amylase, which will fit onto a molecule (starch)
4 | Chapter 1
to catalyse or speed up some reactions, in the case of amylase the breakdown into
smaller molecules such as glucose. When its 3-D structure is somehow changed, the
enzyme activity can be partially or fully lost. Moreover, it may result in undesired
immune responses.
Among the physical and chemical processes which can change the structure
of a protein in the dry state, denaturation is one of the more frequently occurring [2].
Denaturation occurs mainly due to the difference in hydrophilicity of the core and
surface of the protein. Particularly globular proteins will often have more of the non-
polar amino acids towards the core and polar amino acids near the surface, which
improves the solubility of the protein. However, when the protein is dried, the
conditions of the outside are often far less favourable for hydrophilic surfaces and
could actually make it thermodynamically favourable for the hydrophobic core to be
on the outside. And so, in time, the protein will start to unfold and lose its specific
tertiary and secondary structure, and with that also its activity. To prevent this from
happening one could naturally come up with two solutions: either make sure that the
environment around the protein is hydrophilic, or simply prevent or reduce molecular
mobility of the protein. As with many other problems though, nature provides a
solution that covers both: inulin and trehalose.
In nature, there are many plants and organisms that are desiccation resistant.
Upon desiccation, a primary response is to synthesize an increasing amount of
trehalose, which is a disaccharide [3-5]. The trehalose aids in protecting the proteins
and cell membrane by acting as a hydrophilic replacement for water, thereby
maintaining their essential structural features. Furthermore, by incorporating the
protein in a trehalose matrix, the molecular mobility of the protein is reduced as well.
Finally, although trehalose is used as an example, there are several other sugars that
are known to stabilize proteins, amongst which inulin is a promising alternative. It is
important to realize that the sugar should be in its immobile glassy state, as there is
not enough close contact between the protein and the sugar in its crystalline state, and
the translational molecular mobility of the sugar matrix is too high when it is in its
rubbery state (which can also result in crystallization of the sugar). The transition
between both amorphous states occurs at the glass transition temperature; at
temperatures below the glass transition temperature the sugar will be in its glassy state,
and at temperatures above in its rubbery state. Two mechanisms have been proposed
to explain protein stabilization by sugar glasses; i.e. the water replacement and
vitrification theory [6, 7]. However, the exact mechanism and the relative importance
of both mechanisms is not yet fully understood. Therefore, we set out to elucidate the
difference in importance between both the water replacement and vitrification theory
for the stabilizing effect of glassy sugars in chapter 2. Further understanding of these
mechanisms can aid in choosing a suitable formulation for protein stabilization.
Introduction | 5
Besides choosing the correct formulation for optimal storage stability, the
process stability during drying will have to be considered as well. There are several
suitable drying processes that can be used to produce stabilized protein powders. The
most well-known are freeze drying, where the protein solution is frozen and the
solvent is subsequently sublimated under vacuum, and spray drying, where the protein
solution is aerosolized in heated air to evaporate the solvent. From a stability
perspective, freeze drying is often preferred over spray drying due to the absence of
thermal and shear stresses on the protein during drying and this process has been the
focus of many studies [8-19]. Most proteins are quite sensitive and will degrade when
exposed to heat, shear, drying, and interfacial stresses. However, because with freeze
drying the protein is processed under low temperature conditions, denaturation due to
heat does obviously not occur. In addition, the physical characteristics of the
produced powder, such as flow properties and stickiness, are often of less importance,
since the powder itself is in general not transported during the drying process.
However, although using heat to produce stable protein powders might sound
counterintuitive, spray drying can have some interesting advantages over freeze drying,
both from a process as well as a product perspective.
Because spray drying can be executed as a continuous process, it is often
more cost effective on an industrial scale than freeze drying, which is a batch process,
there is a reduced chance of batch to batch variability, processing times are much
shorter, and is more energy efficient. In addition, whereas freeze dried powders are
very difficult to process due to their exceptionally poor flow properties, spray dried
powders may have good flow properties and can be processed and transported more
easily. Finally, from a formulation perspective, it is advantageous that powders can be
produced with a size that is suitable for further processing into dosage forms such as
for example dry powders for inhalation [20].
Unfortunately, the spray drying process does not treat a sensitive protein in
solution gently. If we take a look at our spray dryer, the protein solution is forcefully
pushed through tubing, in close contact to heated air, broken up into droplets through
intensive agitation also resulting in large solvent/air interfaces, subjected to hot air,
stripped from its aqueous environment, and finally left in the harsh conditions at the
outlet for the remainder of the process. However, where and how much the protein
will degrade during spray drying may not only depend on the type of protein, but also
on the configuration of the spray dryer and the used process conditions. Here, it is
important to realize that spray drying is basically a combination of several different
downstream process steps, such as atomization and drying, and each of these process
steps can be changed or improved. While there are numerous studies focused on
protein stability loss during spray drying, many consider the spray dryer as a single
process and focus more on the formulation of the protein solution. However, as
6 | Chapter 1
indicated above, it is also necessary to try and determine the deteriorating effects to
the protein during each step of the whole spray drying process. Doing so will result in
a better understanding on how to develop a spray drying process for protein
stabilization purposes specifically. Therefore, in chapter 3, we try to break down the
spray dryer into different process steps and try to determine the most damaging step
for a specific protein, lactate-dehydrogenase, which was selected based on its heat and
shear sensitivity.
After choosing the formulation and drying process, the drying process
parameters will have to be chosen as well. For spray drying, important parameters
(availability depends on the spray dryer configuration) are the inlet temperature,
atomization air flow, drying air flow, liquid feed flow, and concentration of the
solution. These inlet parameters combined, result in a certain outlet condition that
should be suitable for the desired product. For example, if a sugar glass stabilized
protein powder is intended for inhalation, the outlet conditions should be such that
the temperature will not exceed the glass transition temperature. Because the glass
transition temperature can be lowered by residual moisture, this requires a balance
between relative humidity and temperature at the outlet. This already limits the choice
of inlet parameters that can be chosen. Furthermore, for inhalation, the aerodynamic
diameter of the dry powder should fall within a specified range to prevent throat
deposition of too large particles and exhalation of too small particles. Therefore, the
choice in concentration of the solution, liquid feed flow, and atomization air flow is
also limited. These limitations together form the borders of the design space, and if
one chooses the inlet parameters within this design space, the resulting product will
have the required product specifications.
Such a design space is often obtained with a trial-and-error approach.
Unfortunately, due to the many process parameters that can influence the product
whilst spray drying, this often requires too many experiments that require a lot of time
and money. To aid in this process, the use of a mathematical model that describes the
relation between inlet and outlet parameters of the spray dryer and can predict the
outlet conditions, would help immensely in minimizing the number of empirical
experiments required to optimize the process. Although spray dryer models have been
developed, these models were made available in a shape or form that is mainly suitable
for computational or chemical engineers and less for pharmaceutical technologists
[21-25]. Considering the sheer number of choices for open access publishing, and the
ease with which data can be shared over the internet, this is a missed opportunity.
Therefore, in chapter 4, we set out to develop a user-friendly model to aid
pharmaceutical product development that will be made available freely to everyone.
As mentioned previously, the spray dryer can be seen as a sequence of several
different process steps that eventually make up the whole spray drying process.
Introduction | 7
Although in chapter 3 we try to describe these process steps, one can always zoom in
further. One thing that has fascinated many about research is the different scales in
which a process can be observed. Now we are looking at a spray dryer, and if we
zoom out, we can look at the lab, earth, galaxies, and so on. Vice versa we can also
zoom in further than as described in chapter 3 and look at the processes that make up
the spray dryer, and perhaps even further. However, the level of fundamental
understanding of a process dictates how far we can zoom in on a process and still
accurately describe it. One of the most interesting process steps that may significantly
affect protein stabilization is the drying step, which occurs when the protein solution
is aerosolized in the heated air. In this stage of the process, the sugar matrix around
the protein is formed, and therefore, where storage stability of the protein is
determined. It is also a very complex process that lends itself very well for
mathematical modelling, as measurements on single droplets tend to be rather
cumbersome. In addition, developing a model that is flexible, not process specific, and
able to accurately describe the drying process in general, would also be more widely
applicable, for example in drug inhalation studies. Finally, although models exist to
describe the drying of droplets, none were developed in software that is freely
available and shared in a way that others can easily expand upon it.
In chapter 5 we set out to develop a model that simulates the mass and heat
transfer that occurs between a droplet and the air surrounding. In these simulations
water droplets are considered in which no solutes are dissolved. This model has been
applied to predict what will happen with a droplet when it is inhaled and travels
through the lung. We hope this model will be used by scientists as a starting point for
expansion to increase its accuracy and the application to other processes where
evaporation plays and important role. Furthermore, in chapter 6 we discuss and show
a way to further expand the model to include solutes to also make it suitable for spray
drying protein solutions. With this model the distribution of protein and sugar in
powders prepared by spray drying is studied.
References
1. Walsh G. Biopharmaceutical benchmarks 2014. Nat Biotech.

2014;32(10):992-1000.
2. Chang LQ, Pikal MJ. Mechanisms of protein stabilization in the solid state. J
Pharm Sci. 2009;98(9):2886-908.
3. Zhang Q, Yan T. Correlation of Intracellular Trehalose Concentration with
Desiccation Resistance of Soil Escherichia coli Populations. Appl Environ
Microbiol. 2012;78(20):7407-13.
8 | Chapter 1
4. Leslie SB, Israeli E, Lighthart B, Crowe JH, Crowe LM. Trehalose and
sucrose protect both membranes and proteins in intact bacteria during drying.
Appl Environ Microbiol. 1995;61(10):3592-7.
5. Potts M. Desiccation tolerance of prokaryotes. Microbiol Rev.
1994;58(4):755-805.
6. Carpenter JF, Crowe JH. An infrared spectroscopic study of the interactions
of carbohydrates with dried proteins. Biochemistry (Mosc). 1989;28(9):3916-
22.
7. Hancock BC, Shamblin SL, Zografi G. Molecular Mobility of Amorphous
Pharmaceutical Solids Below Their Glass Transition Temperatures. Pharm
Res. 1995;12(6):799-806.
8. Suzuki T, Imamura K, Yamamoto K, Satoh T, Okazaki M. Thermal
stabilization of freeze-dried enzymes by sugars. J Chem Eng Jpn.
1997;30(4):609-13.
9. Kawai K, Suzuki T. Stabilizing effect of four types of disaccharide on the
enzymatic activity of freeze-dried lactate dehydrogenase: Step by step
evaluation from freezing to storage. Pharm Res. 2007;24(10):1883-90.
10. Chang L, Shepherd D, Sun J, Ouellette D, Grant KL, Tang X, et al.
Mechanism of protein stabilization by sugars during freeze-drying and
storage: Native structure preservation, specific interaction, and/or
immobilization in a glassy matrix? J Pharm Sci. 2005;94(7):1427-44.
11. Wang B, Tchessalov S, Cicerone MT, Warne NW, Pikal MJ. Impact of
sucrose level on storage stability of proteins in freeze-dried solids: II.
Correlation of aggregation rate with protein structure and molecular mobility.
J Pharm Sci. 2009;98(9):3145-66.
12. Ohtake S, Schebor C, Palecek SP, de Pablo JJ. Effect of pH, counter ion, and
phosphate concentration on the glass transition temperature of freeze-dried
sugar-phosphate mixtures. Pharm Res. 2004;21(9):1615-21.
13. Luthra S, Obert JP, Kalcinia DS, Pikal MJ. Impact of critical process and
formulation parameters affecting in-process stability of lactate dehydrogenase
during the secondary drying stage of lyophilization: A mini freeze dryer study.
J Pharm Sci. 2007;96(9):2242-50.
14. Pikal MJ, Dellerman KM, Roy ML, Riggin RM. The Effects of Formulation
Variables on the Stability of Freeze-Dried Human Growth Hormone. Pharm
Res. 1991;8(4):427-36.
15. Abdul-Fattah AM, Dellerman KM, Bogner RH, Pikal MJ. The effect of
annealing on the stability of amorphous solids: Chemical stability of freeze-
dried moxalactam. J Pharm Sci. 2007;96(5):1237-50.
Introduction | 9
16. Chang BS, Beauvais RM, Dong A, Carpenter JF. Physical Factors Affecting
the Storage Stability of Freeze-Dried Interleukin-1 Receptor Antagonist:
Glass Transition and Protein Conformation. Arch Biochem Biophys.
1996;331(2):249-58.
17. Amorij JP, Meulenaar J, Hinrichs WLJ, Stegmann T, Huckriede A, Coenen F,
et al. Rational design of an influenza subunit vaccine powder with sugar glass
technology: Preventing conformational changes of haemagglutinin during
freezing and freeze-drying. Vaccine. 2007;25(35):6447-57.
18. Pikal MJ, Rigsbee D, Roy ML, Galreath D, Kovach KJ, Wang B, et al. Solid
state chemistry of proteins: II. The correlation of storage stability of freeze-
dried human growth hormone (hGH) with structure and dynamics in the
glassy solid. J Pharm Sci. 2008;97(12):5106-21.
19. Wang B, Tchessalov S, Warne NW, Pikal MJ. Impact of sucrose level on
storage stability of proteins in freeze-dried solids: I. correlation of protein–
sugar interaction with native structure preservation. J Pharm Sci.
2009;98(9):3131-44.
20. Saluja V, Amorij JP, Kapteyn JC, de Boer AH, Frijlink HW, Hinrichs WLJ. A
comparison between spray drying and spray freeze drying to produce an
influenza subunit vaccine powder for inhalation. J Control Release.
2010;144(2):127-33.
21. Ivey JW, Vehring R. The use of modeling in spray drying of emulsions and
suspensions accelerates formulation and process development. Computers
& Chemical Engineering. 2010;34(7):1036-40.
22. Baldinger A, Clerdent L, Rantanen J, Yang M, Grohganz H. Quality by design
approach in the optimization of the spray-drying process. Pharm Dev
Technol. 2012;17(4):389-97.
23. Dobry D, Settell D, Baumann J, Ray R, Graham L, Beyerinck R. A model-
based methodology for spray-drying process development. Journal of
Pharmaceutical Innovation. 2009;4(3):133-42.
24. Lebrun P, Krier F, Mantanus J, Grohganz H, Yang M, Rozet E, et al. Design
space approach in the optimization of the spray-drying process. Eur J Pharm
Biopharm. 2012;80(1):226-34.
25. Mezhericher M, Levy A, Borde I. Spray drying modelling based on advanced
droplet drying kinetics. Chemical Engineering and Processing: Process
Intensification. 2010;49(11):1205-13.
10 | Chapter 1
Introduction | 11
12 | Chapter 2
Chapter 2
Unraveling protein stabilization mechanisms: vitrification and water

replacement in a glass transition temperature controlled system
N. Grasmeijer, M. Stankovic, H. de Waard, H.W. Frijlink, W.L.J. Hinrichs
doi:10.1016/j.bbapap.2013.01.020
Unraveling protein stabilization mechanisms | 13

Abstract
The aim of this study was to elucidate the role of the two main mechanisms used to
explain the stabilization of proteins by sugar glasses during drying and subsequent
storage: the vitrification and the water replacement theory. Although in literature
protein stability is often attributed to either vitrification or water replacement, both
mechanisms could play a role and they should be considered simultaneously. A model
protein, alkaline phosphatase, was incorporated in either inulin or trehalose by spray
drying. To study the storage stability at different glass transition temperatures, a buffer
which acts as a plasticizer, ammediol, was incorporated in the sugar glasses. At low
glass transition temperatures (<50 ºC), the enzymatic activity of the protein strongly
decreased during storage at 60 °C. Protein stability increased when the glass transition
temperature was raised considerably above the storage temperature. This increased
stability could be attributed to vitrification. A further increase of the glass transition
temperature did not further improve stability. In conclusion, vitrification plays a
dominant role in stabilization at glass transition temperatures up to 10 to 20 °C above
storage temperature, depending on whether trehalose or inulin is used. On the other
hand, the water replacement mechanism predominately determines stability at higher
glass transition temperatures.
14 | Chapter 2
Introduction
Proteins are applied more and more as therapeutic agents. Unfortunately, many of
these proteins are labile and need to be stabilized. Proteins can be stabilized by drying
a sugar-containing solution of the protein, thereby incorporating the protein into a
matrix of the sugar in the glassy state. Two mechanisms of stabilization have been
described, namely the water replacement theory and the glass dynamics hypothesis or
vitrification theory [1-3]. Although in literature protein stability is often attributed to
either vitrification or water replacement, both could have an effect and should be
considered simultaneously.
The water replacement theory states that in solution, the conformation of the
protein is maintained by the interaction with water, mainly due to hydrogen bonding.
Upon drying, this interaction is lost and replaced by hydrogen bonds between the
protein and the sugar by which the protein structure is maintained upon drying [4]. To
maximize the hydrogen bonding with the protein, the sugar molecules should closely
fit the irregular surface of the protein and should thus be in the amorphous state (and
not in the crystalline state). A closely related theory is the water entrapment theory,
which states that rather than forming hydrogen bonds directly with the sugar, the
protein is coupled to the amorphous sugar matrix through water molecules entrapped
at the interface [5]. Although the water replacement theory and water entrapment
theory describe fundamentally different mechanisms, the stabilization in both theories
is mediated by hydrogen bonding. It is not within the aim of this study to discern
between these two theories and therefore, further discussion regarding the water
replacement theory could also hold true for the water entrapment theory, unless stated
otherwise.
The second theory, the vitrification theory, dictates that the protein will be
immobilized inside the sugar matrix, preventing translational molecular movements
and thereby degradation. The translational molecular mobility within the sample is
mainly determined by the thermodynamic state of the sugar. For an amorphous sugar,
the translational molecular mobility is determined by the difference between the glass
transition temperature and the storage temperature. Although the translational
molecular mobility of an amorphous sugar in its glassy state is greatly reduced
compared to a sugar in its rubbery state, it is not completely absent. As a rule of
thumb, the glass transition temperature should be at least 50 °C above the storage
temperature for the translational molecular mobility to become “insignificant over the
lifetime of a typical pharmaceutical product” [6].
Although such a rule of thumb is useful for designing a stable protein
formulation, further optimization requires more detailed knowledge of the importance
of the glass transition temperature for protein stabilization. Even though it is generally
accepted that the glass transition temperature is important for the stability of the
protein, there is little detailed overview of the relation between the stability and the
glass transition temperature of a given system [7]. To investigate this, typical 3-
component systems, consisting of a model protein (alkaline phosphatase), a buffer
(ammediol), and either inulin or trehalose, were spray dried. These sugars were
selected because trehalose is often considered as the golden standard in protein
stabilization [8], while previous studies have shown that inulin also has good
stabilizing properties [9-13]. Besides the fact that both sugars have a high glass
transition temperature, both sugars are also hydrophilic, have good hydrogen bonding
capacity, are non-toxic and have no reducing groups, which makes them excellent
candidates for the stabilization of proteins [10]. Ammediol was chosen as a buffer
because it is also a good glass former, just like the sugars. Due the low glass transition
temperature of ammediol, it will act as a plasticizer [14]. This enables adjustment of
the glass transition temperature, allowing the stability to be determined for a wide
range of glass transition temperatures, which in turn enabled us to study different
system mobilities, without changing the type of stabilizing sugar. In this study, we
investigated whether this strategy can be used to study the different roles in protein
stabilization of the water replacement and vitrification mechanism.
Materials and methods
Materials
Alkaline phosphatase from bovine intestinal mucosa (10-30 Units/mg, ~160 kDa),
bovine serum albumin, ammediol, and para-nitrophenylphosphate were obtained from
Sigma-Aldrich Co. (St. Louis, Missouri). Magnesium chloride was purchased from
Fluka Chemie GmbH (Buchs, Switzerland). Trehalose was obtained from Cargill B.V.
(Amsterdam, The Netherlands) and inulin with a degree of polymerization of 23 from
Sensus (Roosendaal, The Netherlands). All experiments were performed with
millipore water, type 1.
Spray drying
Protein containing powders were produced by dissolving alkaline phosphatase at a
concentration of 2.5 mg/mL in a 50 mM ammediol at pH 9.8. Either inulin or
trehalose was added in varying concentrations to obtain a range of samples with
different sugar/buffer mass ratios and thus with different glass transition
temperatures. The solutions were spray dried with a B-290 spray dryer (Büchi
Labortechnik AG, Flawil, Switzerland). The inlet air temperature was set at 100 °C,
the aspirator at 100%, pump speed at 7%, and atomizing airflow at 50 mm. These
settings were chosen such, that the outlet temperature (around 58-64 °C) was similar
to the storage temperature. Samples were stored in a desiccator over silica for one
16 | Chapter 2
night at room temperature and then immediately analyzed or stored at 60 °C for
various periods of time. To minimize the moisture content of the samples, open vials
were used during storage, while the relative humidity varied between 4.5% and 6%.
The composition of the different dried powders is shown in Table 1.
Table 1: Stabilized protein sample compositions with varying glass transition temperature
Component mass fractions
Protein sugar
mass ratio 1/1 1/3 1/5 1/7 1/10 1/12 1/15 1/20
Sugar 0.24 0.49 0.62 0.69 0.76 0.79 0.83 0.87
Protein 0.24 0.16 0.12 0.10 0.08 0.07 0.06 0.04
Buffer 0.51 0.34 0.26 0.21 0.16 0.14 0.12 0.09
Differential Scanning Calorimetry (DSC)

Modulated DSC measurements were done with a Q2000 and DSC 2920 differential
scanning calorimeter (TA Instruments, New Castle, United States). Dry samples were
measured after spray drying, while humidified inulin and trehalose sugar glasses were
prepared by storing the samples at a relative humidity of 22%, 33%, and 52% in a
desiccator over a saturated aqueous solution of CH3COOK, MgCl2.6H2O, and
Na2Cr2O7.2H2O, respectively, or at 45%, and 60% in a climate chamber for 1-3 weeks.
Dry samples were weighed in open aluminum pans at ambient conditions and then
preheated at 70 °C for 5 minutes to remove any moisture, after which scanning was
performed at 2.5 °C/min modulated with a period of 40 seconds and amplitude of 0.8
°C. Humidified samples were weighed in closed aluminum pans, then cooled to -20
°C, and finally heated at a rate of 2 °C/min with a modulation period of 60 seconds
and amplitude of 0.316 °C. The glass transition temperature was taken as the
inflection point of the transition.
Enzymatic Activity of Alkaline Phosphatase

The enzymatic activity of alkaline phosphatase was determined using a kinetic assay
based on a method described by Eriksson et al. [14]. The sample was reconstituted in
water to a protein concentration of about 0.03 mg/mL, of which 20 µL was
transferred in triplicate to a BSA coated 96 wells plate, together with 190 µL 50 mM
ammediol with 0.1 mg/mL MgCl2. After warming the solution to 37 °C, 20 µL of a 5
mg/mL p-NPP solution was added to start the reaction. The absorption at a
wavelength of 415 nm was measured on a Benchmark plate reader (BioRad, Hercules,
California). Samples were mixed for 15 seconds and measured every 30 seconds for

5.5 minutes. To calculate the remaining activity, the rate of conversion was compared
to a standard of unprocessed alkaline phosphatase.
Dynamic Vapor Sorption (DVS) analysis

The water sorption isotherms of different powders were measured at ambient
pressure and 25 °C using a DVS-1000 water sorption instrument (Surface
Measurement Systems Limited, London, UK). To estimate the glass transition
temperature of protein containing samples, the moisture content of protein/ammediol
containing powders was determined at a relative humidity of 4.5% or 6.0%, dependent
on the conditions during the stability study, for samples with an initial mass of
approximately 35 mg. To determine the influence of water on the glass transition
temperature of sugar glasses, the moisture content of humidified sugar glasses without
protein/ammediol was measured at relative humidity’s of 0% – 90% with steps of
10%, for samples with an initial mass of around 10 mg. After subjecting the samples
to the specified humidity equilibrium was assumed when the change in mass was less
than 0.9 µg during 10 minutes.
Results
Glass transition temperature

In an attempt to determine the glass transition temperature by DSC, it was found that
alkaline phosphatase interfered with the measurements. DSC measurements
performed with different modulations or higher scan rates of up to 20 °C/min, did
not improve the result. Therefore, the glass transition temperature of the sugar/buffer
matrix, without the protein, was used as the glass transition temperature of the
sample, thereby neglecting the influence of the protein on the glass transition
temperature. However, it is well known that the glass transition temperature can be
greatly influenced by residual moisture or moisture adsorbed from the environment,
which should be taken into account by estimating the glass transition temperature of a
sugar-buffer-water mixture [15]. This was achieved by using the Gordon-Taylor
equation, extended for a ternary mixture (Equation 1), which describes the relation
between the composition of a three component mixture (with mass fractions ws, wb,
and ww) and its glass transition temperature (Tg) [16]. Besides the mass fraction of the
components, the glass transition temperature of the mixture is also dependent on the
glass transition temperature of the individual components (Tg.s, Tg.b, and Tg.w) and two
component-dependent Gordon-Taylor constants (ksb and ksw). The subscripts s, b, and
w are used for sugar, buffer, and water, respectively. To calculate the glass transition
temperature of both the inulin-ammediol-water and trehalose-ammediol-water
18 | Chapter 2
mixtures, the glass transition temperatures of the components, the Gordon-Taylor
constants and the mass fraction of water were determined.
ws Tg.s  k sb  w b  Tg.b  k sw w w Tg.w

Tg (1)
ws  k sb  wb  k sw w w
The glass transition temperature of pure inulin, trehalose, and ammediol was
determined with DSC and found to be 155 °C, 121 °C, and -53 °C, respectively. For
water, a glass transition temperature of -109 °C was used, which is the average of
recently published values [17-19]. Although this value is substantially higher than the
conventionally accepted value of -137 °C [17-19], our calculations indicate that the
choice of either of these glass transition temperatures of water did not have large
influence on the calculated glass transition temperature of the final samples (data not
shown).
The Gordon-Taylor constant, ksb, was determined for both the inulin-
ammediol and trehalose-ammediol mixtures by fitting the Gordon-Taylor equation,
for binary mixtures with ww = 0, with glass transition temperatures of different
compositions measured with DSC. The best fit was determined by using the least
squares method. It was noted that glass transition temperatures below 90 °C were
difficult to determine, due to large shifts in the baseline. These shifts can most likely
be attributed to a highly energetic solid-solid phase transition of the ammediol buffer
at 80 °C, as described by M. Barrio et al. [20]. If the glass transition temperature of the
system is close to the temperature at which this solid-solid phase transition occurs, the
translational molecular mobility might be high enough to allow the phase transition to
start before the glass transition temperature is reached, in turn distorting the
measurement and obscuring the glass transition temperature. Therefore, to determine
the Gordon-Taylor constants, only samples with a glass transition temperature higher
than 90 °C were used.
The Gordon-Taylor constants for the inulin-ammediol and trehalose-
ammediol samples were found to be 1.87 and 1.64, respectively. The correlation
coefficient of 0.999 and 1.000 for both inulin-ammediol (n=8 points) and trehalose-
ammediol (n=8 points) samples, respectively, indicate a perfect fit of the Gordon-
Taylor equation with the experimental data, showing that the buffer was
monomolecularly dispersed within the sugar.
The Gordon-Taylor constant, ksw, for trehalose-water and inulin-water
mixtures was also determined by fitting the Gordon-Taylor equation for binary
mixtures, with wb = 0, with glass transition temperatures of humidified sugar glasses
measured with DSC. The moisture content of sugar glasses that were humidified at

specific relative humidities was determined with DVS analysis. The Gordon-Taylor
constants for the trehalose-water and inulin-water mixtures were found to be 7.90 and
6.57, respectively. The Gordon-Taylor constant for a trehalose-water mixture was
found to be higher than values found in literature (i.e. 5.2, 6.5, and 7.3), due to the
higher glass transition temperature of water we used [21-23].
The mass fraction of water, ww, in the samples was estimated by relating DVS
data to the relative humidity inside the chamber used for storing the samples. The
relative humidity inside the storage chamber was calculated from relative humidity
measurements outside the storage chamber by using the Antoine equation and the
ideal gas law (Equation 2 and Equation 3, respectively).
B
A
C T (2)
p w.sat 10
xw   R ( T  273.15)
pw (3)
Mw
Where pw.sat is the saturated water vapor pressure, A, B, and C are the Antoine
constants of water (10.20, 1730.63, and 233.43, respectively [24]), T is the temperature
(°C), pw is the partial water vapor pressure (Pa), xw is the specific humidity (defined as
the mass of water divided by the mass of dry air), ρ is the density of air, R is the gas
constant, and Mw is the molecular mass of water. The two equations enable the
calculation of the relative humidity (%) (defined as pw/pw.sat·100). When the relative
humidity outside the storage chamber is measured, the specific humidity of the
ambient air can be calculated by multiplying by pw.sat/100 at ambient temperature, and
subsequently dividing by ρ·R·(T+273.15)/Mw. Because the storage chamber is well
ventilated with the ambient air, the humidity inside the storage chamber is equal to
the humidity of the ambient air. However, since the air is heated inside the storage
chamber, the relative humidity decreases. This can be calculated by first calculating pw
at 60 °C, and subsequently the relative humidity by dividing 100·pw by pw.sat at 60 °C.
For example, since the average relative humidity of the ambient air during one of the
experiments was 53% at 20 °C, the relative humidity inside the storage chamber was
approximately 6%. Knowing the relative humidity inside the storage chamber, the
moisture content of the samples could be determined using the water vapor isotherm
obtained with DVS. To take into account any possible influence of protein on water
vapor sorption, DVS was performed with protein containing samples. The moisture
content of the samples was found to be between 1.2 - 2.5% by mass. Subsequently,
20 | Chapter 2
the glass transition temperature of both inulin-ammediol-water and trehalose-
ammediol-water mixtures was calculated with the Gordon-Taylor equation (Equation
1). The glass transition temperatures as calculated with the Gordon-Taylor equation
were used to estimate the different glass transition temperatures of the prepared
protein samples. These results are presented in Table 2.
Table 2: Estimated glass transition temperatures for stabilized protein samples

Glass transition temperature (°C)
protein sugar
mass ratioa 1/1 1/3 1/5 1/7 1/10 1/12 1/15 1/20
Inulin -16 28 51 64 76 81 86 89
Trehalose -20 18 37 49 59 64 69 73
a The glass transition temperature was determined for samples without protein
Varying glass transition temperature

To investigate the contribution of both the vitrification and water replacement
mechanism on the stability of the protein at different glass transition temperatures,
samples with varying glass transition temperatures (and thus varying translational
molecular mobility) were prepared. The enzymatic activity of the protein was
measured after storing these samples in a desiccator overnight and after storing them
for 12 days at 60 °C.
Figure 1: Activity of alkaline phosphatase incorporated in inulin (■) and trehalose (▲) after spray
drying (──) and after 12 days storage at 60 °C (- - -), (n = 3 ± SD).

Samples with the lowest glass transition temperature, below 0 °C, already lost
most of their activity during spray drying (Figure 1). However, when the glass
transition temperature is raised above 30-40 °C (close to the spray dryer outlet
temperature of around 60 °C), the loss of activity was strongly reduced and became
negligible at values over 45 °C. After storing the spray dried samples for 12 days at 60
°C, the stability curve shifted to the right, indicating that a higher glass transition
temperature is required for protein stabilization during storage. Complete activity loss
was observed for samples with a glass transition temperature below 50-60 °C, while at
higher glass transition temperatures the activity loss quickly decreased. Although the
shift is observed for both sugars, trehalose appeared to stabilize the protein already at
somewhat lower glass transition temperatures than inulin.
To investigate the shift of the stability curve in time, inulin samples with
varying glass transition temperatures were stored for various periods of time. After the
first 5 days of storage at 60 °C a similar shift of the stability curve was observed
(Figure 2) as after 12 days of storage (Figure 1). Prolonged storage time intervals did
not reveal any further shift of the curve, but rather an overall decrease of activity
independent from the glass transition temperature. This overall activity loss was most
apparent for samples with a higher glass transition temperature. These higher glass
transition temperatures, in the range of 76 to 89 °C, appear to increase stability only
slightly. This range will further be referred to as the “plateau phase”.
Figure 2: Inulin stabilized protein activity after spray drying (■) and upon storage at 60 °C for 5
(■), 12 (▲), 19 (▲), 26 (X), 33 (X), and 40 (O) days, (n = 3 ± SD).
22 | Chapter 2
Constant glass transition temperature
The decrease of activity within the plateau phase was investigated in more detail. This
was done by measuring the activity loss of samples with varying protein content and a
constant glass transition temperature during storage at 60 °C. The protein content was
varied by preparing samples with varying protein/sugar mass ratios of 1/1, 1/10, and
1/20 (Table 3). A fixed buffer/sugar mass ratio of 0.20 and 0.13 was chosen for inulin
samples to obtain a glass transition temperature of around 80 °C and 93 °C,
respectively. Trehalose samples with similar glass transition temperatures were
obtained by choosing a buffer/sugar mass ratio of 0.06 and 0.02, respectively.
Table 3: Stabilized protein sample compositions with constant glass transition temperature
Component mass fractions
Trehalose (Tg = 81 °C) Trehalose (Tg = 91 °C)
Protein sugar
mass ratio 1/1 1/10 1/20 1/1 1/10 1/20
Trehalose 0.48 0.86 0.90 0.50 0.89 0.93
Protein 0.48 0.09 0.04 0.50 0.09 0.05
Buffer 0.03 0.05 0.06 0.01 0.02 0.02
Inulin (Tg = 78 °C) Inulin (Tg = 95 °C)

Protein sugar
mass ratio 1/1 1/10 1/20 1/1 1/10 1/20
Inulin 0.45 0.77 0.80 0.47 0.81 0.85
Protein 0.45 0.08 0.04 0.47 0.08 0.04
Buffer 0.09 0.15 0.16 0.06 0.11 0.11
Since the stability data of the samples with varying protein/sugar ratios (for
both inulin and trehalose samples) were not significantly different, the results were
combined (Figure 3), which is justified by the overall small standard deviation (<5%).
This shows that, within the range tested, the stability of the protein is independent
from the protein content of the sample. Furthermore, Figure 3 shows that the activity
loss for inulin samples was larger than for the trehalose samples. After 19 days the
activity loss for trehalose samples was almost 10%, while for inulin samples almost
30% was lost. Finally, there is no difference in the activity loss between samples with a
glass transition temperature of 80 °C and 93 °C when the same sugar was used.

Figure 3: Alkaline phosphatase activity upon storage at 60 °C of trehalose (▲) and inulin (■)
samples with a glass transition temperature of around 80 °C (──) and 93 °C (- - -). Data points are
an average of three samples with varying protein/sugar mass ratios of 1/1, 1/10, and 1/20 and scaled to
100% at 0 days (original values were between 100-107%), (n = 9 ± SD).
Discussion
In this study the Gordon-Taylor equation was used to calculate the glass transition
temperature of the samples containing protein, based on DSC data from samples
without protein. It could be argued that the presented glass transition temperatures are
not a reflection of the actual values due to the fact that they are based on a mixture of
the sugar, ammediol, and water, thereby neglecting the presence of the protein.
Especially for the high protein containing samples (having the lowest glass transition
temperatures) this raises questions. However, there are indications in literature that
even a high mass fraction of protein influences the glass transition temperature to only
a limited extent. Imamura et al. showed that the glass transition temperature of a
trehalose glass was increased by 9 °C when the mass content of bovine serum albumin
was increased from 20% to 50%, while the change in glass transition temperature at
lower protein contents was only marginal [25]. Bellavia et al. also investigated the
effect of the protein content on the glass transition temperature and the applicability
of the Gordon-Taylor equation [26]. It was shown that, at high protein contents, the
glass transition temperature of ternary sugar-protein-water mixtures was higher than
predicted by a binary Gordon-Taylor curve for a sugar-water mixture, without protein.
Up to 40% the effect of protein mass content on the glass transition temperature was
only minor. Only at a mass content of 76% a substantial effect was found. Therefore,
24 | Chapter 2
it can be assumed that the addition of protein will only moderately influence the glass
transition temperature of the samples with a protein/sugar mass ratio of 1/1,
corresponding to a protein mass content of 24%. Moreover, the protein activity in the
24% samples was completely lost after storage at 60 °C and conclusions of our study
were not based on these samples but on samples with much lower protein content.
Therefore, it is justified to consider the glass transition temperatures as obtained from
the Gordon-Taylor equation for ternary systems representative for the glass transition
temperatures of the protein containing samples (Table 2).
The employed three-component system for protein stabilization clearly shows
two different phases. In the first phase, protein stability increases rapidly with
increasing glass transition temperature. In the second phase (the plateau phase) the
stability is independent from the glass transition temperature of the sugar matrix but
varies depending on the sugar applied.
In general, the results show an increased stability with an increasing glass
transition temperature up to 10-20 °C above the storage temperature. This
observation is in line with the vitrification mechanism. When the glass transition
temperature of the sugar is below the storage temperature, the material will be in a
mobile rubbery state, which will enable the protein to degrade and lose its activity.
This effect is observed both during spray drying and storage. Of course, considering
the variation in sugar content of the samples, one could argue that a minimum amount
of sugar is required to saturate the surface of the protein, thereby maximizing protein
stability by forming as many hydrogen bonds as possible. Although it is tempting to
explain the lack of stability of samples with a low glass transition temperature, having
a protein/sugar mass ratio of 1/1 and 1/3, by the low amount of sugar, the low
standard deviation shown in Figure 3 shows that the protein stability is independent
of the protein/sugar mass ratio within the range tested. Therefore, the relatively low
sugar content of the samples with a protein/sugar mass ratio of 1/1 and 1/3 cannot
account for the activity loss of these samples, clearly pointing to the glass transition
temperature as the major determinant of the stability in this phase.
During the relatively short spray drying process, the maximum temperature of
the dried samples is assumed to be that of the outlet temperature, which was 58-64
°C. However, even samples with a glass transition temperature as low as 30 °C
exhibited quite good process stability, whereas during storage at 60°C for 12 days, the
protein activity of the samples with a glass transition temperature below storage
temperature was completely lost. Apparently, within the short spray drying time, even
the samples with a glass transition temperature up to 30 °C below the outlet
temperature did not degrade fast enough to show any loss of activity.
When the glass transition temperature is raised above storage temperature, the
amorphous sugar will change to the less mobile glassy state, which should keep the

protein vitrified in its original conformation. Although the onset of protein
stabilization appeared to occur when the glass transition temperature was around the
storage temperature, it was also shown that the protein was still not fully stabilized
immediately after the glass transition temperature was raised above the storage
temperature. The stability appeared to increase exponentially over a 20-30 °C
temperature range around the storage temperature before reaching a plateau phase,
showing maximum stabilization. This finding is in agreement with the result of a study
by Hancock et al. (6), who showed that in the range between the glass transition
temperature and 50 °C below glass transition temperature, the translational molecular
mobility strongly decreases. In addition, from the data presented in the same study,
one can conclude that the change of the translational molecular mobility was much
faster for the relatively small sucrose than for the much larger PVP K90. This resulted
in a difference in mobility of several orders of magnitude when the storage
temperature was 50 °C below the glass transition temperature. A similar difference
was found in this study. The stability of the protein when incorporated in trehalose
increased faster with increasing glass transition temperature than when the protein was
incorporated in the larger inulin. The same phenomenon was also observed in more
detail by others. Wolkers et al. [27, 28] applied FT-IR to show that the wavenumber
temperature coefficient of the OH-stretching mode band position increased with
increasing molecular mass, indicating that the average hydrogen bond length increases
with increasing molecular mass. This results in a more loosely packed glass and a
higher mobility of the sugar matrix.
It should, however, be kept in mind that in order to obtain a glass transition
temperature similar to the trehalose samples, the inulin samples contained higher
amounts of buffer. Therefore, one could argue that the stability difference found to
exist between both sugars was caused by the replacement of sugar with buffer, which
could result in a less favorable interaction with the protein. However, under that
assumption, this could be said for the entire stability curve and this would imply that
the loss of activity at lower glass transition temperatures is not due to increased
translational molecular mobility but due to replacement of sugar by the buffer. In
other words, the protein stability would under such assumption be dominated entirely
by water replacement and never by vitrification. However, it seems unlikely that this is
the case. Although the same difference in stability between inulin and trehalose is
observed at high glass transition temperatures where mobility is expected to be
negligible, at lower glass transition temperatures the rate of activity loss is much faster
and is more likely to be caused by a higher mobility. In addition, the facts that the
sudden exponential increase of stability occurs only at (I) glass transition temperatures
around the storage temperature and (II) shows a good correlation with results from
26 | Chapter 2
literature, strongly suggests that protein stability is dominated by mobility and thus
vitrification at lower glass transition temperatures.
Therefore, the fact that for maximum protein stability a slightly higher glass
transition temperature is required with inulin than with trehalose, should be attributed
to their difference in molecular size. While trehalose has a molecular mass of 342
g/mol, the type of inulin used in this study is a long chained oligosaccharide with a
molecular mass of 3908 g/mol. Therefore, the translational molecular mobility of a
trehalose glass is likely to be smaller than the translational molecular mobility of an
inulin glass, resulting in the slight difference in protein stability with trehalose and
inulin at glass transition temperatures where the mobility dictates the protein stability.
These findings confirm that the protein stability at temperatures near the glass
transition temperature is largely dependent on the translational molecular mobility of
the sugar.
It is important to be aware of a study by Francia et al., which reported an
increased rigidity of a trehalose matrix when its water content is decreased [29]. The
increased rigidity of the matrix was attributed to the anchorage hypothesis, which links
the water replacement theory with the water entrapment theory, stating that at low
hydration levels the water and trehalose molecules form a continuous matrix with a
high rigidity. Therefore, the increased stability of samples when their glass transition
temperature is increased to above storage temperature could have simply been
coinciding with a decrease of the water content in the mixtures. However, the water
content in the trehalose and inulin mixtures used in this study showed only little
variation for samples with a glass transition temperature between 40 °C below and 40
°C above the storage temperature. With an average water/sugar mass ratio of
approximately 0.02, the standard deviation was only around 10%. Therefore the
anchorage mechanism does not appear to play a role in this system, which further
reinforces the statement that the stability increase can be attributed to the vitrification
theory.
At high glass transition temperatures, where the protein stability reaches the
plateau phase, the activity loss was found to be independent from the glass transition
temperature. Instead, a steady decline in activity was observed over time, both for
inulin and trehalose samples. As is shown in the previous section, the protein can be
considered immobile at these glass transition temperatures. At these high glass
transition temperatures degradation can only take place when there is either enough
free volume or insufficient interaction between the protein and the sugar (i.e.
hydrogen bonding). Therefore, the primary stabilization mechanism in this region is
most likely water replacement. This hypothesis is supported by the fact that the
activity loss with inulin is higher than with trehalose. If stabilization is indeed realized
by water replacement, then the difference in activity loss between inulin and trehalose

must be caused by a difference in the ability of the sugar to form hydrogen bonds with
the protein. Since it is likely that the small trehalose molecule will be able to fit the
irregular protein surface more closely than the larger inulin molecules, thereby
forming more hydrogen bonds with the protein, the resulting stability will be higher.
Although the question whether the interaction between the protein and the sugar is
direct or facilitated by a hydration layer is not relevant for the current discussion, an
interesting observation can be made regarding this subject. If water molecules would
be entrapped at the interface between protein and sugar, an influence of the
water/protein molar ratio on the stability would be expected. However, no difference
in stability was observed between samples with a protein/sugar ratio of 1, 10, and 20
(Figure 3), while the water/protein molar ratio differed substantially (3200, 1600, and
160, respectively). Although it is unknown whether there is a minimum water/protein
ratio above which the stability is no longer affected, the results seem to correspond
more with the water replacement mechanism rather than the water entrapment
mechanism.
In addition to the water entrapment theory, another interesting perspective
on the topic is given by the slaving model as proposed by Frauenfelder et al.[30]. In
their study, a clear physical explanation is given for the mobility of a protein
incorporated in a sugar glass. Instead of considering the mobility of the protein as
being autonomous, it is considered to be governed by the surrounding matrix
(“slaved”). Large scale motions of the protein are caused by mobility of the matrix
bulk, while internal protein motions are caused by local movements in the, several
molecules thick, layer surrounding the protein. These motions are designated as
primary (α) and secondary (β) fluctuations, respectively. When the viscosity of the
matrix bulk is low (rubbery state), α fluctuations cause large scale motions of the
protein, which then mainly determine the protein stability. This is in agreement with
the vitrification theory. However, at high viscosity (glassy state), α fluctuations in the
bulk are absent and the protein stability is mainly determined by local β fluctuations.
Although this is a different explanation than given by the water replacement theory,
both theories appear to be in agreement with the results of our study. Instead of
attributing the stability of the protein at high glass transition temperature to the
hydrogen bond formation with the matrix, it is instead attributed to the local mobility
of the surrounding, closely confined, molecules. Therefore, the difference in stability
at high glass transition temperature between trehalose and inulin should then be
explained by their difference in mobility, comparable to the vitrification theory.
If the stability at the high glass transition temperatures is indeed dominated by
the water replacement mechanism, it is important to consider the following. As
mentioned before, in order to compare the activity of trehalose and inulin samples at
the same glass transition temperatures, a relatively higher ammediol/sugar ratio was
28 | Chapter 2
required for inulin due to its inherently higher glass transition temperature of 155 °C.
It could be argued whether or not the protein stability with inulin was different from
trehalose due to a more prevalent interaction of the buffer with the protein and not
just a difference in molecular mass of the sugar. Studies have already shown that
plasticizers can even have a stabilizing effect when added in small amounts, despite
the decrease of the glass transition temperature [31, 32]. It is hypothesized that the
buffer can form hydrogen bonds with the protein at sites that have not been occupied
by the sugar, further stabilizing the protein until the reduced glass transition
temperature increases the mobility of the sugar matrix to an extent that enables the
protein to degrade. Although in our case ammediol did not seem to have a stabilizing
effect on the protein, it is more than likely that a higher ammediol content will cause
replacement of part of the sugar, interacting with the protein, by ammediol instead. If
ammediol is not as good a stabilizer as the sugar, this could decrease the stability of
the protein. However, although it remains unclear whether the difference in stability
with inulin and trehalose is mainly caused by the difference in ammediol content or
molecular mass, both causes are consistent with the water replacement theory.
Conclusion
In conclusion, it is shown that at glass transition temperatures below the storage

temperature protein stabilization is dominated by the vitrification mechanism.
However, when the glass transition temperature is raised significantly above the
storage temperature, the protein becomes immobile and the water replacement
mechanism becomes the dominant mechanism for protein stabilization.
Acknowledgments
This research forms part of the Project P3.02 DESIRE of the research program of the
BioMedical Materials institute, co-funded by the Dutch Ministry of Economic Affairs,
Agriculture and Innovation. This project is part-financed by the European Union,
European Regional Development Fund and The Ministry of Economic Affairs,
Agriculture and Innovation, Peaks in the Delta
References
Pharm Sci. 2009;98(9):2886-908.

3. Sampedro J, Uribe S. Trehalose-enzyme interactions result in structure
stabilization and activity inhibition. The role of viscosity. Mol Cell Biochem.
2004;256-257(1):319-27.
22.
5. Belton PS, Gil AM. IR and Raman spectroscopic studies of the interaction of
trehalose with hen egg white lysozyme. Biopolymers. 1994;34(7):957-61.
Res. 1995;12(6):799-806.
7. Yu L. Pharmaceutical Quality by Design: Product and Process Development,
Understanding, and Control. Pharm Res. 2008;25(4):781-91.
8. Jain NK, Roy I. Effect of trehalose on protein structure. Protein Sci.
2009;18(1):24-36.
9. Kawai K, Suzuki T. Stabilizing effect of four types of disaccharide on the
enzymatic activity of freeze-dried lactate dehydrogenase: Step by step
evaluation from freezing to storage. Pharm Res. 2007;24(10):1883-90.
10. Hinrichs WLJ, Prinsen MG, Frijlink HW. Inulin glasses for the stabilization
of therapeutic proteins. Int J Pharm. 2001;215(1-2):163-74.
11. Amorij JP, Meulenaar J, Hinrichs WLJ, Stegmann T, Huckriede A, Coenen F,
et al. Rational design of an influenza subunit vaccine powder with sugar glass
technology: Preventing conformational changes of haemagglutinin during
freezing and freeze-drying. Vaccine. 2007;25(35):6447-57.
12. Zijlstra GS, J. Ponsioen B, A. Hummel S, Sanders N, Hinrichs WLJ, de Boer
AH, et al. Formulation and process development of (recombinant human)
deoxyribonuclease I as a powder for inhalation. Pharm Dev Technol.
2009;14(4):358-68.
13. Rodríguez Furlán LT, Padilla AP, Campderrós ME. Inulin like lyoprotectant
of bovine plasma proteins concentrated by ultrafiltration. Food Research
International. 2010;43(3):788-96.
14. Eriksson JHC, Hinrichs WLJ, de Jong GJ, Somsen GW, Frijlink HW.
Investigations into the stabilization of drugs by sugar glasses: III. The
influence of various high-pH buffers. Pharm Res. 2003;20(9):1437-43.
15. Hancock BC, Zografi G. The relationship between the glass transition
temperature and the water content of amorphous pharmaceutical solids.
Pharm Res. 1994;11(4):471-7.
30 | Chapter 2
16. Gordon M, Taylor JS. Ideal copolymers and the second-order transitions of
synthetic rubbers. i. non-crystalline copolymers. Journal of Applied
Chemistry. 1952;2(9):493-500.
17. Velikov V, Borick S, Angell CA. The glass transition of water, based on
hyperquenching experiments. Science. 2001;294(5550):2335-8.
18. Miller AA. Glass-transition temperature of water. Science.
1969;163(3873):1325-6.
19. Giovambattista N, Angell CA, Sciortino F, Stanley HE. Glass-Transition
Temperature of Water: A Simulation Study. Phys Rev Lett.
2004;93(4):047801.
20. Barrio M, Font J, López D, Muntasell J, Tamarit J, Cossio F. Polymorphism
in 2-Amino-2-Methyl-l,3 propanediol plastic crystal. Journal of Phase
Equilibria. 1991;12(4):409-15.
21. Crowe LM, Reid DS, Crowe JH. Is trehalose special for preserving dry
biomaterials? Biophys J. 1996;71(4):2087-93.
22. Chen T, Fowler A, Toner M. Literature Review: Supplemented Phase
Diagram of the Trehalose–Water Binary Mixture. Cryobiology.
2000;40(3):277-82.
23. Roos Y. Melting and glass transitions of low molecular weight carbohydrates.
Carbohydr Res. 1993;238(0):39-48.
24. Yaws CL, Yang HC. To estimate vapor pressure easily. Hydrocarbon
Processing. 1989;68(10):65-70.
25. Imamura K, Ohyama K-i, Yokoyama T, Maruyama Y, Kazuhiro N.
Temperature scanning FTIR analysis of secondary structures of proteins
embedded in amorphous sugar matrix. J Pharm Sci. 2009;98(9):3088-98.
26. Bellavia G, Giuffrida S, Cottone G, Cupane A, Cordone L. Protein Thermal
Denaturation and Matrix Glass Transition in Different
Protein−Trehalose−Water Systems. The Journal of Physical Chemistry B.
2011;115(19):6340-6.
27. Wolkers WF, Oliver AE, Tablin F, Crowe JH. A Fourier-transform infrared
spectroscopy study of sugar glasses. Carbohydr Res. 2004;339(6):1077-85.
28. Shirke S, Ludescher RD. Molecular mobility and the glass transition in
amorphous glucose, maltose, and maltotriose. Carbohydr Res.
2005;340(17):2654-60.
29. Francia F, Dezi M, Mallardi A, Palazzo G, Cordone L, Venturoli G.
Protein−Matrix Coupling/Uncoupling in “Dry” Systems of Photosynthetic
Reaction Center Embedded in Trehalose/Sucrose: The Origin of Trehalose
Peculiarity. J Am Chem Soc. 2008;130(31):10240-6.

30. Frauenfelder H, Chen G, Berendzen J, Fenimore PW, Jansson H, McMahon
BH, et al. A unified model of protein dynamics. Proceedings of the National
Academy of Sciences. 2009;106(13):5129-34.
31. Chang L, Shepherd D, Sun J, Tang X, Pikal MJ. Effect of sorbitol and
residual moisture on the stability of lyophilized antibodies: Implications for
the mechanism of protein stabilization in the solid state. J Pharm Sci.
2005;94(7):1445-55.
32. Cicerone MT, Soles CL. Fast Dynamics and Stabilization of Proteins: Binary
Glasses of Trehalose and Glycerol. Biophys J. 2004;86(6):3836-45.
32 | Chapter 2
34 | Chapter 3
Chapter 3
Identifying critical process steps to protein stability during spray drying

using an ultrasonic or a two-fluid nozzle
Niels Grasmeijer, Valeria Tiraboschi, Henderik W. Frijlink, Wouter L.J.

Hinrichs
Critical process steps to protein stability during spray drying | 35

Abstract
The aim of this study was to identify the critical steps to protein stability during spray
drying using two different nozzle types. To achieve this, lactate dehydrogenase was
spray dried with both an ultrasonic perforated mesh nozzle and a standard two-fluid
nozzle in a Büchi B-90 spray dryer. The whole spray drying process was broken down
into several smaller steps and after every step the enzymatic activity of the protein was
measured. It was found that with the ultrasonic nozzle, 78% of activity was lost, of
which about 68% was due to atomizing and heating. Dehydration and circulation of
the liquid together accounted for about 10% activity loss. However, with the two-fluid
nozzle the total activity loss was greatly reduced to 23%, to which atomization,
dehydration, and circulation contributed almost equally. Heating was not an issue due
to the possibility to cool the two-fluid nozzle with water. In conclusion, the type and
the configuration of the nozzle that was used was found to be of importance, since
heating, ultra-sonication, and recirculation of the feed solution were all found to
decrease the protein’s process stability. With the ultrasonic nozzle, atomizing and
heating caused a high activity loss, while with the cooled two-fluid nozzle the activity
loss was largely reduced. Whereas often only the formulation is optimized, this finding
can be used to also further optimize the spray drying process for protein stabilization.
36 | Chapter 3
Introduction
Therapeutic proteins are being used more and more in modern drug therapies.
Although therapeutic proteins are often administered by injection, protein solutions
have a few drawbacks. First and foremost, these solutions are often quite unstable and
require refrigerated conditions during storage and transport. This so-called cold chain
is expensive and troublesome in many third world countries.
Stable solid formulations of the proteins can be prepared by incorporating
them in a glassy sugar matrix by spray drying. The matrix stabilizes the protein by
vitrification as well as by water replacement [1-8]. While this method has been used
successfully to increase the storage stability of various proteins, the process stability
can still remain an issue, since the proteins will be in solution during a significant part
of the spray drying process. In the framework of process control and final product
quality it is therefore important to identify the critical steps in the spray drying process
that can negatively influence protein integrity.
A deeper understanding of the spray drying process can be obtained by
breaking the process down into several steps, namely: transportation of the liquid
solution to the spray head, heating of the liquid solution during transport, atomization
of the liquid solution into droplets, drying of the droplets, and collection of the dried
particles. In each of these steps, the protein is subjected to several stresses that may
cause degradation. Depending on the step, these stresses include, but are not limited
to, heat, shear, drying, and interfacial stresses [9-11].
These stresses are dependent on several spray dryer parameters, such as
temperature and feed flow, which have been researched in the past. However, by
breaking the process down into several steps, the influence of specific process
mechanisms can be investigated in more detail. For example, although it is known that
atomization can be detrimental to process stability, it is also possible to compare
different atomization mechanisms and look at the effect on process stability
throughout the entire spray dryer.
One very popular spray dryer nowadays is the B-90 spray dryer, which
employs an ultrasonic vibrating perforated mesh to disperse the protein solution into
the heated drying air. However, it is known from inhalation studies that the formation
of aerosols using an ultrasonic vibrating perforated mesh can be detrimental to
proteins in solution [12]. Therefore, it will be interesting to investigate whether the
process stability can be improved by using a different type of nozzle.
In this study, the influence of each of the aforementioned steps during spray
drying on the stability of the model protein lactate dehydrogenase (LDH) is
investigated using the spray drying process with two different types of nozzles: an
ultrasonic and a two-fluid nozzle. LDH is an ideal model protein in this case due to its

shear and temperature sensitive nature, which enhances the ability to study the
protein’s process stability in spray-drying processes [11, 13-15].
It should be noted that it is not the intention to optimize the spray drying
process for protein stabilization in this paper. Instead, aggressive spray dryer settings
will be used to make the critical process steps and the difference between nozzle types
more apparent. Although it is expected that this will result in overestimated losses of
protein activity compared to optimized settings, doing so will allow for a more
elaborate discussion regarding the points of attention when stabilizing proteins using a
spray dryer. This, and the relation to possibly optimized settings will be further
discussed.
Materials
L-Lactic dehydrogenase, Type XI from rabbit muscle, 600-1200 units/mg of protein,
bovine serum albumin, sodium pyruvate, and β-nicotonamide adenine dinucleotide,
reduced disodium salt hydrate were purchased from Sigma-Aldrich Co. (St. Louis,
Missouri). Trehalose was obtained from Cargill B.V. (Amsterdam, The Netherlands).
All experiments were performed with Millipore water, type 1.
Spray drying process

Spray drying experiments were performed using a B-90 spray dryer (Büchi
Labortechnik AG, Flawil, Switzerland) in open configuration. A solution of 0.01 wt-%
LDH and 2.5 wt-% trehalose in 10 mM phosphate buffer pH 7.5, filtered with a 0.22
µm PVDF filter was used for all experiments. For each experiment, 10 ml of protein
solution was prepared of which about 6 ml was spray dried after 30 minutes. The
spray dryer was used with either the standard ultrasonic nozzle or a two-fluid nozzle
taken from the B-290 spray dryer (Büchi Labortechnik AG, Flawil, Switzerland)
(Figure 1).
38 | Chapter 3
Figure 1: Schematic overview of the ultrasonic nozzle (left) and two-fluid nozzle (right) in the B-
90 spray dryer. Note that the liquid feed of the two-fluid nozzle is cooled with the cooling water. In
addition, with the ultrasonic nozzle, the liquid feed is only partially atomized and the remainder is
recirculated.
The ultrasonic nozzle was used in conjunction with a 4 µm perforated mesh

spray cap. In addition, the liquid feed was circulated through the ultrasonic nozzle
using the built-in peristaltic pump at a pump setting of 1. Apart from the short tygon
tubing inside the peristaltic pump, all tubing to and from the nozzle consisted of
PTFE to minimize protein adsorption. Finally, the spray was set at 100 %, resulting in
an average atomizing flow rate of 0.2 ml/min. Samples were collected after 30 minutes
of circulating without heating (pump) and circulating with heating (pump + heat),
circulating and atomizing without heating (pump + spray), and the whole drying
process (pump + heat + spray). Because of the recirculating feed system with the
ultrasonic nozzle, samples were also taken from the liquid feed solution after
atomizing without heating (pump + spray (feed)) and whole spray drying process
(pump + heat + spray (feed)). After each step that involved heating, the liquid feed
tubing were flushed with a solution containing sodium dodecyl sulfate, and
subsequently thoroughly rinsed with demineralized water.
The two-fluid nozzle used was equipped with a 0.7 mm diameter nozzle tip.
The liquid feed was transported to the nozzle using a NE300 syringe pump (New Era
Pump Systems Inc., Wantagh, NY, United States of America) set at 0.2 ml/min, to
match the liquid feed flow rate of the ultrasonic nozzle. To minimize protein

adsorption, PTFE tubing was used from the syringe to the tip of the two-fluid nozzle.
In addition, the two-fluid nozzle included a cooling mantle, which enabled cooling of
the solution to be spray dried during feeding. Cooling was established with water of 4-
8 °C which was circulated through the mantle by using the built-in peristaltic pump of
the B-90 spray dryer at speed setting 2. The temperature of the cooling water that
exited the spray dryer was measured at about 30 °C within a few minutes after starting
the process (pump + heat + spray). Finally, the atomizing airflow was provided by a
B-290 spray dryer and was set at 50 mm (600 ln/hr). Samples were collected after
feeding the liquid without heating (pump), atomization without heating (pump +
spray), and the whole spray drying process (pump + heat + spray) for 30 min. Due to
a lack of feed circulation in this configuration, no (pump + heat) samples could be
taken as was possible with the ultrasonic nozzle configuration.
To avoid activity loss during storage, the enzymatic activity was measured the
same day the samples were taken. In addition, collected samples were stored at 4-8 °C
until the enzymatic activity assay was performed.
In addition to the different protein solutions, a 2.5 wt-% solution of pure
trehalose, i.e. without protein, was spray dried using both nozzles to compare the
particle size distribution of powders produced by the two methods.
All experiments were performed with an inlet temperature of 120 °C and a
drying air flow of 150 l/min.
Separating process steps

The method used to calculate the activity loss in the different separated process steps:
transportation, heating, atomization, dehydration, and collection is shown in Table 1.
40 | Chapter 3
Table 1: Calculation method used to estimate activity loss in the different separated process
steps with the two spray dryer configurations using the enzymatic activity measurements.
Ultrasonic nozzle
Separated process step Calculation using activity loss after:
Circulating “Pump”
Heating “Pump + Heat” – “Pump”
Atomizing (feed) “Pump + Spray (feed)*” – “Pump”
Atomizing “Pump + Spray” – “Pump”
Dehydrating + “Pump + Heat + Spray” – Circulating – Heating –
Collecting Atomizing
Two-fluid nozzle
Separated process step Calculation using activity loss after:
Pumping “Pump”
Heating †
Atomizing “Pump + Spray” – “Pump”

Dehydrating +
Collecting + Heating “Pump + Heat + Spray” – Pumping – Atomizing
* Sample was taken from the liquid feed after spray drying, as not all of the feed solution was spray dried
† Heating step cannot be separated with the two-fluid nozzle, since no sample could be collected when
only pumping and heating (and not spraying) were applied
Laser diffraction analysis (LDA)

The particle size distribution of the spray dried trehalose was measured with laser
diffraction analysis. Measurements were performed using a HELOS laser diffraction
sensor with an R3 (0.5/0.9-175 µm) and R1 (0.1/0.18-35 µm) lens, and a RODOS for
powder dispersion (Sympatec GmbH, Clausthal-Zellerfeld, Germany) at 3 bar. Laser
diffraction data were analyzed with the Fraunhofer method. Measurements with the
R3 lens were also performed at 5 bar and gave similar results, indicating that 3 bar was
sufficient for the dispersion of powders produced by spray drying.
Scanning electron microscopy (SEM)

SEM-pictures of spray dried trehalose were taken with a JEOL JSM 6301-F
Microscope (JEOL, Tokyo, Japan). The powders were dispersed on top of double-
sided adhesive carbon tape on aluminum disks and coated with a 10 nm layer of
gold/palladium in a Balzers 120B sputtering device (Balzers UNION, Balzers,
Liechtenstein). An acceleration voltage of 1.5 kV and probe current of 6 was used for
all pictures.

Enzymatic activity assay
LDH enzymatic activity of collected samples was determined directly after spray
drying as described by Bergmeyer et al [16], modified for a 96-wells plate.
Measurements were performed on a BioTek Synergy HT Multi-Mode Microplate
Reader (BioTek Instruments Inc., Bad Friedrichshall, Germany). As a reference, a
sample was taken of the protein solution right before spray drying (starting solution)
and normalized to 100 % activity.
Results
The volume particle size distribution of the trehalose solution spray dried with the
ultrasonic and two-fluid nozzle as measured with laser diffraction were very similar
(Table 2). The volume median diameter (x50) of powders prepared by spray drying
with the ultrasonic and two-fluid nozzle was found to be close to 1 µm and the
difference with the R3 lens between the ultrasonic and two-fluid nozzle was only 0.01
µm. When the powders were analyzed with the R1 lens, which can measure particles
down to 0.1 µm instead of 0.5 µm with the R3 lens, the particle size distribution only
changed slightly. This implies that the detection limit of both lenses was low enough
for a precise measurement of the particle size distribution of the powders.
Table 2: LDA results with R3 and R1 lens for powders produced by spray drying with either the
ultrasonic or the two-fluid nozzle (n=2, less than 5 % deviation between duplicate
measurements).
R3 lens
Nozzle type x10 (µm) x50 (µm) x90 (µm) span* A/V
Ultrasonic 0.60 1.03 2.46 1.81 5.81
Two-fluid 0.60 1.04 2.32 1.65 5.83
R1 lens
Nozzle type x10 (µm) x50 (µm) x90 (µm) span* A/V
Ultrasonic 0.42 0.97 2.02 1.66 7.67
Two-fluid 0.38 1.05 2.25 1.78 8.04
* Span was defined as (x90 – x10) / x50.
With the laser diffraction data, the surface area per volume (A/V) was
calculated for both powders (Table 2). This was done by taking into account the entire
detector range (31 rings) of the laser diffraction as shown in Equation 1. Here, n
indicates the detector ring, Dn and Dn-1 are the particle diameter associated with
42 | Chapter 3
detector ring n and n-1, respectively, and Φcn and Φcn-1 are the cumulative volume
fraction of particles with a diameter smaller than Dn and Dn-1, respectively.
31  
   Dn  Dn 1 
2
A /V           (1)

n 1 
1     D  D 3 c n cn 1

 6 n n 1 
It was found that the surface area per volume was similar for both powders
with both the R1 and R3 lens, which indicates that the total interfacial area of the
particles and perhaps also the droplets is the same with both nozzles. However, to
determine whether the results for the dried particles can be directly extrapolated to the
droplets from which they originate, both powders need to have a similar morphology.
SEM images indicate that indeed the morphology of the particles was similar and close
to spherical (Figure 2). This is important when the particle size distribution is
determined with laser diffraction, in which it is assumed that the particles are perfect
spheres. Furthermore, because the drying conditions in the spray dryer were similar
with both nozzle types, it is reasonable to assume that the droplets generated by both
nozzle types were similar as well. Therefore, it is reasonable to conclude that the
droplet size distribution and consequently also the surface area per volume and thus
the total interfacial area of droplets generated by both nozzle types were similar.
Figure 2: SEM images of trehalose powders produced with the ultrasonic (left) and two-fluid
(right) nozzle at a magnification of 5000x.
Each of the different process steps during spray drying had a clear and
significant effect on the enzymatic activity of LDH for both the ultrasonic and the
two-fluid nozzle (Figure 3). When the ultrasonic nozzle was used, circulation of the

feed through a tube inside the non-heated drying chamber without spraying (pump)
already resulted in an activity loss of about 14 %. When the spray dryer was also
heated (pump + heat), the activity loss increased to about 36 %. When the circulating
feed was only sprayed, but not heated (pump + spray), about 43 % of LDH activity
was lost in the atomized droplets and about 19 % in the feed solution. Finally, when
the spray dryer was both heated and the feed was sprayed (pump + heat + spray), the
dried LDH had lost about 78 % of its initial activity and the feed solution about 90 %.
Figure 3: Remaining enzymatic activity of LDH after specific spray drying process steps with an
ultrasonic (light grey) or a two-fluid (dark grey) nozzle. Data shown as averages, n=3 ±SD.
When the two-fluid nozzle was used instead of the ultrasonic nozzle, the
activity loss of LDH was found to be much lower. Just pumping the feed solution
through the non-heated nozzle, without spraying (pump), resulted in an activity loss of
only about 5 %. When the solution was subsequently sprayed (pump + spray), the
activity loss increased to about 14 %. Even after the entire spray drying process (pump
+ heat + spray), the activity loss of dried LDH was only about 23%. The results show
that by changing the nozzle type, the activity loss of 77% found with the ultrasonic
nozzle was reduced to only 22% with the two-fluid nozzle.
After 30 minutes of processing with the ultrasonic nozzle, the spray head
reached a temperature of around 54 °C under (pump + spray) conditions and 108 °C
under (pump + heat + spray) conditions. In addition, during the latter conditions, the
outlet temperature reached 54 °C after 30 minutes, whereas the outlet temperature
44 | Chapter 3
was slightly lower at around 48 °C after 30 minutes when the two-fluid nozzle was
used. Finally, with both nozzles about the same yield (40 to 65%) was obtained.
To determine the effect of the main stresses imposed on the LDH at each
different step, the activity losses in the combined processes of pump, heat and spray
were separated (Figure 4). When the ultrasonic nozzle was used, the most critical steps
during spray drying appeared to be heating and atomization, which resulted in 22 %
and 29 % activity loss, respectively. Circulation of the protein solution (passing of the
liquid feed over the vibrating perforated mesh) and the combination of dehydrating
and collecting accounted for 14 %, 5 %, and 8 % activity loss, respectively. However,
using the two-fluid nozzle while 1) pumping, 2) atomizing, and 3) the combination of
dehydrating, heating, and collecting only resulted in an activity loss of 5 %, 9 % and 9
%, respectively.
Figure 4: LDH activity loss and remaining activity after spray drying, caused by the different
separate steps that occur during spray drying with an ultrasonic nozzle (left) and two-fluid
nozzle (right). It has to be noted that activity loss due to heating could not be determined separately
with the two-fluid nozzle.
Discussion
The most critical process steps when spray drying protein solution were found to
depend heavily on the choice of nozzle. With the ultrasonic nozzle, the heating and
atomization steps caused the highest activity loss during spray drying. However,
switching to the two-fluid nozzle decreased the activity loss in these steps to a point
where no critical steps could be distinguished anymore. Overall, compared to the
ultrasonic nozzle, the two-fluid nozzle performed much better in each spray drying
step.
One very important observation that should be addressed immediately is that
during spray drying of LDH with the ultrasonic nozzle the activity loss in the liquid
feed was higher than the activity loss in the dried powder (Figure 3, 90 % and 78 %,
respectively). Although it may seem contradictive that the dried LDH appeared to
have a higher activity than the feed from which it was produced, the reason for this
can be found in the circulation of the feed. During this circulation, the feed was

repeatedly subjected to the heat from the drying chamber and to ultra-sonication,
which resulted in a significant decrease of the LDH activity in the feed over time.
Therefore, LDH that was dried at the start of the process had a higher activity than
LDH that was dried at the end, and the remaining activity of the dried and collected
LDH powder is an average of the activity of all powder collected over time. This
average activity will be higher than the remaining activity of the feed, especially
considering the low activity loss due to dehydration and collection relative to the total
activity loss in the feed. This also means that the data presented in Figure 4 according
to the calculations in Table 2 might not be entirely accurate, because the average
activity losses in the powder are directly compared with the cumulative activity losses
in the liquid feed. To improve this, one would have to differentiate between
cumulative activity loss and average activity loss in Figure 3, and subsequently
recalculate the activity loss for each process step by comparing cumulative values with
cumulative values and/or average values with average values.
Since there is no circulating system possible with the two-fluid nozzle, the
values in Figure 4 for the two-fluid nozzle are accurate and do not need to be re-
evaluated.
To recalculate the activity losses in the separate spray drying steps with the
ultrasonic nozzle, first it has to be clear which measured remaining activities are
cumulative and which are average over time: all the measurements that were done of
the liquid feed are cumulative activity values, while measurements of atomized liquid
(non-feed) and powder are average activity values. Let’s start with the circulating step
of the process. This step is relatively straightforward as there is no influence of heating
or atomizing on the activity losses here. Therefore, the cumulative loss is 14 % (equal
to “pump” in Figure 3). Although this is quite a bit higher than the 5 % activity loss
that was found with the two-fluid nozzle, it is reasonable to expect this, since the
liquid feed only passes the nozzle once after which it is dispensed, whereas with the
ultrasonic nozzle the liquid feed is continuously circulated. This will cause repeated
stresses on the protein in the form of shear due to the peristaltic pump and flow
through the tubing, and adsorption to the tubing, although the adsorption is expected
to be very low due to the use of PTFE tubing where possible. However, it is known
that LDH is quite shear sensitive [13, 14] and other less shear sensitive proteins are
expected to degrade much less during circulation.
The next step would be heating. However, if we consider the whole process,
we cannot differentiate between the effect of heating and the effect of the vibrating
perforated mesh on the activity in the circulating liquid feed. Although it was shown
that the activity loss due to heating was 22 %, and due to passing over the vibrating
perforated mesh 5 % (Figure 4), the total of 27 % is much lower than what is seen if
we are both heating and atomizing. In this case the activity loss due to heating and
46 | Chapter 3
passing over the vibrating perforated mesh reaches up to 76 % (90 % (cumulative
activity loss in the liquid feed after the entire process) – 14 % (loss caused by
circulating)) . The difference suggests that there is a synergistic effect on degradation
between heating and atomizing, since the activity loss is greatly increased when
heating is used together with atomization than when both process steps are
considered separately. Therefore, a more accurate estimate of the cumulative activity
loss due to heating and passing over the vibrating mesh would be 76 % instead of 27
%. It should be kept in mind that the average activity loss will be lower, since the
activity loss at the start of the process due to heating is zero and then accumulates
over time. With the two-fluid nozzle, it was not possible to separate the influence of
heating from dehydrating and collecting. However, during spray drying the liquid feed
and the nozzle itself were cooled with water of 4-8 °C. Upon leaving the spray dryer,
the temperature of the cooling water was measured at around 30 °C within a few
minutes after starting the process (pump + heat + spray). Because this was only 5 °C
above the room temperature, the influence of the heating step on the activity loss of
the protein is expected to be negligible over the activity loss that might occur at room
temperature in the starting solution.
To estimate the activity loss due to atomization of the liquid feed into drops,
it is assumed that this activity loss is a fixed percentage of the intact protein that
passes through the perforated mesh. Consequently, the absolute amount of activity
lost in time will decrease due to the accumulation of degraded protein in the liquid
feed. However, as mentioned previously, the measured activity of liquid that is
atomized (“pump + spray”) is an average and can therefore not be directly compared
to the activity losses in the liquid feed. It is thus preferable to correct the measured
activity loss of atomized liquid (43 %) for the average activity loss in the liquid feed
due to circulating and passing over the vibrating perforated mesh. To do so, it is
necessary to estimate the average activity loss in the liquid feed by calculating the
accumulating activity loss over time and then calculating the average. This can be done
by assuming that the degradation of the protein is a first-order reaction, as described
for protein denaturation in literature (Equation 2), with A(t) as the activity loss as a
function of processing time t, A0 the remaining activity at t = 0 (=100 %), and k the
rate of degradation constant [17-20]. This constant can be calculated by fitting the
equation to the cumulative activity loss at time t = 30 minutes. Note that increasing
the reaction to a higher order of 1.25 and 1.5 only had a minor effect on the results
and did not change the conclusion.
A  t   100%  A0  e  kt (2)

Using this equation with “pump + spray (feed)” with a cumulative activity
loss of 19 % gives a rate of degradation constant of 0.007. By calculating the
cumulative activity loss in time and then taking the average, we obtain an average
activity loss of 10 % due to circulating and passing over the vibrating perforated mesh.
We can now correct the activity loss after atomizing (43 % - 10 %) and divide that by
the average activity that was left when the liquid feed went through the perforated
mesh (100 % - 10%) to obtain the relative activity that is lost when the liquid is
atomized into drops, which is 37 %. To further clarify, at the beginning of the process
(t = 0) practically 100 % of the protein is active when it is atomized and 37 % is lost
due to atomizing. However, at the end of the process (t = 30 minutes), only 10 % of
the protein is active when it is atomized and only 3.7 % would be lost due to
atomizing. With the two-fluid nozzle, activity loss due to atomization was much lower
with only 9 %. The question remains, whether this difference in activity loss is solely
due to the different atomization mechanisms or due to other factors. When the
protein solution is dispersed as small droplets in the drying air, the interfacial area
between air and water is greatly increased. It is well known that proteins tend to
denature when present at such interfaces. Therefore, after atomization, the chances
that the protein will degrade and lose its activity at the interface are also greatly
increased. When the droplets generated by both nozzles would be different, the
interfacial stress would be different as well. Not only that, a different droplet size
would also imply a different energy input to the protein solution resulting in a
difference in multiple stresses that are not necessarily related to the atomization
mechanism itself. Therefore, to enable a direct comparison of the two nozzle types,
the droplets that are generated should be of the same size. This is shown to be the
case in this study by measuring the particle size distribution of powder produced with
both nozzle types (Table 2). Therefore, the difference in activity loss between both
nozzle types in the atomizing step cannot be attributed to a difference in interfacial
stress, but mainly to the difference in atomization mechanism of the ultrasonic and
two-fluid nozzle.
Finally, the activity loss due to dehydration and collection needs to be
determined. Again, we can assume that this is a fixed percentage of the remaining
activity after the liquid feed is atomized into drops. Because we are able to estimate
the activity loss in every process step at each point in time with Equation 2, it is
possible to sum the losses and determine the average activity loss over the whole
process. This value should equal the average activity loss that was measured in the
produced LDH powder (78 %), but is lower (75 %) because the activity loss due to
dehydrating and collecting is not yet taken into account. By adding an estimated 10 %
activity loss over the remaining activity after atomization of the liquid feed into drops,
the average activity loss in the powder of 78 % is obtained. Therefore, the relative
48 | Chapter 3
activity loss due to dehydrating and collecting is estimated at 10 % of the remaining
activity. With the two-fluid nozzle, dehydration and collection combined accounted
for 9 % activity loss. It can be expected that there is no relevant difference in these
two steps, since the steps in the process after atomization are hardly dependent on the
choice of nozzle as long as the generated droplets are identical. The only difference is
the addition of the atomizing gas, which enters the drying chamber through the two-
fluid nozzle. This resulted in a slightly lower outlet temperature (48 °C instead of 54
°C), which appears to have an insignificant effect on the activity loss.
A more detailed overview of the estimated activity loss in time due to the
different process steps with the ultrasonic nozzle is shown in Figure 5 and the average
activity losses in each process step in Figure 6. Note that although the actual
degradation rates might differ from those shown here, the relative effects of the
different process steps on the activity will remain largely the same.
Figure 5: Estimated cumulative activity loss in time with the ultrasonic nozzle due to 1) pumping
(), 2) heating and passing over the vibrating perforated mesh (), 3) atomizing (), and 4)
dehydrating and collecting (). Values are stacked. Note that the black diamond line also indicates
the remaining activity of powder produced at each specific point in time.

Figure 6: Estimated average LDH activity loss and remaining activity after spray drying, caused
by the different separate steps that occur during spray drying with an ultrasonic nozzle.
First and foremost, our results show that heating and atomizing are the most
critical steps when spray drying with the ultrasonic nozzle. Although it appears as
though atomizing only has a small influence compared to heating, this is because there
is not much intact protein left to degrade once it passes through the vibrating mesh
and part of the influence of the atomizing step is inseparable from the heating step. If
one would cool the liquid feed in order to minimize the activity loss due to heating,
the amount of activity lost due to atomizing would be increased (to an estimated
maximum of 37 % of activity left when passing through the perforated mesh). When
looking at the conditions that were used for spray drying, it is tempting to conclude
that the large difference in activity loss between the ultrasonic and two-fluid nozzle is
exacerbated by the high inlet temperature of 120 °C. Indeed, the chosen condition
would normally not be used when spray drying protein solutions, especially those that
are thermally sensitive. Even more so, lowering the inlet temperature would most
certainly lower the difference in activity loss between the ultrasonic and two-fluid
nozzle. To visualize this, the effect of heating in the previous calculations could be
neglected to estimate the effect of the other spray drying steps in the hypothetical case
that the activity loss due to heating is eliminated (Figure 7). However, the spray drying
conditions were deliberately chosen such, that any weak points in the characteristics of
the spray drying process would become apparent as clearly as possible. Therefore, to
compare the effects of heat transfer from the system to the protein solution during
spray drying, a high inlet temperature had to be used. This implies that the activity
losses due to heating encountered during this study are not necessarily an indication of
the activity loss found during a spray drying process with optimized conditions.
However, what it does show are the critical process parameters during spray drying
and more importantly, that the critical process steps depend heavily on the chosen
nozzle type.
50 | Chapter 3
Figure 7: Estimated average activity loss and remaining activity upon spray drying with the
ultrasonic nozzle, in the hypothetical scenario where the effect of heating is completely
neglected. Note the relatively high activity loss due to atomizing compared to the case where the effect
of heating is considered (Figure 6).
Secondly, the results clearly show that any time dependent processes during
spray drying should be avoided, since they can make the assessment of process and
product quality unnecessarily complex. For example, since the two-fluid nozzle, lacks
recirculation of the feed, activity losses will not be time dependent. Therefore, apart
from the standard activity losses during storage in the liquid feed and in the collector,
with the two-fluid nozzle the processing time and thus batch size do not have to be
taken into account when optimizing the spray drying process.
Finally, results with the ultrasonic nozzle suggest that when only a small
fraction of the liquid feed is spray dried, the activity loss would be much lower.
Indeed, if the processing time is lowered relative to the liquid feed volume, the
produced protein powder will have a higher activity because the cumulative activity
loss in the liquid feed is still low. However, this is not a realistic scenario, since a
prepared protein solution will always be spray dried wholly. So, despite the fact that
processing time and volume can be varied independently, they are tied together with
the atomizing flow rate. If somehow the atomizing flow rate would increase, the
activity loss at the end of the process would most likely be reduced. Unfortunately,
this is not possible with the ultrasonic nozzle, unless the 4 µm perforated mesh spray
cap is exchanged for one with a smaller mesh size , thereby also increasing the particle
size of the resulting powder.
Besides the possibility to cool the nozzle and the lack of circulation, there are
a few other advantages of using the two-fluid nozzle over the ultrasonic nozzle in
combination with the B-90 spray dryer (the nano spray dryer). Although the
combination of the two-fluid nozzle with the B-290 spray dryer is able to produce a
powder with a similar particle size distribution, the yield will be lower due to a lower
collection efficiency of a cyclone, even when using the high performance cyclone. In
addition, the electrostatic collector that is included in the B-90 spray dryer is better
able to collect smaller batch sizes efficiently. Although not evident from the results

described in this paper, even when a total of 50 mg of protein and excipients dissolved
in 10 ml of water was spray dried, a yield of up to 35 mg (70 %) was obtained, which
has never been possible with the B-290 spray dryer. This is especially useful for
expensive and potent proteins, which for research purposes are often only needed in
small amounts. Finally, with the two-fluid nozzle, the liquid feed and atomizing gas
flow (and thus particle size) can be varied freely, whereas with the ultrasonic nozzle
there are only three perforated mesh sizes available that can be used to change the
particle size, each with their fixed liquid feed flow. This makes it a lot easier to
optimize the process and to control the outlet conditions, especially the outlet
temperature and humidity, when the two-fluid nozzle is used.
Conclusion
This study showed that every different process step in the spray drying of versatile
proteins may result in a loss of activity of the protein. However, the extent to which
the different steps add to the total process stability of the protein may significantly
differ for different process equipment used and for different proteins. With regard to
the process equipment and configuration, the avoidance of any time dependent
process was found to be essential to maintain the highest possible fraction of the
protein intact; especially during up scaling this maybe an aspect of utmost importance.
With the model protein chosen in this study we have shown that the choice of the
droplet generating principle (the nozzle) determines to a large extent the process
stability of the protein. Heating of the feed solution (due to the use of a non-cooled
nozzle), the exposure to ultra-sonication, as well as the recirculation of fluid through
the nozzle should all be prevented. This makes the two-fluid nozzle used in our study
much better suited for spray drying of proteins than the ultrasonic nozzle tested.
Acknowledgments
The authors thank Hans van der Meer from Sympatec for performing the R1 LDA
measurements.
References
1. Grasmeijer N, Stankovic M, de Waard H, Frijlink HW, Hinrichs WLJ.

Unraveling protein stabilization mechanisms: Vitrification and water
replacement in a glass transition temperature controlled system. Biochimica et
Biophysica Acta (BBA) - Proteins and Proteomics. 2013;1834(4):763-9.
52 | Chapter 3
Pharm Sci. 2009;98(9):2886-908.
2004;256-257(1):319-27.
22.
Res. 1995;12(6):799-806.
8. Tonnis WF, Mensink MA, de Jager A, van der Voort Maarschalk K, Frijlink
HW, Hinrichs WLJ. Size and Molecular Flexibility of Sugars Determine the
Storage Stability of Freeze-Dried Proteins. Mol Pharm. 2015;12(3):684-94.
9. Maltesen MJ, van de Weert M. Drying methods for protein pharmaceuticals.
Drug Discovery Today: Technologies. 2008;5(2–3):e81-e8.
10. Ameri M, Maa Y-F. Spray Drying of Biopharmaceuticals: Stability and
Process Considerations. Drying Technology. 2006;24(6):763-8.
11. Adler M, Lee G. Stability and surface activity of lactate dehydrogenase in
spray-dried trehalose. J Pharm Sci. 1999;88(2):199-208.
12. Khatri L, Taylor KMG, Craig DQM, Palin K. An assessment of jet and
ultrasonic nebulisers for the delivery of lactate dehydrogenase solutions. Int J
Pharm. 2001;227(1–2):121-31.
13. Niven RW, Brain JD. Some functional aspects of air-jet nebulizers. Int J
Pharm. 1994;104(1):73-85.
14. Niven RW, Ip AY, Mittelman SD, Farrar C, Arakawa T, Prestrelski SJ.
Protein nebulization: I. Stability of lactate dehydrogenase and recombinant
granulocyte-colony stimulating factor to air-jet nebulization. Int J Pharm.
1994;109(1):17-26.
15. Hertel S, Pohl T, Friess W, Winter G. That’s cool! – Nebulization of
thermolabile proteins with a cooled vibrating mesh nebulizer. Eur J Pharm
Biopharm. 2014;87(2):357-65.
16. Bergmeyer HU, Bernt E. Lactate Dehydrogenase. In: Bergmeyer HU, editor.
Methods of Enzymatic Analysis (Second Edition): Academic Press; 1974. p.
574.

17. Law AJR, Leaver J. Effect of Protein Concentration on Rates of Thermal
Denaturation of Whey Proteins in Milk. J Agric Food Chem.
1997;45(11):4255-61.
18. Lyster RLJ. The denaturation of α-lactalbumin and β-lactoglobulin in heated
milk. J Dairy Res. 1970;37(02):233-43.
19. Hillier RM, Lyster RLJ. Whey protein denaturation in heated milk and cheese
whey. J Dairy Res. 1979;46(01):95-102.
20. Dannenberg F, Kessler H-G. Reaction Kinetics of the Denaturation of Whey
Proteins in Milk. J Food Sci. 1988;53(1):258-63.
54 | Chapter 3
56 | Chapter 4
Chapter 4
A user-friendly model for spray drying to aid pharmaceutical product

development
Niels Grasmeijer, Hans de Waard, Wouter L.J. Hinrichs, Henderik W. Frijlink
doi:10.1371/journal.pone.0074403
A user-friendly model for spray drying | 57

Abstract
The aim of this study was to develop a user-friendly model for spray drying that can
aid in the development of a pharmaceutical product, by shifting from a trial-and-error
towards a quality-by-design approach. To achieve this, a spray dryer model was
developed in commercial and open source spreadsheet software. The output of the
model was first fitted to the experimental output of a Büchi B-290 spray dryer and
subsequently validated. The predicted outlet temperatures of the spray dryer model
matched the experimental values very well over the entire range of spray dryer settings
that were tested. Finally, the model was applied to produce glassy sugars by spray
drying, an often used excipient in formulations of biopharmaceuticals. For the
production of glassy sugars, the model was extended to predict the relative humidity at
the outlet, which is not measured in the spray dryer by default. This extended model
was then successfully used to predict whether specific settings were suitable for
producing glassy trehalose and inulin by spray drying. In conclusion, a spray dryer
model was developed that is able to predict the output parameters of the spray drying
process. The model can aid the development of spray dried pharmaceutical products
by shifting from a trial-and-error towards a quality-by-design approach.
58 | Chapter 4
Introduction
Pharmaceutical product development can be a costly and time consuming process.

Although fairly simple processes allow researchers to base their development on a
trial-and-error approach, more complex processes will quickly increase the required
number of experiments to unfeasible heights. To allow products to be developed with
reasonable resources on more complex processes, a shift towards a quality-by-design
approach is desired. Even more so, a quality-by-design approach can not only improve
the development stage, but will also tremendously aid in the quality control of the end
product since it forces researchers to acquire a more detailed and fundamental
understanding of the processes used for production. This is also the main reason why
the FDA and EMA advocate the use of quality-by-design in drug development, and a
framework for this approach has been developed in the ICH guidelines Q8, Q9, and
Q10 [1, 2]. Moving from a trial-and-error approach to a quality by design approach,
requires the development of a model of the production process linking variables to
critical quality attributes of the final product. The use of such a process model can aid
pharmaceutical product development in several ways.
First, determining the design space can be considerably more efficient, since a
model allows output parameters to be calculated without performing a considerable
amount of experimental work. One might think that the time required to develop such
a model will hardly ever compensate the time gained during product development by
shifting from a trial-and-error to a quality-by-design approach. Indeed, when a model
would be specifically designed for one product, the time required to develop the
model could easily be longer than the time gained. On the other hand, the model
could still improve the quality of the product and could therefore be advantageous.
However, a model that is more generally applicable to the process or even multiple
processes would be advantageous, since such a model would only have to be
developed once and can be used for other products and future research as well.
Secondly, the use of even a basic model can greatly increase the understanding
of a process. Although the user may not have taken part in the development of the
model, the use of the model does allow one to quickly see effects of changes in
various parameters on the output of the process. However, the detail and number of
affected parameters does depend on the complexity of the model.
Finally, a good model can also give additional relevant process information
that is not provided inline during processing. There can be many parameters that are
not measured inline, but they can be very useful for the researcher developing a
process. Although most of these parameters can be determined offline, this would
require additional experiments. Therefore, although a basic model can be developed to
provide the user with just one critical output parameter of a process, other parameters
that are usually not measured inline can be calculated and added to the result, thereby
expanding the usefulness of the model. The addition of otherwise unknown output
parameters can be invaluable in the development stage of a pharmaceutical product.
A process that can significantly benefit from being modeled is spray drying.
The many input and output parameters make it a complex process to optimize by
trial-and-error during product development. Furthermore, several process parameters
and product properties can be very difficult to measure inline. Several papers have
been published on the development of a spray dryer model to facilitate a shift from
trial-and-error to quality-by-design [3-6]. Unfortunately, the models that are presented
in these studies are often either based on very complex computational fluid dynamics
(CFD) [7] or they are developed in expensive and specialized software [3]. Although
these models are useful for many applications, the details of the models and the
complexity of the software far exceeds that required for standard pharmaceutical
product development. Therefore, a pharmaceutical scientist cannot easily use or even
understand the models that were studied, which does not add to a fundamental
understanding of the modeled process. In that respect, a model that is developed in
software that is common amongst pharmaceutical scientists would be preferred. In
addition, it would be even more useful if said model was easily adaptable for different
spray dryers and individual user needs.
Therefore, in this study, common mass and energy balances were used as a
basis to develop a user-friendly spray dryer model that can aid in the development of
spray dried products. The model will be presented in such a way, that it can be used
by those less familiar with spray drying and specialized software, and that it can be
adapted for different spray dryers. Furthermore, public access to the model is ensured
by developing the model in an open source software package and making it available
through an open access journal (Supporting Information File S1–S3).
As an example we will use spray drying of an aqueous trehalose and aqueous
inulin solution. Spray drying of these disaccharides is of interest in modern
pharmaceutics, since both have been shown to be good stabilizers for
biopharmaceuticals such as therapeutic peptides, proteins, or vaccines [8-14]. Using
the adequate drying process is of paramount importance here since stabilization of the
incorporated biopharmaceutical will in general only be obtained when the sugar is in
the glassy state. To obtain glassy material from the drying process requires specific and
well controlled process conditions and adequate process understanding. This makes
the example of trehalose and inulin interesting for many development scientists.
However, the aim of the model is not to determine the influence of excipients on the
outcome of the spray drying process, but rather the influence of the process
conditions on the final product. Therefore, the example of glassy sugar production,
which can be applied to protein stabilization, is merely used to show the application of
the model in the specific field of protein stabilization, where glassy sugars are desired.
60 | Chapter 4
In fact, the spray dryer model can be applied to numerous spray drying applications
due to the general setup of the model. It will, however, not predict whether a protein
will be stabilized, as it will also depend on the type of sugar used, but rather the
optimal spray drying conditions to stabilize a protein.
The model development will be divided into three separate stages. First, a
basic model will be developed that will enable us to calculate the spray dryer outlet
temperature. Then, for the purpose of obtaining glassy sugars by spray drying, the
model will be extended to include a relative humidity calculation, which is an essential
parameter. Finally, this extended model will be used to predict whether glassy
trehalose and inulin can be obtained successfully by spray drying at specific inlet
conditions.
Materials
Trehalose was obtained from Cargill B.V. (Amsterdam, The Netherlands). Inulin was
kindly provided by Sensus (Roosendaal, The Netherlands) and had a degree of
polymerization of 11. All experiments were performed with millipore water, type 1.
Spray drying process

Model validation and sample preparation were done by performing several spray
drying experiments with a B-290 spray dryer in conjunction with a high performance
cyclone and a B-295 dehumidifier (Büchi Labortechnik AG, Flawil, Switzerland). All
results were obtained with the spray dryer in closed loop configuration. Furthermore,
spray drying experiments that included a liquid feed flow were performed using water
only, except for the experiments with trehalose or inulin, which were performed using
an aqueous solution containing 2.5 % w/v trehalose or inulin. For model fitting,
validation, and relative humidity measurements, the inlet temperature was varied
between 50 °C and 200 °C, the liquid feed flow between 0 mL/min and 4.9 mL/min,
and the aspirator flow between 50 % and 100 %. Trehalose and inulin were spray
dried at a constant inlet temperature of 70 °C, while the aspirator flow was set at 100
%. The liquid feed flow for the aqueous trehalose solution was set at 4.1 and 5.1
mL/min, and for the aqueous inulin solution at 4.5 and 5.7 mL/min, using a syringe
pump. The atomizing airflow was kept constant for all experiments at 600 Ln/hr,
which corresponds to a setting of 50 mm (normal liter (Ln) is the volume at 0 °C and 1
atm). Specific spray dryer settings used for fitting, validation, and measuring the
relative humidity at the outlet are shown in Table 1 and 2. During the spray drying
experiments, equilibrium of outlet conditions was assumed to exist when the
temperature did not change more than 0.5 °C during 5 minutes.
Table 1. Spray dryer settings used for model Table 2. Spray dryer settings used for
fitting and validation. relative humidity measurements.
Inlet Aspirator Liquid Inlet Aspirator Liquid
temperature flowa feed flow temperature flowb feed flow
(°C) (m3n/hr) (mL/min) (°C) (m3n/hr) (mL/min)
50 12 0 50 22 0
50 22 0 50 22 1.3
100 12 0 50 22 2.7
100 22 0 50 22 3.6
100 22 1.3 70 12 0
100 22 2.7 70 12 1.3
100 22 3.6 70 12 2.7
100 22 4.9 70 22 0
150 12 0 70 22 1.3
150 22 0 70 22 2.7
150 22 1.3 70 22 3.6
150 22 2.7 70 22 4.9
150 22 3.6 90 12 0
150 22 4.9 90 12 1.3
200 12 0 90 12 2.7
200 22 0 90 12 3.6
200 22 1.3 90 22 0
200 22 2.7 90 22 1.3
200 22 3.6 90 22 2.7
200 22 4.9 90 22 3.6
a Aspirator flow of 12 and 22 m3n/hr 90 22 4.9
corresponded to a setting of 50 % and 100 %, b Aspirator flow of 12 and 22 m3n/hr
respectively (determined with the flow rate - corresponded to a setting of 50 % and 100 %,
pressure drop relationship over the cyclone and respectively (determined with the flow rate -
filter as described in the flow rate – pressure pressure drop relationship over the cyclone
drop relationship section in materials and and filter as described in section 2.3).
methods).
Flow rate – pressure drop relationship

Since the aspirator flow of the B-290 spray dryer is expressed in percentage, the mass
flow of the system had to be determined with respect to the given percentage in order
to use the aspirator flow in the model. To minimize the influence on the spray dryer
process, the aspirator flow was determined using the flow rate - pressure drop
relationship, where the flow rate through a system is related to the square root of the
pressure drop over the same system. The pressure drop over the cyclone and the filter
was measured with a HBM PD1 differential pressure transducer in conjunction with a
HBM MC2A measuring converter (Hottinger Baldwin Messtechnik, Darmstadt,
Germany) at flow rates between 0 and 150 Ln/min, after which the slope of the
62 | Chapter 4
relation between the flow rate and the square root of the pressure drop could be
determined (R). The flow rate through the cyclone and the filter was determined using
a Brooks 5863S mass flow meter (Brooks Instruments B.V., Ede, The Netherlands).
Subsequently, the pressure drop (Δp) across the cyclone and the filter was measured
inside the spray dryer at aspirator settings between 50 and 100 %. The aspirator flow
inside the spray dryer (QV.g) with respect to the spray dryer setting was then calculated
(Equation 1). The slope and intercept of the linear relationship between the aspirator
setting in percentage and actual flow rate were used in the model. Because the
pressure drop of the high performance cyclone may differ between copies, the flow
rate – pressure drop relationship will have to be determined separately for every
cyclone used.
QV . g  R 2  p (1)
Relative humidity
Relative humidity measurements were performed using a Testo 650 handheld device
with a standard climate sensor (Testo B.V., The Hague, The Netherlands). The sensor
had a relative humidity range of 0 % to 100 % (±2 %) and a temperature range of -20
°C to +70 °C (±0.5 °C), limiting the inlet temperature to a maximum of 90 °C with
the chosen liquid feed flow and aspirator flow (Table 2). Measurements were done
directly behind the outlet temperature sensor of the B-290 spray dryer.
Differential Scanning Calorimetry (DSC)

Modulated DSC measurements were done with a DSC 2920 differential scanning
calorimeter (TA Instruments, New Castle, United States). Humidified spray dried
trehalose and inulin samples were prepared by storing the samples at a relative
humidity of 22 %, 33 %, and 52 % in a desiccator over a saturated aqueous solution of
CH3COOK, MgCl2.6H2O, and Na2Cr2O7.2H2O, respectively, or at 45 %, and 60 % in
a climate chamber for 1-3 weeks. Humidified samples were weighed in closed
aluminum pans, then cooled to -20 °C, and finally heated at a rate of 2 °C/min with a
modulation period of 60 seconds and amplitude of 0.316 °C. The glass transition
temperature was taken as the inflection point of the transition in the reversing heat
flow versus temperature curve.
Dynamic Vapor Sorption (DVS) analysis

The water sorption isotherms of spray dried trehalose and inulin were measured at
ambient pressure and 25 °C using a DVS-1000 water sorption instrument (Surface
Measurement Systems Limited, London, UK). The moisture content was determined

at relative humidity’s ranging from 0 – 90 % in 10 % increments, for a sample with an
initial mass of approximately 10 mg. After subjecting the samples to the specified
humidity, equilibrium was assumed when the change in mass was less than 0.9 µg
during 10 minutes.
Model development
The spray dryer model was developed and tested using both commercial and open
source spreadsheet software packages, namely: Office 2003 and 2010 (Microsoft),
Libreoffice 3.4 (The Document Foundation), and OpenOffice.org 3.3 (The Apache
Software Foundation). The model was based on the B-290 spray dryer, which was
simplified for the development of the model, as shown schematically in Figure 1.
Figure 1. Schematic representation of a spray dryer (left) and the simplified spray dryer model
(right). Output variables include but are not restricted to the outlet temperature (Tout).
The entire spray dryer was considered to be a cylinder (right of Figure 1) with
a diameter and wall thickness that could be measured directly from the device used by
the researcher. Although the actual spray dryer is more complex than a simple
cylinder, inner flow characteristics are not considered and the drying gas is considered
to be continuously and ideally mixed. Due to these assumptions, the main parameters
that determine the outlet temperature are mainly limited to the surface area and
properties of the wall and surrounding medium. Therefore, the complex shaped spray
dryer can be modeled as a straight cylinder.
Whereas the inlet of the spray dryer consists of three separate streams: the
atomizing airflow, aspirator airflow, and liquid feed flow, the outlet consists of one
single gas stream. Using several input parameters, the outlet temperature can be
calculated using basic thermodynamic equations. As shown in Figure 1, the outlet
temperature is determined by the heat loss through both conduction and evaporation.
64 | Chapter 4
The heat loss through conduction (Qh.con) can be calculated using a basic heat transfer
equation for flat surfaces, whereas the heat loss through evaporation (Qh.evap) is
straightforward, as shown in Equation 2 and Equation 3 respectively.
Tw  A (2)
Qh.con 
d glass d air

 glass air
Qh.evap  H evap  QV .l   l (3)
Where ΔTw is the log mean temperature difference across the wall over the
entire length of the spray dryer, A is the surface area of the wall, dglass is the thickness
of the wall, dair is the boundary layer of air on the outside, λglass and λair are the heat
conductivity of the glass and air respectively, Hevap is the heat of evaporation of the
liquid, QV.l is the liquid feed flow, and ρl is the density of the liquid.
Except for ΔTw and dair, all parameters are known. The unknown parameter,
ΔTw, can be calculated using Equation 4.
Tw 
Tin  Tair   Tout  Tair  (4)
 T  Tair 
ln in 
 Tout  Tair 
Where Tin and Tout are the temperature of the heated drying air at the spray
dryer inlet and outlet respectively, and Tair is the temperature of the ambient air, which
is assumed to be constant. The unknown input parameter, dair, was used as a fitting
parameter, to match the output values of the model to experimentally determined
values that were obtained by running the spray dryer under various conditions.
Finally, the outlet temperature can be calculated by subtracting the heat flow
due to evaporation and conduction from the heat capacity of the aspirator flow, as
shown in Equation 5.
Qh.con  Qh.evap (5)

Tout  Tin 
C p. g  QV . g   g
Where Cp.g is the heat capacity if the drying gas under constant pressure, QV.g
is the drying gas flow rate, and ρg is the density of the gas. However, because ΔTw (and
thus heat loss due to conduction, Qh.con) is dependent on the outlet temperature of the
spray dryer, the calculation was repeated, or iterated, until the output was constant
(Figure 2).
Figure 2. Overview of iteration steps in our spray dryer model. Details are left out for clarity. Model
expansion for relative humidity at the spray dryer outlet (RHout) will be discussed in the results.
As shown in Figure 2, the relative humidity at the outlet (RHout) can also be
calculated when the outlet temperature is known. The relative humidity can be
calculated with the Antoine equation and the ideal gas law (Equation 6 and Equation 7
respectively).
B
A
p w.sat  10 C T (6)
xw    R  T  273.15 (7)
pw 
Mw
Where pw.sat is the saturated water vapor pressure, A, B, and C are the Antoine
constants of water (10.20, 1730.63, and 233.43, respectively [15]), T is the temperature
(°C), pw is the partial water vapor pressure (Pa), xw is the specific humidity, ρ is the
density of air, R is the gas constant, and Mw is the molecular mass of water. Since all
the parameters are known during the iterated calculation in the spreadsheet software,
the relative humidity, which is defined as pw/pw.sat·100, can be calculated.
Results
Model basis
A basic model was developed, as described in the model development section in
materials and methods, using only freely available software. First the model was fitted
to experimental values by running the spray dryer under various conditions to
determine the value of the fitting parameter, dair. Since the fitting parameter, dair, solely
determines the heat loss due to conduction and not due to evaporation, the
experiments were conducted without a liquid feed. In other words, only a heated gas
flow through the spray dryer was considered. By using the least squares method on
66 | Chapter 4
the modeled outlet temperature and experimental outlet temperature, the optimum
value for dair was found to be 1.97 mm. With this value, the modeled outlet
temperature matched the experimental outlet temperature well for all settings, with a
difference ranging between -2.5 and +1.5 °C (Figure 3).
Figure 3. Modeled (grey) and experimental (black) outlet temperature used to determine dair.
Aspirator flow was set at either 12 m3n/hr (circle) or 22 m3n/hr (triangle), while the liquid feed flow was
kept constant at 0 mL/min. Thickness of the lines indicate the 95 % confidence interval of the modeled
outlet temperature.
For the confidence assessment of the fitting parameter, dair, a protocol

described by Kemmer et al was used [16]. Based on this protocol, the 95 %
confidence interval of dair was calculated to be between 1.91 and 2.03 mm. The
thickness of the line in Figure 3 indicates the range of modeled outlet temperatures
corresponding to this 95 % confidence interval. Furthermore, the mean difference of
the modeled and experimental values (modeled values minus experimental values) was
-0.3 °C, indicating a slight bias of the model towards a lower outlet temperature.
Finally, the mean absolute difference (the mean of the absolute difference between
modeled and experimental values) was found to be 0.9 °C, which indicates a good
precision.
Subsequently, the fitted model was validated. This was done by comparing the
model output to several spray drying measurements that included a liquid feed flow
(Figure 4). The modeled outlet temperature was found to match the experimentally
determined outlet temperature very well. Both at a high and low inlet temperature of
200 °C and 100 °C, respectively, the outlet temperature matched the experimentally
determined outlet temperature even at the highest liquid feed flow of 5 mL/min, with

a difference ranging between -2.5 and +1 °C. It should be noted, however, that some
of the experimentally determined values lie outside the 95 % confidence interval,
which is most likely due to the low number of data points used for fitting the model.
Despite this deviation, the mean difference was found to be -0.8 °C, indicating a slight
underestimation of the modeled outlet temperature. In addition, the mean absolute
difference was 1.1 °C, which shows that the precision of the model is high.
Figure 4. Modeled (grey) and experimental (black) outlet temperature. Inlet temperature was set at
100 °C (circle), 150 °C (triangle), or 200 °C (square). Aspirator flow was kept constant at 22 m3n/hr.
Thickness of the lines indicate the 95 % confidence interval of the modeled outlet temperature.
Although the model was able to predict the outlet temperature of the B-290
spray dryer very well, it would be even more useful if the model could also be used for
other spray dryers. Therefore, an attempt was made to adapt the model for another
type of spray dryer. Although the B-290 spray dryer complicated the development,
due to the necessary conversion of aspirator setting in percentage to actual volume
and mass flow rate, it was possible to adapt the model for a B-90 spray dryer, which
reports the aspirator flow in L/min. Although not implied by the name, the B-90
spray dryer is very different from the B-290 spray dryer. Not only does the nozzle
consist of an ultrasonic sprayhead, without the atomizing airflow, but the B-90 spray
dryer also uses an electrostatic collector instead of a cyclone. In addition, the spray
dryer is shaped like a cylinder instead of the more complex system of components that
composes the B-290 spray dryer.
The model was fitted to the spray dryer by simply measuring the dimensions
of the drying column (length, diameter, glass thickness) and performing 6 spray drying
experiments without a liquid feed flow. Spray drying was performed at an inlet
68 | Chapter 4
temperature of either 50, 90, or 120 °C and an aspirator flow of either 85 or 165
L/min. It was found that the modeled outlet temperature again matched the
experimentally determined outlet temperature very well for all settings, with a
difference between ±2 °C (data not shown). The mean difference was 0 °C, indicating
that there is no bias of the model, whereas the precision of the model was similar to
the fit of the B-290 spray dryer with a mean absolute difference of 1.3 °C.
Model extension
An interesting application of spray drying is the stabilization of biopharmaceuticals
with sugar glasses. When a biopharmaceutical is incorporated in a matrix of a glassy
sugar, it can retain its conformation for prolonged periods in the dry state. The
conformation of the biopharmaceutical can be retained due to several mechanisms.
Although various mechanisms are proposed to play a role, one of the most often
considered hypothesis is the vitrification theory [13, 17-19]. The vitrification theory
states that the biopharmaceutical is immobilized when it is incorporated in a sugar.
Since most degradation pathways require molecular mobility, the degradation rate is
strongly reduced. To immobilize the biopharmaceutical, it is important that the sugar
can accommodate the irregular surface of the biopharmaceutical. Therefore, the sugar
should be in the amorphous state and not in the crystalline state [20]. More
specifically, the amorphous sugar should be in the glassy state and not in the rubbery
state for three important reasons. Firstly, in the glassy state the translational molecular
mobility is low, which is required for vitrification, whereas in the rubbery state the
translational molecular mobility is relatively high, which facilitates degradation of the
biopharmaceutical [21]. Secondly, in the rubbery state the sugar can easily crystallize.
Thirdly, what is highly relevant for the spray drying process is that in the rubbery state
the sugar also tends to be sticky. As a consequence, the rubbery sugar is more likely to
stick to the cyclone wall, reducing the yield of the product [22]. Therefore, it is
important that the glass-rubber transition temperature of the product is higher than
the surrounding temperature.
The relative humidity of the drying air is an important parameter during the
production of amorphous sugars by spray drying. Not only does humid air cause the
water droplets to evaporate slower, it also lowers the glass transition temperature of
amorphous sugars as adsorbed moisture acts as a plasticizer. Therefore, usually, a dry
product with a moisture content as low as possible is aimed for. Although the relative
humidity, and thus the product moisture content, can be lowered by simply increasing
the temperature, exposing the material to excessive temperatures should in general be
avoided to prevent thermal degradation. To find the optimum balance between
relative humidity and outlet temperature, while also maximizing the throughput,
optimization is required. Therefore, any information on relative humidity prior to

spray drying can be highly relevant to the development scientist. Unfortunately, in
most commercially available lab-scale spray dryers a relative humidity measurement is
not included and would therefore be a desirable addition to a spray dryer model.
Therefore, the basic model was extended to include the relative humidity in order to
increase the usefulness of the model, as described in the model development section
in materials and methods.
To validate the results of the extended model, the relative humidity at the
spray dryer outlet was measured at various spray dryer settings (Table 2), and
compared to the relative humidity predicted by the model (Figure 5). During the
measurements, the reported temperature of the relative humidity sensor was only
slightly higher than the temperature that was reported by the spray dryer itself (1 – 3
°C). Therefore, the fitting parameter, dair, was not adjusted to fit the model to the
temperature reported by the relative humidity sensor, but kept at 1.97 mm. This
resulted in a slight underestimation of the modeled outlet temperature, averaging
around -2 °C, when compared to the outlet temperature measured with the relative
humidity probe. On the other hand, the modeled outlet temperature matched the
outlet temperature reported by the B-290 spray dryer very well with a mean difference
of +0.2 °C, and a mean absolute difference of 1.3 °C..
The relative humidity was predicted well by the model. The difference
between the modeled and experimentally determined relative humidity ranged
between -3.5 and +2.1 % RH, with the largest difference at a liquid feed flow of 4.9
mL/min. However, when calculating the modeled relative humidity according to the
95 % confidence interval that was determined based on the fitting results, it was found
that most of the experimental relative humidity values were outside this confidence
interval (data not shown). In addition, calculating the confidence interval based on the
relative humidity data did not yield a correct confidence interval either (data not
shown). Therefore, a more general approach was applied as described by Brown, by
which the 95 % confidence interval is determined directly from the modeled and
experimental value [23]. With this method a 95 % confidence interval of 3.1 % RH
was calculated, which appears to fit all the experimental values (Figure 5).
Furthermore, the mean difference between the modeled and experimental relative
humidity was 0 % RH, indicating no bias of the model. Finally, the mean absolute
difference was 1.2 % RH, which is considered precise.
70 | Chapter 4
Figure 5. Modeled (grey) and experimental (black) relative humidity. Measurements were done at
an inlet temperature of 90 °C (circle), 70 °C (triangle), or 50 °C (square). Results shown at the top (A)
were obtained with an aspirator flow of 12 m3n/hr and the results on the bottom (B) with 22 m3n/hr.
Thickness of the lines indicate the 95 % confidence interval of the modeled relative humidity (3.1 % RH).
Model application
To show the applicability of the model we took the example of trehalose and inulin,
both suitable excipients for stabilization of biopharmaceuticals during spray drying.
Although the glass transition temperature of trehalose and inulin are relatively high
(121 °C and 130 °C, respectively), it can be greatly reduced by adsorbed moisture, as
discussed in the model extension section in the results. Therefore, knowledge of the
hygroscopicity and quantification of the reduction of the glass transition temperature
due to adsorbed moisture is key to understanding the outcome of the spray drying

process. Therefore, the model was used in conjunction with DVS and DSC analyses
to determine the optimal settings for spray drying both a trehalose and an inulin
solution.
The glass transition temperature dependence on the moisture content can be
described by the Gordon-Taylor equation (Equation 8), which describes the relation
between the composition of an ideal and homogeneous mixture consisting of two
components (with mass fractions ws, and ww) and its glass transition temperature (Tg)
[24]. Besides the mass fraction of the components, the glass transition temperature of
the mixture is also dependent on the glass transition temperature of the individual
components (Tg.s, and Tg.w) and a component-dependent Gordon-Taylor constant (ksw).
The subscripts s, and w are used for sugar, and water respectively. To calculate the
glass transition temperature of a trehalose-water and inulin-water mixture with the
Gordon-Taylor equation, the glass transition temperature of trehalose and inulin, the
Gordon-Taylor constants and the mass fraction of water were determined.
ws  Tg .s  k sw  ww  Tg .w (8)
Tg 
ws  k sw  ww
The glass transition temperature of trehalose and inulin were determined with
DSC and found to be 121 °C and 130 °C, respectively. For water, a glass transition
temperature of -109 °C was used, which is the average of recently published values
[25-27]. Although this value is substantially higher than the conventionally accepted
value of -137 °C [25-27], our calculations indicated that the choice of either of these
glass transition temperatures of water did not have a large influence on the calculated
glass transition temperature of the final samples.
The Gordon-Taylor constant, ksw, for a trehalose-water and inulin-water
mixture was determined by fitting the Gordon-Taylor equation with glass transition
temperatures of humidified sugar glasses measured with DSC. The mass fraction of
water (ww) of these humidified sugar glasses was determined with DVS analysis. The
Gordon-Taylor constant for the trehalose-water and inulin-water mixture were found
to be 7.90 and 7.40, respectively. The Gordon-Taylor constant for a trehalose-water
and inulin-water mixture was higher than values found in literature (i.e. 5.2, 6.5, and
7.3 and between 5.9 and 6.4, respectively), due to the higher glass transition
temperature of water we used [8, 28-30].
The mass fraction of water, ww, at a spray dryer setting of choice, was
determined by relating the relative humidity output of the model to DVS data of
trehalose or inulin. Thereby, it is assumed that the moisture content of the spray dried
sugar is in equilibrium with the outlet conditions of the spray dryer.
72 | Chapter 4
Depending on the requirements of the product and process, the inlet
temperature, liquid feed flow and aspirator can be varied in the model to find the
optimum settings, where the glass transition temperature of the sugar is higher than
the outlet temperature of the spray dryer. An example is shown, where the liquid feed
flow is changed slightly to determine the effect on the yield of spray dried trehalose
and inulin (Table 3). Assuming that the moisture content is in equilibrium with the
outlet conditions, at a liquid feed flow of 5.1 mL/min the glass transition temperature
of trehalose (25 ºC) is expected to be below the outlet temperature (35 ºC). A liquid
feed flow of 5.7 mL/min was used to obtain the same difference between the glass
transition temperature of inulin (22 ºC) and the outlet temperature (32 ºC). Because,
under these conditions, the glass transition temperature is lower than the outlet
temperature, trehalose and inulin are expected to be in a rubbery, sticky, state. In
contrast, at a liquid feed flow of 4.1 mL/min, a lower relative humidity is expected,
resulting in a glass transition temperature of trehalose (49 ºC) above the outlet
temperature (39 ºC). A liquid feed flow of 4.5 mL/min was used to obtain the same
difference between the glass transition temperature of inulin (47 ºC) and the outlet
temperature (37 ºC). Because, under these conditions, the glass transition temperature
is higher than the outlet temperature, trehalose and inulin are expected to be in a
glassy, non-sticky state.
Table 3. Trehalose and inulin yield depending on spray drying conditions.a

Trehalose Inulin
Liquid feed flow (mL/min) 5.1 4.1 5.7 4.5
RH model (%) 41 29 52 34
Tout model (°C) 35 39 32 37
Estimated Tg (°C)b 25±5 49±7 22±5 47±4
Measured Yield (%) 4 68 8 75
aThe inlet temperature was set at 70 ºC and the aspirator was set at 22 m3n/hr (100%).
b The margin of error for the estimated Tg is based on the 95 % confidence interval of the modeled
relative humidity (3.1 % RH)
It was found that the experimental observations agreed with the modeled
conditions. At a liquid feed flow of 5.1 and 5.7 mL/min, when the amorphous
powder was expected to be in its sticky rubbery state, the yield was very low (4 and 8
% of trehalose and inulin, respectively) and the cyclone wall was completely covered
with powder. However, at a liquid feed flow of 4.1 and 4.5 mL/min, when the
amorphous powder was expected to be in its glassy state, the yield was much higher
(68 and 75 % of trehalose and inulin, respectively) and hardly any powder was visible
on the cyclone wall after spray drying. DSC confirmed that both trehalose and inulin
were in an amorphous state.

Discussion
A spray dryer model is presented that is developed in an open source spreadsheet

program and made freely available, which enables the use of the model by virtually
everyone. Additional information on the use of the model is also made available
online (Supporting Information File S1-S3). The development of this model was
explained to provide the basic knowledge required to understand the model.
Furthermore, the model allows the user to make adaptations in case one’s process
deviates from the spray drying process that is used to develop the model. Such a user-
friendly model for spray drying can aid pharmaceutical product development by
shifting from a trial-and-error to a quality-by-design approach. We confirmed this by
the successful application of the model to spray drying of trehalose and inulin.
The model was developed while keeping in mind that the end user may not
be familiar with mathematical software programs, energy and mass balances or even
the spray drying process. Therefore, it was decided to develop the model in an open
source spreadsheet program. One could argue that the use of a program such as
Mathcad or an open source alternative would be more appropriate due to the visibility
of the symbolic equations. Indeed, a large disadvantage of spreadsheet software is that
the equations are not shown symbolically, which often makes the equations difficult to
read, especially for those not involved in the development. However, symbolic
equation editors are less often used than spreadsheet software by pharmaceutical
researchers. Therefore, although the equations might be more difficult to read, it will
be easier to apply and adapt the model, since the user will be more familiar with the
software. In addition, during the development of the model, equations can be easily
clarified by adding comments, descriptions, and pictures, as we did with the presented
model.
Compared to commercial software packages, open source software allows
those with less extensive budgets to be able to use the model as well. The difference
between the open source software, such as Libreoffice or OpenOffice, and
commercial software, such as Microsoft Office, is rather small. Anyone that has
experience with any of these packages will be able to find their way in the other
packages as well. However, during the development of the model a difference was
found in the way iterative calculations are handled that does make the modeling in the
open source packages slightly more challenging. First, the number of iteration steps in
Excel is limited to 10000, whereas the limit in OpenOffice and Libreoffice is only
1000. Therefore, it can be more challenging to let the iterative calculations converge in
the open source software. Secondly, there is a difference in the handling of iteration
convergence. All software options allows one to choose a maximum change value,
which is the maximum amount a value is allowed to change between iterations before
it is considered converged. However, in Excel this value will only determine whether
74 | Chapter 4
or not the iterative calculation will stop before the iteration step limit is reached. If
convergence is not reached, the last calculated values are shown. In the open source
alternatives the value is only shown when convergence is reached. When it is not
reached, the cell will return an error value. Although this does make it clear that the
iterative calculation did not finish, it severely hinders the debugging of such
convergence issues, since the source of the problem cannot be identified. Although
most convergence issues have been solved during the testing and use of the model,
sometimes the model will not converge when the input is changed radically or when
the input would result in outlet conditions with a relative humidity close to 100 %. In
most situations where the model has a problem to converge, simply forcing a
recalculation will suffice. However, in case this is not sufficient, additional tips on how
to solve or prevent convergence issues are discussed in the short guide accompanying
the model (Supporting Information File S3).
When comparing the modeled outlet temperature to the experimental results,
it was shown that the model was able to predict the outlet temperature of the spray
dryer quite well (Figures 3-4). In addition to the results with the B-290 spray dryer, a
good prediction of the outlet temperature of a B-90 spray dryer was found, showing
that the model can be used for other types of spray dryers as well. The adaptation of
the model was found to be rather straightforward, mainly due to the aspirator flow
that was reported in the proper units by the B-90 spray dryer (L/min instead of
percentage). Although some knowledge of spreadsheet software is required to be able
to change the calculation of the aspirator flow when the spray dryer does not report
the proper units, this will generally not be a major problem for most pharmaceutical
scientists.
Although the modeled outlet temperature shows only a minor
underestimation (< 1 °C), and a good precision to around 1 °C, some of the
experimental values that were used to validate the model were found outside the 95 %
confidence interval (Figure 4). Therefore, it could be concluded that the 95 %
confidence interval shown here is simply too small, which is most likely true.
However, the main reason for the underestimation of the confidence interval is most
likely the small dataset that was used to fit the model and calculate the confidence
interval. This was deliberately done to show the flexibility of the model and how quick
the model can be fitted to a particular spray dryer, while still obtaining a reasonable
accuracy. When we consider the small mean (absolute) differences that were found, it
can only be concluded that the model is able to predict the outlet temperature quite
accurately even when only a small dataset of 8 measurements is used to fit the model
to the spray dryer.
To enable the application of the model to the spray drying of amorphous
materials, as is often used for stabilizing biopharmaceuticals, the model was extended

to calculate the relative humidity. The relative humidity calculation in the model was
shown to give a good estimate of the experimentally determined outlet condition
(Figure 5). Especially the mean (absolute) difference showed that the model has no
bias to under- or overestimate the relative humidity, and that the modeled values are
fairly precise to around 1.2 % RH. However, the 95 % confidence interval that was
determined from the outlet temperature dataset used for fitting the model, was clearly
too small (data not shown). In addition, even when the confidence interval of the
fitting parameter was calculated from the relative humidity instead of the outlet
temperature, the resulting confidence interval of the modeled relative humidity would
be too small (data not shown). This is most likely due to the physical relation between
the fitting parameter, dair, the outlet temperature, and the relative humidity. Whereas
the outlet temperature is directly influenced by the thickness of the boundary layer for
heat conduction (dair), the relative humidity is only influenced indirectly. Therefore, dair
might not be a suitable fitting parameter for the relative humidity and therefore also
does not give a suitable confidence interval based on the method described by
Kemmer et al. [16]. However, using a more general method that directly calculates the
95 % confidence interval of the modeled relative humidity based on the experimental
values was shown to include all experimental values. Although the interval of 3.1 %
RH may seem rather large, the difference in moisture content of trehalose or inulin at
the outlet is only about 0.6 %. This translates into a difference of the glass transition
temperature of 4-8 °C. Although this could be the difference between a rubbery or a
glassy sugar, the difference between the outlet temperature and the glass transition
temperature of the sugar should be much larger to minimize the molecular mobility
and therefore maximize protein stability [21]. Therefore, the 95 % confidence interval
of 3.1 % RH should be sufficient for relative humidity sensitive operations, such as
protein stabilization.
Although not clearly pronounced in Figure 5, many of the higher deviations
from the modeled relative humidity were found in cases where the outlet temperature
is low. At these low temperatures, the relative humidity is much more sensitive to
small deviations in moisture content of the humid air. Especially the assumptions
made regarding the inlet conditions of the air coming from the dehumidifier (0 °C and
100 % relative humidity) are of great influence. When the inlet condition was changed
by only 1 or 2 °C, the relative humidity could change up to 2 % RH. The same could
also be the said for the temperature difference between that reported by the relative
humidity sensor and the B-290 spray dryer. Although the difference was around 2 °C,
the influence on the relative humidity could be significant. However, this was tested
by fitting the model to the outlet temperature measured with the relative humidity
probe and the difference was found to be minor (up to 3 % RH, with an average
difference of 1% RH). In addition, since the relative humidity describes the moisture
76 | Chapter 4
content in the air, the liquid feed flow and aspirator flow are of importance. Although
the liquid feed was facilitated by a roller pump, the pulsation in the liquid feed flow is
not expected to have influenced the relative humidity measurement, since the
measurement was done over a period of 30 seconds; much longer than the pulsation
period, which was around 2 seconds. However, the aspirator flow had to be
determined with the flow rate – pressure drop relationship, due to the aspirator flow
being expressed in percentage instead of units for flow rate. The equipment that was
used, only allowed the pressure drop to be measured at a flow up to 150 Ln/min,
whereas the aspirator flow that was estimated using this method varied between 190
and 370 Ln/min, which is outside the reference measurement range. However, the
error (if any) must be quite small, since a wrong estimation would result in a deviation
of the relative humidity for all experiments, which was not apparent in the results, as
shown by the small mean difference (0 % RH).
Although the relative humidity was found particularly useful for production
of amorphous materials by spray drying, the relative humidity was also used as a
variable to improve the output of the model. Without information on relative
humidity, it remains unknown how much liquid can actually be evaporated. Therefore,
the assumption was made that all the liquid fed into the spray dryer was evaporated.
This also meant that when the liquid feed flow in the model was set higher than what
could in practice be evaporated due to relative humidity limitations at the outlet, the
model would simply use this information and return an outlet temperature based on
unrealistic circumstances. In other words, the model would predict a combination of
liquid feed flow and outlet temperature that would in practice result in a wet product.
However, since the relative humidity calculation was added to the model, the
assumption that all liquid is evaporated was no longer necessary. Instead, the relative
humidity could be coupled to the amount of liquid evaporated to cap the relative
humidity at 100 %. If the calculated relative humidity would be higher than 100 %, the
model would simply reduce the amount of evaporated liquid until a relative humidity
of 100 % is reached. Therefore, the model will no longer predict a relative humidity
above 100%, and does not overestimate the amount of liquid evaporated in case too
much liquid is sprayed into the modeled spray dryer. Although such conditions are
very unlikely to be sought after in a spray drying process, the addition of such
calculations does help in reducing the amount of misinformation that could otherwise
be obtained by using the model. In addition to the liquid feed flow, knowing the
relative humidity also allows one to calculate the adiabatic saturation temperature,
which is close to the wet bulb temperature and could be used as an indication of the
product temperature during evaporation of the liquid prior to crust formation.
However, no experiments have been performed to validate these additions. Besides
the relative humidity, extending the model was also found to be quite useful for less

common spray dryer configurations. For example, spray drying is usually performed
on a single liquid solution. However, there are many interesting applications in which
two separate solutions are introduced into the spray dryer to form a mixture with the
use of a 3 or 4-fluid nozzle [31, 32]. Therefore, a second liquid stream was added to
the model, which enabled the prediction of the outlet conditions depending on the
ratio of the two liquid feed flows.
Finally, the good estimate of the relative humidity at the outlet of the spray
dryer enabled the prediction of dried product conditions. By combining the modeled
outlet conditions with DVS and DSC measurements, the influence on the yield of the
dried product could be predicted. However, one could argue that the outlet
temperature of the spray dryer did not correspond to the temperature at which the
DVS measurements were performed. Therefore, the moisture content of trehalose
and inulin that was calculated could deviate significantly from the actual value.
However, DVS isotherms of trehalose measured at 45 and 65 °C indicated that the
moisture content did not change with temperature (data not shown). Therefore, DVS
measurements at 25 °C could be used to calculate the glass transition temperature of
trehalose and inulin at the spray dryer outlet temperature between 32 and 39 °C.
When the glass transition temperature of trehalose or inulin at the modeled outlet
conditions was predicted to be lower than the outlet temperature, the yield was lower
than when the glass transition temperature was predicted to be higher than the outlet
temperature. This shows that the model allows a wide variety of assessments to be
made before spray drying experiments are performed, when the model is combined
with other analytical techniques.
Conclusion
A spray dryer model is presented that is both clear to understand for experienced and
novice users, and also readily available online for everyone. Due to the use of open
source software for the development, free use of the model is ensured. It was shown
that the model can predict the outlet conditions very well for a wide range of spray
dryer settings, which enables the user to move from a trial-and-error approach to a
quality-by-design approach. In addition, the model can easily be adapted for other
types of spray dryers and combined with other analytical techniques such as DSC and
DVS to get a better indication of the product properties prior to spray drying.
References
1. Pharmaceutical quality for the 21st century A risk-based approach progress

report. U.S. Food and Drug Administration; 2007 [cited 2013 15 august];
Available from:
78 | Chapter 4
http://www.fda.gov/AboutFDA/CentersOffices/OfficeofMedicalProductsa
ndTobacco/CDER/ucm128080.htm.
2. Pharmaceutical Development Q8, Quality Risk Management Q9,
Pharmaceutical Quality System Q10. International Conference on
Harmonisation of Technical Requirements for Registration of
Pharmaceuticals for Human Use (ICH); 1996 - 2012 [cited 2013 15 August];
Available from:
http://www.ich.org/products/guidelines/quality/article/quality-
guidelines.html.
3. Ivey JW, Vehring R. The use of modeling in spray drying of emulsions and
suspensions accelerates formulation and process development. Computers
& Chemical Engineering. 2010;34(7):1036-40.
4. Baldinger A, Clerdent L, Rantanen J, Yang M, Grohganz H. Quality by design
approach in the optimization of the spray-drying process. Pharm Dev
Technol. 2012;17(4):389-97.
5. Dobry D, Settell D, Baumann J, Ray R, Graham L, Beyerinck R. A model-
based methodology for spray-drying process development. Journal of
Pharmaceutical Innovation. 2009;4(3):133-42.
6. Lebrun P, Krier F, Mantanus J, Grohganz H, Yang M, Rozet E, et al. Design
space approach in the optimization of the spray-drying process. Eur J Pharm
Biopharm. 2012;80(1):226-34.
7. Mezhericher M, Levy A, Borde I. Spray drying modelling based on advanced
droplet drying kinetics. Chemical Engineering and Processing: Process
Intensification. 2010;49(11):1205-13.
8. Crowe LM, Reid DS, Crowe JH. Is trehalose special for preserving dry
biomaterials? Biophys J. 1996;71(4):2087-93.
9. Geeraedts F, Saluja V, ter Veer W, Amorij JP, Frijlink HW, Wilschut J, et al.
Preservation of the Immunogenicity of Dry-powder Influenza H5N1 Whole
Inactivated Virus Vaccine at Elevated Storage Temperatures. AAPS J.
2010;12(2):215-22.
10. Francia F, Dezi M, Mallardi A, Palazzo G, Cordone L, Venturoli G.
Protein−Matrix Coupling/Uncoupling in “Dry” Systems of Photosynthetic
Reaction Center Embedded in Trehalose/Sucrose: The Origin of Trehalose
Peculiarity. J Am Chem Soc. 2008;130(31):10240-6.
11. Liao YH, Brown MB, Nazir T, Quader A, Martin GP. Effects of sucrose and
trehalose on the preservation of the native structure of spray-dried lysozyme.
Pharm Res. 2002;19(12):1847-53.
12. Lai MC, Topp EM. Solid-state chemical stability of proteins and peptides. J
Pharm Sci. 1999;88(5):489-500.

Processing. 1989;68(10):65-70.
16. Kemmer G, Keller S. Nonlinear least-squares data fitting in Excel
spreadsheets. Nat Protocols. 2010;5(2):267-81.
Pharm Sci. 2009;98(9):2886-908.
2004;256-257(1):319-27.
20. Costantino HR, Andya JD, Nguyen PA, Dasovich N, Sweeney TD, Shire SJ,
et al. Effect of mannitol crystallization on the stability and aerosol
performance of a spray-dried pharmaceutical protein, recombinant
humanized anti-IgE monoclonal antibody. J Pharm Sci. 1998;87(11):1406-11.
Res. 1995;12(6):799-806.
22. Goula AM, Adamopoulos KG. Spray Drying Performance of a Laboratory
Spray Dryer for Tomato Powder Preparation. Drying Technology.
2003;21(7):1273-89.
23. Brown AM. A step-by-step guide to non-linear regression analysis of
experimental data using a Microsoft Excel spreadsheet. Comput Methods
Programs Biomed. 2001;65(3):191-200.
24. Gordon M, Taylor JS. Ideal copolymers and the second-order transitions of
synthetic rubbers. i. non-crystalline copolymers. Journal of Applied
Chemistry. 1952;2(9):493-500.
1969;163(3873):1325-6.
80 | Chapter 4
2004;93(4):047801.
28. Chen T, Fowler A, Toner M. Literature Review: Supplemented Phase
Diagram of the Trehalose–Water Binary Mixture. Cryobiology.
2000;40(3):277-82.
29. Roos Y. Melting and glass transitions of low molecular weight carbohydrates.
Carbohydr Res. 1993;238(0):39-48.
30. Kawai K, Fukami K, Thanatuksorn P, Viriyarattanasak C, Kajiwara K. Effects
of moisture content, molecular weight, and crystallinity on the glass transition
temperature of inulin. Carbohydrate Polymers. 2011;83(2):934-9.
31. de Waard H, Grasmeijer N, Hinrichs WLJ, Eissens AC, Pfaffenbach PPF,
Frijlink HW. Preparation of drug nanocrystals by controlled crystallization:
Application of a 3-way nozzle to prevent premature crystallization for large
scale production. Eur J Pharm Sci. 2009;38(3):224-9.
32. Chen R, Tagawa M, Hoshi N, Ogura T, Okamoto H, Danjo K. Improved
Dissolution of an Insoluble Drug Using a 4-Fluid Nozzle Spray-Drying
Technique. Chem Pharm Bull (Tokyo). 2004;52(9):1066-70.

82 | Chapter 5
Chapter 5
An adaptable model for growth and/or shrinkage of droplets in the

respiratory tract during inhalation of aqueous particles
Niels Grasmeijer, Henderik W. Frijlink, Wouter L.J. Hinrichs
Accepted for publication (2015)
Model for growth and / or shrinkage of inhaled droplets | 83

Abstract
The site of deposition of pulmonary delivered aerosols is dependent on the aerosol’s

droplet size distribution, which may change during inhalation. The aim of this study
was to develop a freely accessible and adaptable model that describes the growth (due
to condensation) and shrinkage (due to evaporation) of inhaled droplets as a function
of the distance from the airway wall during various inhalation conditions, for a laminar
flow scenario. This was achieved by developing a model with which the evaporation
of water from a droplet surface or condensation of water onto the droplet surface can
be calculated. This model was then applied to a second model that describes the heat
and mass transfer from the airway wall to the inhaled aerosol. The latter was based on
the Weibel model. It was found that the growth and shrinkage of inhaled droplets
markedly differs, depending on the distance from the airway wall. Droplets near the
wall start to grow immediately due to fast water vapor transfer from the wall to the
cold inhaled air. This growth continues until the air reaches body temperature and is
fully saturated. However, droplets in the center of the airway first evaporate partly,
due to a delay in water vapor transfer from the airway wall, before they start to grow.
Depending on the conditions during inhalation, the droplet size distribution can
widen considerably, which may affect the lung deposition and thereby the efficacy of
the inhalation therapy. In conclusion, the model was able to show the effect of the
conditions in the respiratory tract on the growth and shrinkage of inhaled droplets
during standard inhalation conditions. Future developments can be aimed at
expanding the model to include turbulent flow and hygroscopic growth, to improve
the accuracy of the model and make it applicable to both droplets of solutions and dry
particles.
84 | Chapter 5
Introduction
Inhalation of drugs is widely applied in the treatment of asthma and COPD.

Furthermore, the pulmonary route is currently targeted for the treatment of diseases
that are not directly located or even associated with the respiratory tract. Due to the
proximity of the capillaries and blood vessels with the lining in the respiratory tract
and the leaky nature of the pulmonary membranes (especially in the alveoli), the
absorption of active pharmaceutical ingredients can be extremely rapid [1, 2].
However, the rate and amount of absorption heavily depend on the location of
deposition of the inhaled substance [3-5].
During a standard inhalation procedure, the largest particles (either dry
powder or droplets, although in this paper only aqueous droplets will be considered)
are deposited in the mouth and the back of the throat, meaning their effect is
essentially lost. The rest of the particles are deposited throughout the respiratory tract,
where the spread of deposition is mainly dependent on the particle size distribution of
the liquid aerosol or dry powder. This aerosol’s particle size distribution is ideally
tailored to obtain deposition at a location where it is desired for a specific application.
As a general rule of thumb, particles with an aerodynamic diameter larger than 5 µm
are deposited in the throat and smaller than 1 µm are exhaled after inhalation [3].
Anything in between is deposited in the lung, with smaller particles having a higher
chance to reach the peripheral parts of the lung and even the alveoli and larger
particles being deposited more centrally in the respiratory tract. However, in reality it
is more complex.
When particles are inhaled, they are subjected to quickly changing conditions.
When the particles exit the inhaler together with the ambient air, conditions
surrounding the particles closely match those of the ambient air, which means
relatively low relative humidity and room temperature. During the passage through the
respiratory tract, the relative humidity quickly increases as well as the temperature. For
both dry powder and liquid aerosols, this could affect their aerodynamic diameter, as
the increased humidity could cause the dry powder particles to get wet and grow,
while droplets in the liquid aerosol could grow due to condensation or shrink due to
evaporation. If it would be possible to tell whether, and if so, how much the
aerodynamic diameter of inhaled particles is changed while traversing the respiratory
tract, it could aid in optimizing formulations for inhalation. By taking into account
growth and/or shrinkage of the particles, the deposition location in the respiratory
tract could be further tailored for each treatment.
Indeed, a lot of effort has gone into modeling the condensational growth of
solid submicron particles during inhalation to allow targeted deposition in the
respiratory tract, mainly by Longest et al. [6-17] but others as well [18, 19]. In addition,
a lot is known about the conditions to which inhaled particles and droplets are
subjected when inhaled [13, 16, 20]. Most of these studies revolve around the use of
computational fluid dynamics (CFD), which is an excellent method to accurately
model the complex flow profiles that exist inside the airways. However, such models
lack openness and accessibility for researchers who are not directly involved in such
fields. The same can be said for more analytical models that use pre-formulated sets of
equations that describe, for example, the evaporation rate from a droplet.
Furthermore, the complete models are not readily available online. Although simpler
models using standard heat and mass balances usually result in less detailed results,
their use often allows more openly accessible models to be developed that can be used
and adapted by others. In addition, given enough effort, the complexity of the model
can be increased to obtain more accurate results. Furthermore, we believe that
understanding the basic mechanics underlying such a model can also greatly increase
the understanding of the system under consideration.
Therefore, the aim of this study is to investigate the behavior, i.e. growth and
shrinkage, of inhaled particles in the respiratory tract with a theoretical model based
on standard heat and mass balances. This will be done by primarily focusing on
aqueous particles. Although dissolved components such as drugs and excipients will
be left out of the model to prevent it from becoming overly complex, the model will
be constructed such that it can be expanded by anyone who wishes to do so.
Therefore, the model will be developed in an open source platform called GNU
Octave and be made available freely. Consequently, a model will be obtained that can
form a basis for further expansion and development to create a more complex but
also more complete model of dry particle and droplet behavior in the respiratory tract
without resolving to CFD. Finally, the clinical relevance of the results of this study
towards inhalation practices will also be discussed.
Open source software

The model was developed entirely in GNU Octave, which is freely available [21].
Furthermore, the developed model itself is available online as supplementary data.
Model input parameters

The basis for the input of the models was obtained from the Respimat® soft mist
inhaler (Boehringer Ingelheim, Ingelheim, Germany). The inhaled jet that is ejected
from the inhaler contains about 15 mg of water and lasts for about 1.2 seconds.
Furthermore, the starting droplet diameter is 4.5 µm [22]. For the simulations, an
inhalation flow of 40 L/min was chosen, although for comparison, a fast inhalation
with an inhalation flow of 100 L/min will also be considered [23]. Finally, a room
86 | Chapter 5
temperature of either 293 K or 303 K, and a relative humidity of either 35, 50, or 70
% will be used. Note that all of these parameters can be easily customized in the
developed model, if desired.
Using these parameters, the volume of air that is surrounding each droplet
can be estimated. To do so, it is both assumed that each droplet will be of equal size
and separated by an equal distance from each other and therefore, the volume of air
surrounding each droplet will be the same. To simplify the model, the layer of air
around each droplet is considered to be a spherical shell instead of cubical. The
volume of air around each droplet is then simply calculated by dividing the amount of
air inhaled during the ejection of the jet from the inhaler by the amount of droplets
generated. Furthermore, the droplet is assumed to move at the same velocity as the
surrounding air, which implies that the air surrounding each droplet can be considered
as stagnant.
Modeling strategy
The complete model for predicting the behavior of droplets in the respiratory tract
consists of two separate models that work together. The first (droplet) model
describes the mass transfer of water and energy from and to a water droplet, which
will be described in the first section. The second (respiratory tract) model will describe
the conditions inside the respiratory tract, and will also be based on mass and energy
balances. With the respiratory tract model, the exchange of heat and mass between the
airway wall and the inhaled air is first calculated. The changes in relative humidity and
temperature are then passed on to the droplet model to calculate the exchange of heat
and mass between the droplet and the inhaled air. By keeping the models separated, it
is very easy to vary the amount of droplets that are considered during the calculation.
The development of the respiratory tract model will be described in further detail in
the second section.
Droplet model
The droplet model was developed as a 1-D model, meaning that the system is
completely symmetrical and changes in morphology are not considered.
The primary aim of the droplet model is to calculate the changes in droplet
radius as a function of the conditions to which the droplet is exposed inside the
respiratory tract. To achieve this, the basic equation shown in Equation 1 was used to
start the calculation.
∙∆
∙ (1)
∙

Where rt is the radius of the droplet at time t, mt-1 is the mass of liquid at the
previous time point, Φm is the mass flow rate from the droplet to the environment (or
vice versa when negative), Δt is the time step for the iterative calculation, and ρ is the
density of the liquid (1000 kg/m3 for water, assumed to be constant). Because the
mass flow rate (Φm) is still unknown, Equation 2 (derived from Fick’s first law) has to
be used to calculate the rate of mass flow from or towards the droplet.
∙ ∙ ∙∆
∙ (2)
∙ ∙
Where D is the diffusion coefficient of water molecules in the gas phase, Mw

is the molecular mass of the liquid, Δp is the pressure difference between the partial
water vapor pressure at the surface of the droplet and the partial water vapor pressure
in the gas phase, R is the gas constant, T is the temperature at the surface of the
droplet (K), rd is the radius of the droplet, and ra is the total radius of the droplet and
gas phase surrounding the droplet. Here, it is assumed that the partial water vapor
pressure at the surface of the droplet is equal to the saturation pressure. Therefore, the
pressure gradient, Δp, can be calculated with both the Antoine equation and the ideal
gas law (Equation 3 and Equation 4, respectively).
10 (3)
∙ ∙ ∙
(4)
Where psat is the saturated vapor pressure (Pa), A, B, and C are the Antoine
constants (for water these are 10.20, 1730.63, and -39.72, respectively [24]), T is the
temperature (K), pvap is the partial water vapor pressure (Pa), x is the specific humidity
(defined as the mass of water divided by the mass of dry air), and ρg is the density of
the gas phase. Besides the partial water vapor pressure gradient, the above equations
can also be used to calculate the relative humidity (RH), which is defined as pvap / psat ·
100.
The temperature in the equations for the mass flow rate (Equation 2) and
both the vapor pressures (Equation 3 and 4) can be calculated with Equation 5.
∙∆
(5)
∙
Where Tt-1 is the temperature at the previous time point, Q is the heat flow
rate, Cp is the specific heat capacity of the considered phase, and m is the mass. The
88 | Chapter 5
heat flow rate consists of heat flow through conduction and evaporation, which can
be calculated using Equation 6 (derived from Fourier’s law) and Equation 7,
respectively. Note that equation 2 and 6 are similar due to the fact that the base
equations for mass and heat transfer (Fick’s law and Fourier’s law, respectively) are
similar as well.
∙ ∙∆
∙ (6)
∙ (7)
Where λ is the thermal conductivity (W/m/K), ΔT is the temperature

difference between the temperature at the surface of the droplet and the temperature
at the boundary of the gas layer surrounding the droplet, and Hevap is the heat of
evaporation of the liquid. Note that the heat of evaporation is equal to the negative of
the heat of condensation. Whether or not the heat is added to (condensation) or
subtracted from (evaporation) the heat of the droplet purely depends on whether
mass is flowing towards or from the droplet, respectively. By iterating the Equation 1 -
7, the radius of the droplet in time can be calculated as a function of the conditions to
which the droplet is subjected.
Respiratory tract model

To simulate the conditions inside the respiratory tract to which the droplet is
subjected, a second model was developed. Whereas the droplet model is used to
calculate the exchange of heat and water between a droplet and the surrounding gas
phase, the respiratory tract model calculates the exchange of heat and water between
the airway wall and the inhaled mist. Furthermore, because the dimensions of the
airways that a droplet is passing through changes with each generation (sections of the
airway between bifurcations, with the trachea being generation zero), these are taken
into account as well. These dimensions are given by the Weibel lung model, and are
shown together with the cumulative residence time of the droplets in table 1 [25].
In this respiratory tract model, the heat and mass transfer that occurs during
passage from the inhaler through the mouth to the beginning of the trachea is not
considered. This is done to simplify the model, since the dimensions of the mouth
during inhalation are not well defined. The aerosol that is inhaled will be at ambient
conditions, meaning room temperature and RH, and will start to heat up the moment
it reaches the trachea. During the travel through the airways heat and mass is
transferred from the airway wall to the center of the airway, increasing both the
temperature and the RH of the air surrounding the droplets. This can then lead to

evaporation or condensation from or onto the droplets, until equilibrium is reached or
droplets are evaporated. In reality, heat and water will also be transferred to the
inhaled air during travel to the trachea, which will affect the generation where the
equilibrium is reached [13, 14, 20].
Table 1: Respiratory tract dimensions according to the Weibel lung model [25].
Generation Diameter Length Number Cumulative
(mm) (mm) residence
time (ms)a
0 18.0 120.0 1 45.8
1 12.2 48.0 2 62.6
2 8.3 19.0 4 68.8
3 5.6 8.0 8 71.2
4 4.5 13.0 16 76.1
5 3.5 10.7 32 81.1
6 3.2 9.9 64 88.9
7 3.0 9.1 128 101.0
8 2.7 8.2 256 119.2
9 2.4 7.4 512 146.0
10 2.2 6.6 1024 184.0
11 1.9 5.8 2048 235.4
12 1.7 5.0 4096 301.1
13 1.4 4.2 8192 378.6
14 1.1 3.3 16384 460.5
15 0.9 2.5 32768 533.0
16 0.6 1.7 65536 580.2
17 0.6 1.5 131072 652.9
18 0.5 1.2 262144 761.3
19 0.5 1.0 524288 915.7
20 0.5 0.9 1048576 1159.6
21 0.5 0.8 2097152 1534.8
22 0.4 0.6 4194304 2092.6
23 0.4 0.5 8388608 2883.2
a Cumulative residence time for an inhalation flow rate of 40 L/min.
For this respiratory tract model, the airway is considered to be a smooth

cylinder consisting of concentric cylinders to divide the area inside the airway (see
Figure 1). The content of each of the cylinders are considered to be ideally mixed,
meaning that there is no variation in conditions within each cylinder at any time point.
90 | Chapter 5
This will result in a stepwise gradient of temperature and RH from the airway wall to
the center of the airway, with the number of steps equal to the number of chosen
concentric cylinders. Furthermore, at each bifurcation (when a new generation is
reached) the dimensions are changed such that the same number of cylinders will be
kept the same, together with the conditions in each ring at that time point. Although
in reality the flow might be split through the middle at each bifurcation leaving half
the inner cylinder on the outside, in this model the distribution of the concentric
cylinders inside the airway does not change when a bifurcation is passed. How this
might affect the result will be discussed. In addition, loss of droplets due to impaction
is not taken into account. Finally, the mass and heat transfer is modeled for a laminar
flow, thereby neglecting the slightly turbulent flow in the first two to four stages and
the effect of droplets on the flow (one-way coupling). Although this will impose an
error on the final estimated behavior of droplets inside the respiratory tract, the
importance and magnitude of this error will also be discussed.
Figure 1: Schematic overview of the modeled airway with a top view on the left and a side view
on the right. Left: Each concentric cylinder transfers mass and heat but there is no variation in
conditions inside each cylinder. The arrows depict the transfer of mass and heat from the airway wall
through each cylinder to the center of the airway. Right: The airway is considered to be a straight smooth
cylinder without bifurcations but with sudden changes in diameter, through which a slice of the inhaled
mist is followed.
The calculations for the respiratory tract model are comparable to the
calculations that are done in the droplet model. The mass transfer between the

concentric cylinders is calculated using Equation 8 and the heat conduction is
calculated using Equation 9, derived from Fick’s law and Fourier’s law, respectively.
∙ ∙ ∙ ∙∆
∙ (8)
∙ ∙
∙ ∙ ∙∆
∙ (9)
Where h is the height of the considered layer of mist moving through the
respiratory tract (arbitrarily chosen as the diameter of the droplet with its surrounding
gas layer), rn and rn+1 are the radii of two adjacent concentric cylinders n and n+1 in
the airway, respectively, and Δp and ΔT are the partial water vapor pressure and
temperature of cylinder n and n+1, respectively.
Finally, the evaporation of water from the airway wall to the airway is
assumed to occur through a small 1 µm boundary layer with 99.5% RH at body
temperature [16], and for the heat transfer to the airway, the thickness of the airway
wall decreases from 2.75 mm in the trachea to about 1 mm at generation 11, after
which it is assumed to remain constant [26]. The thermal conductivity, λ, used for the
airway wall is about 0.28 W/m/K [27]. It is important to realize that the outer
cylinder, due to the ideally mixed conditions, will not be at the body temperature from
the start.
Combined model
The droplet model and the respiratory tract model are combined to calculate the
growth and shrinkage of a droplet in each concentric cylinder in the respiratory tract
model. For each droplet, the heat and water transfer from or onto the droplet is
calculated. In addition, the heat and water transfer between each concentric cylinder in
the airway is also calculated. By adding the heat and water transfer from the airway
and the droplet, the total change in temperature and relative humidity in each
concentric cylinder and droplet is obtained after each time step. Furthermore, after the
simulated time in each generation surpasses the residence time of that specific
generation (Table 1), the dimensions of the concentric cylinders are instantly changed
to that of the next generation. To keep calculation times limited, while still obtaining a
good resolution of the size distribution of droplets inside the airway, the airway was
divided into 30 cylinders and thus 30 droplets were modeled between the airway wall
and the center of the airway.
92 | Chapter 5
Results
Droplet model
A model was developed that describes the heat and mass transfer to and from a liquid
droplet. To validate the accuracy of the droplet model alone (without external
influences, e.g. as imposed by the respiratory tract model), the output as shown in
Table 2 was compared to data found in a psychrometric chart. This chart gives
information on the temperature of a droplet dispersed in air with a specific
temperature and humidity. When a droplet is exposed to air with a moisture content
below the dew point, the droplet will evaporate. Due to the evaporation, the
temperature of the droplet will decrease to a specific temperature, the wet bulb
temperature, which is dependent on the rate of evaporation. This effect is referred to
as evaporative cooling. The rate of evaporation, and consequently the equilibrium
temperature of the droplet, is dependent on the relative humidity, the ambient
temperature, the type of liquid (in this case only water is considered), and the velocity
difference between the ambient air and the droplet. For this case it was assumed that
the velocity of the droplet is equal to the velocity of the surrounding air, i.e. there is a
stagnant layer of air surrounding the droplet. Furthermore, the droplet diameter at the
start of the simulation was 4.5 µm. Finally, the droplet temperature (Tw model) was
calculated at three different relative humidities, i.e. 0 %, 50 %, and 95%, at an ambient
temperature of 293 K, and compared to values found in a psychrometric chart at
similar conditions (Tw chart). It was found that the calculated droplet temperatures
were in full agreement with values found in psychrometric charts. Therefore, it can be
concluded that the mass and heat transfer in the droplet model is modeled correctly
and accurately. Since the same calculations were used for the mass and heat transfer
between the airway wall and the airway, it can be assumed that the results will also
correspond well to the conditions set for this model (i.e. laminar flow).
Table 2: Comparison of calculated droplet temperature in stagnant air at 0 %, 50 %, and 95 %

RH with the wet bulb temperature found in psychrometric charts.
RH (%) Tw model (K)a Tw chart (K)

0 278.7 279
50 286.8 287
95 292.6 293
a The wet bulb temperature was reached within 1 ms, after which the droplet temperature remained
constant
To further show the application of the droplet model to the growth of

droplets due to condensation, a simulation was run for droplets with a diameter of 0.9

µm suspended in air of 37 °C with a relative humidity of 99.5, 100, 101, or 104 %. The
volume of air surrounding the droplets was high enough to negate any effect of
evaporation and condensation of the droplets on the relative humidity. These settings
were specifically chosen to be able to compare the results to data reported by Longest
et al. [17], to determine whether the rate of condensation is correctly modeled (see
Figure 2). Although the values reported by Longest et al. include the effect of
hygroscopic growth (starting droplets consisted of a 0.5 % w/v albuterol sulfate
solution), at higher relative humidity it is expected that growth due to condensation
will be high, causing a high dilution of the solution and in turn reducing the effect of
hygroscopic growth.
Figure 2: Condensational growth of droplets suspended in air at 37 °C and a relative humidity of

99.5 % (- and o), 100 % (- - and Δ), 101 % (- · - and ), or 104 % (· · · and ). Lines depict values
calculated with the droplet model, whereas symbols represent values reported by Longest et al. [17].
Values reported by Longest et al. include hygroscopic effects. Original droplet diameter at start of
simulation was 0.9 µm.
At a relative humidity of 104 % the growth rate corresponds well with the
values reported by Longest et al. for condensational growth of droplets containing
94 | Chapter 5
albuterol sulfate. When the relative humidity is 101 %, the effect of hygroscopic
growth is noticeable as the condensational growth rate calculated with our droplet
model is slightly lower. However, at a relative humidity of 100 % and 99.5 % the
difference is much more pronounced: droplets without solutes evaporate, whereas
values reported by Longest et al. still show growth due to hygroscopic effects.
Droplet growth and shrinkage – Conditions in the respiratory tract

The airway was divided into 30 concentric cylinders and for each cylinder the behavior
of a droplet was modeled. For the default simulation, the ambient air temperature and
relative humidity used were 293 K and 35 %, respectively. In addition, the starting
diameter for each droplet was 4.5 µm. For each of the droplets the surrounding air
temperature and relative humidity are shown in time (Figure 3). Furthermore, the
diameter of each of the droplets in time is shown in relation to the generation where
the droplets are located during the simulation (Figure 4). Finally, the droplet size
distribution was calculated when the system reached equilibrium and the droplet
diameter did not change any further (Figure 5).
The relative humidity and the temperature close to the wall of the airway
increased very rapidly. However, due to the assumption that the flow is laminar, the
temperature in the center of the airway does not increase immediately. Only after
about 0.075 s (less for locations closer to the airway wall) the temperature in the
center of the airway increased to body temperature. Therefore, depending on the
location inside the airway, inhaled droplets are subjected to different conditions.
A noticeably high relative humidity was found during the initial entry of the
colder inhaled air into the trachea of about 240 % closest to the wall, implying
supersaturation. This quickly decreases again as the temperature increases and the
water vapor diffuses to the center of the airway. During the stage where the
temperature of the airway has not yet reached body temperature, the simulated relative
humidity never decreased below 110 %. In fact, a slight increase to about 125 % is
seen at the same time when the temperature of the air further from the airway wall
starts to increase rapidly. This is due to a quick decrease in the diameter of the airway,
by which the diffusion rate of water vapor and heat rapidly increases. Another effect
that results from a change in the diameter of the airway (and thus the modeled
concentric cylinders) is the slight drop in temperature that can be observed at several
time points. Here, the heat diffusion towards the center is suddenly increased,
decreasing the temperature near the airway wall until a new equilibrium is reached and
the temperature increases again. The effect of the supersaturation, i.e. relative
humidities higher than 100%, on the behavior of the droplets will be investigated
further below in the results section entitled: “Droplet behavior – Maximum relative

humidity”, as nucleation and formation of separate water droplets which might occur
by this supersaturation will affect the results shown in Figures 3 - 5.
Figure 3: Relative humidity (black) and temperature (grey) in the airway during inhalation. The
results are shown for a simulation with 30 concentric cylinders with 30 modeled droplets. Locations
closest to the airway wall are shown at the top and locations closer to the center of the airway are shown
at the bottom. Inhaled air was assumed to have 35 % RH and 293 K. Inhalation flow rate was 40 L/min.
A clear influence of the location of the droplet in the airway on the droplet
behavior was found. Droplets closest to the airway wall grew, while the droplets in the
center of the airway first shrunk and after a while started to grow. This is clearly the
result of the higher relative humidity and temperature near the airway wall and the
lower relative humidity and temperature at the center of the airway. Air near the
airway wall is quickly humidified and heated by the airway wall, which causes
condensation on the cooler droplets. On the other hand, moisture and heat from the
airway wall takes a while to reach the air in the center of the airway (farthest from the
airway wall). Therefore, droplets near the center of the airway will first evaporate until
the air is saturated, which is when the droplet size reaches a plateau value. Due to
evaporative cooling, the temperature of the air will also decrease. When the heat and
moisture from the airway wall reaches these air layers, the relative humidity and
temperature increases rapidly, resulting in condensation on the droplets. Finally, no
96 | Chapter 5
large changes in droplet size occur after generation 9-10 because equilibrium was
reached, although there is a slight and steady evaporation of the droplets due to the
lower relative humidity of the airway wall (99.5 %).
Figure 4: Change in droplet size (black) while traveling through the airways depending on the
distance from the airway wall. The stepwise line (grey) depicts changes in generation of the airway.
Droplets closest to the airway wall are shown at the top, while droplets in the center are shown at the
bottom. Original droplet diameter at start of simulation was 4.5 µm, inhaled air with 35 % RH, 293 K, 40
L/min.
When the actual droplet size distribution after reaching equilibrium is

considered (Figure 5), a large widening of the droplet size distribution is found, ending
up with a major volume fraction of the droplets larger than the originally inhaled
droplets. Under the simulated circumstances (inhaled air with 35 % RH, 293 K, 40
L/min) about 63 % of the total droplet volume had a diameter above 4.5 µm, which
was the original size of the inhaled droplets.

Figure 5: Droplet size distribution of modeled droplets at the end of generation 10. The original
droplet diameter at the start of the simulation was 4.5 µm, inhaled air with 35 % RH, 293 K, inhaled at 40
L/min.
Droplet behavior – Maximum relative humidity

In the previous simulation, the maximum relative humidity in the airway was not
limited. In reality, however, the relative humidity might be limited by nucleation and
formation of separate water droplets (mist). Although in the current model, the
formation of these droplets is not considered, the maximum relative humidity can
easily be changed. Because the relative humidity has a large influence on the
condensation and evaporation rate (and thus droplet growth and shrinking), a
simulation was run to compare three relative humidity settings: unlimited (same as in
Figures 3-5), limited to 104 %, and limited to 100 % (Figure 6). The relative humidity
of 104 % was added to match that used by Longest et al. [15, 17], in addition, it was
added to serve as a scenario for modest supersaturation, i.e. between no
supersaturation (100 % RH) and unlimited supersaturation.
98 | Chapter 5
Figure 6: Comparison of equilibrium droplet size distribution with relative humidity in the
airway being: limited to 100 % (-), limited to 104 % (- -), and unlimited (- · -) relative humidity.
The original droplet diameter at the start of the simulation was 4.5 µm, inhaled air with 35 % RH, 293K,
inhaled at 40 L/min.
When the maximum relative humidity is limited to a maximum of 100 % or

104 %, a clear overall decrease of the droplet size was found when compared to
droplets at unlimited maximum relative humidity. However, it is questionable whether
limiting the maximum relative humidity would result in a better representation of
conditions during inhalation. It is well known that the degree of supersaturation
required for nucleation depends strongly on purity of the liquid and the air. Under
ideal circumstances (no impurities), the relative humidity corresponding to the
supersaturation required for homogeneous nucleation can be as high as 2200 % [28].
The presence of any nuclei, either droplets or other particles, will lower the relative
humidity at which condensation occurs. Therefore, in our situation, a lot of the water
might actually condensate on the inhaled droplets instead of forming new droplets. In
addition, by limiting the relative humidity in the model, a lot of the water that will
evaporate from the airway wall to the cooler and drier inhaled air is simply ignored. In
other words, all the extra water that evaporates from the airway wall nucleates and

does not interact with the droplets in any way. Most likely, in reality some or perhaps
most of the moisture may condense onto or agglomerate with the droplets. Finally,
the time during which the relative humidity reaches values well above 100 % is quite
short. The initial burst in relative humidity before being dissipated by transport and
condensation is less than 0.02 s (Figure 3), and the total time before equilibrium is
reached with the airway wall is about 0.12 s. Although it is unknown what time scales
are involved with the formation of nuclei, it might be possible that the high relative
humidities observed in Figure 3 can be reached for such short duration, even when
impurities are present in the inhaled air (besides the droplets themselves). Therefore, it
is expected that the results with the unlimited model are more accurate than the model
with limitations with respect to RH. However, below both unlimited and limited
relative humidity will be considered to better assess the importance of different
conditions during inhalation on the behavior of inhaled droplets in the respiratory
tract.
Droplet behavior – Inlet conditions

Because the conditions at which an inhaler is used can vary significantly, the
equilibrium droplet size distribution of droplets inhaled under varying conditions was
calculated (Figure 7). The conditions chosen were an inhaled air temperature of 293 K
with a relative humidity of 35 %, 50 %, and 70 %. Furthermore, the influence of
inhaled air temperature was determined with an air temperature of 303 K and a
relative humidity of 35 %. These conditions were simulated with a relative humidity
that was either not limited, limited to a maximum of 104 %, or limited to a maximum
or 100 %. Finally, all conditions were simulated during a slow inhalation (40 L/min)
and a fast inhalation (100 L/min).
Although the ambient relative humidity did not have a great effect on the
maximum droplet diameter, there was a noticeable effect on the smallest droplet size.
When air with a higher relative humidity was chosen, the diameter of the smallest
droplets increased. This is caused by a faster saturation of the air closer to the center
of the airway when it is not yet heated by the airway wall. Therefore, the droplets
evaporate less and the minimum droplet diameter increases.
When the inhaled air temperature was 303 K instead of 293 K, a large overall
decrease of the droplet diameter was observed. For droplets closest to the airway wall
the higher inhaled air temperature resulted in a lower super saturation of the air with
moisture (data not shown). Therefore, the condensation rate onto the droplets was
lower than onto droplets in air with a lower temperature. This effect was absent when
the relative humidity in the airway was limited to a maximum of 104 % and 100 %,
and the maximum droplet size does not change with a change in inhaled air
temperature. Droplets closer to the center of the airway became smaller with increased
100 | Chapter 5
inhaled air temperature, due to the higher moisture capacity. Air with a fixed relative
humidity but a higher temperature will have a higher absolute moisture content before
becoming saturated. Therefore, droplets can evaporate more before the air is saturated
and equilibrium is reached.
When the inhalation is performed quickly, i.e. at 100 L/min instead of 40
L/min, an overall decrease in droplet size was found. In addition, the effects of
changing the inhalation temperature and relative humidity were more pronounced as
well. At the same relative humidity and temperature of the inhaled air, the total
volume of air surrounding each droplet is increased when the inhalation flow is
increased. Therefore, much more water can evaporate before the air is saturated,
resulting in smaller droplets near the center of the airway. The effect is smaller near
the airway wall, where it takes a bit longer for the air to be saturated and cause slightly
slower condensation before equilibrium is reached. If besides the inhalation flow also
the temperature or the relative humidity is changed, the effect on droplet size
distribution becomes more pronounced as the change in moisture capacity due to
both parameters also depends on the volume of air. The same change in temperature
from 293 K to 303 K or a change in relative humidity from 35 % to 70 % will result in
a much bigger change in moisture capacity at a high inhalation flow rate than at a
lower inhalation flow rate.
When the relative humidity in the respiratory tract is limited to a maximum of
104 %, there is an overall decrease in the droplet size distribution compared to the
scenario where the relative humidity in the respiratory tract is unlimited. Although
droplets in the center of the airway only reduce less than 0.5 µm in size, due to the
larger decrease in droplet size near the airway wall, there is a noticeable reduction in
the width of the droplet size distribution. Furthermore, there is a decrease of the
volume fraction of droplets with a size larger than the originally inhaled droplets of
4.5 µm.
Finally, when the relative humidity was limited to 100 %, no growth of
droplets is observed. Instead, the droplet diameter does not change when close to the
airway wall and decreases when closer to the center of the airway. Compared to the
scenario where the relative humidity is limited to 104 %, the width of the droplet size
distribution remains the same, although the distribution shifts towards smaller
droplets with no fraction larger than the original droplet size.

Figure 7: Comparison of equilibrium droplet size distribution after slow inhalation of air (40
L/min, left) or fast inhalation of air (100 L/min, right), with a relative humidity that was not
limited (top), limited to a maximum of 104 % (middle), or limited to a maximum of 100 %
(bottom). Simulated inhalation conditions were a relative humidity of 35 % (-), 50 % (- -), and 70
% (- · -) at 293 K, and 35 % at 303 K (···). The original droplet diameter at the start of the simulation
was 4.5 µm.
102 | Chapter 5
Discussion
It is shown that the developed droplet model can be used to accurately estimate the
shrinkage or growth of a droplet by evaporation or condensation, respectively. When
this droplet model is coupled to the Weibel model of the lung to obtain the
respiratory tract model, estimations can be made regarding the behavior of inhaled
droplets in the respiratory tract. Our simulations indicate that whether a droplet grows
or shrinks is dependent on where the droplet is situated with regard to the airway wall
(Figures 3 - 5). When close to the wall, the relatively cold droplet is quickly subjected
to an environment of higher temperature and humidity which promotes
condensational growth, resulting in larger droplets. However, droplets farther away
from the wall will first evaporate before the heat and moisture from the airway wall
will reach these droplets. These changes in droplet size distribution happen rather
quickly, as it was found that at generation 9 – 10 the conditions inside the airway will
have reached equilibrium and no further change in droplet size occurs. Therefore,
there could be a substantial effect of this change in droplet size distribution on the
deposition site. Instead of reaching the lower airways, the larger droplets (which are,
as shown in the results, already closer to the airway wall) will be deposited higher in
the respiratory tract than might be originally intended, thereby reducing the efficacy of
the inhalation therapy, while the smaller droplets will deposit deeper in the respiratory
tract than originally intended. Knowing the behavior of the droplets inside the
respiratory tract can help in countering the effect with either changes in formulation,
design of the inhalation device, or inhalation instructions.
Although the idea of droplet growth and shrinkage in the respiratory tract is
not new, we have developed a basic model that enables one to estimate the effect of
various conditions on inhaled droplets. The model demonstrates the relevance of
temperature and relative humidity of the inhaled air, and inhalation flow rate to the
behavior of inhaled droplets in the respiratory tract. Furthermore, the model is made
available freely for anyone to use and adapt. Due to the step by step approach using
discrete equations, we hope to enable users to easily understand and expand the model
to further improve the accuracy and perhaps complexity. While a more accurate and
complete result might be obtained with a full computational fluid dynamics (CFD)
model, the simplicity of this model is easier to understand and adapt by researchers
that are not involved in the field of CFD.
The conditions at which the aerosol is inhaled determine the magnitude of the
effects on changes in size distribution. Both the relative humidity and the temperature
of the inhaled air were found to have a noticeable effect on the fraction of droplets
smaller than the inhaled droplets, and thus on the deposition in the respiratory tract.
With increasing relative humidity, the deposition will occur at the higher regions of
the respiratory tract, while with increasing temperature, the deposition will shift
toward lower regions of the respiratory tract. Although the conditions during
inhalation will change inevitably with location and season, it is something to keep in
mind when developing devices for worldwide applications. It could, for example, be
worthwhile to look into air conditioning methods that are built into the device to
reduce the effects when the deposition site is critical.
Also the rate at which the aerosol is inhaled is of importance. Especially in the
case of the Respimat® soft mist inhaler, which ejects the aerosol in a fixed timeframe,
the concentration of the droplets in the air mainly influences the evaporation of
droplets in the center of the airway. With more diluted droplets (fast inhalation), it
takes longer for the air in the center of the airway to saturate, allowing droplets to
evaporate more. This in turn results in a smaller droplet size than when the aerosol is
slowly inhaled. Therefore, depending on the device, the instructions should clearly
specify which type of inhalation is required. Furthermore, the resistance of the inhaler
can also play an important role here. A high resistance inhaler will promote slow
inhalation, whereas a low resistance inhaler will allow for a fast inhalation, but not
necessarily enforce one. Finally, whether or not the inhalation is performed fast or
slow will also affect the flow regime inside the respiratory tract. When the air is
inhaled slowly, the flow regime will be closer to laminar, also in the upper airways.
However, when the air is inhaled faster, the flow will be more turbulent, by which the
behavior of the droplets inside the respiratory tract will change. Also worth
mentioning is that the same result of a change in inhalation flow rate can be obtained
by an equal inverse change of the inhaled amount of droplets of the same size,
because the concentration of droplets in the inhaled air will in that case be the same.
Because the concentration of the generated droplets in the inhaled air can
affect the behavior of the droplets in the respiratory tract, the aerosol ejection profile
from the inhaler and the inhalation flow profile can also play a role. To minimize
variations due to growth and shrinkage of droplets during inhalation, a constant
inhalation flow and a square ejection profile, where the maximum throughput is
reached instantly and also ends instantly, is preferred. However, in reality there will be
a gradual increase and decrease in the ejected volume from the device, the droplets at
the beginning and at the end of the ejection will be diluted more than droplets ejected
in the middle, when the ejected flow is higher. In part, this could be counteracted by
the inhalation flow profile, which will also not be square. At the start of the inhalation
procedure the inhalation flow will increase until a maximum is reached and then
slowly decrease again. This will also affect the behavior of inhaled droplets in the
respiratory tract differently, similar to the inhalation flow rate. These notices could
also be taken into account when designing a new generation of Adaptive Aerosol
Delivery systems [29], with an even further improved control over lung deposition.
104 | Chapter 5
It has to be noted that there are several limitations to the model that should
be kept in mind. First of all, the model proposed in this study assumes a fully laminar
flow. While this is true for the largest part, the flow in the first few (two to four)
generations may be turbulent [20]. This will have an effect on the behavior of the
droplets in the respiratory tract. It can be expected that under turbulent conditions,
the difference between the center of the airway and the layers close to the airway wall
is smaller. Both the temperature and the relative humidity gradients will be smaller and
there will be some mixing of the droplets throughout the airway. This should result in
a smaller droplet size distribution, as droplets near the airway wall will grow less and
droplets near the center of the airway will evaporate less. In fact, in an ideally mixed
system, the droplets would all have the same size. This system is the opposite extreme
compared to our fully laminar system. However, when the flow becomes laminar after
2 - 4 generations, the droplet size distribution will most likely become wider again due
to the formation of a temperature and relative humidity gradient. It is possible that,
due to increased mass and heat transfer under turbulent conditions from the airway
wall to the airway, the average droplet size will increase when turbulent conditions are
considered in the first few generations. Therefore, it may be worthwhile to expand the
model in the future to also include the various flow regimes that are encountered in
the subsequent generations of the respiratory tract.
Furthermore, the heat and moisture transferred to the inhaled air during
passage from the mouth to the trachea that was omitted could change the location in
the respiratory tract where equilibrium is reached. If fully laminar flow were to be
assumed for this region as well then the resulting changes in droplet size would most
likely change very little, because the conditions would be similar as in the trachea.
More specifically, the droplets near the center will shrink to the same size, because the
air will be saturated as this also happens in the trachea (Figures 3 and 4) and only
droplets near the wall might have a bit more time to grow. However, equilibrium will
be reached sooner, perhaps after only a few generations depending on the inhalation
flow. However, it is known that the flow in this region is slightly turbulent and more
accurate results would be obtained by taking this into account when expanding the
model [20].
Another aspect that is not considered in the model, but might be very
important under laminar flow conditions, is the influence of the bifurcations. In our
model, the airway simply reduces in size (following the Weibel model) but retains the
internal distribution of the droplets. However, under fully laminar conditions, the
airway is split in half and the droplets at the center of the airway are suddenly close to
the airway wall and exposed to conditions that enhance condensational growth.
Therefore, the droplets that are originally located in the center will have less time to
evaporate before they start to grow after the bifurcation. This could cause an overall

increase in the droplet size due to limited evaporation of droplets in the center
compared to what is simulated in our model. However, it was also shown that the
evaporation of the droplets near the center of the airway reaches equilibrium due to
saturation of the air with moisture (Figures 3 and 4). This equilibrium is under most
conditions reached before the end of generation zero, the trachea, thus before any
bifurcations are encountered. Therefore, although the droplets may start to grow
sooner (upon entering generation one) the effect on the overall droplet size
distribution will be small.
Furthermore, the hygroscopicity of solutes is not included in the model. A lot
is known about the influence of solutes on the growth on both particles and droplets,
and it is possible to model these influences [6-17]. The model without hygroscopicity
can be used for conditions where droplet growth is mainly caused by condensation
instead of hygroscopicity, i.e. during inhalation of aerosols where the relative humidity
tends to reach over 100 %. However, for inhalation of solid particles or aerosols with
solutes at lower relative humidity there will be an underestimation of droplet growth
due to the sizeable contribution of hygroscopicity to particle growth. Because of the
stepwise calculation using discrete equations, the addition of such influence should be
quite straightforward and is intended to be done during future development.
Therefore, although current results will give a slight underestimation of the droplet
size due to the absence of solute hygroscopicity (Figure 2), future expansions of the
model can improve the accuracy. In addition to more accurate results for aerosols, this
will also make the model more suitable for particle behavior in the respiratory tract.
However, current results indicate that the same effects can be expected for dry
particles, since both relative humidity and temperature will dictate how much water
will be taken up by dry particles. Therefore, it can be assumed that particles near the
airway wall will grow more than particles in the center of the airway when the airflow
is laminar.
Finally, it is unknown exactly how much the exclusion of nucleation at an RH
> 100% will influence the behavior of inhaled droplets inside the respiratory tract. It is
clearly shown that the reduction of the maximum relative humidity due to possible
nucleation will definitely affect the droplet growth near the airway wall, as can be
expected. However, it is unknown how much water will nucleate and form separate
droplets that do not coalesce with the inhaled droplets. An experimental setup, as was
used by Longest et al. and Hindle et al. [10, 15, 17, 30], could help to tune the model
to more exact results. Despite the fact that the actual size distribution of the droplets
can shift depending on the effect of nucleation at high relative humidity, the practical
implications of the results remain unchanged.
106 | Chapter 5
Conclusion
The developed model was used to predict the influence of several parameters on the
behavior of inhaled aqueous droplets in the respiratory tract. It was found that, under
laminar conditions, the location of the droplet with respect to the airway wall
determines whether a droplet will grow or shrink. Droplets in the center of the airway
will first evaporate until the air is saturated, while droplets near the airway wall will
immediately start to grow due to condensation. The extent of the effects is also shown
to be largely influenced by the relative humidity and temperature of the inhaled air,
and the inhalation flow rate. Understanding how these factors influence the behavior
of inhaled droplets can help during formulation, design of the inhalation device, or
inhalation instructions. Furthermore, the model can easily be adapted to include
phenomena such as turbulent flow, hygroscopic growth, and others that can help to
improve the accuracy of the model. In addition, these additions could also make the
model applicable to dry powders.
Acknowledgements
The authors are especially thankful for the help of Paul Hagedoorn and Floris
Grasmeijer. Their ideas and expertise in the field of inhalation helped to shape this
paper.
References
1. Patton JS. Mechanisms of macromolecule absorption by the lungs. Advanced

Drug Delivery Reviews. 1996;19(1):3-36.
2. Rabinowitz JD, Lloyd PM, Munzar P, Myers DJ, Cross S, Damani R, et al.
Ultra-fast absorption of amorphous pure drug aerosols via deep lung
inhalation. J Pharm Sci. 2006;95(11):2438-51.
3. Labiris NR, Dolovich MB. Pulmonary drug delivery. Part I: Physiological
factors affecting therapeutic effectiveness of aerosolized medications. Br J
Clin Pharmacol. 2003;56(6):588-99.
4. Zanen P, Go LT, Lammers J-WJ. The optimal particle size for β-adrenergic
aerosols in mild asthmatics. Int J Pharm. 1994;107(3):211-7.
5. Usmani OS, Biddiscombe MF, Barnes PJ. Regional Lung Deposition and
Bronchodilator Response as a Function of β2-Agonist Particle Size. Am J
Respir Crit Care Med. 2005;172(12):1497-504.
6. Golshahi L, Tian G, Azimi M, Son Y-J, Walenga R, Longest PW, et al. The
Use of Condensational Growth Methods for Efficient Drug Delivery to the

Lungs during Noninvasive Ventilation High Flow Therapy. Pharm Res.
2013;30(11):2917-30.
7. Worth Longest P, Spence BM, Holbrook LT, Mossi KM, Son Y-J, Hindle M.
Production of inhalable submicrometer aerosols from conventional mesh
nebulizers for improved respiratory drug delivery. Journal of Aerosol Science.
2012;51(0):66-80.
8. Tian G, Longest PW, Li X, Hindle M. Targeting Aerosol Deposition to and
Within the Lung Airways Using Excipient Enhanced Growth. Journal of
Aerosol Medicine and Pulmonary Drug Delivery. 2013;26(5):248-65.
9. Longest PW, Tian G, Li X, Son Y-J, Hindle M. Performance of Combination
Drug and Hygroscopic Excipient Submicrometer Particles from a Softmist
Inhaler in a Characteristic Model of the Airways. Ann Biomed Eng.
2012;40(12):2596-610.
10. Longest PW, Hindle M. Condensational Growth of Combination Drug-
Excipient Submicrometer Particles for Targeted High Efficiency Pulmonary
Delivery: Comparison of CFD Predictions with Experimental Results. Pharm
Res. 2012;29(3):707-21.
11. Longest PW, Hindle M. Numerical Model to Characterize the Size Increase of
Combination Drug and Hygroscopic Excipient Nanoparticle Aerosols.
Aerosol science and technology : the journal of the American Association for
Aerosol Research. 2011;45(7):884-99.
12. Longest PW, Tian G, Hindle M. Improving the Lung Delivery of Nasally
Administered Aerosols During Noninvasive Ventilation—An Application of
Enhanced Condensational Growth (ECG). Journal of Aerosol Medicine and
Pulmonary Drug Delivery. 2011;24(2):103-18.
13. Tian G, Longest PW, Su G, Hindle M. Characterization of Respiratory Drug
Delivery with Enhanced Condensational Growth using an Individual Path
Model of the Entire Tracheobronchial Airways. Ann Biomed Eng.
2011;39(3):1136-53.
14. Hindle M, Longest PW. Evaluation of Enhanced Condensational Growth
(ECG) for Controlled Respiratory Drug Delivery in a Mouth-Throat and
Upper Tracheobronchial Model. Pharm Res. 2010;27(9):1800-11.
15. Longest PW, Hindle M. CFD simulations of enhanced condensational growth
(ECG) applied to respiratory drug delivery with comparisons to in vitro data.
Journal of Aerosol Science. 2010;41(8):805-20.
16. Ferron GA. The size of soluble aerosol particles as a function of the humidity
of the air. Application to the human respiratory tract. Journal of Aerosol
Science. 1977;8(4):251-67.
108 | Chapter 5
17. Longest PW, McLeskey JT, Hindle M. Characterization of Nanoaerosol Size
Change During Enhanced Condensational Growth. Aerosol Science and
Technology. 2010;44(6):473-83.
18. Kleinstreuer C, Zhang Z. Airflow and Particle Transport in the Human
Respiratory System. Annual Review of Fluid Mechanics. 2010;42(1):301-34.
19. Broday DM, Georgopoulos PG. Growth and Deposition of Hygroscopic
Particulate Matter in the Human Lungs. Aerosol Science and Technology.
2001;34(1):144-59.
20. Olson DE, Sudlow MF, Horsfield K, Filley GF. Convective patterns of flow
during inspiration. Arch Intern Med. 1973;131(1):51-7.
21. Eaton JW, Community, GNU Octave, Version: 3.8.2, Available from:
https://www.gnu.org/software/octave/index.html, Last accessed: 23 March
2015.
22. Dalby RN, Eicher J, Zierenberg B. Development of Respimat(®) Soft Mist™
Inhaler and its clinical utility in respiratory disorders. Medical Devices
(Auckland, NZ). 2011;4:145-55.
23. Brand P, Hederer B, Austen G, Dewberry H, Meyer T. Higher lung
deposition with Respimat(®) Soft Mist(™) Inhaler than HFA-MDI in COPD
patients with poor technique. Int J Chron Obstruct Pulmon Dis.
2008;3(4):763-70.
Processing. 1989;68(10):65-70.
25. Weibel ER. Morphometry of the human lung: Springer Berlin Heidelberg;
1963.
26. Liu X, Chen D, Tawhai MH, Hoffman EA, Sonka M, editors. Measurement,
evaluation and analysis of wall thickness of 3D airway trees across
bifurcations. Second international workshop on pulmonary image processing;
2009 2009; London.
27. Poppendiek HF, Greene ND, Chambers JE, Feigenbutz LV, Morehouse PM,
Randall R, et al. Thermal and Electrical Conductivities of Biological Fluids
and Tissues.1964 23 March 2015.
28. Hinds WC. Aerosol Technology: Properties, Behavior, and Measurement of
Airborne Particles. 2nd ed: Wiley-Interscience; 1999.
29. Dhand R. Intelligent Nebulizers in the Age of the Internet: The I-neb
Adaptive Aerosol Delivery (AAD) System. Journal of Aerosol Medicine and
Pulmonary Drug Delivery. 2010;23(S1):S1-S10.

30. Hindle M, Longest PW. Condensational Growth of Combination Drug-
Excipient Submicrometer Particles for Targeted High Efficiency Pulmonary
Delivery: Evaluation of Formulation and Delivery Device. The Journal of
pharmacy and pharmacology. 2012;64(9):1254-63.
110 | Chapter 5
112 | Chapter 6
Chapter 6
Model to predict inhomogeneous protein-sugar distribution in powders

prepared by spray drying
Niels Grasmeijer, Henderik W. Frijlink, Wouter L.J. Hinrichs
Protein-sugar distribution in powders prepared by spray drying | 113

Introduction
Depending on their application, spray dried products should often meet specific
physical properties. These physical properties can be a specific particle size
distribution and morphology, whether the particles are crystalline, glassy or rubbery,
and presence or absence of concentration gradients inside the particle, in case the
starting solution contains more than one solute. A typical field where such
information is particularly useful, is protein stabilization, accomplished with the aid of
a sugar. For such products, it is important that the protein is fully encapsulated in a
sugar matrix which should be in its glassy state. Therefore, understanding and the
ability to predict the final product properties may determine whether a product and
process can be developed successfully or not.
There are two main stages during the drying of a droplet in a spray dryer that
can be distinguished. The first stage begins at the moment that water starts
evaporating from the droplet. The evaporation is rapid and depends fully on the
relative humidity and temperature of the drying air around the droplet, and the
presence of dissolved components in the droplet (assuming a stagnant air layer,
otherwise relative air velocity is important as well). During evaporation, the solute
concentration in the droplet will increase until a critical threshold is reached after
which it starts to solidify. From that moment on, the evaporation rate slows down due
to the slower diffusion of moisture through the growing shell of solidified solute at
the surface of the droplet, which is the second drying stage.
The solidification of solute at the transition from first to second stage will
usually happen at the surface of the droplet due to several reasons. In general, solute
molecules will diffuse slower in solution than water molecules. During evaporation,
the surface gradually recedes towards the center of the droplet. To maintain a
homogenous mixture, the solute will have to diffuse towards the center, away from
the receding surface. However, this diffusion will be slower than the surface recession
resulting in accumulation at the surface until the concentration is high enough to form
a solid. Furthermore, when a solute is surface active, it will have a surface excess
concentration at the liquid-air interface, which may also influence how fast a solid
shell will be formed at the surface of a drying droplet. For protein stabilization this
issue is important to consider. First of all, many proteins are surface active and thus
have a tendency to be at the interface. This in turn promotes unfolding of the protein
because it is thermodynamically favorable when the more lipophilic inner parts of the
protein are in contact with relatively hydrophobic air surrounding the droplet.
Secondly, a solution will usually consist of multiple solutes. Specifically in the case of
protein stabilization, a standard formulation will at least contain a protein, a sugar, and
perhaps a buffer and salts or other excipients. Because proteins are larger molecules,
they diffuse much slower than other components, such as sugars. During drying of the
114 | Chapter 6
droplet, this can result in a separation of the solutes, where the protein will accumulate
much faster at the surface than the sugar. Therefore, when the drying is complete,
there can be a large fraction of protein at the surface that is not fully encapsulated in a
sugar matrix. Consequently, protein instability due to accumulation at the surface of a
drying droplet can occur because of both its surface active properties and its relatively
slow diffusion.
When the solid shell starts to form at the surface of the droplet at the end of
the first stage, evaporation is slowed down considerably due to a gradual viscosity
increase of the solid shell. Furthermore, during the second stage, the morphology of
the dried particles is determined as well. Although generally the morphology of spray
dried particles is very important from a product engineering perspective, the focus in
this chapter will be more on protein stabilization. Therefore, current efforts are more
focused on building the basis for a model that includes the diffusion characteristics of
the solutes, i.e. protein and sugar, and the glass transition temperature. Although
models exist that include the glass transition temperature, they are not used for
pharmaceutical application such as protein stabilization [1]. In addition, the glass
transition temperature is in these cases secondary to the formation of the dried
particle, whereas we believe that it is the primary property that determines the
transition from the first to the second stage.
The main goal will be to obtain a model that can predict the drying of the
droplet both during the first as well as the second stage in one continuous model,
without creating two separate models for the first and second stage. For this model,
an aqueous solution of trehalose and bovine serum albumin (BSA) is considered.
Trehalose was used as an example, since it is a well-known protein stabilizer, the
diffusion coefficient as a function of the concentration and temperature is known
from literature, and the Gordon-Taylor constants of water/trehalose and
trehalose/BSA mixtures are known [2-4]. The model will be used to follow the
distribution of solutes inside the droplet during evaporation using empirically
determined diffusion coefficients for trehalose and water, and an estimated diffusion
coefficient for BSA [2].
Methods
Open source software

Similar to chapter 5, the model was further developed in GNU Octave, which is freely
available [5]. Furthermore, the developed model itself is available online as
supplementary data as well as in Appendix B of this thesis.
General model structure and assumptions

The droplet model that was introduced in chapter 5 was further expanded to include
concentric spherical shells inside the droplet, as was also done by Gac and Gradorí [6].
In each shell, the diffusion of a solute, either only trehalose or both trehalose and
bovine serum albumin (BSA), and water is calculated, and together with the
evaporation, the concentration gradients inside the droplet are predicted. Here, the
exchange of mass and heat is calculated from the surface of a subshell to the surface
of a neighboring subshell (Figure 1). Furthermore, each subshell is ideally mixed,
implying that any heat or mass diffusing into a subshell is instantly homogeneously
mixed. Finally, the system is considered closed, meaning that there is no heat or mass
transfer to or from the outside. As a consequence, changes in temperature and
humidity changes in the air can occur.
Figure 1: Schematic representation of a model droplet with a surrounding stagnant layer of air.
Droplet and air layer are divided into concentric subshells, and heat and mass transfer occurs between the
surfaces, as depicted by the arrows. Initial subshell volumes of the droplet are the same. Note that the
ratio of air to droplet is not to scale, and the depicted number of subshells in the droplet and air layer are
not the same as used in further calculations.
116 | Chapter 6
Due to the evaporation of water, the droplet will shrink. During the
simulation, the volumes of the individual subshells of the droplet will be kept the
same, instead of letting the outer subshell volume decrease more than the other
subshells. In order to achieve this, the stepwise calculation is performed as follows.
First, the mass of water that evaporates from the outer subshell and the diffused mass
of solute and liquid throughout the droplet in a time step (Δt) are calculated. Next, the
new solute and liquid mass in each subshell can be calculated. Assuming that the
mixtures are ideal, the intermediate subshell volumes (sum of component masses
divided by their respective densities) are then also known. With this information, the
required volume changes that are needed to equalize the subshell volumes are
calculated. The result is that subshells that otherwise would have been smaller than
other subshells have received a small amount of liquid and solute mass from
neighboring subshells, whereas subshells that otherwise would have been larger than
other subshells have lost a small amount of liquid and solute mass to neighboring
subshells. This will result in an error, as there is some non-diffusional mass transfer
introduced which in reality does not take place. The significance of this error was
tested by comparing the results using the aforementioned method with those
generated using a second method, with which the non-diffusional mass transfer is
smaller or absent. The first method will be referred to as the ‘shrink method’, whereas
the latter will be referred to as the ‘merge method’. Although the ‘merge method’ may
be more accurate, the ‘shrink method’ is preferred due to its shorter calculation time.
Of importance is that with both methods the droplet will keep shrinking even after a
solid shell has been formed. This will be discussed further below.
Calculation of glass transition temperature

With the calculated component mass fractions and the aid of the secondary or ternary
Gordon-Taylor equation, the glass transition temperature inside the droplet can be
calculated to determine when the surface of the drying droplet becomes glassy
(Equation 1).
∙ . ∙ ∙ . ∙ ∙ .
(1)
∙ ∙
Here, subscripts s, w, and p refer to trehalose, water, and BSA, respectively.

Furthermore, w is the weight fraction, Tg is the glass transition temperature (394, 164,
and 150 K for trehalose, water, and BSA, respectively [3, 7-10]), and k is the Gordon-
Taylor constant of a trehalose/water mixture (ksw, 7.90 [3]) or a trehalose/BSA

mixture (ksp, 0.08 [4]). Finally, for trehalose only, wp becomes zero by which the
equation reduces to the secondary Gordon-Taylor equation (Equation 1 without the
the bracketed sections).
Calculation of diffusion coefficients

During the calculations, the diffusion coefficient of the different components will
change constantly with the temperature and concentration. To take these changes
into account, we designed a tailor made equation to fit the data of a study by Ekdawi-
Sever et al. with which the diffusion coefficient for trehalose and water can be
calculated as a function of the solute mass fraction and the temperature (Equation 2
and 3, Figure 2) [2]. Here, xsolute is the mass fraction of solute and T is the temperature
of the droplet (K). Note that no exact diffusion coefficient is known for a ternary
mixture that includes a protein, such as BSA. In this scenario, the ratio between the
radius of gyration of trehalose (3.4 Å, [11]) and BSA (31.5 Å, [12]) was calculated (31.5
/ 3.4 = 9.26), and the diffusion coefficient of trehalose was divided by this ratio to
obtain a rough estimate of the diffusion coefficient of BSA. In addition, the mass
fraction of solute (xsolute in equations 2 and 3) was then taken as the sum of the mass
fraction of trehalose and BSA.
∙ . ∙
∙
∙ . ∙ (2)
∙ . ∙
∙
∙ . ∙ (3)
Figure 2: Diffusion coefficient of water (black) and trehalose (grey) at 303 K (left) and 358 K
(right) as a function of the concentration. Solid symbols indicate literature values, while open symbols
are calculated values using equation 2 and 3 [2].
118 | Chapter 6
Both at 303 K and 358 K the function appears to correlate well with the
literature values of the diffusion coefficient over the entire concentration range. It is
therefore believed that the calculated values will allow us to correctly estimate the
diffusion of trehalose and water during drying.
Calculation of vapor pressure and moisture content

Because the model describes the moisture exchange between air and the
droplet/particle, it is important to take into account the effect of dissolved
components on the vapor pressure of water. Assuming an ideal solution, the influence
of dissolved components on the vapor pressure can be calculated using Raoult’s law,
in which the saturation vapor pressure is multiplied with the molar fraction of water in
the outer subshell of the droplet (xm,w). However, this will only slow down evaporation
due to lowering of the saturation vapor pressure, but not take into account the
hygroscopic properties of the solute at higher concentrations. To correct for this, the
saturation vapor pressure is further multiplied with the Margules function, which is a
measure of the deviation from ideality (Equation 4).
∙ ∙
, , (4)
The empirical constants A and B were determined with the aid of dynamic
vapor sorption (DVS) data for trehalose, and were found to be -5.0 and 4.6,
respectively [3]. This was achieved by calculating the moisture uptake of a dry particle
in air with various relative humidities, while changing the constants A and B. The
resulting moisture contents according to DVS and our model are shown in Figure 3.

Figure 3: Moisture content of trehalose in equilibrium with different relative humidities (RH)
according to DVS (square) and the developed model using the Margules constants A = -5.0 and
B = 4.6 (circle).
Starting conditions
The volume ratio of air / water was set at 1.5*105 and the starting droplet radius was
set at 3.75 µm, based on properties of a typical lab-scale spray dryer. Furthermore, the
starting concentration of trehalose was set at 5 mg/mL, the starting concentration of
BSA was set at either 0.01 or 1 mg/mL, the starting temperature of the droplet at 293
K, and the starting temperature and relative humidity of air at 333 K and 0 %,
respectively.
Results & discussion
Temperature gradient inside the droplet

A simulation was run in which a temperature gradient inside the droplet was taken
into account. This was done to ensure that the choice to omit this temperature
gradient in further simulations was justified (i.e. has a negligible effect on the result).
The omission of the temperature gradient results in a large decrease of the calculation
time, and is therefore desirable. To increase the temperature gradient for visibility, the
initial air temperature was increased to 393 K, whereas the initial droplet temperature
was decreased to 274 K. A simulation of the first 1 ms is shown in Figure 4. An
aqueous 5 mg/mL trehalose solution without BSA was considered for these
simulations.
120 | Chapter 6
Figure 4: Temperature of each of the droplet subshells when a droplet of 274 K is exposed to air
of 393 K and 0% RH. The insert shows an enlargement of the temperature gradient in the droplet in the
first 0.04 ms. The bottom curve represents the inner subshell, whereas the top curve represents the outer
subshell.
It was found that indeed the temperature difference between the subshells is
hard to distinguish even at these exacerbated conditions. Only at a very small
timescale, a difference between the surface and the core of the droplet of about 2.5 K
can be seen. Considering the much milder conditions in further simulations and the
much longer timescales, the assumption that no temperature gradient inside the
droplet exists will not significantly affect the result and therefore appears justified.
Optimum number of subshells and choice for either the ‘shrink method’ or the
‘merge method’
Further simulations were performed to choose the optimal number of droplet
subshells. The droplet radius at the formation of the solid shell was calculated at
different numbers of subshells for an aqueous 5 mg/ml trehalose solution (no BSA).

Obviously, an infinite number of subshells would yield optimal results, but also
requires infinitely long calculation times. With a restricted number of subshells,
however, an error is introduced. Due to the assumption that each individual subshell
is homogeneously mixed, a larger outer subshell should result in the formation of the
solid shell at a later stage (at a smaller droplet radius) due to a larger error in the
concentration gradients inside the droplet. This is most pronounced when the droplet
is modeled as a single subshell, as no concentration gradients inside the droplet is
considered and thus no diffusion. Therefore, the whole droplet will solidify when it is
close to the maximum density of trehalose (~1580 kg/m3). In between a single
subshell (fast but resulting in the smallest particle size) and infinite subshells (slow but
resulting in the largest and most accurate particle size) is an optimal number of
subshells with a negligible deviation from the result with infinite subshells while still
fast. Furthermore, two methods to handle the receding surface of the droplet were
compared: one where the subshell volumes were equalized after each iteration (‘shrink
method’), and another where the outer subshell was merged with the neighboring
subshell only when the thickness was less than 10 nm and the mass fraction difference
of water was less than 0.001 (‘merge method’). These limits were chosen arbitrarily,
although they had to be low enough to prevent constant merging. To make the effect
of the chosen number of subshells more pronounced, the drying rate was increase by
increasing the initial air temperature to 373 K (see Figure 5).
It was found that with the ‘merge method’ less subshells were required to
reach the maximum radius at which the solid shell is formed than with the ‘shrink
method’. Due to the generally smaller outer subshell during the simulation with the
‘merge method’, the error introduced with the assumption of homogeneously mixed
subshells is reduced. This reduction will of course depend on the chosen minimum
thickness and concentration difference at which a merge of the two outer subshells is
allowed. If these limits would be reduced further, the maximum radius would be
reached at a lower number of subshells, but the calculation time would increase as
well. However, due to the smaller volume of the outer subshell, the time steps during
the simulation become smaller to ensure the temperature and mass changes in each
time step stay small enough to prevent diverging or oscillating results. Therefore, the
total calculation times for simulations with the ‘merge method’ were actually longer
than with the ‘shrink method’ for the number of subshells that would result in
equivalent radii (for example, 20 subshells with the ‘shrink method’ took about 165 s,
whereas 10 subshells with the ‘merge method’ took about 200 s even though they
resulted in the same radius at solid shell formation). Furthermore, the same maximum
radius at which a solid shell is formed is obtained at 40 subshells. If indeed the ‘shrink
method’ would introduce a significant amount of mass transfer not attributed to
diffusion but to the method used to deal with the receding droplet surface, the outer
122 | Chapter 6
subshell would be diluted more and thus form a solid shell at a smaller radius. This
appears not to be the case. Due to the fact that the ‘merge method’ is slower, the
‘shrink method’ was preferred and used for subsequent simulations. Furthermore,
above 40 subshells, the increase in radius at solid shell formation was found to be
negligible. Therefore, 40 subshells is considered to be the optimum number of
subshells as this appears to give the best balance between accuracy and calculation
time.
Figure 5: Influence of the number of droplet subshell on the radius at which the solid shell is first
formed. Simulations were performed either with the ‘shrink method’ (circles) or the ‘merge method’
(squares). The initial air temperature was set at 373 K.
Pure trehalose
To get a better understanding of what is happening inside the droplet, a simulation
with standard conditions was run and the changes in mass fraction of the solute in
each subshell of the droplet were tracked in time. and the results of this simulation are
shown in Figure 6. The outer radius of each of the subshells is shown as well,
indicating the size of the droplet during the drying process.

Figure 6: Mass fraction of trehalose in the droplet in each modeled subshell (indicated by the
steps) at different time points. The end of the line corresponds to the outer radius of the droplet
during drying while the center of the droplet is at radius 0.
It was found that the model indeed predicted the accumulation of the solute
near the surface of the droplet. After about 17.9 ms, the mass fraction of the solute in
the outer subshell was high enough for the glass transition temperature to surpass the
temperature of the droplet at the surface, changing from a rubbery to a glassy state.
For reference, the temperature of the droplet and the glass transition temperature in
each shell are shown in Figure 7. It is shown that the increase in glass transition
temperature is fast and happens less than one ms before the first solid shell is formed.
Furthermore, an increase in the droplet temperature is found already before the outer
subshell turned into a glass. Due to the rapidly increasing concentration of solute, the
saturation vapor pressure at the surface of the droplet also decreases (Raoult’s law).
This lower saturation vapor pressure results in a decreased evaporation rate, which
124 | Chapter 6
thus reduces the evaporative cooling effect that kept the droplet at a lower
temperature in the first place (the wet bulb temperature).
Figure 7: Temperature (black) and glass transition temperature (grey) inside the droplet in each
modeled subshell. The full simulation is shown on the left, whereas more detailed view of the solid shell
formation is shown on the right. The bottom curve represents the inner subshell, whereas the top curve
represents the outer subshell.
What is most striking is the large separation between the glass transition
temperature of the outer subshell and the other subshells. When the mass fraction of
solute in the outer subshell starts to increase rapidly, the evaporation of water is also
inhibited due to the much slower diffusion through the concentrated solute solution.
Although once in the glassy state the outer subshell will remain as such due to
evaporation and slower diffusion, the concentration differences in the other subshells
will start to decrease, as the diffusion there is still much faster. This can be seen by a
decrease in the differences between the glass transition temperatures of these
subshells. However, during further evaporation, the glass transition temperature of
these subshells will diverge again, due to the difference in distance over which water
has to diffuse which results in increasing differences of the solute concentrations in
the subshells . If the number of subshells would be increased from a finite to an
infinite number, the difference in glass transition temperature between the subshells
would become smaller and eventually gradual. However, since we consider a finite
number of subshells, the artifact of the large difference in glass transition temperature
between the outer and inner subshells remains.
This artifact is caused by the low diffusion rate of water through the solidified
outer shell. However, the question remains whether this low diffusion rate will be
found in reality as well. There are at least three scenario’s where the diffusion rate
would be higher: 1) when the solid shell will be formed much later due to other
wetting mechanics, such as capillary forces, 2) when the solid shell will still deform

and break up due to low mechanical strength, and 3) when diffusion through the solid
shell is faster due to porosity. This implies that the drying time with the current model
is a worst case scenario, which will immediately be shortened when any other method
is used to estimate the diffusion rate of water through the subshells that would result
in an effectively faster diffusion rate through the solid shell. Whereas the developed
model considers only homogeneous subshells, all three scenarios are based on
inhomogeneity’s in the subshells caused by different mechanisms.
For the first scenario one could imagine that the increased viscosity of highly
concentrated solution at the surface of the drop in combination with inhomogeneity
results in possible channels where mass transfer proceeds faster. These channels
would allow water to pass through the shell much faster and perhaps also wet it much
faster than is possible when only the diffusion through the concentrated solution is
considered. However, no attempt was made to model this scenario as its complexity
would require a separate study that is outside the scope of this paper.
The second scenario was also discussed in a study by Ekdawi-Sever et al. [2].
Upon formation of the solid shell, it is uncertain whether the shell is thick enough to
stay in the position where it was formed. For example, with 40 modeled shells the
outer layer shell is only 6 nm thick when the first glassy shell is formed (Figure 8). Due
to the possibly low mechanical strength of such a layer, it could still shrink further
until the layer has a sufficient thickness to withstand deformation. It could also simply
break forming ruptures, through which water can diffuse more rapidly than through
the intact shell. From a modeling perspective this would mean that at some critical
solid shell thickness, determined by the mechanical characteristics of the considered
solutes, shrinking of the droplet stops and further evaporation results in a vacuole in
the core of which the under pressure is withstood by the solid shell, resulting in a
hollow particle. This will have to be investigated further experimentally with the aid of
mechanical/structural engineering.
126 | Chapter 6
Figure 8: Solid shell thickness during drying when the droplet is assumed to keep shrinking. The
steps corresponding to each subshell were smoothed, resulting in a closer approximation when an infinite
number of subshells would be used.
The third scenario, can be roughly modeled by assuming that the maximum
porosity of the solidified subshell is equal to the water mass fraction upon
solidification. When the water evaporates after the solid shell has formed, the resulting
pores will be used for water to diffuse through as vapor. This assumption is based on
a model used by Stephen, et al. [13], but should not be considered an accurate
prediction of the diffusion through the solid shell. Instead it merely serves to illustrate
the influence of different scenario’s on the drying characteristics (Figure 9). The
diffusion rate through the porous subshell is calculated according to equation 5, where
C is a constrictivity/tortuosity factor (3·10-5), ε is the porosity, Dw,vapor is the diffusion
coefficient of water vapor in air, and Dwater is the diffusion coefficient of water in the
droplet as calculated with equation 3.
, ∙ ∙ , 1 ∙ (5)

Figure 9: Temperature (black) and glass transition temperature (grey) inside the droplet in each
modeled subshell for the scenario where diffusion of water through the solid shell is faster due to
porosity.
With the faster diffusion of water through the solid shell, the predicted glass
transition temperature gradient was indeed found to be much smoother. This also
resulted in a more gradual heating of the droplet. Whereas the increase in droplet
temperature with the slow diffusion was divided in one to two steps due to the large
difference in glass transition temperature of the two outer subshells (see Figure 7),
here the increase in temperature is divided into multiple steps. The increase of the
droplet temperature in multiple steps is an artifact of the rather rapid increase in water
diffusion and thereby increased evaporation rate once the shell transitions from a
rubber to a glass. When more subshells would be modeled, the steps become smaller
and the temperature increase smoother. We expect that this prediction overestimates
the rate at which water diffuses through the solid shell, as the porosity that is used
here is most likely an overestimate of the actual porosity that would be available for
water diffusion through the subshell. Instead, the solution that correlates best with
128 | Chapter 6
experimental values is expected to be between the scenario depicted in Figures 7 and
9.
Further simulations of the ternary water, trehalose, BSA mixture will be
calculated using both methods to determine whether the rate of diffusion through the
solid shell will be relevant from a protein stabilization perspective, although by default
the unmodified diffusion model will be presented unless stated otherwise.
Protein stabilization
To predict the distribution of components in a droplet during protein stabilization by
spray drying, BSA was added as a third component. The diffusion coefficient of BSA
was estimated with the radius of gyration as described in the materials and methods.
To get a good indication of the mass distribution of BSA and trehalose inside the
dried particle, a simulation was run with a mass ratio of trehalose to protein of both 5
and 500. The mass fraction of protein and trehalose of the total protein and trehalose
content, respectively, is shown in each subshell (Figure 10).
Figure 10: Prediction of trehalose (solid line) and BSA (dashed line) mass distribution
throughout a dried particle. Depicted is the mass fraction of solute that resides in each subshell of the
total amount of that specific solute in the particle (i.e. the mass fractions of each solute add up to 1). The
total mass ratio trehalose/protein was either 5 (left) or 500 (right).
A clear separation of trehalose and BSA was found with both trehalose/BSA
mass ratio’s. Due to the slower diffusion of the protein, accumulation of protein is
indeed predicted near the surface. Furthermore, the increased protein content near the
surface replaces the trehalose, which would otherwise be equally distributed
throughout the particle (in case of pure trehalose each subshell of equal volume would
contain 1 / 40 = 0.025 of the total trehalose content). With a trehalose/BSA mass
ratio of 500, this reduction in trehalose content near the surface is not observed due
to the low absolute mass of protein in the particle. Most striking is that the
accumulation of BSA near the surface is already very pronounced, despite the fact that
the surface excess concentration of surface active components (such as proteins) is
not taken into consideration. Doing so would result in an even more pronounced
accumulation near or at the surface of the dried particle. Of further note is that, due to
the accumulation of BSA at the surface relative to trehalose, there will be a difference
in the trehalose/BSA mass ratio throughout the dried particle. Because the stability of
the protein is in part dependent on this ratio, it could influence the effectiveness of
the protein stabilization process if only the initial trehalose/BSA mass ratio would be
taken into account during product development. The protein accumulation near the
surface was more pronounced with the trehalose/BSA mass ratio of 500 than with the
lower trehalose/BSA mass ratio of 5. However, with the starting trehalose/BSA mass
ratio of 500 the protein may still be sufficiently protected as the minimum
trehalose/BSA mass ratio is still about 250, whereas with a low starting trehalose/BSA
mass ratio of 5 the minimum is only about 3.5. To get a better overview of
distribution of the trehalose/BSA mass ratio throughout the particle, the cumulative
distribution is shown in Figure 11.
Figure 11: Cumulative protein mass fraction with a mass ratio protein/trehalose below or equal
to depicted values. The total mass ratio trehalose/protein was either 5 (left) or 500 (right).
With both scenario’s, close to 60 % of the total protein mass content is

encapsulated in less trehalose than when both components would have been
homogeneously distributed. Furthermore, results obtained with simulations where a
porous solid shell was considered did not deviate much from this result (data not
shown). This can be expected, since the distribution of the components will, for the
most part, already have occurred before the solid shell is formed. The main difference
was that the distribution was slightly wider, leading to a lower minimum and a higher
130 | Chapter 6
maximum trehalose/BSA ratio at the surface and the center of the particle,
respectively. This is most likely caused by faster evaporation during the second stage,
resulting in less redistribution of components inside the droplet after the initial solid
shell was formed.
Whether or not and how much the lower trehalose/BSA mass ratio near the
surface will affect the stability cannot be said, however, it is likely that it will be
reduced. This will also depend largely on the protein that is used, as some protein are
much more sensitive than others. Furthermore, the extent to which the sugar/protein
mass ratio will change is dependent on the spray drying conditions. For example,
faster drying would result in more accumulation at the surface, and thus possible
reduced stabilization. In addition, the molecular weight and shape of the sugar and
protein used will also play a role. The smaller the difference in the diffusion rate
between the sugar and the protein, the smaller the change in sugar/protein mass ratio.
In case a small protein is chosen that diffuses faster than the chosen sugar, one might
even find that sugar accumulates at the surface and protein in the center of the drying
droplet. Finally, when the surface excess concentration would be considered, the
amount of protein that will not be properly encapsulated will increase even further.
Therefore, it should be kept in mind when developing stabilized protein formulations
by spray drying, that adding surfactants to prevent accumulation of protein at the
surface due to its surface active nature, may actually not be sufficient. Instead,
simulations such as performed here can be used to more accurately adapt the initial
composition of the solution to ensure proper stabilization of the protein, for example
by simply increasing the amount of sugar or, if this is limited by the desired loading,
changing the drying conditions to slow down the drying rate.
Conclusions
The presented model gives an insight into the distribution of components with respect
to protein stabilization during spray drying. Furthermore, the model can easily be
adapted to suit other needs and change components. However, the method strongly
depends on the use of reliable diffusion coefficient data and Gordon-Taylor
coefficients, which are vital for accurate predictions of the component distribution
inside the dried particle. Especially for multi-component mixtures, this can become
quite complex. Therefore, at a later stage, a welcome addition would be the use of
molecular dynamics models to predict the interaction between different solutes and
water, and couple these to the model of a drying droplet.
Furthermore, a better understanding of the morphology changes and the
mass transfer during the second stage is welcomed, to further improve the accuracy of
predictions in this region. However, it was shown that for protein stabilization, the
most important factor is the distribution of the components inside the droplet, which
is mainly determined during the first drying stage. Even with the fast and slow drying
during the second stage that was considered in this study, the difference in the
predicted effect on protein encapsulation was minor.
References
1. Adhikari B, Howes T, Lecomte D, Bhandari BR. A glass transition

temperature approach for the prediction of the surface stickiness of a drying
droplet during spray drying. Powder Technology. 2005;149(2–3):168-79.
2. Ekdawi-Sever N, de Pablo JJ, Feick E, von Meerwall E. Diffusion of Sucrose
and α,α-Trehalose in Aqueous Solutions. The Journal of Physical Chemistry
A. 2003;107(6):936-43.
4. Bellavia G, Giuffrida S, Cottone G, Cupane A, Cordone L. Protein Thermal
Denaturation and Matrix Glass Transition in Different
Protein−Trehalose−Water Systems. The Journal of Physical Chemistry B.
2011;115(19):6340-6.
5. Eaton JW, Community, GNU Octave, Version: 3.8.2, Available from:
https://www.gnu.org/software/octave/index.html, Last accessed: 23 March
2015.
6. Gac JM, Gradoń L. A distributed parameter model for the spray drying of
multicomponent droplets with a crust formation. Advanced Powder
Technology. 2013;24(1):324-30.
1969;163(3873):1325-6.
2004;93(4):047801.
10. Lee AL, Wand AJ. Microscopic origins of entropy, heat capacity and the glass
transition in proteins. Nature. 2001;411(6836):501-4.
11. Lerbret A, Bordat P, Affouard F, Descamps M, Migliardo F. How
Homogeneous Are the Trehalose, Maltose, and Sucrose Water Solutions? An
Insight from Molecular Dynamics Simulations. The Journal of Physical
Chemistry B. 2005;109(21):11046-57.
132 | Chapter 6
12. Santos SF, Zanette D, Fischer H, Itri R. A systematic study of bovine serum
albumin (BSA) and sodium dodecyl sulfate (SDS) interactions by surface
tension and small angle X-ray scattering. J Colloid Interface Sci.
2003;262(2):400-8.
13. Werner SRL, Edmonds RL, Jones JR, Bronlund JE, Paterson AHJ. Single
droplet drying: Transition from the effective diffusion model to a modified
receding interface model. Powder Technology. 2008;179(3):184-9.

134 | Chapter 7
Chapter 7
General discussion and perspectives
General discussion and perspectives | 135

A dried protein can be stabilized by incorporating the protein into a sugar matrix. This
stabilization is often explained using either of two mechanics: vitrification and water
replacement. Vitrification occurs when the molecular mobility of the protein is
inhibited by the rigid sugar matrix, whereas stabilization by water replacement occurs
by hydrogen bonding with the sugar to replace hydrogen bonds lost when water was
evaporated during drying. To achieve optimal stabilization, the sugar should be in a
glassy state. This allows the sugar matrix to conform to the rough surface of the
protein, whereas a crystalline matrix could not, and facilitates maximum vitrification
due to the low translational molecular mobility compared to a sugar matrix in the
rubbery state. These two stabilization mechanisms are often considered separately.
However, in chapter 2 we have seen that stabilization by vitrification and water
replacement are both important mechanisms that act in balance. Without vitrification,
no level of hydrogen bonding between the sugar and the protein can keep it from
unfolding due to the high translational molecular mobility; the sugar matrix should
form a foundation to which the protein can be anchored. On the other hand, when
the sugar matrix is vitrified but there is not enough hydrogen bonding between the
sugar and the protein, it can still unfold, albeit at a much lower rate than in a non-
vitrified sugar matrix. This would also explain why certain sugars are better stabilizers
than others and why sometimes a sugar such as sucrose can still provide stabilization
despite a relatively low glass transition temperature. Finally it explains why sometimes
an increase in stability is found when a small amount of plasticizer is present instead
of absent.
If we compare trehalose to inulin, the former is a smaller molecule that
should more easily follow the irregular (rough) contours of the surface of a protein
and form more hydrogen bonds. This will in turn form a stronger anchor to the
vitrified matrix to prevent the protein from unfolding. However, when the glass
transition temperature of trehalose (121 °C) is depressed far enough for thesugar
matrix to become rubbery, e.g. under humid conditions, stability is lost. In these
conditions, inulin may still be in a glassy state due to its higher glass transition
temperature (155 °C) and thus stabilize the protein. This of course doesn’t mean that
any sugar with a higher glass transition temperature is better for protein stability.
Dextran, for example, is known to be a poor stabilizer despite its high glass transition
temperature (> 200 °C), due to its poor water replacement properties. Furthermore,
sucrose, when used at temperatures well below its glass transition temperature of
about 65 °C, is known to have better stabilizing properties than trehalose in certain
formulations. It is hypothesized that sucrose has less self-interactions than trehalose,
also often referred to as preferential exclusion effects. However, the practical
application of stabilized protein formulations should be kept in mind, and
formulations with a high glass transition temperature will give significantly less
136 | Chapter 7
problems during storage. The tricky part here is that there is no fixed glass transition
temperature of the formulation, unless it is stored at a predefined constant relative
humidity and temperature. Finally, interesting things may occur at low plasticizer
contents. Cases are known where adding a few percent of moisture can actually
increase the stability of the protein, despite the plasticizing effect. This could also be
explained in a similar way as was the difference in stabilizing properties of differently
sized sugars. Even trehalose will not be able to cover the entire protein surface,
especially where small gaps or intrusions are present in the protein molecular
structure. Water, being a much smaller molecule might be able to reach those
locations and form hydrogen bonds, either further anchoring the protein to the
vitrified sugar matrix, or to other parts of the protein itself. This would also imply
that, depending on the protein surface roughness, the optimal amount of plasticizer
may be protein specific.
Although this makes for a reasoning where everything seems to fit, the true
challenge is to actually show what is happening around the protein in these protein
formulations. The results in chapter 2 show how the processes involved in protein
stabilization (protein-sugar-plasticizer interaction and vitrification) result in either a
stable protein formulation or not. However, a more fundamental understanding of the
intermolecular interactions that takes place between the protein, the sugar, and other
excipients, will aid in developing stabilized protein formulations more effectively. For
example, understanding the influence of small concentrations of different plasticizers
may result in formulations that consist of several stabilizers of different molecular
weight that more effectively keep the protein vitrified. Furthermore, this could then
also be tailored to each specific protein to optimize the amount of plasticizer added.
Perhaps molecular dynamics simulations could be a great way to investigate these
interactions and visualize them. In a similar fashion this also holds true for the
formation of these interactions during drying (chapter 6), as this is when storage
stability is determined.
With regard to spray drying of protein solutions, a lot can be done to improve
the quality of stabilized protein products by optimizing each separate process step
during spray drying. Whereas an understanding of the stabilization mechanisms will
mainly lead to optimized formulations for storage stability, a further understanding of
the influence of various conditions during specific steps within the spray drying
process (as well as the interactions between them) on the protein activity, will lead to
improved process stability. For example, as was found in chapter 3, there is a very
large influence of the type of nozzle that is chosen on the process stability of lactate
dehydrogenase. Finding the most suitable nozzle, pump, tubing, collector, and so on,
may greatly improve the ability to stabilize sensitive proteins by spray drying.
Furthermore, a more fundamental understanding of the effect of the atomization

mechanism (for example) on protein activity loss is key to also being able to fully
tailor a spray dryer towards optimization of protein stabilization.
Besides the spray dryer system itself, optimization of the spray dryer settings
can be improved by the use of a model. As shown in chapter 4, the model enabled the
accurate prediction of the state of the sugar matrix, glassy or rubbery, at a specific set
of input parameters, like the chosen process conditions. This not only reduces the
amount of experiments that need to be performed in order to optimize a spray dryer
process for protein stabilization, but also has a built-in quality check. When the
outcome of the spray dryer is unexpected, the chances are that the model does not
take into account all the variables that are relevant for the spray dried formulation. In
this case the product was simply a sugar, without protein or any other excipients, and
so the current model was enough to accurately predict the relevant product quality
parameters. Further development could be focused specifically on protein
stabilization, to include more relevant parameters such as the atomization mechanism
and the formation of the sugar matrix during drying (chapter 5 and 6).
In its current form, the model introduced in chapter 4 is very useful for
general optimization of spray drying processes. Especially at the start of the
development, an initial estimate of the optimal settings is valuable. Whereas in the
field of chemical engineering the use of models is quite common, in the
pharmaceutical technology field it is less so. Therefore, bringing together both fields
to increase the awareness and use of models for process and product optimization, or
even moving further towards a quality by design approach in research would be highly
beneficial. Even more so, I believe that models in general could be used so much
more, even in daily practice in the labs. Hopefully in the future, a model will no longer
be seen as some vague and difficult to understand number generator, but be treated
no different than a calculator, a pen and paper. Especially in this mobile era, everyone
is walking around with computers that are far more powerful than most realize and
models as described in chapter 4 and perhaps chapter 5 and 6 can or will be able to
run on mobile phones on the spot, while preparing the experiments.
With the research presented in this thesis we have made great strides to
predictively develop new stabilized protein products. However, there are still many
unexpected results when trying to develop new formulations, and many times a trial-
and-error approach is still required for reliable optimization. This is fine, as that is the
nature of research: answering one question usually brings up multiple other questions.
Improving a process based on new findings for one product may or may not cause
new challenges somewhere down the line with other products. However, it goes to
show that there is still a lot to learn.
138 | Chapter 7
140 | Appendix A
Appendix A
Summary
Summary | 141
There is an increasing interest for therapeutic proteins in pharmacy. These proteins
can be formulated as a dry powder or as an aqueous solution. Dried protein
formulations may offer many advantages over the currently used aqueous
formulations. They are easier to store, may have a longer shelf life and they can be
used for solid dosage forms such as dry powder inhalers. However, the stability of
proteins in the solid state and stabilizing mechanisms underlying various stabilization
technologies is not yet fully understood, which may therefore offer options for
improvement.
A general approach used to keep proteins stable in the solid state is to
incorporate them into a sugar matrix by drying a solution of the protein and a sugar.
The sugar matrix, which may consist of materials such as trehalose or inulin, is
believed to stabilize the protein by forming hydrogen bonds during drying. Upon
doing so, the sugar acts as a replacement for water that is removed during drying, and
the theory behind this mechanism is therefore called the water replacement theory.
Furthermore, the sugar matrix also prevents unfolding of the protein by keeping it
vitrified, known as the vitrification theory. For this to work, the sugar needs to be in a
glassy state, to both maximize the contact with the irregular protein surface, and to
minimize the translational molecular mobility of the matrix. Although usually either
the water replacement theory or the vitrification theory are considered to explain the
stabilization of a protein, it was found that both mechanisms play a role depending on
the storage conditions (chapter 2). When the storage temperature is close to or above
the glass transition temperature of the formulation, the storage stability of the protein
is dependent on the temperature. However, when the storage temperature is well
below the glass transition temperature, the stability depends on the water replacement
properties of the sugar. Therefore, when developing protein formulations, it is
important to use formulations of which the glass transition temperature is well above
the storage temperature, while taking into account any plasticizing effect of moisture.
There are several drying techniques that can be used to produce stable protein
formulations. Although freeze drying is one of the most commonly used drying
processes, spray drying offers some interesting advantages: it is a continuous process
instead of a batch process, reduced variability in product properties, decreased
production costs, and shorter processing times. In addition, spray dried powders are
well suited for dry powder inhalation, which is getting more and more attention for
the treatment of not only lung diseases but also conditions that are systemic and not
specifically related to the lung. However, spray drying is a relatively harsh process for
sensitive proteins, as they are subjected to shear, heat, drying, and interfacial stresses.
For a shear sensitive protein such as lactate dehydrogenase, in chapter 3 it is shown
that especially the heating step and the atomization step of the spray dryer causes a
substantial activity loss. However, this could be reduced greatly by choosing a more
142 | Appendix A
suitable atomization technique: upon switching from an ultrasonic perforated mesh
nozzle to a two-fluid nozzle, the process stability of the protein was improved greatly.
Therefore, an important step in developing stabilized protein products is to also tailor
the drying process and its sub processes to the specific protein, as every protein
responds differently to the different stresses during drying.
To further optimize the production process of stabilized protein
formulations, the spray dryer settings will also need to be chosen such that the
product adheres to the required specifications. This is often done using a trial-and-
error approach, however, this can be time consuming, costly, and may result in
suboptimal results. To move towards a quality-by-design approach, a fundamental
understanding of the process is required. To this end a model that describes the outlet
condition of the spray dryer as a function of the inlet parameters was developed
(chapter 4). This model was found to accurately predict the outlet conditions and
could also be used in conjunction with other analytical techniques, such as DVS and
DSC. This enabled the prediction of whether the product would be in a glassy or
rubbery state, which can help to ensure a high yield and storage stability of the
protein. The model was developed in commonly used software, i.e. spreadsheet
software, which can be downloaded freely from the internet. Furthermore, it was
made freely available for anyone by publishing it in an open access journal. This model
is particularly suitable for those who have less background in chemical engineering or
are less familiar with the process and is accessible to anyone.
A more detailed model was developed to describe the behavior of a droplet in
a surrounding where evaporation or condensational growth can occur (chapter 5).
This model was used to predict the behavior of an inhaled droplet under laminar flow
conditions. The model was found to accurately predict the drying rate of a pure water
droplet under various conditions.
It was found that the droplet can either grow or shrink depending on the
location inside the airway relative to the airway wall. Droplets close to the airway wall
were found to increase in size rapidly, while droplets near the center of the airway first
evaporated partly, resulting in a reduction in size compared to the inhaled droplets.
Furthermore, several parameters were found to influence the amount of change in the
droplet size distribution, amongst which the relative humidity and temperature during
inhalation, and the concentration of the dispersed droplets in the air. This can have
serious implications for the efficacy of inhalation therapies based on aerosols, but also
on dry powder inhalation, because a change in the particle size distribution can affect
the deposition site within the lung. Finally, because the model was again developed in
freely available software and the source code was made available freely, the model can
serve as a framework for further expansion to improve the accuracy and complexity of
the model by others.
Summary | 143
In chapter 6, the droplet model has been further expanded to make it more
useful for spray drying protein formulations by taking into account the diffusion rates
of the various excipients during drying. It has been found that differences in the
diffusion rates of the various solutes results in concentration gradients and (phase)
separation of components, which may negatively affect the storage stability of proteins
if they are not properly incorporated into the sugar matrix. A further understanding of
this process can help in improving the ability to produce stable protein formulations
by spray drying.
144 | Appendix A
Summary | 145
146 | Appendix B
Appendix B
Samenvatting
Samenvatting | 147
Er is een toenemende belangstelling voor therapeutische eiwitten in de farmacie. Deze
eiwitten kunnen in een droog poeder geformuleerd worden of als een waterige
oplossing. Gedroogde eiwit formuleringen kunnen vele voordelen bieden boven de
momenteel veelal gebruikte waterige oplossingen. Ze zijn gemakkelijker op te slaan,
langer houdbaar en kunnen worden gebruikt voor vaste toedieningsvormen zoals in
droogpoederinhalatoren. De stabiliteit van eiwitten in de vaste toestand en de
mechanismen die ten grondslag liggen aan verschillende stabilisatietechnologieën
worden nog niet volledig begrepen en dit vormt dien ten gevolge tot mogelijkheden
voor verbeteringen.
Een algemene benadering om eiwitten stabiel te houden in de vaste toestand
is deze in te sluiten in een suikermatrix door een oplossing van het eiwit en een suiker
te drogen. Aangenomen wordt dat de suikermatrix, die bijvoorbeeld kan bestaan uit
trehalose of inuline, het eiwit stabiliseert door de vorming van waterstofbruggen
tijdens het drogen. Hierbij fungeert het suiker als vervanging voor water dat tijdens
het drogen wordt verwijderd, en de theorie achter dit mechanisme wordt daarom de
watervervangingstheorie genoemd. Bovendien voorkomt de suikermatrix ook dat het
eiwit kan ontvouwen door verglazing, bekend als de zogenaamde verglazingstheorie.
Voor deze twee mechanismen moet het suiker in een glasachtige toestand zijn om
zowel optimaal contact met het onregelmatige eiwitoppervlak te realiseren alsmede de
translationele moleculaire mobiliteit van de suikermatrix te minimaliseren. Hoewel
meestal enkel de watervervangingstheorie of de verglazingstheorie wordt beschouwd
om de stabilisatie van een eiwit te verklaren, is gebleken dat beide mechanismen een
rol spelen, afhankelijk van de opslagomstandigheden (hoofdstuk 2). Bij een
opslagtemperatuur dichtbij of boven de glasovergangstemperatuur van de formulering
is de opslagstabiliteit van het eiwit afhankelijk van de temperatuur. Wanneer de
opslagtemperatuur ver beneden de glasovergangstemperatuur ligt, is de stabiliteit
afhankelijk van de watervervangingseigenschappen van het suiker. Daarom is het bij
het ontwikkelen van eiwitformuleringen belangrijk om ervoor te zorgen dat de
glasovergangstemperatuur van de formulering ver boven de opslagtemperatuur ligt,
daarbij rekening houdend met het weekmakende effect van geabsorbeerd vocht.
Er zijn verscheidene droogtechnieken die gebruikt kunnen worden om
stabiele eiwitformuleringen te produceren. Hoewel vriesdrogen één van de meest
gebruikte droogprocessen is, biedt sproeidrogen enkele interessante voordelen: het is
een continu proces in plaats van een batchproces, verminderde variabiliteit in
producteigenschappen, verlaagde productiekosten en kortere verwerkingstijden.
Bovendien zijn gesproeidroogde poeders uitermate geschikt voor
droogpoederinhalatie, welke steeds meer gebruikt wordt voor de behandeling van niet
alleen longziekten maar ook systemische aandoeningen die niet specifiek betrekking
op de longen hebben. Echter, sproeidrogen is een relatief ruw proces voor gevoelige
148 | Appendix B
eiwitten, omdat zij worden blootgesteld aan afschuif-, warmte-, droog- en
grensvlakspanningen. Voor afschuifgevoelige eiwitten zoals lactaat dehydrogenase
wordt in hoofdstuk 3 aangetoond dat vooral verwarming en de atomisering tijdens
sproeidrogen leidt tot activiteitsverlies. Dit kon echter sterk worden verminderd door
het kiezen van een geschikte vernevelingstechniek: bij het overschakelen van een
ultrasone sproeikop met een geperforeerd membraan naar een pneumatische
sproeikop werd de procestabiliteit van het eiwit sterk verbeterd. Daarom is een
belangrijke stap tijdens de ontwikkeling van gestabiliseerde eiwitproducten ook het op
maat afstemmen van het droogproces en de bijbehorende subprocessen op het
specifieke eiwit; immers, elk eiwit reageert anders op de verschillende spanningen
tijdens het drogen.
Om het productieproces van gestabiliseerde eiwitformuleringen verder te
optimaliseren, zullen de instellingen van de sproeidroger ook zodanig moeten worden
gekozen dat het product voldoet aan de vereiste specificaties. Dit wordt vaak gedaan
via een “trial-and-error” aanpak, welke tijdrovend en kostbaar kan zijn en bovendien
kan leiden tot suboptimale resultaten. Om deze aanpak te veranderen naar “quality-by-
design”, is echter een fundamenteel begrip van het proces vereist. Hiertoe is een
model ontwikkeld dat de condities aan de uitgang van de sproeidroger voorspeld als
functie van de instellingen (hoofdstuk 4). Dit model bleek de condities aan de uitgang
goed te voorspellen en kon ook worden gebruikt in combinatie met andere analytische
technieken zoals DVS en DSC. Hierdoor kon voorspeld worden of het product in een
glasachtige of rubberachtige toestand verkeerde, wat kan helpen om een goede
opbrengst en opslagstabiliteit van het eiwit te garanderen. Het model werd ontwikkeld
in veelgebruikte software, dat wil zeggen spreadsheet software, die vrij gedownload
kan worden van het internet. Daarnaast is het model vrij beschikbaar voor iedereen
doordat het gepublicereerd werd in een open access tijdschrift. Het model is met
name geschikt voor diegenen die geen of weinig achtergrond hebben in de
scheikundige technologie of die minder vertrouwd zijn met het sproeidroogproces.
Een gedetailleerder model werd ontwikkeld om het gedrag van een druppel in
een omgeving te beschrijven waar krimp van de druppel door verdamping of groei
door condensatie kunnen voorkomen (hoofdstuk 5). Dit model werd gebruikt voor
het voorspellen van het gedrag van een geïnhaleerde druppel onder laminaire
omstandigheden. Het model bleek het gedrag van een zuivere waterdruppel onder
verschillende inhalatieomstandigheden goed te voorspellen. Het bleek dat de druppel
groter of kleiner kan worden, afhankelijk van de locatie binnen de luchtweg ten
opzichte van de longwand. Druppels nabij de longwand bleken snel in omvang
toenemen, terwijl druppels nabij het centrum van de luchtweg eerst gedeeltelijk
verdampen, wat resulteert in een verkleining ten opzichte van de ingeademde
druppels. Bovendien werden verscheidene parameters gevonden die de mate van
Samenvatting | 149
verandering in de druppelgrootteverdeling konden beïnvloeden, waaronder de
relatieve vochtigheid en temperatuur tijdens het inhaleren en de concentratie van de
gedispergeerde druppels in de lucht. Dit kan ernstige gevolgen hebben voor de
effectiviteit van inhalatietherapieën op basis van verneveling, maar ook voor
droogpoederinhalatie, omdat een verandering in de deeltjesgrootteverdeling de plaats
van afzetting in de longen kan beïnvloeden. Tenslotte, omdat het model wederom
werd ontwikkeld in vrij beschikbare software en tevens de broncode vrij beschikbaar
is gesteld, kan het model dienen als een basis voor verdere uitbreiding door anderen
om de nauwkeurigheid en de complexiteit van het model verder te verbeteren.
Afsluitend is in hoofdstuk 6 het voorgaande druppelmodel verder uitgebreid
om het geschikter te maken voor het sproeidrogen van eiwitformuleringen door
rekening te houden met de diffusiesnelheden van de verschillende hulpstoffen in de
druppel tijdens het drogen. Hierbij is gebleken dat de verschillen in de
diffusiesnelheden van de verscheidene opgeloste stoffen leidt tot
concentratiegradiënten en (fase) scheiding van componenten wat een negatieve
invloed kan hebben op de opslagstabiliteit van eiwitten als deze niet goed in de
suikermatrix zijn ingesloten. Een verder begrip van dit proces kan helpen bij het
verbeteren van het vermogen om stabiele eiwitformuleringen te produceren met
behulp van sproeidrogen.
150 | Appendix B
Samenvatting | 151
152 | Appendix C
Appendix C
Model discussed in chapter 5
Model discussed in chapter 5 | 153

154 | Appendix C
156 | Appendix C
158 | Appendix C
160 | Appendix C
162 | Appendix D
Appendix D
Model discussed in chapter 6

164 | Appendix D
166 | Appendix D
168 | Appendix D
170 | Appendix D
172 | Appendix D
174 | Appendix D
176 | Appendix D
178 | Appendix E
Appendix E
Biography and list of publications
Biography and list of publications | 179

Niels Grasmeijer finished high school at the ‘Hogeland College’ in Warffum in 2002.
Due to a general interest in natural sciences and technology, Niels decided to study
Pharmacy with the intention to follow the master curriculum of Pharmaceutical
Technology and Chemical Engineering. In 2009 his study was finalized with a master’s
thesis on the semi-continuous production of drug nanocrystals using a three-fluid
nozzle under the supervision of Wouter L.J. Hinrichs and Hans de Waard at the
Department of Pharmaceutical Technology and Biopharmacy of the University of
Groningen. He started his PhD study on protein stabilization by spray drying that year
at the same department. After finishing his research in 2013, he wrote his thesis while
holding a post-doctoral position where he did research on the pulsatile release of
protein from polymeric formulations until late 2015. Currently, Niels is working at
InnoCore Pharmaceuticals as a Scientist Formulation and Process Development and
at the Department of Pharmaceutical Technology and Biopharmacy of the University
of Groningen.
Journal articles
de Waard H, Grasmeijer N, Hinrichs WLJ, Eissens AC, Pfaffenbach PPF, Frijlink
HW. Preparation of drug nanocrystals by controlled crystallization: Application of a 3-
way nozzle to prevent premature crystallization for large scale production. Eur J
Pharm Sci. 2009;38(3):224-9.
Grasmeijer N, Stankovic M, de Waard H, Frijlink HW, Hinrichs WLJ. Unraveling

protein stabilization mechanisms: Vitrification and water replacement in a glass
transition temperature controlled system. Biochimica et Biophysica Acta (BBA) -
Proteins and Proteomics. 2013;1834(4):763-9.
Grasmeijer N, de Waard H, Hinrichs WLJ, Frijlink HW. A User-Friendly Model for

Spray Drying to Aid Pharmaceutical Product Development. PLoS ONE.
2013;8(9):e74403.
Oral presentations
Grasmeijer N, de Waard H, Hinrichs WLJ, Frijlink HW. A user-friendly model for
spray drying to aid pharmaceutical product development. BMM-TerM Annual
Meeting, Ermelo, The Netherlands, 17-18 September, 2012
Grasmeijer N, Stankovic M, de Waard H, Frijlink HW, Hinrichs WLJ. Stabilization of

proteins during spray drying. Nederlandse Werkgroep Drogen, Utrecht, The
Netherlands, 25 November, 2013
180 | Appendix E
Grasmeijer N, de Waard H, Hinrichs WLJ, Frijlink HW. A user-friendly model for
spray drying to aid pharmaceutical product development. 9th World Meeting on
Pharmaceutics, Biopharmaceutics and Pharmaceutical Technology, Lisbon, Portugal,
31 March - 3 April, 2014
Poster presentations
Grasmeijer N, Stankovic M, Hinrichs WLJ, Frijlink HW. Importance of the glass
transition temperature for protein stabilization by spray drying. BMM-TerM Annual
Meeting, Ermelo, The Netherlands, 25-26 May, 2011
Grasmeijer N, Stankovic M, de Waard H, Hinrichs WLJ, Frijlink HW. Importance of

the glass transition temperature for protein stabilization by spray drying. 8th World
Meeting on Pharmaceutics, Biopharmaceutics and Pharmaceutical Technology,
Istanbul, Turkey, 19-22 March, 2012
Grasmeijer N, Stankovic M, de Waard H, Hinrichs WLJ, Frijlink HW. Importance of

the glass transition temperature for protein stabilization by spray drying. BMM-TerM
Annual Meeting, Ermelo, The Netherlands, 17-18 September, 2012
Grasmeijer N, Tiraboschi V, Frijlink HW, Hinrichs WLJ. Unraveling critical process

steps during spray drying of protein solutions using two different nozzles. Meeting of
the Belgian-Dutch Biopharmaceutical Society, Ghent, Belgium, 18 December, 2013
Biography and list of publications | 181

182 | Appendix F
Appendix F
Dankwoord
Dankwoord | 183
Aan alle goede dingen komen een eind, zo ook dit promotietraject. De afgelopen jaren
heb ik met ontzettend veel plezier mogen werken aan dit onderwerp en heb ik veel
mensen leren kennen, die in meer of mindere mate allemaal een bijdrage hebben
geleverd aan dit proefschrift. In het bijzonder zijn er een aantal personen zonder wie
dit proefschrift misschien nooit tot stand was gekomen en voor wie ik bij deze de
gelegenheid wil nemen om hen te bedanken. Voor ieder ander: bedankt voor je
aandacht, zijn er nog vragen?
Erik Frijlink, promotor en baas, jij hebt mij de gelegenheid gegeven om het
onderzoek uit te voeren op jouw afdeling. Nu bleek daar later wel wat eigenbelang
achter te zitten, daar je mij regelmatig optrommelde voor diverse uitdagingen op
computergebied, afijn. Je hebt mij altijd de vrijheid gegeven om te doen wat ik wilde,
waardoor soms je geduld wel erg op de proef werd gesteld. Ik waardeer de manier
waarop jij hiermee om bent gegaan en denk dat, doordat jij op eenzelfde manier de
afdeling leidt, het altijd zo’n gezellige en productieve werkomgeving is geweest. Ik ben
je erg dankbaar voor de mogelijkheden die jij mij hebt gegeven. Dank je.
Wouter Hinrichs, mijn co-promotor. Ik heb je mogen zien groeien van beginner tot
absolute IT pro over de laatste jaren. In ruil daarvoor heb jij mij erg veel geleerd en
heb ik met ontzettend veel plezier onder jouw begeleiding mijn onderzoek gedaan.
Ook jouw geduld werd regelmatig op de proef gesteld, maar je wist altijd hoe je mij
kon motiveren. Niet zelden draaiden onze discussies om de pragmatisering van mijn
ietwat perfectionistische experimentele plannen, om binnen redelijke tijd toch tot een
uitkomst te komen. Bedankt voor alle tijd die je hebt genomen om mij te begeleiden
en de waardevolle discussies die we hebben mogen voeren. Ik waardeer onze
samenwerking zeer en hoop dit in de toekomst te kunnen blijven doen. Dank je.
Hans de Waard, we hebben elkaar helaas al een tijdje niet meer gesproken, maar het
was onder jouw begeleiding dat ik zowel mijn master als een groot deel van mijn
promotieonderzoek heb mogen doen. Jij hebt je altijd ingezet om mij buiten mijn
‘comfort zone’ te laten stappen, om verder te kijken. Hoewel dit misschien niet altijd
even succesvol is geweest, heb ik dit altijd zeer gewaardeerd en ben ik hierdoor zeker
gegroeid. Ik wens jou en je gezin alle geluk. Dank je.
Milica Stankovic, ik heb het genoegen gehad om het eerste jaar met jou samen te
mogen werken. Jouw vrolijkheid, inzet (soms misschien wat teveel) en passie voor het
onderzoek waren zeer inspirerend. Ik wens jou en je gezin het allerbeste. Dank je.
184 | Appendix F
Anko Eissens, ik in jouw kantoor binnenlopen was als een klein kind in een
snoepwinkel. De indrukwekkende voorraad onderdelen en gereedschap die jij huisvest
bleek vaak zeer goed van pas te komen voor mijn ondernemingen op de afdeling.
Ondanks dat ik soms meerdere malen op een dag kwam vragen om meer, was je altijd
bereidt om mij te helpen. Dank je.
Paul Hagedoorn, de teleurstelling op jouw gezicht als ik langskwam en weer op zoek

bleek te zijn naar Anko was hartverscheurend. Gelukkig was je altijd meer dan bereidt
om een uitgebreide uitleg te geven over allerhande onderwerpen, van laserdiffractie tot
auto’s. Het enthousiasme waarmee jij dingen uit kan leggen en vertellen is zeer
inspirerend en laat zien hoeveel hart jij voor het onderzoek hebt. Dank je.
Marcel Hoppentocht, collega en goede vriend. Opvallend dat de meest interessante

gespreksstof, waar ik op dit moment niet nader op in zal gaan, vaak alleen in jouw
aanwezigheid naar boven kwam borrelen. De middagwandelingen en de koffiepauze
waren in jouw aanwezigheid zeker een welkome onderbreking van de dag. Maar de
meeste waardering heb ik voor je openheid, en hoe je mij altijd wist te motiveren om
mijn horizon te verbreden, zowel op professioneel als persoonlijk vlak. Dank je.
Voor mijn promotieonderzoek heb ik in de afgelopen jaren ook met zeer veel plezier
het master onderzoek van Valeria Sassu, Alessia Pavone, Valeria Tiraboschi en
Eline Azou mogen begeleiden. Jullie inzet en gezelligheid was onmisbaar voor het
onderzoek. Ik wens jullie ontzettend veel geluk in jullie verdere carriere en
persoonlijke leven.
Prof. dr. Francesco Picchioni, Prof. dr. ir. Wim E. Hennink, and Prof. dr. Guy
van den Mooter, my assessment committee, thank you for taking the time to read
and comment on this thesis.
Floris, broer(tje), tja... wat valt er eigenlijk nog te zeggen. Elke dag in de auto, thuis en
op het werk spreken we elkaar en er is weinig dat onbesproken blijft. Hoewel het
soms vanzelfsprekend lijkt, is dit voor mij onmisbaar geweest en had ik zonder jouw
steun en commentaar zeker mijn verstand verloren. In ons geval was jij toch de
stabiliserende factor en ik het meer instabiele element... Hoewel ik het misschien nog
wel eens voor lief neem, voor twee broers om zo’n hechte vriendschap te hebben is
best uniek en dat koester ik zeer. Dank je.
Germ, mam en Maryse, voor mij zijn jullie nog altijd een van de belangrijkste
onderdelen van mijn leven. Jullie steun, tomeloze inzet en betrokkenheid met al het
goede en minder goede dat ons aangaat is voor mij onmisbaar. Bedankt voor alles.
Dankwoord | 185

Complete Thesis

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Complete Thesis

Uploaded by

Copyright:

Available Formats

University of Groningen

Improving protein stabilization by spray drying

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Download date: 05-11-2023

Cover: Scanning Electron Microscope (SEM) image of inulin particles prepared

ISBN: 978-90-367-8519-8 (printed version)

Printing: Ridderprint BV, The Netherlands

ter verkrijging van de graad van doctor aan de

De openbare verdediging zal plaatsvinden op

maandag 11 januari 2016 om 11:00 uur

geboren op 22 juni 1984

Chapter 2 Unraveling protein stabilization mechanisms: vitrification 13

Chapter 3 Identifying critical process steps to protein stability during 35

Chapter 4 A user-friendly model for spray drying to aid pharmaceutical 57

Chapter 5 An adaptable model for growth and/or shrinkage of droplets 83

Chapter 6 Model to predict inhomogeneous protein-sugar distribution 113

Chapter 7 General discussion and perspectives 135

Appendix A Summary 141

Appendix B Samenvatting 147

Appendix C Model discussed in chapter 5 153

Appendix D Model discussed in chapter 6 163

Appendix E Biography and list of publications 179

Appendix F Dankwoord 183

1. Walsh G. Biopharmaceutical benchmarks 2014. Nat Biotech.

Unraveling protein stabilization mechanisms: vitrification and water

N. Grasmeijer, M. Stankovic, H. de Waard, H.W. Frijlink, W.L.J. Hinrichs

Unraveling protein stabilization mechanisms | 13

Materials and methods

Differential Scanning Calorimetry (DSC)

Enzymatic Activity of Alkaline Phosphatase

Unraveling protein stabilization mechanisms | 17

Dynamic Vapor Sorption (DVS) analysis

Glass transition temperature

ws Tg.s  k sb  w b  Tg.b  k sw w w Tg.w

Unraveling protein stabilization mechanisms | 19

Table 2: Estimated glass transition temperatures for stabilized protein samples

Varying glass transition temperature

Unraveling protein stabilization mechanisms | 21

Inulin (Tg = 78 °C) Inulin (Tg = 95 °C)

Unraveling protein stabilization mechanisms | 23

Unraveling protein stabilization mechanisms | 25

Unraveling protein stabilization mechanisms | 27

In conclusion, it is shown that at glass transition temperatures below the storage

Unraveling protein stabilization mechanisms | 29

Unraveling protein stabilization mechanisms | 31

Identifying critical process steps to protein stability during spray drying

Niels Grasmeijer, Valeria Tiraboschi, Henderik W. Frijlink, Wouter L.J.

Critical process steps to protein stability during spray drying | 35

Critical process steps to protein stability during spray drying | 37

Materials and methods

Spray drying process

The ultrasonic nozzle was used in conjunction with a 4 µm perforated mesh

Critical process steps to protein stability during spray drying | 39

Separating process steps

Atomizing “Pump + Spray” – “Pump”

only pumping and heating (and not spraying) were applied

Laser diffraction analysis (LDA)

Scanning electron microscopy (SEM)

Critical process steps to protein stability during spray drying | 41

Critical process steps to protein stability during spray drying | 43

Critical process steps to protein stability during spray drying | 45

Critical process steps to protein stability during spray drying | 47

Critical process steps to protein stability during spray drying | 49