Engineering Aspects of Food Emulsification and Homogenization

Food and Culinary Science
RAYN E R • DE JME K
Emulsions are found in a wide variety of food products, pharmaceuticals, paints,
and cosmetics, thus emulsification is a truly multidisciplinary phenomenon.
Therefore, understanding of the process must evolve from the combination of (at
Engineering Aspects
least) three different scientific specializations. Engineering Aspects of Food
Emulsification and Homogenization describes the state-of-the-art technology
E N G I N E E R I N G A S P E C T S O F F O O D E M U L S I F I C AT I O N A N D H O M O G E N I Z AT I O N
and brings together aspects from physical chemistry, fluid mechanics, and
chemical engineering. The book explores the unit operations used in emulsifica-
of
tion and homogenization processes, using fundamental theory from different fields
to discuss design and function of different emulsification techniques.
This book summarizes the present understanding of the involved physical–chemi-

cal processes as well as specific information about the limits and possibilities for
the different types of emulsifying equipment. It covers colloidal chemistry and Food Emulsification
engineering aspects of emulsification and discusses high-energy and low-energy
and
Homogenization
emulsification methods. The chapters highlight low-energy emulsification process-
es such as membrane emulsification that are now industrially feasible. Dramatical-
ly more energy-efficient processes are being developed, and this book clarifies
their present limitations, such us scale-up and achievable droplet sizes.
Features
• Describes state-of the-art technology of emulsification and homogenization

processes
• Brings together aspects from physical chemistry, fluid mechanics, and
chemical engineering
• Presents formulation aspects of emulsions with respect to stability
and function
• Brings together fundamental theory from different fields to discuss
design and function of different emulsification techniques
• Compares high-energy, low thermodynamic efficiency methods
with alternative low-energy, higher-efficiency processes
The present literature on emulsification is, to a large degree, influenced by the

division between physical chemistry, fluid dynamics, and chemical engineering.
Written by experts drawn from academia and industry, this book brings those
areas together to provide a comprehensive resource that gives a deeper under-
standing of emulsification and homogenization in food product development.
EDITED BY
K16909
Marilyn R ayner
ISBN: 978-1-4665-8043-5
90000 P et r D ejm ek
9 781466 580435
Engineering Aspects
of
Food Emulsification
and
Homogenization
Contemporary Food Engineering
Series Editor
Professor Da-Wen Sun, Director
Food Refrigeration & Computerized Food Technology
National University of Ireland, Dublin
(University College Dublin)
Dublin, Ireland
http://www.ucd.ie/sun/
Engineering Aspects of Food Emulsification and Homogenization,

edited by Marilyn Rayner and Petr Dejmek (2015)
Handbook of Food Processing and Engineering, Volume II: Food Process
Engineering, edited by Theodoros Varzakas and Constantina Tzia (2014)
Handbook of Food Processing and Engineering, Volume I: Food Engineering
Fundamentals, edited by Theodoros Varzakas and Constantina Tzia (2014)
Juice Processing: Quality, Safety and Value-Added Opportunities, edited by
Víctor Falguera and Albert Ibarz (2014)
Engineering Aspects of Food Biotechnology, edited by José A. Teixeira
and António A. Vicente (2013)
Engineering Aspects of Cereal and Cereal-Based Products, edited by
Raquel de Pinho Ferreira Guiné and Paula Maria dos Reis Correia (2013)
Fermentation Processes Engineering in the Food Industry, edited by
Carlos Ricardo Soccol, Ashok Pandey, and Christian Larroche (2013)
Modified Atmosphere and Active Packaging Technologies, edited by
Ioannis Arvanitoyannis (2012)
Advances in Fruit Processing Technologies, edited by Sueli Rodrigues and
Fabiano Andre Narciso Fernandes (2012)
Biopolymer Engineering in Food Processing, edited by Vânia Regina Nicoletti
Telis (2012)
Operations in Food Refrigeration, edited by Rodolfo H. Mascheroni (2012)
Thermal Food Processing: New Technologies and Quality Issues, Second
Edition, edited by Da-Wen Sun (2012)
Physical Properties of Foods: Novel Measurement Techniques and
Applications, edited by Ignacio Arana (2012)
Handbook of Frozen Food Processing and Packaging, Second Edition,
edited by Da-Wen Sun (2011)
Advances in Food Extrusion Technology, edited by Medeni Maskan and
Aylin Altan (2011)
Enhancing Extraction Processes in the Food Industry, edited by Nikolai Lebovka,
Eugene Vorobiev, and Farid Chemat (2011)
Emerging Technologies for Food Quality and Food Safety Evaluation,
edited by Yong-Jin Cho and Sukwon Kang (2011)
Food Process Engineering Operations, edited by George D. Saravacos and
Zacharias B. Maroulis (2011)
Biosensors in Food Processing, Safety, and Quality Control, edited by Mehmet
Mutlu (2011)
Physicochemical Aspects of Food Engineering and Processing, edited by
Sakamon Devahastin (2010)
Infrared Heating for Food and Agricultural Processing, edited by Zhongli Pan
and Griffiths Gregory Atungulu (2010)
Mathematical Modeling of Food Processing, edited by Mohammed M. Farid (2009)
Engineering Aspects of Milk and Dairy Products, edited by Jane Sélia dos Reis
Coimbra and José A. Teixeira (2009)
Innovation in Food Engineering: New Techniques and Products, edited by Maria
Laura Passos and Claudio P. Ribeiro (2009)
Processing Effects on Safety and Quality of Foods, edited by Enrique Ortega-
Rivas (2009)
Engineering Aspects of Thermal Food Processing, edited by Ricardo Simpson
(2009)
Ultraviolet Light in Food Technology: Principles and Applications,
Tatiana N. Koutchma, Larry J. Forney, and Carmen I. Moraru (2009)
Advances in Deep-Fat Frying of Foods, edited by Serpil Sahin and Servet Gülüm
Sumnu (2009)
Extracting Bioactive Compounds for Food Products: Theory and Applications,
edited by M. Angela A. Meireles (2009)
Advances in Food Dehydration, edited by Cristina Ratti (2009)
Optimization in Food Engineering, edited by Ferruh Erdoǧdu (2009)
Optical Monitoring of Fresh and Processed Agricultural Crops, edited by
Manuela Zude (2009)
Food Engineering Aspects of Baking Sweet Goods, edited by Servet Gülüm
Sumnu and Serpil Sahin (2008)
Computational Fluid Dynamics in Food Processing, edited by Da-Wen Sun
(2007)
Engineering Aspects
of
Food Emulsification
and
Homogenization
EDITED BY
Marilyn Rayner
Petr Dejmek
Boca Raton London New York
CRC Press is an imprint of the

Taylor & Francis Group, an informa business
Cover Image: Structural ensemble of the cyclin-dependent kinase inhibitor p27Kip1 (shown by colored
ribbons) bound to the Cdk2/cyclin complex (blue-white surface). 100 residue long p27 tail coordinates
were derived from the molecular dynamics simulation of the full protein [J. Mol. Biol. 2008, vol. 376,
827-838].
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2015 by Taylor & Francis Group, LLC
CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works

Version Date: 20150318
International Standard Book Number-13: 978-1-4665-8044-2 (eBook - PDF)
This book contains information obtained from authentic and highly regarded sources. Reasonable
efforts have been made to publish reliable data and information, but the author and publisher cannot
assume responsibility for the validity of all materials or the consequences of their use. The authors and
publishers have attempted to trace the copyright holders of all material reproduced in this publication
and apologize to copyright holders if permission to publish in this form has not been obtained. If any
copyright material has not been acknowledged please write and let us know so we may rectify in any
future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced,
transmitted, or utilized in any form by any electronic, mechanical, or other means, now known or
hereafter invented, including photocopying, microfilming, and recording, or in any information stor-
age or retrieval system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copy-
right.com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222
Rosewood Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that pro-
vides licenses and registration for a variety of users. For organizations that have been granted a photo-
copy license by the CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
Dedication
For our families, teachers, and mentors.

“Love many, trust few, and always paddle your own canoe.”
Contents
Series Preface ............................................................................................................xi
Series Editor........................................................................................................... xiii
Preface...................................................................................................................... xv
Editors .....................................................................................................................xix
Contributors ............................................................................................................xxi
Synopsis ............................................................................................................... xxiii
Section i emulsion Fundamentals
Chapter 1 Scales and Forces in Emulsification .....................................................3

Marilyn Rayner
Chapter 2 Emulsion Formation and Instability ................................................... 33

Björn Bergenståhl
Chapter 3 Formulation of Emulsions .................................................................. 51

Marie Wahlgren, Björn Bergenståhl, Lars Nilsson,
and Marilyn Rayner
Chapter 4 Particle-Stabilized Emulsions .......................................................... 101

Malin Sjöö, Marilyn Rayner, and Marie Wahlgren
Section ii High-energy Processes
Chapter 5 Droplet Breakup in High-Pressure Homogenizers .......................... 125

Andreas Håkansson
Chapter 6 High-Pressure Homogenizer Design ................................................ 149

Fredrik Innings
ix
x Contents
Chapter 7 High-Pressure Homogenization with Microstructured Systems ...... 169

Karsten Köhler and Heike Schuchmann
Chapter 8 Rotor–Stator Devices ....................................................................... 195

Section iii Low-energy Processes
Chapter 9 Microchannel Emulsification: Aspects of Droplet Generation,

Channel Materials, Operating Conditions, and Scaling-Up
Strategies ..........................................................................................209
Isao Kobayashi, Marcos A. Neves, and Mitsutoshi Nakajima
Chapter 10 Emulsification with Microsieves and Other Well-Defined

Microstructured Systems.................................................................. 235
Karin Schroën and Akmal Nazir
Chapter 11 Formation and Modification of Dispersions Using Shirasu

Porous Glass Membranes ................................................................. 255
Goran T. Vladisavljević
Index ...................................................................................................................... 297
Series Preface
CONTEMPORARY FOOD ENGINEERING
Food engineering is a multidisciplinary field of applied physical sciences combined
with the knowledge of product properties. Food engineers provide the technological
knowledge transfer essential to the cost-effective production and commercialization
of food products and services. In particular, food engineers develop and design pro-
cesses and equipment to convert raw agricultural materials and ingredients into safe,
convenient, and nutritious consumer food products. Food engineering topics are con-
tinuously undergoing changes to meet diverse consumer demands, and the subject is
being rapidly developed to reflect market needs.
In the development of food engineering, one of the many challenges is to employ
modern tools and knowledge, such as computational materials science and nano-
technology, to develop new products and processes. Simultaneously, improving
food quality, safety, and security continues to be a critical issue in food engineer-
ing studies. New packaging materials and techniques are being developed to pro-
vide more protection to foods, and novel preservation technologies are emerging to
enhance food security and defense. Additionally, process control and automation
are among the top priorities identified in food engineering. Advanced monitoring
and control systems are developed to facilitate automation and flexible food manu-
facturing. Furthermore, energy saving and minimization of environmental prob-
lems continue to be important food engineering issues, and significant progress is
being made in waste management, efficient utilization of energy, and reduction of
effluents and emissions in food production.
The Contemporary Food Engineering Series addresses some of the recent
developments in food engineering. The series covers advances in classical unit
operations in engineering applied to food manufacturing as well as such topics as
progress in the transport and storage of liquid and solid foods; heating, chilling,
and freezing of foods; mass transfer in foods; chemical and biochemical aspects of
food engineering and the use of kinetic analysis; dehydration, thermal processing,
nonthermal processing, extrusion, liquid food concentration, membrane processes,
and applications of membranes in food processing; shelf life and electronic indica-
tors in inventory management; sustainable technologies in food processing; and
packaging, cleaning, and sanitation. The books in this series are aimed at pro-
fessional food scientists, academics researching food engineering problems, and
graduate-level students.
The editors of these books are leading engineers and scientists from different
parts of the world. All the editors were asked to present their books to address the
market’s needs and pinpoint cutting-edge technologies in food engineering.
All chapters have been contributed by internationally renowned experts who have
both academic and professional credentials. All authors have attempted to provide
critical, comprehensive, and readily accessible information on the art and science of a
xi
xii Series Preface
relevant topic in each chapter, with reference lists for further information. Therefore,
each book can serve as an essential reference source to students and researchers in
universities and research institutions.
Da-Wen Sun
Series Editor
Series Editor
Born in southern China, Dr. Da-Wen Sun is a world
authority in food engineering research and education; he
is a member of the Royal Irish Academy (RIA), which is
the highest academic honor in Ireland; he is also a mem-
ber of Academia Europaea (The Academy of Europe) and
a fellow of the International Academy of Food Science and
Technology. His main research activities include cooling,
drying, and refrigeration processes and systems, quality
and safety of food products, bioprocess simulation and optimization, and computer
vision technology. Especially, his many scholarly works have become standard ref-
erence materials for researchers in the areas of computer vision, computational fluid
dynamics modeling, vacuum cooling, and so on. Results of his work have been pub-
lished in over 800 papers, including more than 380 peer-reviewed journal papers (Web
of Science h-index = 60). He has also edited 14 authoritative books. According to
Thomson Reuters Essential Science IndicatorsSM, based on data derived over a period
of 10 years from the ISI Web of Science, there are about 4500 scientists who are among
the top 1% of the most cited scientists in the category of agriculture sciences. For many
years, Dr. Sun has consistently been ranked among the top 50 scientists in the world
(he is at the 25th position in March 2015).
He received a first class in both bachelor’s (honors) and master’s degree programs
in mechanical engineering, and a PhD in chemical engineering in China before
working in various universities in Europe. He became the first Chinese national
to be permanently employed in an Irish university when he was appointed college
lecturer at the National University of Ireland, Dublin [University College Dublin
(UCD)], in 1995, and was then continuously promoted in the shortest possible time
to senior lecturer, associate professor, and full professor. Dr. Sun is now the professor
of Food and Biosystems Engineering and the director of UCD Food Refrigeration
and Computerised Food Technology.
As a leading educator in food engineering, Dr. Sun has significantly contributed
to the field of food engineering. He has trained many PhD students who have made
their own contributions to the industry and academia. He has also delivered lec-
tures on advances in food engineering on a regular basis in academic institutions
internationally and delivered keynote speeches at international conferences. As a
recognized authority in food engineering, he has been conferred adjunct/visiting/
consulting professorships from 10 top universities in China, including Zhejiang
University, Shanghai Jiaotong University, Harbin Institute of Technology, China
Agricultural University, South China University of Technology, and Jiangnan
University. In recognition of his significant contribution to food engineering world-
wide and for his outstanding leadership in the field, the International Commission
of Agricultural and Biosystems Engineering (CIGR) awarded him the CIGR Merit
Award twice in 2000 and in 2006, the Institution of Mechanical Engineers based
xiii
xiv Series Editor
in the United Kingdom named him Food Engineer of the Year 2004. In 2008, he
was awarded the CIGR Recognition Award in honor of his distinguished achieve-
ments as the top 1% of agricultural engineering scientists in the world. In 2007, he
was presented with the only AFST(I) Fellow Award by the Association of Food
Scientists and Technologists (India), and in 2010, he was presented with the CIGR
Fellow Award; the title of Fellow is the highest honor in CIGR and is conferred
to individuals who have made sustained, outstanding contributions worldwide. In
March 2013, he was presented with the You Bring Charm to the World award by
Hong Kong–based Phoenix Satellite Television with other award recipients includ-
ing Mr. Mo Yan—the 2012 Nobel Laureate in Literature and the Chinese Astronaut
Team for Shenzhou IX Spaceship. In July 2013, he received the Frozen Food
Foundation Freezing Research Award from the International Association for Food
Protection for his significant contributions to enhancing the field of food freezing
technologies. This is the first time that this prestigious award was presented to a
scientist outside the United States.
He is a fellow of the Institution of Agricultural Engineers and a fellow of Engineers
Ireland (the Institution of Engineers of Ireland). He is also the editor-in-chief of
Food and Bioprocess Technology—An International Journal (2012 impact fac-
tor = 4.115), former editor of Journal of Food Engineering (Elsevier), and editorial
board member for a number of international journals, including the Journal of Food
Process Engineering, Journal of Food Measurement and Characterization, and
Polish Journal of Food and Nutritional Sciences. He is also a chartered engineer.
On May 28, 2010, he was awarded membership in the RIA, which is the high-
est honor that can be attained by scholars and scientists working in Ireland; at the
51st CIGR General Assembly held during the CIGR World Congress in Quebec
City, Canada, on June 13–17, 2010, he was elected incoming president of CIGR
and became CIGR president in 2013–2014; the term of his CIGR presidency is six
years, two years each for serving as incoming president, president, and past presi-
dent. On September 20, 2011, he was elected to Academia Europaea (The Academy
of Europe), which is functioning as the European Academy of Humanities, Letters
and Sciences, and is one of the most prestigious academies in the world; election to
Academia Europaea represents the highest academic distinction.
Preface
MOTIVATION IN SHORT FOR THIS BOOK
• Emulsions are used widely and produced in large volumes. Thus, emulsion
formation or emulsification is an important unit operation.
• Emulsification is of interest for a broad audience, both because of its influ-
ence on the functionality of emulsion-based products and because it is gen-
erally energy intensive running at low efficiency.
• Today, there is no comprehensive text on emulsification available in English
describing the state-of-the-art technology and bringing together aspects
from physical chemistry, formulation, fluid mechanics, and chemical engi-
neering. Together these aspects are the foundation needed for understanding
emulsification at more than a rudimentary level.
EMULSIONS AND EMULSIFICATION

Emulsions can be found in a wide variety of food products, such as milk, cream,
spreads, ice-creams, dressings, and sauces. Emulsions are also common in many
related areas such as pharmaceuticals (e.g., topical formulation and nutritional emul-
sions) and many household products (e.g., paints and cosmetics).
Emulsions bring many different forms of functionality to these products via
emulsion drops. Emulsion drops can be used to design bulk properties such
as appearance, solubilization, mouthfeel, rheology, and electrical properties.
Emulsions can also be used for bringing highly specific functionality to prod-
ucts, as in controlled delivery and in the release of pharmaceuticals, in increasing
the bioavailability of nutrients and in delivering flavor. All the aforementioned
product properties are highly influenced by the characteristics of the emulsion.
Emulsion character includes many different aspects, such as drop size distribu-
tion and structure and type and amount of adsorbed emulsifiers. The properties
of a given emulsion are influenced to a large extent by the emulsification pro-
cess through which it is created. Intensity and spatial/temporal distribution of the
applied energy, time allowed for adsorption of surface-active material, and type
of force acting to destabilize the interface will all influence the properties of the
final emulsion. Hence, creating functional emulsions requires a fundamental and
general understanding of the involved physical–chemical processes as well as
specific information about the limits and possibilities for the different types of
emulsifying equipment.
Emulsions are produced in large quantities. As an example, approximately
270 million tons of liquid dairy products are annually treated with high-pressure
homogenization (Tetra Pak Dairy Index Report, 2011). Emulsification is also a rather
energy-demanding unit operation, in absolute terms and particularly in relation to the
theoretical energy requirements. The thermodynamically required energy for creat-
ing an emulsion is determined by the increase in free energy due to surface energy.
xv
xvi Preface
For micrometer-sized emulsion drops, this would amount to an order of magnitude of

103 J/m3, whereas the actual energy requirement for high-pressure homogenization is
of the order 107 J/m3. Taking the example of the dairy market, this would correspond
to roughly 600 GWh of energy, of which only 0.01% is needed from a thermody-
namic viewpoint. The conclusion from these large volumes and low efficiency must
be that even a small incremental improvement in process operation would lead to
substantial savings. Thus, there is a broad and large demand for a better understand-
ing of emulsification processes.
The low thermodynamic efficiency is characteristic to all the high-energy meth-
ods such as rotor–stator-based technologies, high-pressure homogenizers, micro-
fluidization, ultrasonic systems, and colloidal mills. Alternative low-energy flow
processes, such as membrane-based technologies, have higher efficiencies but have
not been able to yield the same productivity and extremely small drop size that high-
energy technologies have been able to. However, new methods are developed con-
tinuously and an up-to-date comparison is much in need.
THE EMULSIFICATION PROCESS

The outcome of an emulsification process is generally a combination of two compet-
ing processes: disruption of the drop interface from dynamic destabilizing forces
and thermodynamically driven coalescence. Studying the processes one at a time is
not uncomplicated; the coupling is highly complex and poorly understood.
Emulsification also encompasses a large range of timescales, from the very fast
velocity fluctuations in high-pressure homogenization (100 MHz), ultrasonication
(100 kHz), the adsorption times of surfactants (1 MHz), and macromolecular emulsi-
fiers to the slower process of shear and drop detachment and all the way to the equi-
librium thermodynamics, which determines the long-term fate of every emulsion.
From the discussion above, it is clear that emulsification is truly a multidisci-
plinary phenomenon, an understanding of which must, therefore, spring from the
combination of (at least) three different scientific specializations:
• Physical chemists have long studied interfaces between liquids and their
relation to surface-active molecules. This will be a key aspect in the process
of coalescence as well as to understanding how surface-active agents can
aid in disruption.
• Fluid mechanics has developed theoretical and experimental methods for
understanding the interplay between hydrodynamic forces and the drop
interface. This is vital for understanding how differences in design can
bring about different emulsification results.
• Research in chemical engineering has led to relations between operating
conditions and emulsion characteristics and devised methods for measur-
ing fragmentation or coalescence rates in bulk. This has obvious practical
importance, but can also be used for comparison with predictions from the
more fundamental disciplines.
A comprehensive discussion on emulsification methods and their application must

bring together aspects from all these different specialties.
Preface xvii
THE NEED FOR NEW LITERATURE ON EMULSIFICATION

The present work on emulsification is, to a large degree, influenced by the above-
mentioned division between the different topics. Many of the more general text-
books on surface and colloidal chemistry contain chapters on emulsions (e.g.,
Cosgrove, 2010; Goodwin, 2004; Holmberg et al., 2003). There is also a rather
voluminous literature on emulsions per se, both in general (e.g., Sjöblom, 2006;
Binks, 1998; Becher, 1983) and for food applications (Dickinson and Stainsby,
1982; Friberg and Larsson, 1997; McClements, 2005). These works, however, to a
large extent focus on the physiochemical aspects of emulsions; the emulsification
process in itself is not treated in sufficient detail. Similarly, books from the fluid
dynamics field dealing with multiphase flow sometimes discuss implications on
drop formation (e.g., Clift, Grace, and Weber, 1978); however, they do so with-
out the coupling to technical considerations/design and the advances in physical
chemistry.
Steps toward this synthesis of different aspects can be found in the older litera-
ture (e.g., Gopal, 1968) and substantial advances were made by Walstra in a series
of book chapters and review articles during the 1980s and 1990s (Walstra, 1983,
1993; Walstra and Smulders, 1998); however, these works are now gradually being
outdated, especially in relation to the coupling between process design and the emul-
sification result obtained. Today, there is no comprehensive treatment of emulsifica-
tion including the state-of-the art developments and integration of all three aspects
available in English. This book is aimed at filling that void. The perspective will be
that of the unit emulsification process, using fundamental theory from different fields
to discuss the design and function of different emulsification techniques.
In summary, our motivation for writing this book is as follows:
• Emulsions are used widely and produced in large volumes. Thus, emulsion
formation or emulsification is an important unit operation.
• Emulsification is of interest for a broad audience, both because of its influ-
ence on the functionality of emulsion-based products and because it is gen-
erally energy intensive running at low efficiency.
• Today, there is no comprehensive text on emulsification available in English
describing the state-of-the-art technology and bringing together aspects
from physical chemistry, formulation, fluid mechanics, and chemical engi-
neering. Together these aspects are the foundation needed for understand-
ing emulsification at more than a rudimentary level.
REFERENCES
Becher, P. (1983). Encyclopedia of Emulsion Technology. Marcel Dekker Inc., New York.
Binks, B.P. (1998). Modern Aspects of Emulsion Science. Royal Society of Chemistry, Cambridge.
Clift, R., Grace, J.R., Weber, M.E. (1978). Bubbles, Drops and Particles. Academic Press,
New York.
Cosgrove, T. (2010). Colloid Science: Principles, Methods and Applications. John Wiley & Sons,
Chippenham, UK.
Dickinson, E., Stainsby, G. (1982). Colloids in Food. Applied Science Publishers, London.
xviii Preface
Friberg, S., Larsson, K. (1997). Food Emulsions. Marcel Dekker Inc., New York.
Goodwin, J.W. (2004). Colloids and Interfaces with Surfactants and Polymers. John Wiley & Sons,
TJ International, Padstow, Cornwall.
Gopal, E.S.R. (1968). Principles of emulsion formation. In: P. Sherman (ed.), Emulsion Science.
Academic Press, London, pp. 2–75.
Holmberg, K., Jönsson, B., Kronberg, B., Lindman, B. (2003). Surfactants and Polymers in
Aqueous Solution. John Wiley & Sons, Guildford, UK.
McClements, J.D. (2005). Food Emulsions. CRC Press, Boca Raton, FL.
Sjöblom, J. (2006). Emulsions and Emulsion Stability. Taylor & Francis, Boca Raton, FL.
Tetra Pak Dairy Index Report, Issue 4, July 2011, “Emerging Middle Class” Tetra Pak AB,
Lund, Sweden.
Walstra, P. (1983). Formation of emulsions. In: P. Beacher (ed.), Encyclopedia of Emulsion
Technology, Volume I: Basic Theory. Marcel Dekker Inc., New York, pp. 57–127.
Walstra, P. (1993). Principles of emulsion formation. Chemical Engineering Science 48,
333–349.
Walstra, P., Smulders, P.E.A. (1998). Emulsion formation. In: B.P. Binks (ed.), Modern Aspects
of Emulsion Science. Royal Society of Chemistry, Cambridge, pp. 56–99.
Petr Dejmek
Lund University
Marilyn Rayner
Lund University
Editors
Marilyn Rayner graduated with a bachelor’s degree in
biological engineering, with a food engineering special-
ization from the University of Guelph, Canada, in 1999
and earned her PhD in food engineering from Lund
University in 2005 on modeling droplet formation in
membrane emulsification. Dr. Rayner was awarded the
Food Engineer of the Year 2005 from the Institution of
Mechanical Engineers, United Kingdom, for her work on
predicting the effects of pore geometry on droplet size
in membrane emulsification processes. Since then, Dr. Rayner has worked in the
area of multiphysics modeling, unit operation, interfacial phenomena, and particle-
stabilized emulsions, and is currently associate professor in food engineering at
Lund University and the founder of two spin-off enterprises in the area of membrane
design and particle-stabilized formulations.
Petr Dejmek, a native Czech, graduated as a process

engineer from the Technical University in Aachen,
Germany, and earned his PhD in food engineering at
Lund University, Sweden. After 10 years with Alfa Laval
companies in Sweden and Denmark, Dr. Dejmek returned
to Lund University to eventually hold a research chair as
professor of dairy technology, full professor of food engi-
neering, and department head. As a visiting professor,
Dr. Dejmek worked at Tokyo University, ENSIA Massy,
Japanese National Food Research Institute, University of Wisconsin, and University
of California, Davis. Dr. Dejmek was one of the editors of the International Dairy
Journal for 10 years. He is the author of more than 100 publications on varying top-
ics on food technology and engineering and the founder of two spin-off companies.
xix
Contributors
The collective of authors contributing to this volume comprise a major portion of the
researchers who have in recent years been most active in fundamental research on
the process of emulsification aimed at food industries.
Björn Bergenståhl Lars Nilsson

Department of Food Technology, Department of Food Technology,
Engineering and Nutrition Engineering and Nutrition
Lund University Lund University
Lund, Sweden Lund, Sweden
Andreas Håkansson Marilyn Rayner

Food and Meal Science School of Department of Food Technology,
Education and Environment Engineering and Nutrition
Kristianstad University Lund University
Kristianstad, Sweden Lund, Sweden
Fredrik Innings Karin Schroën

Tetra Pak Processing Systems Food Process Engineering Group
Lund, Sweden Department of Agrotechnology
and Food Sciences
Isao Kobayashi Wageningen University
Food Engineering Division Wageningen, the Netherlands
National Food Research Institute
Heike Schuchmann
(NARO)
Karlsruhe Institute of Technology
Tsukuba, Japan
Karlsruhe, Germany
Karsten Köhler Malin Sjöö
Beiersdorf AG Department of Food Technology,
Hamburg, Germany Engineering and Nutrition
Lund University
Mitsutoshi Nakajima Lund, Sweden
University of Tsukuba
Tsukuba, Japan Goran T. Vladisavljević
Chemical Engineering Department
Akmal Nazir Loughborough University
Department of Food Engineering Loughborough, United Kingdom
University of Agriculture
Faisalabad, Pakistan Marie Wahlgren
Department of Food Technology,
Marcos A. Neves Engineering and Nutrition
University of Tsukuba Lund University
Tsukuba, Japan Lund, Sweden
xxi
Synopsis
SECTION I—EMULSION FUNDAMENTALS
Chapter 1
Scales and Forces in Emulsification
Marilyn Rayner
An overview is provided of the characteristics of the types of equipment used in the
mechanical production of emulsions and the basic governing physics of emulsifica-
tion in the limits of high-energy emulsification in processes such as vat mixers, col-
loid mills, high-pressure homogenizers, microfluidizers, ultrasound transducers, and
low-energy emulsification in channel devices of membrane, sieve, or microchannel.
Droplet size distributions, the classical Taylor droplet breakup criterion, the dimen-
sionless Weber number, timescales of deformation and the Kolmogorov–Hintze
theory of turbulent breakup are introduced.
Chapter 2
Emulsion Formation and Instability
Björn Bergenståhl
This chapter provides a reasoned explanation of the instabilities based on cream-
ing/sedimentation, flocculation, coalescence, and Ostwald ripening. Stokes’s settling
equation is discussed with respect to hindered settling and non-Newtonian viscosity.
The arrested state of both attractive and repulsive colloidal glasses is introduced,
with size and volume fraction effects. Brownian, shear-induced, and sedimentation-
induced collisions are discussed, as well as the interface parameters governing
coalescence stability; van der Waals, solvation, and electrostatic forces; and poly-
mer-induced steric repulsion, depletion attraction, and bridging.
Chapter 3
Formulation of Emulsions
Marie Wahlgren, Björn Bergenståhl, Lars Nilsson, and Marilyn Rayner
An overview is provided of the effects of compositional choices on the end-user func-
tionality of emulsions, including colloidal, microbiological, and oxidative stability; fla-
vor release; and freeze–thaw stability. The properties affecting emulsion functionality
are reviewed in detail for lipids, low-molecular emulsifiers, proteins, gelling and non-
gelling polysaccharides, protein–polysaccharide complexes, and food-based particles.
Emulsion characterization methods such as emulsification capacity, emulsion sta-
bility index, creaming index, accelerated characterization methods based on critical
osmotic pressure, and emulsion rheological test are described together with their
theoretical underpinnings.
xxiii
xxiv Synopsis
Chapter 4
Particle-Stabilized Emulsions
The theoretical basis of the high stability of particle-stabilized emulsions based on
detachment energy is shown, and the literature on the specifics of particle-stabilized
emulsions with respect to creaming, coalescence, Ostwald ripening, surface tension,
and rheology are reviewed in detail.
Examples of food-grade stabilizing particles, including modified starch granules,
lipid particles, soy and zein particles, egg yolk particles, cellulose, chitin, and flavo-
noids are given.
SECTION II—HIGH-ENERGY PROCESSES

Chapter 5
Droplet Breakup in High-Pressure Homogenizers
Andreas Håkansson
A detailed review is provided of the current understanding of the physical causes of
droplet breakup in high-pressure homogenizers, by laminar shear in the gap inlet
and its boundary layers, by local turbulence in the gap exit jet, and by cavitation,
with quantitative predictions.
Experimental evidence of the effects of homogenization pressure, Thoma num-
ber, dispersed and continuous phase viscosity, and dispersed phase volume fraction
is summarized and discussed in relation to what it reveals in terms of what mecha-
nism dominates high-pressure homogenization.
Chapter 6
High-Pressure Homogenizer Design
Fredrik Innings
Starting with the history of high-pressure homogenizers, the pros and cons of design
choices in modern commercial homogenizers are clarified, with a focus on effi-
ciency, process line integration, wear, and investment and running costs.
Chapter 7
High-Pressure Homogenization with Microstructured Systems
The development of the orifice-type high-pressure homogenization is presented from
both theoretical and technical perspectives, including patent literature. In addition to
the approaches of the previous chapters, attention is paid to the role of spatial varia-
tion of the flow field in different orifice geometries and the methods available for its
study, in particular modeling and simulation. Experimental results from different
orifice geometries are compared.
Synopsis xxv
A novel, more efficient homogenizer design suitable for partial homogenization is

introduced, where remixing of the excess continuous phase occurs within microsec-
onds of droplet breakup.
Chapter 8
Rotor–Stator Devices
Emulsification in pumps, stirred vessels, colloid mills, toothed-rim dispersers,
double rotor dispersers, and extruders is described. The unifying parameter of
energy density is described, and the positive consequences of batch processing, in
particular for emulsions in which the droplets are only slowly being stabilized, are
pointed out.
SECTION III—LOW-ENERGY PROCESSES

Chapter 9
Microchannel Emulsification: Aspects of Droplet Generation, Channel
Materials, Operating Conditions, and Scaling-Up Strategies
Isao Kobayashi, Marcos A. Neves, and Mitsutoshi Nakajima
The principle of channel shape/surface tension–driven emulsification, which is
the basis of low-energy emulsification methods, is introduced and the limits of the
operating parameter space in which effective monodispersity can be achieved are
described. Particular attention is paid to the geometry of the channels and the chan-
nel materials, and the viscosities of the phases and temperature effects on the perfor-
mance of different surfactants.
Different approaches to scale up the microchannel devices beyond the present L/h
scale are discussed.
Chapter 10
Emulsification with Microsieves and Other Well-Defined
Microstructured Systems
In membrane emulsification, one of the main issues is control of the droplet size. Due to
the polydispersity of the membrane pores, polydisperse emulsions are obtained; there-
fore, the use of membranes with equally sized pores is seen to control droplet size. In
this chapter, microsieves and other devices with monodispersed pores are presented;
their performance in cross-flow emulsification is compared to that in regular mem-
branes and more classic emulsification devices. Further, premix emulsification with
metal sieves with uniform pores, and the combination of metal sieves with glass bead
beds, is described as a means to prevent low pore activation, which is a drawback of
using devices with uniform pores. The pros and cons of all methods are summarized.
xxvi Synopsis
Chapter 11
Formation and Modification of Dispersions Using
Shirasu Porous Glass Membranes
Goran T. Vladisavljevic’
Shirazu porous glass (SPG) membrane morphology, chemical composition, and
methods to modify its surface chemistry, as well as the common setups for SPG
membranes are presented. Emulsification performance parameters of the system,
both in cross-flow and direct emulsification and in premix, straight-through emulsi-
fication, are explained and quantified.
Examples are given of the use of PSG membranes for the formation of W/O, O/W,
and multiple emulsions, as well as emulsion-based particles and gas bubbles.
Section I
Emulsion Fundamentals
1 Scales and Forces
in Emulsification
Marilyn Rayner
contents
1.1 Introduction ......................................................................................................3
1.2 Droplet Size Distributions ................................................................................4
1.3 Overview of Emulsification Machines and Homogenization Devices .............7
1.4 Droplet Disruption Fundamentals .................................................................. 11
1.4.1 Interfacial Forces ................................................................................ 12
1.4.2 Disruptive Forces ................................................................................ 13
1.4.3 Flow Types .......................................................................................... 15
1.5 Flow Regimes ................................................................................................. 16
1.5.1 LV Flow Regime ................................................................................. 18
1.5.2 Laminar Interfacial Tension-Driven Regime ..................................... 19
1.5.3 TI Regime ........................................................................................... 23
1.5.4 TV Flow Regime ................................................................................24
1.5.5 Cavitation Inertial Regime .................................................................25
1.6 Comparison of Emulsification Efficiency .......................................................26
1.7 Complications and Concluding Remarks ....................................................... 29
References ................................................................................................................ 30
ABSTRACT This chapter introduces the mechanical aspects of the formation and
breakup of emulsion droplets. An overview of the most common types of emulsifica-
tion machines and homogenization devices is provided. Droplet disruption mecha-
nisms are presented considering the forces, length, and timescales involved in the
various types of flow regimes encountered in the mechanical production of emulsions
by different categories of equipment. Finally, the energy efficiency of emulsification
methods is discussed, highlighting some complications with the theory presented, in
addition to some general comments on the future in the area.
1.1 IntroductIon
In order to make an emulsion, four basic ingredients are required: two immiscible
phases (often oil and water), energy to create the oil–water interface of the emulsion
drops, and surfactants to stabilize this interface to prevent the drops from coalesc-
ing immediately upon contact. By varying the amounts and composition of these
four basic ingredients, we can control key emulsion properties such as emulsion
3
4 Engineering Aspects of Food Emulsification and Homogenization
type (i.e., oil-in-water [O/W] versus water-in-oil [W/O]), droplet volume fraction,
ϕ (also known as “dispersed phase faction”), droplet size and size distribution, and
the nature of the stabilizing layer surrounding the droplets. These properties deter-
mine many of the central organoleptic properties of emulsion-based food products
such as shelf life, appearance, rheology, texture, and flavor. How we convert energy
from bulk mechanical stresses during emulsification to generate and stabilize the
new interfacial area created as a result of the emulsion drops is, of course, central to
the study of emulsification technologies and the focus of much of this book. A wide
variety of processing equipment and technologies are available for the generation of
emulsions, many of which will be thoroughly discussed in Chapters 5 through 11.
In line with the aim of this book, we consider emulsification in the context of food
emulsions, and thus only the mechanical production of emulsions will be discussed.
In the chemical industry, there are other methods to produce emulsions that do not
rely on mechanical generation of droplets, such as precipitation of the dispersed
phase previously dissolved in the continuous phase and the phase inversion tem-
perature method. These and other nonmechanical methods, however, are outside the
scope of this book. This chapter is an overview of the physics and technologies used
in the mechanical production of emulsions. For a more comprehensive treatment of
the topic, the reader is referred to Walstra’s eminent works in the area (Walstra 1993,
2003, Walstra 2005, Walstra and Smulders 1998) as well as Schubert and Ax (2003),
Karbstein and Schubert (1995), and McClements (2005).
The basic physical process of emulsification is illustrated in Figure 1.1, which
depicts the two main types of emulsification processes: high energy and low energy. In
both cases, the four main ingredients of emulsions are shown: (1) a continuous phase,
(2) a dispersed phase, (3) an emulsifier or a stabilizer, and (4) the necessary energy
input to create more oil–water interface. In the case of high-energy emulsification,
intense mechanical input is used to break up a coarse premix of emulsion drops. This
mechanical energy may arise from stirring, shearing, turbulent eddies, or even from
ultrasonic waves. This can be practically achieved through a wide range of different
technical designs and apparatus, however what is similar for all high-energy methods
is that droplet disruption is determined by the strength and duration of the applied
mechanical energy. High-energy emulsification processes are discussed in detail in
Chapters 5 and 6, on high pressure homogenization; Chapter 7, on microfluidization;
and Chapter 8, on rotor stator devices.
In the case of low-energy emulsification processes, a coarse premix is not nec-
essary. However, in some cases, the dispersed phase consists of an emulsion itself.
Droplets are formed directly in the continuous phase by injection through a porous
membrane or microchannel. Low-energy emulsification processes are discussed in
detail in Chapters 9 through 11.
1.2 droplet sIze dIstrIbutIons

Since creating fine emulsions is the objective of mechanical emulsification, the
droplet size distribution is the key variable used to evaluate the effectiveness of the
process. Small droplet size is strongly correlated with high emulsion stability, as
creaming is often a precursor to coalescence. For this reason, the majority of the
Scales and Forces in Emulsification 5
Premixer Emulsification machine
4. Mechanical energy Intense mechanical energy input
1. Continuous phase
Deformation
2. and disruption
Dispersed Fast stabilization
phase with sufficient
concentrations
3. Emulsifier
stabilizer
Slow stabilization
or deficient
concentrations
Coalescence
Premixing— Emulsification in the droplet disruption zone— Stabilization of new interface—

coarse emulsion fine emulsion stable emulsion
(a)
1. Continuous phase 4. Energy input for

continuous phase flow Fast stabilization
with sufficient
concentrations
Slow stabilization or deficient

Formation concentrations
and detachment
3. Emulsifier/stabilizer
2. Dispersed phase 4. Energy input to press dispersed

phase through pores/channels Jetting and coalescence
Phases Emulsification in the droplet formation zone and stabilization of new interface—
generally separate uniform drops
(b)
FIGure 1.1 (a) High- and (b) low-energy emulsification processes. (High-energy processes
redrawn based on Karbstein, H., and Schubert, H., Chem. Eng. Process., 34, 3, 205–211, 1995.)
studies on formulation, production, and evaluation of emulsion focus on droplet size

and droplet size distributions.
Because size distributions are of a statistical nature, the mean of these distributions
can be presented in several ways. A pedagogic treatment of particle size distributions
can be found in Walstra’s (2003) book, but in brief most literature reports either the
volume frequency average diameter, d43, or the volume/surface average (also called
Sauter mean), d32, of droplet size distributions. These means are generated as follows:
if the number frequency of droplets as a function of droplet diameter, d, is given by
f(d), then the nth moment of the distribution is
∞
∫
Sa ≡ d a f ( d )∂d
0
(1.1)
Thus, any type of average dab is given by

1/( a − b )
S 
dab =  a  (1.2)
 Sb 
In the case of discrete values, such as measuring droplets from microscope images,
the equations for d32 and d43 are as follows:
∑
n
di3
d32 = i =1
(1.3)
∑
n
di2
i =1
∑
n
di4
d43 = i =1
(1.4)
∑
n
di3
i =1
where:
di is the diameter of the ith measured drop of a total of n drops measured
Furthermore, in many texts, the coefficient of variation (CV) of a given mean as a

percentage of the standard deviation is also used:
Stdev
CV = × 100% (1.5)
dab
n 1/ 2
 (di − dab )2 
Stdev = ∑ i =1

 n −1 
 (1.6)
Another emulsion variable directly related to the droplet size distribution is the spe-
cific surface area of the emulsion, AS:
6φ
As = (1.7)
d32
where:
ϕ is the volume fraction of the disperse phase droplets and has the unit m−1, that
is, m2 interfacial area of emulsions droplets per m3 of emulsion
This specific surface area is important as it relates to both the amount of energy put
into the system to create this interface as well as the amount emulsifier or surfactant
required to stabilize the emulsion.
There are many different methods to assess particle size distributions, such as
microscopy (light, confocal, electron, etc.), particle counters (i.e., Coulter counter),
light scattering (i.e., Malvern Mastersizer), nuclear magnetic resonance (via diffusion
times), and sedimentation/centrifugation (Walstra 2005). They vary with respect to
particle size ranges covered, measurement principles, and limitations. The interested
reader is directed to McClements’s (2007) comprehensive review on emulsion char-
acterization techniques for more details.
1.3 overvIew oF emulsIFIcatIon machInes

and homoGenIzatIon devIces
Common emulsification apparatuses are schematically illustrated in Figure 1.2 and
their main features are summarized in Table 1.1. In high-speed mixers and agitated
tanks, the average energy density input per unit volume that can give rise to the
necessary local stresses required to break large drops into smaller ones is rather
low, unless one is working on a very small volume scale. In agitated tanks, droplet
disruptive stresses are a function of power per unit volume and are determined by
the tank and stirrer geometry, the viscosity of the liquids, and the rotational speed
Inlet (premix)
Stator
Rotor
Outlet
(a) (b)
FIGure 1.2 Schematic illustration of various designs of emulsification machines and

homogenization devices: (a) high-speed mixers and agitated tanks are often used in the food
and cosmetic industry to directly homogenize oil and water and (b) rotor–stator homogenizers
such as colloid mills are more applicable for intermediate and high viscosity fluids. (Continued )
Outlet
Ultrasound probe
Outlet
Inlet
Valve
Impact ring
Valve seat
Inlet
(c) (d)
Piston pump
Inlet Dispersed phase is pressed through the

membrane or microchannels
Channels
Circulation of the
continuous phase
Impact
zone
Fine emulsion
(e) (f)
FIGure 1.2 (Continued ) Schematic illustration of various designs of emulsification

machines and homogenization devices: (c) high-pressure homogenizers are used to produce
fine droplet sizes mostly in low viscosity products such as cream and milk; (d) ultrasonic
homogenizers use sound waves to generate intensely disruptive forces through cavitation;
(e) microfluidizers can produce extremely fine emulsions at high flows; and (f) membrane and
microchannel emulsification processes produce uniform droplets at low energy inputs.
of the stirrer. Thus, mixed tanks are generally limited to the production of coarse
emulsions batchwise as a pre-step to other emulsification processes.
In rotor–stator type equipment such as colloid mills and toothed-disc dispersing
machines, drops are disrupted in the gap between the rotating rotor and the station-
ary stator. In colloid mills, drops are disrupted in the conical gap, which can be either
smooth or serrated with various designs. Here, the droplet disruptive stresses are
determined by the gap width (typically 100–3000 µm), rotor radius, rotational rate
(typical peripheral speeds between 5 and 40 m s−1), and the liquid flow rate through
the gap, which can range between 4 and 20,000 l h−1 (Karbstein and Schubert 1995,
McClements 2005). Colloid mills are most suitable for production of intermediate
to high viscosity products and can achieve droplet diameters between 1 and 5 µm
(McClements 2005). Toothed disc dispersing machines are similar to a colloid
mill, except that the flow is not specifically bounded, consisting of single or several
table 1.1
some Features of common methods and machines to produce emulsions
bounded or energy relative energy droplet sizes
homogenizer types Flow regime un-bounded Flow density (J m−3) efficiency achieved viscosity typical volumes
Stirred tanks TI, TV, LV U Low-high Low 2 µm and larger Low-to-medium Batches up to
103–106 several m3
Colloid mill LV (TV) B Low-high Intermediate 1–5 µm Medium-to-high 4–20,000 l h−1
103–108
Toothed-disc disperser TV B Low-high Intermediate 1–10 µm Low-to-medium Batches cm3 up to
(e.g., Ultraturrax) 103–108 several m3

High-pressure TI, TV, (CI), LV U Medium-high High 0.1 µm Low-to-medium 100–20,000 l h−1
homogenizer 106–108
Ultrasonic probe CI U Medium-high Low 0.1 µm Low-to-medium Batches <100 cm3
106–108
Ultrasonic jet CI U Medium-high High 1 µm Low-to-medium 1–500,000 l h−1
106–108
Microfluidization TI, TV B/U Medium-high High <0.1 µm Low-to-medium Up to 12,000 l h−1
106–108
Membrane and Injection STB B Low 103 Exceptionally 0.3 µm—often Low-to-medium Batch or
microchannel high larger semicontinuous
10’s l h−1
CI, cavitation inertial; LV, laminar viscous; STB, spontaneous transformation based, TI, turbulent inertial; TV, turbulent viscous.
9
pairs of concentrically arranged split discs, one of which rotates at high speed, with
typical peripheral speeds up to 40 m s−1 (Karbstein and Schubert 1995).
High-pressure homogenizers, in contrast to mixers and rotor–stator type machines,
do not have moving parts acting in the droplet disruption zone. High-pressure valve
homogenizers are currently the most common method of producing fine emulsions
in the food industry. Here, a high-pressure pump delivers a coarse premix emulsion
through a narrow orifice or valve, which can have various designs. High-pressure
drops are achieved (typically between 3 and 20 MPa) where the emulsion flows
at very high speed (up to 2000 ms−1). Droplets are disrupted due to elongational
flow, eddies, stress fluctuations inside turbulent jets, or in some cases, by cavitation
(Karbstein and Schubert 1995, Schultz et al. 2004). High-pressure homogenization
is most suitable for low-to-medium viscosity products (such as milk and cream) and
submicron-sized droplets can be attained by a single pass through the homogenizer
with industrial production rates ranging from 100 to 20,000 l h−1. Most commercial
homogenizers use a spring-loaded standard valve design such that the gap through
which the emulsion is passed can be adjusted (typically between 15 and 300 µm)
(McClements 2005). Decreasing the gap height increases the pressure drop across
the valve, thereby increasing the intensity of the flow and droplet disruptive forces,
which in turn can produce smaller droplets. The relationship between pressure drop
and droplet size is roughly linear in the log–log scale, that is, log d ∝ log P, the
slope depending on flow conditions within the homogenizer, which in turn are deter-
mined by the valve geometry and fluid properties (Karbstein and Schubert 1995,
Phillips 1985).
Ultrasonic homogenizers use high-intensity sound waves to generate intense
shear and pressure gradients within the liquid that disrupts droplets mainly by
cavitation and turbulent effects (McClements 2005). There are two methods com-
monly used in the industry to produce ultrasonic waves. Piezoelectric transducers
are used for small batch volumes ranging from a few cubic centimeters to a few hun-
dred cubic centimeters, and liquid jet generators are used on a larger scale where
a jet coarse emulsion is pumped to impinge on a sharp-edged blade. This jet flow
causes the blade to vibrate rapidly, thus generating the ultrasonic field that breaks
droplets in its immediate vicinity. Ultrasonic jet-type homogenizers can produce
larger volumes continuously with fluid flow rates ranging from 1 to 500,000 l h−1.
The factors that govern droplet disruption are intensity, duration, and frequency of
the ultrasonic waves in relationship to the volume of emulsion they are applied on
(McClements 2005).
Microfluidizers are similar to high-pressure homogenizers in that they do not
have moving parts in the droplet disruption zone. This type of homogenizer consists
of fluid inlet(s), some type of powerful pumping device, and an interaction chamber
containing channels through which the fluids are forced to flow and interact with
each other in an impact zone. A coarse premix emulsion (or in some cases the indi-
vidual oil and water phases depending on the design) is accelerated to exceptionally
high velocities within the channels of the microfluidizer and then led to simulta-
neously impinge on each other on a solid surface inside the interaction chamber
(McClements 2005). Microfluidizers are available in a wide range of production vol-
umes ranging from 10 ml up to 12,000 l h−1, maintaining an operating pressure of
up to 270 MPa, thus having the ability to produce very fine emulsions with droplet
sizes <0.1 µm (Lee and Norton 2013, McClements 2005).
Membrane and microchannel emulsification methods involve pressing a dispersed
phase (either as a single phase or as a premix) through a microporous membrane or
a microchannel chip. This process, invented by Nakashima’s group (Nakashima,
Shimizu, and Kukizaki 1991), makes use of microporous membranes, where the
continuous phase flows tangentially to the membrane surface and the dispersed
phase is pressed through the membrane. Droplets of dispersed phase are formed at
the openings of the membrane pores and are continuously swept away by the flowing
continuous phase. The key feature of the membrane emulsification process, which
sets it apart from conventional emulsification technologies, is that the size distri-
bution of the resulting droplets is primarily governed by the choice of membrane
rather than the development of turbulent or extensional droplet breakup (Peng and
Williams 1998). Hydrophilic membranes are used in making O/W emulsions and
hydrophobic membranes are used in making W/O emulsions, the point being that the
dispersed phase should not wet the membrane surface, that is, the oil–membrane–
water contact angle should be >145°. A close relative to membrane emulsification is
microchannel emulsification, which operates on the same principle of extruding the
dispersed phase into the continuous phase, but differs with respect to the membrane.
As the name implies, microchannel emulsification uses a microfabricated channel
array. These plates can be stacked to create a module to increase production output.
Membrane and microchannel emulsification methods are carried out at very low-
energy input rates, and although energy is required for the generation of droplets,
they are not subsequently broken by flow conditions in the continuous phase. Here,
droplet size is determined mainly by the geometry of the pore or channel rather than
by the flow conditions in the continuous phase, thus the energy transfer for creating
a new interfacial area (i.e., the area of emulsion droplets) is much more efficient than
in high-energy processes, which rely on turbulent or viscous shearing by the flowing
continuous phase. For this reason, low-energy emulsification processes often have
narrow droplet sizes and are more suitable for sensitive ingredients, but generally
have larger droplet sizes and much lower production rates.
1.4 droplet dIsruptIon Fundamentals

The exact nature of the processes that take place during emulsification depends
largely on the type of emulsification equipment used, because the flow profiles,
geometry, and energy dissipation are determined by the equipment design, which in
turn determine the mechanical forces and stresses acting on droplets that can lead
to their disruption into smaller ones. The formation of emulsion drops is a relatively
simple process. We can make a coarse emulsion by simply shaking or stirring a
surfactant solution with some oil. However, making the droplets sufficiently small
to create stable emulsion products is the real challenge, the reason being that the
Laplace pressure of the drops, which gives rise to the resistance to deformation and
breakup of drops, is inversely proportional to the droplet radius. As the Laplace pres-
sure increases with decreasing drop diameter, making successively smaller drops
requires successively higher external stresses applied by mechanical means and/or
successively higher concentrations of surfactant to lower the interfacial tension. In

most emulsification technologies, a coarse dispersion of droplets is successively
broken down into finer ones. Droplets are disrupted if the Laplace pressure, pL (i.e.,
the interfacial forces acting to keep the droplets together), is locally exceeded by
the deformation stresses, σ (i.e., the disruptive forces acting to pull droplets apart
into smaller ones), for a period of deformation time long enough for droplet break-
age, t br, to occur (i.e., the local disruptive forces act long enough on the droplet for
breakup to occur), which exceeds some critical droplet breakage time, t br,cr (Walstra
and Smulders 1998).
σ > pL and tbr > tbr,cr (1.8)
In the following sections, the scales and forces associated with emulsion formation
will be discussed and several simplifications have been made. Here, the interfa-
cial tension, γ, is assumed constant. This means that either no surfactant is present
(not ever practically the case) or that there is a large excess. Furthermore, we have
assumed the phases to have a Newtonian viscosity (true for oil and dilute water
phases) as well as relatively low volume fractions; thus emulsification processes for
high internal-phase emulsions are not considered.
1.4.1 InterfacIal forces

The Laplace pressure is defined as the difference between the inside and the outside
of a drop dispersed in a continuous phase and is proportional to the interfacial ten-
sion, γ, between these phases.
2γ
pL = (1.9)
R
where:
γ is the interfacial tension
R is the radius of a spherical drop
For any general closed surface, the Laplace pressure is a function of the mean curva-
ture, H, defined by its two principal radii of curvature, R1 and R2:
 1 1 
pL = 2 γH = γ  +  (1.10)
 R1 R2 
One way to understand the mechanical basis of the Laplace pressure is to consider a
differential change in surface energy dEsurf = γdA and the corresponding differential
change in volume for a drop of a given size.
dEsurf γdA
pL = = (1.11)
dV dV
Recall the following geometric relationships for a sphere and the chain rule of
calculus:
Area:
dA
A = 4πR2 , = 8πR
dR
Volume:
4 3 dV
V= πR , = 4πR2
3 dR
Thus,
dA dR dA 2
⋅ = =
dR dV dV R
Substituting into Equation 1.11 we get the expression of Laplace pressure for a spher-
ical drop:
dEsurf γdA 2 γ
pL = = = (1.12)
dV dV R
To break an existing drop into smaller ones in the context of emulsification, it must
be strongly deformed, and any deformation from a spherical shape increases the
Laplace pressure. As seen in Figure 1.3, R1 and R2 are the principal radii of curva-
ture in Equation 1.10. This deformation can be achieved via shear and elongational
flow in the continuous phase or by extrusion through a noncircular pore. According
to Taylor’s (1934) theory of deformation-induced breakup of droplets, the droplet
deformation can be described by
L − B R1 − R2
δTaylor = = (1.13)
L + B R1 + R2
His work showed that under idealized conditions the breakup of a droplet is only
possible when the deformation achieves a critical value δTaylor > 0.5. If we set
Equation 1.13 equal to 0.5 and solve for L, we get L > 3B; meaning, if a drop is
deformed into an ellipsoid to such an extent that its length is more than three times
its breadth, it can break up into two daughter drops. This criterion seems to hold in
shear stress induced by viscous flows as well as in extrusion through membranes and
microchannels. The latter phenomenon has been termed spontaneous transformation
based (STB) droplet formation. In this case, the interfacial forces are also driving
the droplet formation, rather than being merely a source of resistance that disruptive
forces need to overcome. This is discussed further in Section 1.5.2.
1.4.2 DIsruptIve forces

As mentioned above, there are many different methods and machines to make emul-
sions. These differ in several aspects, including energy intensity and efficiency,
whether operated in batch or continuous mode, the minimum size of drops attainable,
the viscosities that can be handled as well as the flow profiles/conditions present in
L
R1
R1
R2 B R2
x
dVx
(a) Droplet breakup in simple shear flow, G = = shear rate (velocity gradient)
dy
L
x R1 R1
R2 B
R2
dVx
(b) Droplet breakup in elongational flow, G = = elongation rate
dy
Cross section of oil in a pore where

B aspect ratio: L/B > 3
Membrane or
microchannel
Droplet necking inside pore
(c) Droplet breakup in spontaneous transformation based droplet formation
FIGure 1.3 Illustration of droplet formation in laminar flows: Simple shear flow (a) and
elongational flow (b). As a spherical droplet is deformed it takes on an ellipsoidal shape with
two principal radii of curvature, R1 and R2. Once deformation exceed a certain extent it will
break into two or more daughter droplets. A similar deformation criteria can been seen in
extruding through pores and channels with high aspect ratio geometries (c).
the machine (see Table 1.1). It is the latter that is the basis for the external stresses that
lead to the generation of disruptive forces for droplet breakup. In most emulsification
processes, the velocity or pressure gradients required for droplet disruption forces are
delivered via flow conditions in the continuous phase; the smaller the desired drops,
the more intense this flow needs to be. There are two types of external forces causing
the disruption of drops during emulsion formation: frictional and inertial.
Frictional forces (also called as viscous or shear forces) are due to the flow
of the continuous phase along the surface of the drop, thus viscous forces
mainly act parallel to the drop surface. The local stress generated is a func-
tion of the strain rate, G (velocity gradient), and the continuous phase vis-
cosity, ηc, that is, σ ∼ Gηc (see Figure 1.3).
Inertial forces are generated by the local pressure fluctuations caused by local
velocity fluctuations in turbulent flows. They generally act perpendicu-
lar to the surface of drops. The local stress generated is a function of the
continuous phase density and the mean and local velocity differences, that
is, σ ∼ ρcu ∆u .
1.4.3 flow types

There are several ways to consider flows; they can be either laminar or turbulent and
they can be either unbounded or strongly confined flows. In unbounded flows, any
droplet is surrounded by a sizable amount of flowing liquid, or in other words, the
confining walls of the apparatus are far away from the majority of the droplets. The
droplet-disrupting forces are in this case frictional (i.e., viscous) or inertial (Walstra
and Smulders 1998). In strongly confined flows (bounded) flows, the smallest dimen-
sion of the apparatus in which the droplets are formed or disrupted (e.g., a slit or
a membrane pore or the gap in a colloid mill) is comparable to the droplet size.
Here, the droplets are deformed by interactions between a confining geometry and
a pressure-driven flow, which causes the oil–water interface to deform when it is in
contact with the solid confinement. This is typical characteristic of membrane and
microchannel emulsification methods.
The main distinction made between flow types is whether the flow is laminar to
turbulent via the Reynolds number, which is the most important way to describe any
flow, or movement of a drop within a liquid:
inertial forces ULρ

Re = = (1.14)
viscous forces η
where:
U is linear velocity
L the characteristic length
ρ the density
η the viscosity of the flowing fluid
The choice of U and L depend on the whether we are considering the flow in relation
to the apparatus it is flowing through, or the flow experienced by a drop in a continu-
ous medium. For example, for the flow in a pipe, U is the mean linear liquid velocity,
that is, u, and L is the diameter of a pipe. In the case of flow around a suspended
drop, U is the drops velocity relative to the continuous phase directly surrounding
it and L is the drop diameter. In the case of a mixer in a vessel, Rei is related to the
diameter and rotational speed of the impellor.
Ni Di2ρ
Rei = (1.15)
η
For each type of flow, there exists one or more flow regimes that can describe the
source and intensity of disruptive forces acting on droplets. Within each flow regime,
there is an essential variable that describes the intensity of the disruptive forces
(or stresses) acting upon the droplets. This is often a mechanical stress, σ, generated
by the flow. The ratio of the disruptive forces (pulling drops apart) to the interfacial
forces (holding drops together) is represented by the Weber number in turbulent
flows and the capillary number, Ca, in laminar flows (also known as the laminar
Weber number, WeL).
inertial force
We = (1.16)
interfacial tension force
viscous force
Ca = We L = (1.17)
interfacial tension force
The origin, magnitude, and efficacy of the inertial and viscous forces to achieve
droplet disruption are determined by the type of flow conditions and the properties
of the phases.
1.5 Flow reGImes

In all methods of droplet formation and breakup, there is liquid flow and thus
emulsification processes are often classified according to their prevailing flow
regimes (Tables 1.1 and 1.2). The flow of liquids generates external forces act-
ing upon the drops, which generally act via the continuous phase (Walstra and
Smulders 1998). Here, the forces are derived from the mechanical energy, from
the pumping, stirring, or shearing of the liquids. Depending on the geometry and
velocity, different flow regimes are present, specifically, laminar viscous (LV),
interfacial tension driven, turbulent inertial (TI), turbulent viscous (TV), and
cavitational.
In LV flow (Re < 1000), droplets are deformed and broken by simple shear flow
(Figure 1.3a) or elongational flow (see Figure 1.3b), where the droplet breakup origi-
nates from extension and tip streaming (Windhab et al. 2005). In low-energy emulsi-
fication processes such as membrane and microchannel emulsification, the geometry
of the pore and interfacial effects play a key role (Figure 1.3c). Here, in contrast to
hydrodynamically driven droplet disruption, droplet detachment is, at least in part,
driven by the interfacial tension.
In cases where the flow regime is turbulent (Re > 2000), droplet breakup is either
from the action of turbulent eddies or cavitation (Figure 1.6). Droplet breakup in
turbulent flow has been described by the Kolmogorov–Hinze theory (Hinze 1955,
Kolmogorov 1949); the two main types of the droplet breakup are identified: TI and
TV (Vankova et al. 2007, Walstra and Smulders 1998).
In the following sections, the different flow regimes generally encountered in
various mechanical emulsification processes are described with respect to their
disruptive forces, characteristic times, and expected droplet diameters based on
physical properties and process conditions. Typical cases for common emulsification
processes are summarized in Table 1.2.
table 1.2
summary of equations for typical cases for common emulsification processes
Flow regimes laminar-viscous shear or elongational turbulent-viscous shear Forces turbulent-Inertial Forces
Reflow <1000 >approx. 2500 >approx. 2500
Re-droplet <1 <1 >1a
Forces
3
External stress acting on droplets (σ) ηc G ε ηc ε 2 d 2 ρc
length scales
1/ 5
Mean diameter (d≈)b (2 γ We cr )/(ηc G) γ/( ε ηc )  γ 3 /(ε2 ρc )
 
timescales
Droplet deformation timescale (τDEF ) ηd /(ηc G) ηd / ε ηc ηd / 3 ε 2 d 2 ρc

( ) ( )
1/ 5
Duration of disruptive stresses (τDIS ) 1/G ηc /ε (1/2) ( γ 2ρc )/ε3 
Surfactant adsorption timescale (τADS ) (6 π Γ)/(d mc G) [(6 π Γ)/(d mc )] ( ηc /ε ) (Γ/mc )  3 ρc /(d ε) 

 
—
Droplet collision timescale (τCOL ) π/(8 G φ) [1/(15 φ)]  3 (d 2 ρc )/ε 
 
Sources: Adapted from McClements: D.J., Food Emulsions: Principles, Practices, and Techniques, CRC Press, Boca Raton, FL, 2005; Walstra, P., Chem. Eng. Sci., 48, 2,
333–349, 1993; Walstra, P., and Smulders, P.E.A., Modern Aspects of Emulsion Science, The Royal Society of Chemistry, Cambridge, pp. 56–99, 1998.
Symbols: We, Weber number (see text); Re, Reynolds number (see text); Γ, surface excess of surfactant (mol m−2); ε, power density (J s−1 m−3); d, droplet diameter (m);
γ, interfacial tension (J m−2); η, viscosity (Pa s); G, velocity gradient (s−1); mc, surfactant concentration in the continuous phase (mol m−3); τ, characteristic time (s);
σ, stress (pa); ρ, density (kg m−3).
Subscripts: d, dispersed phase; c, continuous phase; DEF, deformation; ADS, adsorption; cr, critical value for droplet break-up.
a For d > η2 /( γ ρ ).
c c
d c
b Only if η  η .
17
1.5.1 lv flow regIme

Re flow < 1000, Re drop < 1
Here, the droplets are deformed by viscous shear stresses acting on the droplets
generated by a velocity gradient inside a viscous liquid. There are two extremes
with respect to velocity gradients: simple shear and purely elongational (Figures 1.3).
Simple shear causes rotation as the fluid is moving faster on the top of the drop rela-
tive to the bottom of the drop, which also causes the liquid inside the drop to rotate.
For small Weber numbers, the Taylor deformation (Equation 1.13) is equal to the
Weber number in laminar flow:
σ G ηc d
We L = = (1.18)
0.5 pL 2γ
where:
d is the droplet diameter (Walstra 2003)
For larger values of We, the drop becomes deformed and will break if We exceeds a
critical value Wecr. Numerous experimental studies have considered how flow condi-
tions and viscosity ratio of the continuous and dispersed phases affect the critical
Weber number required for the droplet breakup. Figure 1.4 is a typical plot of the
effect of viscosity ratio on Wecr for various types of laminar flow. By taking Wecr
from the plot and inserting it into Equation 1.18, the largest drop with a given viscos-
ity and interfacial tension that could remain unbroken in a given applied flow field
can be estimated. For simple shear the following equation gives a good approxima-
tion for the maximum droplet sizes that can persist under steady-state conditions in,
for example, a simple shear flow found in colloid mills (Walstra 1993, Walstra and
Smulders 1998):
α=0
10 Shear
Critical Weber number, Wecr
α = 0.2
α=1
Elongational α = 0.6
0.1
Viscosity ratio, ηd/ηc
FIGure 1.4 The effect of viscosity ratio ηd/ηc (dispersed phased over continuous phase) on
the critical Weber number required for drop breakup in the laminar flow with varying degrees
of elongation, from simple shear (α = 0) to purely plane hyperbolic elongation (α = 1).
2γ We cr
dmax = (1.19)
ηcG
However, in practice, the average droplet sizes obtained are somewhat smaller
(Walstra 2003). The viscosity ratio has a significant effect on Wecr in a simple shear
flow and if the dispersed phase has a viscosity more than four times that of the con-
tinuous phase, it does not disrupt at all, but rather rotates in the flow field. The reason
for this lies in relative magnitude the timescales of droplet deformation and rotation.
The time required for deformation in simple shear is as follows:
ηd
τDEF = (1.20)
ηc G
The time required for a drop to turn half a rotation is π/G; thus setting this equal to
Equation 1.20 yields ηd = πηc, which approaches 4ηc (Walstra 2003). Aside from the
strong effect viscosity ratio has on whether the droplet breakup seen in Figure 1.4 can
be achieved, the type of flow also plays a significant role where elongational flows
have much lower critical Weber numbers. In elongational flow (plane hyperbolic),
the denominator of Equation 1.19 would be 2ηcG. Furthermore, there is no internal
rotation of droplets in elongational flow and becomes highly elongated independent
of the viscosity ratio. Laminar flow patterns can also be combinations of shear and
elongation (curves shown in Figure 1.4 for 0 < α < 1), and such types of intermediate
flow are common in stirred vessels and mixers (Walstra 2003).
In many practical situations, different emulsifying equipment is compared by
their power density, that is, how much energy per unit time and volume is available
for droplet disruption. In simple laminar flows, this is a function of the viscosity and
local velocity gradient with the units of J s−1 m−3.
ε = ηcG 2 (1.21)
For the case of a mixer or stirrer in a vessel with laminar flow conditions (Rei < 10)
the power density, ε, is proportional to the revolution rate squared, Ni2 and the
impellor diameter to the power 3 divided by the fluid volume in the vicinity of the
impellor Vi ∼ Di (the term Di cancels out).
3
ε = klηN i2 (1.22)
The constant k l varies depending on the impellor type and Reynolds number. Plots
of these constants are found in texts on industrial mixing (Doran 1995); for example,
at Rei = 1, for Rushton turbines k l equals 70, for paddle type stirrers 35, for pitched
propellers 40, and for helical ribbons 1000.
1.5.2 lamInar InterfacIal tensIon-DrIven regIme

In Section 1.4.1, the effects of interfacial tension have been regarded as the resistance
to droplet breakup, where the disruptive forces must overcome the Laplace pres-
sure. This is, of course, generally the case in conventional emulsification processes.
However, in many types of membrane and microchannel emulsification methods,

the interfacial tension and the free energy associated with it may, in fact, provide
the actual means to drive droplet detachment. Sugiura et al. (Sugiura et al. 2001,
Sugiura, Nakajima, and Seki 2002) presented a mechanism for interfacial tension-
driven droplet formation. This mechanism termed spontaneous transformation
based droplet formation is described by considering the change in Gibbs free energy,
ΔG, of the system. In their setup, the droplet was deformed by the rectangular geom-
etry of the microchannel causing it to have a disc-like shape, which is unstable from
the viewpoint of Gibbs free energy, as it has a much greater interfacial area than a
sphere of equivalent volume. The ability for a droplet to spontaneously form was
calculated from the reduction in total interfacial area from before, and after, the
droplet forms through estimating the interfacial areas from video images. Sugiura
et al.’s estimation of the free energy of droplet formation showed that STB droplet
formation was favorable. They also found that the geometry of the microchannel
played a critical role in STB droplet formation as it is essential that the droplet is
deformed from its spherical, lowest energy shape. This type of droplet deformation
does not solely take place in microchannel emulsification, but is also observed in
Shirasu porous glass (SPG) membrane emulsification (Yasuno et al. 2002) and in
arrays of straight through holes in silicon plates (Kobayashi and Nakajima 2002,
Kobayashi, Nakajima et al. 2002). SPG membranes, which have been used to make
emulsion droplets of uniform size, do not have circular pores. They can be described
as a network of tortuous ellipsoidal cylinders (Figure 1.5). This noncircular shape
increases the surface area of the droplet-to-volume ratio, making the oil–water inter-
face farther away from its lowest energy configuration, which may indeed be the key
to their performance.
Kobayashi et al. (Kobayashi, Mukataka, and Nakajima 2004, Kobayashi et al.
2011) explicitly showed that it is not only the overall size of the pore, but its aspect
ratio that determines the location of the droplet necking during detachment and the
ability of the membrane to form droplets with a narrow droplet size distribution.
They found that there was a critical aspect ratio (pore width divided by its breadth)
Microporous glass High aspect ratio

SPG membrane microsieve
FIGure 1.5 Images of STB droplet-forming structures. Left: illustration of the pore
structure of a microporous glass membrane. (Courtesy of SPG instruction manual, ISE
Chemical Company, Japan.) Right: microchannel with rectangular-shaped pores. (Courtesy
of Aquamarijn BV.)
around 3. Droplets formed spontaneously from pores having an aspect ratio above 3,
whereas pores with aspect ratios below 3 had poly-dispersed droplets detaching with
the help of the continuous phase flow. Pores having an aspect ratio at 3 displayed a
mixed mechanism, where it was observed that some of the pores produced droplets
spontaneously whereas others exhibited large droplet inflation. This phenomenon
has been observed in microscope video images of droplets forming from round pores
by van der Graaf et al. (2004) and Kobayashi et al. (Kobayashi, Mukataka, and
Nakajima 2004, Kobayashi, Uemura, and Nakajima 2006, Kobayashi, Yasuno et al.
2002) as well as in computational fluid dynamics (CFD) simulations performed by
Abrahamse et al. (2001). Conversely, Kobayashi et al. (2004) did not observe neck-
ing outside the pore for their oblong straight through microchannels with aspect
ratio 3.8, but did see continuous outflow from pores with aspect ratios of 1, 1.9, and
2.7. The location of droplet necking can explain why markedly different emulsifica-
tion results are obtained depending on the shape of the pore. Pores with aspect ratios
greater than 3 produce droplets of regular size because necking occurs inside the
pore, which is illustrated in Figure 1.3c. Rounder pores with aspect ratio less than
3 produce poly-dispersed droplet-size distributions because the necking takes place
outside the pore exposed to the flow of the continuous phase. This flow, although
often laminar at these dimensions, is still rather chaotic. The shear stress to which any
given droplet is exposed to during necking may be drastically different depending on
when a neighbor droplet is detached and the path it took as it moved away from the
membrane surface, that is, droplets can shield each other from the shearing liquid.
An explanation as to why an aspect ratio of 3 is important can be found again in
Taylor’s (1934) description of the theory of deformation-induced breakup. Taylor
defines the deformation of a spherical oil droplet into an ellipsoid, as the droplet is
subject to shear-inducing flow inside a narrow gap, recall Equation 1.13.
L−B
δTaylor =
L+B
Here, L and B are the lengths of the major and minor axes directions. It was estab-
lished that the droplet breakup is only possible when the deformation achieves a
critical value δcrit > 0.5. If we set δTaylor = δcrit and solve for L, one obtains L > 3B.
Therefore, if Taylor’s ellipsoidal deformation concept is applied to the cross section
oil protruding inside a membrane pore or microchannel, then L/B is, by definition,
its aspect ratio and if it is greater than 3, breakup (i.e., necking) can occur in the pore
or microchannel.
The diameter of the resulting droplets can be estimated based on the interfacial
tensions, the critical pressure of the membrane pore or microchannel determined by
its geometry, and the three-phase contact angle (Rayner 2005, Rayner, Trägårdh, and
Trägårdh 2005). Equations and details are found in Table 1.3. If the aspect ratio is less
than 3, then further deformation of the droplet via drag forces generated by the flow-
ing continuous phase tangential to the pore are required for droplet detachment. This
type of shear-assisted droplet detachment is similar to LV, as it is the deformation of
the droplet via the wall shear stress that leads to droplet breakup (Peng and Williams
1998, Schroder and Schubert 1999, Timgren, Trägårdh, and Trägårdh 2009, 2010).
22
table 1.3
droplet sizes Generated in continuous emulsification processes for various Flow conditions
Flow regimes membrane/microchannel rotor stator/colloid mill high-pressure homogenizer
Laminar interfacial tension driven d = [(8 γ )/Pcrit ] ( γ o /γ ) – –
(Redroplet < 1, Reflow < 1000a) Pcrit = 2γ cosθ H
Laminar viscous—shear or elongational d = ( 4 γ Dpore )/(6 kwall σ wall ) d max = A1 EV−1 –

(Redroplet < 1, Reflow < 1000)
Turbulent inertial – d max ∝ PV−0.4 d max = A3 EV−0.6

−0.3
(Redroplet > 1, Reflow > 2500) d max ∝ t
d max = A2 EV−0.35
Turbulent viscous – d max ∝ PV−0.5 d max = A5 EV−0.75
(Redroplet < 1, Reflow > 2500) d max ∝ t −0.3
d max = A2 EV−0.4
Sources: Membrane and micro-channel correlations modified from Peng, S.J., and Williams, R.A., Chem. Eng. Res. Des., 76, A8, 894–901, 1998; Rayner, M., and
Trägårdh, G., Desalination, 145, 1–3, 165–172, 2002; Rayner, M. et al., Colloids Surf., A, Physicochem. Eng. Aspects, 266, 1–3, 1–17, 2005. Other from Walstra,
P., and Smulders, P.E.A., Modern Aspects of Emulsion Science, The Royal Society of Chemistry, Cambridge, pp. 56–99, 1998; Schubert, H., and Ax, K., Texture
in Food, Woodhead Publishing Ltd, New York, 2003.
Symbols: d, droplet diameter (m); γ, interfacial tension (J m−2) of the surfactant covered interface; γo, interfacial tension of the bare oil–water interface; Pcrit, critical break-
through pressure of extrusion of oil through the membrane; θ, oil–water–membrane contact angle; H, mean curvature of the pore or channel (see Equation 1.10);
Dpore, the diameter of a membrane pore of the microchannel, kwall, 1.7; σwall, wall shear stress (Pa); Pv, net power density (J s–1 m–3); EV, energy density
(see Equation 1.36 ) (J m−3); t, time in the homogenizer; A1, A2, . . ., A5 are constants that are a function of physiochemical properties of the fluids, that is,
Ai = f (ηd , ηc , ρc , γ,), in general d max = Ai EV− b, where b is an experimentally determined exponent and in this case dmax is the asuter mean diameter, d32.
Engineering Aspects of Food Emulsification and Homogenization
a Flow of dispersed phase for injection.

1.5.3 tI regIme
Reflow > 2500, Redrop > 1
Turbulent flow is typified by the occurrence of eddies (swirls and vortices), thus the
local velocity at a given point in the flow has a local velocity that generally differs
from its time-average value of u. This local velocity fluctuates chaotically and the
average difference over time between u and u will equal zero; however, the root
mean square of these fluctuations are finite (Walstra 2003).
u′ ≡ (u − u )21/ 2 (1.23)
It is the magnitude of these fluctuations on the length scale of turbulent eddies that
give rise to disruptive forces on drops. TI breakup takes place when the droplet
deformation is caused by the smallest scale eddies in the system and, therefore, drop-
lets are of the same order of magnitude to the eddy size. In the TV breakup regime
(Section 1.5.4), droplet sizes are reduced below the size of the smallest eddies in the
system from the shearing forces created within these eddies.
Turbulent flow has a spectrum of eddy sizes (l). The larger the eddy, the higher
the value of u′. Large eddies transfer their kinetic energy to smaller eddies. Although
smaller eddies have a lower degree of velocity fluctuations (u′), they have a higher
velocity gradient (u′/l ). Thus, the small eddies have a high-specific kinetic energy
and are called energy-bearing eddies having the size le. As such, the local velocity
on the size scale of these eddies is given by:
u′(le ) ≈ ε1/ 3le1/ 3ρc−1/ 3 (1.24)

The eddy size, life time, and energy are given by the Kolmogorov–Hinze theory
(Hinze 1955, Kolmogorov 1949), which are a set of scaling laws. The size scale of
the energy-carrying eddies can be estimated by
le ≈ ε −1/ 4η3c / 4ρc−3/ 4 (1.25)

having a timescale (eddy lifetime) on the order of
le
τ(le ) ≈ = le2 / 3ε −1/ 3ρ1c/ 3 (1.26)
u′(le )
Therefore, the smaller the eddy, the shorter its lifetime, but, at the same time, have a
higher local power density. Inside a small eddy, the local Re is small; hence the flow
on this scale is laminar. Here, as in macroscopic laminar flow, the power density is
ε ≈ ηcG 2, but in this case, the velocity gradient is u′(le )/le (Walstra 2003).
The local velocity fluctuation on the size scale of le (Equation 1.24) provides the
disruptive forces specifically taking place in the TI regime. In the TI regime, the
diameter of the drops are larger than the smallest turbulent eddies, le, and deform
under the action of fluctuations in the hydrodynamic pressure, schematically illus-
trated in Figure 1.6a, and can be estimated by the following:
2
∆p(le ) ≈ ρc u′(le )  ≈ ε2 / 3le2 / 3ρ1c/ 3 (1.27)
To disrupt a spherical droplet, a pressure must be externally applied over a distance

R (the drop radius) to overcome the Laplace pressure, which results in a pressure
gradient on the order of 2γ/R2. Assuming 1 µm drops and an interfacial tension of
10 m Nm−1, the local pressure gradient would need to be of the order 1010 Pa m−1,
thus a considerable amount of power needs to be dissipated over very small scales
(Walstra 2003).
If the pressure fluctuation ∆p(le ) near a drop is larger than the Laplace pressure,
the drop will be disrupted, that is, Wecr criterion is fulfilled as the inertial disruptive
forces exceed the interfacial retaining forces. Because drop disruption via pressure
fluctuations is most effective at the scale d = le (Walstra and Smulders 1998), and by
setting the Laplace pressure equal to Equation 1.27 (in a parallel way as done for
the Laminar Wecr earlier), substituting d for le, we obtain the maximum size a drop
can survive without breakup in the given turbulent field as a function of the applied
power density:
dmax ≈ ε −2 / 5 γ 3/ 5ρc−1/ 5 (1.28)
Due to the chaotic nature of turbulent flow, the droplet disruption likely occurs via an
abrupt local protrusion or indent on a drop. There is no reason to believe that drops
will break into to two equal volumes, hence the resulting droplet size distributions
are fairly wide, as observed experimentally. For valve-type high-pressure homoge-
nizers, the homogenizing pressure, pH, can be varied and the resulting power density
is proportional to p1H.5. For the case of a mixer or a stirred vessel the power density,
ε, is proportional to the revolution rate cubed, Ni3 the impellor diameter to the power
5 divided by the fluid volume in the vicinity of the impellor Vi ∼ Di . The constant k P
3
is called the power number and varies depending on the impellor geometry, but is
constant for a given geometry at Rei > 105, k P is equal to 5–6 for Rushton turbines, 2
for paddle-type stirrers, and 0.35 for pitched propellers (Doran 1995).
ε = kP ρc Ni3 Di2 (1.29)
For other types of machines and a more rigorous theoretical treatment of the topic,
the interested reader is referred to Vankova et al. (2007) and Walstra’s works and
references therein.
1.5.4 tv flow regIme

Reflow > 2500, Redrop > 1
Droplet deformation and breakup in TV flow is mechanistically similar to laminar

(Walstra 2003) and typically occurs when the viscosity ratios of the dispersed phase
to the continuous phase ηd/ηc is in the range of 0.1–5 (Lee et al. 2013, Walstra and
Lyklema 2005). In the TV regime, droplets are smaller than the smallest eddy size in
the prevailing hydrodynamic conditions, as illustrated in Figure 1.6b. Here, the drop-
lets are deformed and disrupted under the action of viscous stresses existing within
and between eddies. Because Redrop < 1, the flow near a drop is laminar. The criteria
for the transition between TI and TV regimes depend on the size of the smallest
eddies (Equation 1.25) and drop size. Here, the limits of Redrop can be estimated
using Equation 1.30 (Walstra 2003). It also gives the limit for condition for the larg-
est droplet in the Redrop in the TI regime (d > η2c /γρc ) in Table 1.2.
γ1/ 2ρ1c/ 2 d 1/ 2
Re drop = (1.30)
ηc
In the TV regime, the scale of the droplets is much smaller than that of the energy-
carrying eddies (d  le ); however, the local velocity needed at the scale of the
droplet is
u′(d ) ≈ ε1/ 2ηc−1/ 2 d (1.31)
because the shear stress acting on the drop is given by viscosity time and the velocity
gradient, that is,
u′(d )
σ = ηc (1.32)
d
As the flow between eddies is likely elongational (i.e., plan hyperbolic), Wecr will
not strongly depend on the viscosity ratio and the maximum droplet size in the TV
regime can be estimated by Equation 1.33 (McClements 2005, Walstra 2003).
dmax ≈ γε −1/ 2ηc−1/ 2 (1.33)
As what could be anticipated, the continuous phase viscosity is a variable governing

the droplet size in the TV regime, where instead it is the density in the TI regime
(Equation 1.28).
1.5.5 cavItatIon InertIal regIme

Cavitation can occur in fluids that are subjected to rapid changes in pressure, and
are of practical importance in ultrasonic and high-pressure homogenization. Fluids
contract when the pressure increases and expand when the pressure decreases. If
the instantaneous pressure acting on a fluid falls below a critical value (cavitational
threshold), a cavity forms. As the fluid continues to expand, this cavity grows and the
surrounding fluid is vaporized. During a subsequent increase in pressure (i.e., com-
pression due to hydrodynamic conditions or by the action of sonic waves), the cavity
catastrophically collapses generating an intense shock wave that propagates locally
into the surrounding fluid (see Figure 1.6c). This causes droplets in the immediate
vicinity of this event to be disrupted (McClements 2005). This is the operating prin-
ciple of ultrasonic homogenizers, and cavitation can also be contributing to droplet
disruption in some types of valve homogenizers. However, in the case of the latter,
there is some debate as to how much cavitation contributes to droplet disruption
(Freudig, Tesch, and Schubert 2003) and whether or not using operating conditions
that lead to cavitating is desirable from an equipment wear viewpoint (Innings et al.
2011, Phillips 1985).
Eddies
pressure
fluctuations
Drops are larger than the smallest eddies and deform by the
(a) action of hydrodynamic pressure fluctuations
shockwave
Drops are smaller than the smallest

eddies and deform under the action
of viscous stress inside and
(b) between the eddies (c) Drops are disrupted by shockwaves
FIGure 1.6 Illustration of droplet formation in turbulent flows: (a) droplet breakup in
turbulent inertial regime, (b) droplet breakup in turbulent viscous regime, and (c) droplet
breakup in cavitation.
1.6 comparIson oF emulsIFIcatIon eFFIcIency

Homogenization efficiency—the energy efficiency of a homogenizer, ΦH, can be cal-
culated by comparing the minimum amount of energy theoretically required to form
an emulsion (Emin) with the actual amount to energy that is expended per unit volume
expended during homogenization to create the given emulsions, the energy density (EV).
Emin
ΦH = 100% (1.34)
EV
The minimum theoretical energy is calculated based on the interfacial tension,

γ (J m−2), the specific surface area of an emulsion (see Equation 1.35) of a given
droplet size, d32, and the volume fraction of oil, ϕ, in the emulsion.
6φ
Emin = γ (1.35)
d32
In emulsification, as the droplets are broken into smaller ones, the interfacial area
becomes significantly larger, and the free energy of the system is increased by an
amount equal to γAS. If we consider a typical emulsion with a droplet diameter of

1 µm and an oil content of 20% (ϕ = 0.2) with an interfacial tension of 0.01 N m−1, the
change in surface free energy is no greater than 12 kJ m−3. This is a vanishingly small
amount of energy—it is equivalent to heating it a few thousandths of a degree Kelvin.
However, the actual energy required to create this emulsion using traditional high-
energy emulsification processes is at least 10 s of MJ m−3. The reason for this large
discrepancy is that there are viscous losses in the liquids, and the mechanical energy
applied is not intimately coupled to creating new interface. Specifically, in almost all
emulsification processes, both the dispersed phase droplets and continuous phase are
set into vigorous motion, where the gradients of their velocities via shear or turbulent
eddies is what act to create disruptive forces for droplet breakup during emulsifica-
tion (unless low-energy emulsification processes such as membrane emulsification
can be used). The required energy input, EV, for producing emulsions via membrane
emulsification has been reported to be in the range of 103–106 J m−3 compared to con-
ventional mechanical methods having 106 –109 J m−3 (Charcosset, Limayem, and Fessi
2004). The energy input for conventional mechanical methods is higher because only
a fraction of it is used for the droplet breakup. For example, in a high-pressure homog-
enizer, approximately 99.8% of the energy supplied is converted into heat by viscous
dissipation (Gijsbertsen-Abrahamse, Van Der Padt, and Boom 2004). As mentioned
above, the total amount of energy supplied to an emulsion during its homogenization
process is referred to as the energy density. It has been defined as the energy input per
unit volume of emulsion or the power per volumetric flow rate of emulsion (Karbstein
and Schubert 1995, Schubert and Ax 2003, Schubert, Ax, and Behrend 2003):
∫
EV = PV t dt (1.36)
where:
PV is the net power density
t is the duration of the emulsification procedure
However, PV needs to exceed some critical value for droplet disruption to take place,
which depends on the Laplace pressure of the droplet to be broken. For most of the
common types of homogenizers used in the food industry, theoretical or semiempiri-
cal equations are available to estimate the energy density (Karbstein and Schubert
1995, Schubert and Ax 2003, Stang and Schubert 2001). For example, in a high-
pressure valve homogenizer, the energy density is equal to the operating pressure,
that is, EV = PH (Stang and Schubert 2001). Alternatively, the net energy consump-
tion can be found experimentally by measuring the increase in temperature dur-
ing homogenization (as well as the newly created droplet interface) as 99.9% of the
energy lost is lost as heat (Berg and Lundh 1978) or by monitoring the electrical
power requirements of the homogenizer (Abismaïl et al. 1999). The energy density
concept is practically useful in comparing emulsification processes and machine
designs as power input into a machine and volumetric flow rate are readily accessible
operating parameters. A comparison of various emulsification processes showing
the effect of energy density on droplet size is shown in Figure 1.7, which is a basic
illustration of the energy density concept, that given equal energy densities, different
100
ϕ = 30%
Mean droplet diameter, d32 (μm)

Membrane emulsification Colloid mill
ϕ = 1% ϕ = 10% ϕ = 50%
10
HPH-v
HPH-sev
1
US
HPH-ov
MF
0.1 3
10 104 105 106 107 108
−3
Energy density, Ev (Jm )
FIGure 1.7 Comparison of various emulsifying processes based on the energy density
concept. Vegetable oil-in-water emulsions with excess surfactant. Symbols: ϕ is oil-volume
fraction, HPH refers to high-pressure homogenizers with standard valve (v), sharp-edged
valve (sev), and orifice valve (ov) designs, respectively. US refers to ultrasound; MS refers
to Microfluidizer®. (Redrawn from Schubert, H., and Ax, K., Texture in Food, Woodhead
Publishing Ltd., New York, 2003.)
emulsifying equipment produces very different droplet sizes. Most notable is the
difference between low-energy processes such as membrane emulsification and
high-energy processes such as valve homogenizers, microfluidizers, and ultrasonic
probes. Membrane emulsification has an energy density about 3 orders of magnitude
lower than that of conventional high-energy processes, with an energy efficiency
approaching the theoretical limit. Furthermore, as the majority of the energy sup-
plied to making droplet via membrane or microchannel emulsification processes
goes directly into the generation of interfacial area (rather than to viscous dissipa-
tion), the energy density is strongly correlated with the oil-volume fraction, ϕ, as seen
in Figure 1.7. In Table 1.3, some approximations of the maximum expected droplet
sizes as a function of EV are given for high-energy emulsification processes as well
as equivalent correlations for membrane and microchannel droplet formation under
STB and shear-induced droplet formation mechanisms.
Decreasing energy use and costs is always an objective, and several strategies
to improve the energy efficiency of homogenizers have been suggested, including
(McClements 2005, Walstra and Smulders 1998)
• Increasing the power intensity and reducing the processing time.

• Increasing the dispersed volume fraction during homogenization, subse-
quently diluting it to the desired concentration. In high-pressure homog-
enizers, energy loss due to friction is proportional to the overall volume
being processed.
• Increasing the emulsifier concentration; this further decreases the interfa-
cial tension and faster adsorption, thereby helping droplet breakup and pre-
venting droplet coalescence.
• Combining techniques that are most efficient for a given size range, for
example, a high-speed blender in combination with a high-pressure
homogenizer.
1.7 complIcatIons and concludInG remarks

As mentioned in the beginning of the chapter, the interfacial tension has been
assumed to be constant. This is, of course, a major simplification. Furthermore, the
effects of droplet collisions may lead to coalescence if sufficient levels of emulsi-
fier are not present. Here, we have been working under the premises that either we
have a very dilute system (low ϕ) or that the time of adsorption is much smaller
than the collision time, that is, τADS /τCOL  1 (see Table 1.2 and equations within).
In general, coalescence should decrease with decreasing volume fraction, increasing
droplet diameter, decreasing surface excess concentrations, and increasing emul-
sifier concentrations. Adsorption kinetics and collision rates also play a key role
(Håkansson, Trägårdh, and Bergenståhl 2009, Karbstein and Schubert 1995, Stang,
Karbstein, and Schubert 1994). Other complications not considered include non-
Newtonian fluids or any sort of interfacial rheology or the types and characteristics
of the surfactant and emulsifiers used in stabilizing emulsions. This is to be found in
Chapters 2, 3, and 4.
As a general comment to this chapter, much of the underlying theory presented
has been developed more than 30 years ago, and a large proportion of it has been
tested empirically since then. However, continuing advances in computing capac-
ity had made way for a new generation of numerical studies possible through CFD.
This will enable one to study and design more efficient homogenization processes
in the future. Not only to optimize for generating the highest energy intensities for
droplet disruption but also to take into consideration subsequent individual droplet
events such as the competitive processes of disruption, collision, and coalescence.
Indeed, given the extremely low-energy efficiency of homogenization processes in
the food industry, which also rely on large production volumes, there will be an
added incentive to understand, model, predict, and design the most efficient emul-
sification processes possible. Furthermore, process conditions also have a signifi-
cant impact on the bulk physiochemical properties of the resulting products, and
more gently processes can also be desirable. This could be fulfilled in part by the
further development of low-energy emulsification processes such as membrane and
microchannel emulsification. Up until recently these processes have had two major
drawbacks: low production volumes and cost. However, recent advances in micro-
fabrication and nanofabrication technologies have opened up many new possibilities
with respect to design and manufacturing microengineered structures for generating
emulsions. Examples include high-capacity microsieves that are easier to clean with
more robust materials, cost-efficient membranes, and higher degrees of paralleliza-
tion for greatly increased production rates (Brans et al. 2006, Maan, Boom, and
Schroën 2013, Matos et al. 2013, Nazir, Schroën, and Boom 2011). Only time will
tell if these advances will enable membrane and microchannel emulsification pro-
cesses to be applied on a wider industrial scale.
reFerences
Abismaïl, B., J.P. Canselier, A.M. Wilhelm, H. Delmas, and C. Gourdon. 1999. “Emulsification
by ultrasound: Drop size distribution and stability.” Ultrasonics Sonochemistry 6
(1–2):75–83.
Abrahamse, A.J., A. van der Padt, R.M. Boom, and W.B.C. de Heij. 2001. “Process fundamentals
of membrane emulsification: Simulation with CFD.” AIChE Journal 47 (6):1285–1291.
doi:10.1002/aic.690470606.
Berg, E.T., and G. Lundh. 1978. “Functional characterization of protein stabilized emulsions:
Standardized emulsifying procedure.” Journal of Food Science 43 (5):1553–1558. doi:
10.1111/j.1365-2621.1978.tb02541.x.
Brans, G., R.G.M. van der Sman, C.G.P.H. Schroën, A. van der Padt, and R.M. Boom. 2006.
“Optimization of the membrane and pore design for micro-machined membranes.”
Journal of Membrane Science 278 (1–2):239–250.
Charcosset, C., I. Limayem, and H. Fessi. 2004. “The membrane emulsification process—A
review.” Journal of Chemical Technology and Biotechnology 79 (3):209–218.
Doran, P.M. 1995. “7—Fluid flow and mixing.” In Bioprocess Engineering Principles, edited
by P.M. Doran, 129–163. London: Academic Press.
Freudig, B., S. Tesch, and H. Schubert. 2003. “Production of emulsions in high-pres-
sure homogenizers—Part II: Influence of cavitation on droplet breakup.” Chemical
Engineering and Technology 26 (6):266–270.
Gijsbertsen-Abrahamse, A.J., A. Van Der Padt, and R.M. Boom. 2004. “Status of cross-flow
membrane emulsification and outlook for industrial application.” Journal of Membrane
Science 230 (1–2):149–159.
Håkansson, A., C. Trägårdh, and B. Bergenståhl. 2009. “Studying the effects of adsorption,
recoalescence and fragmentation in a high pressure homogenizer using a dynamic simu-
lation model.” Food Hydrocolloids 23 (4):1177–1183.
Hinze, J.O. 1955. “Fundamentals of the hydrodynamic mechanism of splitting in dispersion
processes.” AIChE Journal 1 (3):289–295. doi:10.1002/aic.690010303.
Innings, F., E. Hultman, F. Forsberg, and B. Prakash. 2011. “Understanding and analysis of
wear in homogenizers for processing liquid food.” Wear 271 (9–10):2588–2598.
Karbstein, H., and H. Schubert. 1995. “Developments in the continuous mechanical pro-
duction of oil-in-water macro-emulsions.” Chemical Engineering and Processing 34
(3):205–211. doi:10.1016/0255-2701(94)04005-2.
Kobayashi, I., S. Mukataka, and M. Nakajima. 2004. “Effect of slot aspect ratio on droplet for-
mation from silicon straight-through microchannels.” Journal of Colloid and Interface
Science 279 (1):277–280.
Kobayashi, I., and M. Nakajima. 2002. “Effect of emulsifiers on the preparation of food-grade
oil-in-water emulsions using a straight-through extrusion filter.” European Journal of
Lipid Science and Technology 104 (11):720–727. doi:10.1002/1438-9312(200211)104:
11<720::aid-ejlt720>3.0.co;2-e.
Kobayashi, I., M. Nakajima, K. Chun, Y. Kikuchi, and H. Fukita. 2002. “Silicon array of elonga-
ted through-holes for monodisperse emulsion droplets.” AIChE Journal 48 (8):1639–1644.
doi:10.1002/aic.690480807.
Kobayashi, I., K. Uemura, and M. Nakajima. 2006. “CFD study of the effect of a fluid
flow in a channel on generation of oil-in-water emulsion droplets in straight-
through microchannel emulsification.” Journal of Chemical Engineering of Japan
39 (8):855–863.
Kobayashi, I., G.T. Vladisavljević, K. Uemura, and M. Nakajima. 2011. “CFD analysis of micro-
channel emulsification: Droplet generation process and size effect of asymmetric straight
flow-through microchannels.” Chemical Engineering Science 66 (22):5556–5565.
Kobayashi, I., M. Yasuno, S. Iwamoto, A. Shono, K. Satoh, and M. Nakajima. 2002. “Microscopic
observation of emulsion droplet formation from a polycarbonate membrane.” Colloids
and Surfaces A: Physicochemical and Engineering Aspects 207 (1–3):185–196.
Kolmogorov, A.N. 1949. “On the breakage of drops in a turbulent flow.” Mathematics and
Mechanics 1:339–343.
Lee, L.L., N. Niknafs, R.D. Hancocks, and I.T. Norton. 2013. “Emulsification: Mechanistic
understanding.” Trends in Food Science and Technology 31 (1):72–78. doi:10.1016/j.
tifs.2012.08.006.
Lee, L., and I.T. Norton. 2013. “Comparing droplet breakup for a high-pressure valve
homogeniser and a Microfluidizer for the potential production of food-grade nanoemul-
sions.” Journal of Food Engineering 114 (2):158–163.
Maan, A.A., R. Boom, and K. Schroën. 2013. “Preparation of monodispersed oil-in-water
emulsions through semi-metal microfluidic EDGE systems.” Microfluidics and
Nanofluidics 14 (5):775–784.
Matos, M., M.A. Suárez, G. Gutiérrez, J. Coca, and C. Pazos. 2013. “Emulsification with
microfiltration ceramic membranes: A different approach to droplet formation mecha-
nism.” Journal of Membrane Science 444:345–358.
McClements, D.J. 2005. Food Emulsions: Principles, Practices, and Techniques, edited by
F.M. Clydesdale, Chapter 6, pp. 233–268. CRC Press Series in Contemporary Food
Science. Boca Raton, FL: CRC Press.
McClements, D.J. 2007. “Critical review of techniques and methodologies for characterization
of emulsion stability.” Critical Reviews in Food Science and Nutrition 47 (7):611–649.
Nakashima, T., M. Shimizu, and M. Kukizaki. 1991. “Membrane emulsification by micropo-
rous glass.” Key Engineering Materials 61–62:513–516.
Nazir, A., K. Schroën, and R. Boom. 2011. “High-throughput premix membrane emulsifica-
tion using nickel sieves having straight-through pores.” Journal of Membrane Science
383 (1–2):116–123.
Peng, S.J., and R.A. Williams. 1998. “Controlled production of emulsions using a cross-
flow membrane. Part I: Droplet formation from a single pore.” Chemical Engineering
Research and Design 76 (A8):894–901. doi:10.1205/026387698525694.
Phillips, L.W. 1985. The High Pressure Dairy Homogenizer. Vol. 6, NIRD Technical Bulletins.
28pp. Reading, England: The National Institute in Dairying.
Rayner, M. 2005. “Membrane emulsification: Modelling interfacial and geometric effects
on droplet size.” PhD Dissertation, Department of Food Technology, Engineering and
Nutrition, Faculty of Engineering, Lund University, Sweden.
Rayner, M., and G. Trägårdh. 2002. “Membrane emulsification modelling: How can we
get from characterisation to design?” Desalination 145 (1–3):165–172. doi:10.1016/
s0011-9164(02)00403-4.
Rayner, M., G. Trägårdh, and C. Trägårdh. 2005. “The impact of mass transfer and interfacial
expansion rate on droplet size in membrane emulsification processes.” Colloids and
Surfaces A: Physicochemical and Engineering Aspects 266 (1–3):1–17.
Schroder, V., and H. Schubert. 1999. “Production of emulsions using microporous, ceramic
membranes.” Colloids and Surfaces A: Physicochemical and Engineering Aspects 152
(1–2):103–109. doi:10.1016/s0927-7757(98)00688-8.
Schubert, H., and K. Ax. 2003. “Engineering food emulsions.” In Texture in Food, edited by
D.M. McKenna. New York: Woodhead Publishing Ltd.
Schubert, H., K. Ax, and O. Behrend. 2003. “Product engineering of dispersed systems.” Trends
in Food Science and Technology 14 (1–2):9–16. doi:10.1016/s0924-2244(02)00245-5.
Schultz, S., G. Wagner, K. Urban, and J. Ulrich. 2004. “High-pressure homogenization as a
process for emulsion formation.” Chemical Engineering and Technology 27 (4):361–368.
doi:10.1002/ceat.200406111.
Stang, M., H. Karbstein, and H. Schubert. 1994. “Adsorption-kinetics of emulsifiers at oil-water

interfaces and their effect on mechanical emulsification.” Chemical Engineering and
Processing 33 (5):307–311. doi:10.1016/0255-2701(94)02000-0.
Stang, M., and H. Schubert. 2001. “Emulsification in high-pressure homogenizers.” Chemical
Engineering and Technology 24 (10):151–157.
Sugiura, S., M. Nakajima, S. Iwamoto, and M. Seki. 2001. “Interfacial tension driven
monodispersed droplet formation from microfabricated channel array.” Langmuir 17
(18):5562–5566.
Sugiura, S., M. Nakajima, and M. Seki. 2002. “Prediction of droplet diameter for microchan-
nel emulsification.” Langmuir 18 (10):3854–3859.
Taylor, G.I. 1934. “The formation of emulsions in definable fields of flow.” Proceedings of
the Royal Society of London: Series A 146 (858):501–523. doi:10.1098/rspa.1934.0169.
Timgren, A., G. Trägårdh, and C. Trägårdh. 2009. “Effects of cross-flow velocity, capillary
pressure and oil viscosity on oil-in-water drop formation from a capillary.” Chemical
Engineering Science 64 (6):1111–1118.
Timgren, A., G. Trägårdh, and C. Trägårdh. 2010. “A model for drop size prediction during
cross-flow emulsification.” Chemical Engineering Research and Design 88 (2):229–238.
van der Graaf, S., C.G.P.H. Schroën, R.G.M. van der Sman, and R.M. Boom. 2004. “Influence
of dynamic interfacial tension on droplet formation during membrane emulsifica-
tion.” Journal of Colloid and Interface Science 277 (2):456–463. doi:http://dx.doi.org/
10.1016/j.jcis.2004.04.033.
Vankova, N., S. Tcholakova, N.D. Denkov, I.B. Ivanov, V.D. Vulchev, and T. Danner. 2007.
“Emulsification in turbulent flow. 1. Mean and maximum drop diameters in inertial and
viscous regimes.” Journal of Colloid and Interface Science 312 (2):363–380.
Walstra, P. 1993. “Principles of emulsion formation.” Chemical Engineering Science 48
(2):333–349. doi:10.1016/0009-2509(93)80021-h.
Walstra, P. 2003. Physical Chemistry of Foods. New York: CRC Press.
Walstra, P. 2005. “8 Emulsions.” In Fundamentals of Interface and Colloid Science, edited by
J. Lyklema, 1–94. Amsterdam, the Netherlands: Elsevier Academic Press.
Walstra, P., and P.E.A. Smulders. 1998. “Emulsion formation.” In Modern Aspects of Emulsion
Science, edited by B.P. Binks, 56–99, Chapter 2. Cambridge: The Royal Society of Chemistry.
Windhab, E.J., M. Dressler, K. Feigl, P. Fischer, and D. Megias-Alguacil. 2005. “Emulsion
processing—From single-drop deformation to design of complex processes and prod-
ucts.” Chemical Engineering Science 60 (8–9 SPEC. ISS.):2101–2113.
Yasuno, M., M. Nakajima, S. Iwamoto, T. Maruyama, S. Sugiura, I. Kobayashi, A. Shono, and
K. Satoh. 2002. “Visualization and characterization of SPG membrane emulsification.”
Journal of Membrane Science 210 (1):29–37.
2 Emulsion Formation
and Instability
Björn Bergenståhl
contents
2.1 Introduction .................................................................................................... 33
2.2 Interfaces ........................................................................................................34
2.3 Stability and Instability of Droplets ............................................................... 35
2.4 Creaming/Sedimentation ................................................................................ 36
2.5 Flocculation ....................................................................................................40
2.6 Coalescence .................................................................................................... 41
2.7 Ostwald Ripening ........................................................................................... 43
2.8 Surface Interactions in Emulsion Systems .....................................................44
2.8.1 Van der Waals Interactions .................................................................44
2.8.2 Solvation Interactions .........................................................................44
2.8.3 Electrostatic Repulsion .......................................................................44
2.8.4 Polymer-Induced Interactions ............................................................. 47
2.8.5 Bridges due to a Third Phase .............................................................. 48
References ................................................................................................................ 48
ABSTRACT The general conditions for emulsification and droplet formation

are discussed. The basic theory for droplet aggregation and emulsion destabiliza-
tion is reviewed. The role of surface interactions for controlling the instability is
introduced.
2.1 IntroductIon
Emulsions are dispersions of two liquid phases within each other: a more polar phase,
usually termed water phase, as it typically is an aqueous solution, and an oil phase,
as it typically consists of a more or less nonpolar liquid with a low solubility in the
water phase. One of the phases is continuous whereas the other is dispersed. The
structures are commonly named oil in water or water in oil, depending on whether
we have oil droplets in water or water droplets in oil.
Emulsions are, by their nature, unstable and are defined as consisting of at least
two phases. Technically, we want to be able to control the stability, or more cor-
rectly, the kinetics of the destabilization, when the emulsion is formed. The emul-
sion processing and composition is aimed to create an emulsion with a suitable
stability.
33
2.2 Interfaces
All liquids display cohesiveness through intermolecular interactions. An interface
between two phases displays the step change in composition as well as in the inten-
sity of the intermolecular interactions. An interfacial tension is obtained that reflects
the difference. The interfacial tension is the energy cost of creating the surface. The
interface between two pure liquids appears liquid.
Various molecular species present at the interface may moderate the interaction
contrast and thereby reduce the interfacial tension. The reduction of surface tension
shows the attraction of the interface toward the molecules, their surface activity. The
surface activity leads to a flow of surface active molecules toward the interface until
it becomes saturated. Molecules present at the interface generate a pressure along
the surface. If the concentration is uneven, there are pressure gradients that may
lead to flow along the surface. If the surface displays a significant surface pressure,
it appears more or less as a solid. All together, these basic features create a highly
dynamic situation when new surfaces are formed during an emulsification event.
An efficient emulsification is achieved if we have unstable interfaces and obtain
stable droplets, whereas stable interfaces and unstable droplets make emulsification
difficult.
The stability respective instability, of interfaces respectively droplets, is deter-
mined both by static forces as well as by dynamic forces, as illustrated in Figure 2.1.
Dynamic effects can be summarized on the basis of Gibbs–Marangoni effect,
whereas static effects can be summarized on the basis of interfacial tension and
surface forces (Walstra 1993, 2003). Flow across and along the surfaces contributes
strongly to the emulsification and interfacial instability if the interfacial tension is
low. The flow to newly formed interfaces is obviously essential and critical for the
emulsification efficiency (Maldonado-Valderrama et al. 2008).
Spontaneous emulsification can be obtained due to the dynamic effects in systems
with very low interfacial tension, with low viscosity, and with large flow across the
Interfaces Droplets
Surface tension Surface forces

Static
Diffusion to and
across interfaces. Flux Flux along
along interfaces interfaces
Dynamic
fIGure 2.1 Interactions destabilizing interfaces and stabilizing droplets.

Emulsion Formation and Instability 35
interface (Lopez-Montilla et al. 2002). Examples of this type of emulsification can

be found widely in the industry. It is commonly employed in plant protection for-
mulations that are emulsified at the farm just before the distribution event. It is also
commonly utilized in cleaning formulations. Spontaneous emulsification leads to a
normal two-phase emulsion (oil and water phases) that displays a habitual range of
thermodynamic instability. Microemulsions, which are also spontaneously formed,
are single-phase systems (with oil and water domains) and represent a thermody-
namic equilibrium condition.
2.3 stabIlIty and InstabIlIty of droplets

The thermodynamic instability of emulsions is, fundamentally, caused by the inter-
facial tension. The interfacial tension ranges from about 40 mN/m for hydrocarbons
in water to lower values, down to about 1 mN/m in systems with high amounts of
surfactant and/or high amount of water or oil in both phases.
Emulsions represent large interfacial areas as a consequence of the small particle
size.
6φ
A= (2.1)
d
where:
A is interfacial area (m2/m3)
ϕ is volume fraction dispersed phase
d is droplet diameter (m)
Despite the large area, typically 600 m2/liter (in an emulsion of 10% oil and 1 µm in
droplet diameter), the total interfacial energy is just 6 J/liter (assuming an interfacial
tension of 10 mN/m)—indeed is a quite small energy. By experience, we may also
compare emulsions with comparable, large interfacial tensions with emulsions with
comparable, low interfacial tensions; it can be concluded that the interfacial tensions
do not reflect the instability, although it is a major driving force for the instability.
Instability is caused by different instability mechanisms, which describe the loss
of the dispersed state by overcoming the threshold energies that keep the emul-
sions stable. Typically, we distinguish between creaming/sedimentation, floccula-
tion, coalescence, and Ostwald ripening (Figure 2.2) (McClements 2004a, Walstra
1993, 2003).
Creaming/sedimentation. This refers to separation due to gravity. Most com-

monly, the oil phase is lighter than the water phase and the separation goes
upward, that is, creaming. Creaming/sedimentation leads to a change in
concentration in space. The change in concentration may change the rate of
other destabilizing mechanisms.
Flocculation. Collisions between particles may lead them to aggregation if the
contact between the droplets is adhesive. Such aggregation is termed floc-
culation as long as the discrete droplets remain unaltered.
Coalescence
0.2
ϕ
Flocculation Creaming
1 10
Diameter (μm)
fIGure 2.2 Volume fractions and particle sizes, where different instability processes are
of particular importance.
Coalescence. This refers to fusion between oil droplets. It may happen as a

consequence of collision events or as a consequence of long time contact in
the flocculate state or in a tightly packed cream layer. The coalescence rate
is influenced by the presence of additional phases.
Ostwald ripening. Differences in the internal pressure between smaller and
larger droplets may lead to diffusive transport of the dispersed phase
from smaller droplets to the larger droplets. This process is essential
because the dispersed phase has significant solubility in the continuous
phase.
All the mechanisms occur in parallel in all emulsion systems. However, various factors
influence the rate and thereby determine the mechanisms that are most critical in a
particular system. Most decisive are particle size and volume fraction, as indicated in
Figure 2.2.
2.4 creamInG/sedImentatIon
Creaming is an obvious source of instability, as anybody who has observed the rapid
separation between vinegar and oil when a salad dressing is mixed would confirm.
However, when considering the large variability of technical emulsion, we can see
that there are systems where gravity-induced separation proceeds very readily and
there are systems that remain stable for a long time. Many factors influence the rate
(McClements 2004a). Here, we will look into droplet size, concentration, rheology,
and aggregation state.
Particle size. The role of particle size is clearly expressed in the Stokes law for
settling particles.
g ⋅ d 2 ⋅ ∆ρ
vStokes = (2.2)
18 ⋅ η
where:
vStokes is the settling or creaming velocity (m/s)
g is the gravitational acceleration (m/s2)
∆ρ is the density difference (kg/m3)
η is the viscosity (Pa s)
Stokes law is valid for spherical particles. Normally, emulsion droplets

remain quite spherical in a gravity field as the stress is much smaller than
the interfacial tension. The typical range of the Stokian velocity is within
50 nm/s for a micrometer-sized droplet in water. To evaluate the role of the
settling, it has to be compared with the diffusion.
D
vdiffusion = (2.3)
2⋅t
kB ⋅ T
D= (2.4)
3⋅ π⋅η⋅ d
where:
vdiffusion is the average Brownian velocity (it is provided as the absolute
value) (m/s)
D is the diffusion constant (m2/s)
t is the time (s)
k B is the Boltzmann’s constant (J/K)
T is the absolute temperature (K)
The diffusion constant for a micrometer-sized emulsion droplet is about

4·10 –13 m2/s, which gives a Brownian velocity of 400 nm/s (at a second
scale). This is much higher than the Stokian velocity. However, over longer
timescales, the sedimentation wins as the Brownian velocity drops with
increasing timescale. In our example, the timescales will be about equal
over a timescale of minutes and, clearly, the creaming will dominate at a
timescale of hours. However, as the Stokes velocity is proportional to d2
and the Brownian velocity to d−1, the ratio between them is proportional
to d3. Thus, a smaller droplet diameter makes the actual timescale for the
creaming to win over the diffusion much longer; at sizes below 0.5 µm, most
emulsions can be assumed to be stable against creaming.
Concentration. One assumption in Stokes law is that the movement of one
particle is independent of that of other particles. However, in an emulsion,
particles are not alone; with micrometer-sized droplets, the number concen-
tration is about 1014 droplets/liter. The average distance between the par-
ticles can be estimated assuming cubic close packing to
  π  13

δ = d 1 −    (2.5)
  φ ⋅ 6  
where:
δ is the distance between the emulsion droplets (surface to surface) (m)
The distance clearly scales against diameter. It also becomes quite small
when the volume fraction approaches 0.5. The interparticle distance in a
typical emulsion with a volume fraction of about 0.1 and a droplet size of
1 μm is about 1 µm. The particles in a typical emulsion are polydisperse and
different particles display different Stokes velocities.
If a particle moves with a creaming velocity of 50 nm/s, it may encounter
another particle within 20 s (if the next particle is significantly smaller).
The frequent particle encounters reduce the average creaming compared
to the prediction according to Stokes law significantly. The phenomenon
is usually referred to as hindered sedimentation. Hunter (1986) made
a prediction of the effect as a function of volume fraction, as showed in
Figure 2.3. The hindered creaming/settling explains the apparent stability
that we observe for various concentrated practical systems.
Rheology. The flow properties of the continuous phase are described by vis-
cosity in the Stokes equation. Therefore, there is an underlying assumption
that the viscosity is Newtonian. In the presence of polymers in the continu-
ous phase, this assumption may be valid (McClements 2004a). For non-
Newtonian fluids, the viscosity is an apparent property that depends on the
shear rate. We may assume that shear is developed over a distance comparable
to the radius of the particle. Thus, the shear rate in our example emulsion,
rising with the velocity of about 50 nm/s, is about 0.1/s. However, the
Brownian motion of the same particle in this timescale is about 10 times
higher and results in a shear rate of 1/s. For larger particles, the shear field
created by the sedimentation will be higher, and more important. But the
present field will never be smaller than about 1–0.1/s. In addition, we may
have non-Newtonian liquids that display yielding properties (if the mol-
ecules in the continuous phase have developed a network that demands a
certain stress to flow). In that case, the yield value needs to be stronger than
the stress generated by the rising force on the particle, which, for our exam-
ple particle, is about 1 mPa. For larger particles, the stress by the rising
movements increases approximately proportionally to the diameter. Thus,
v/vStokes
0.5
0 0.1 0.2 0.3 0.4 0.5

ϕ
fIGure 2.3 The creaming rate of emulsions as a function of volume fraction dispersed
phase. The figure is based on the semiempirical equation by Hunter using the constants
by Chanamai and McClements. (Data from Hunter, R.J., Foundations of Colloid Science,
Oxford University Press, Oxford, 1986; Chanamai, R. and McClements, D.J., Colloids and
Surfaces A: Physicochemical and Engineering Aspects 172, 79–86, 2000.)
a yield stress around 0.01 Pa can be expected to provide sufficient stabiliz-

ing power to prevent creaming of most emulsions.
Aggregation. In the Stokes law, the droplets are assumed to move freely. Freely
moving particles mean that the emulsion appears like a liquid and that drops
in one position of the container may diffuse to another position. The opposite
refers to the movements of the droplets being restricted to a limited volume.
Usually, this is termed an arrested state and the emulsion can be viewed
as a solid (with very low droplet diffusion). The state can be considered
glasslike if there is no long-range order and, hence, we may term the state
as a colloidal glass (Figure 2.4) (Dawson 2002). A colloidal glass typically
displays plastic shear properties with a clear yielding character. Examples of
such structures are mayonnaise emulsions. There are two principal types of
conditions that may lead to the formation of a colloidal glass: repulsive inter-
action between the particles and attractive interactions between the particles.
Repulsive interactions lead to a liquid–glass transition when the interparticle
distance becomes comparable with the distance of the repulsive interactions,
thereby allowing the surrounding droplets to form a cage by restricting fur-
ther movements. Typically, this occurs close to a volume fraction between
0.5 and 0.7. Attractive interactions lead to a liquid–glass transition when the
droplets are entrapped in loose aggregated structures that completely fill the
volume of the emulsion. Attractive glasses can be formed down to quite low
volume fractions, possibly as low as about 0.1.
Interaction
Repulsive
0.1 μm 10 μm colloidal
Repulsive glass
Colloidal
liquid
Colloidal
liquid
Attractive Attractive
colloidal
glass
Volume fraction
fIGure 2.4 Principal phase diagram showing when colloidal liquids and colloidal glasses
are formed as a function of volume fraction dispersed phase and repulsive attractive interac-
tions, respectively. The effects of particle size are outlined.
2.5 flocculatIon
Flocculation is caused by collision events between moving droplets. The collision
may be caused by different sources of droplet movement: Brownian, shear-, and
gravity-induced collisions. The different collision mechanisms are compared in
Figure 2.5.
Brownian flocculation. Brownian diffusion provides a source of random drop-

let movements that may lead to collisions if the movements are compa-
rable with the size of the particle (McClements 2004a). From the previous
estimations of the mobility of our standard emulsion, we can observe that
it takes a few seconds for a droplet to encounter another droplet due to
Brownian motion in an emulsion with micrometer-sized droplets. Thus, the
collision frequency will be quite high in such a system, and if every colli-
sion leads to aggregation, the system will aggregate within a minute or less.
When the particle becomes larger, the Brownian movements are slower
and the distances become longer, leading to much longer timescales. As
a consequence, this destabilization mechanism only appears in a finely
dispersed system below, maybe, 3 µm. Classically, the aggregation rate is
described as a reaction between two droplets giving a rate proportional to n2
(n, the number concentration). However, the number concentration is not an
obvious property. It depends very strongly on size and becomes large when
the size is small. Thus, the function is strongly size dependent. This strong
size dependence is illustrated in Figure 2.5 with the change in the diameter
as a function of time.
Shear-induced flocculation. The exposure of emulsions to a shear field also
leads to collisions between droplets. The collision frequency depends
Time
10 years
1 year
Gravity
1 month
1 week
1 day
Brownian
1 hour
1 minute
Shear
1s
0.1 0.2 0.5 1 2 5 10 20 50 100
Droplet size in μm
fIGure 2.5 The kinetics of different aggregation mechanisms in emulsions. The diagram
shows the timescale of significant instability (in terms of the timescale for a doubling of
the diameter when assuming immediate droplet fusion after a collision event) as a function
of particle size (droplet diameter). The timescale of instability is shown for Brownian floc-
culation, shear-induced flocculation, and gravity-induced flocculation. Assumptions: Every
encounter leads to coalescence (w = 1), volume fraction 0.1, aqueous environment at 25°C,
10% density difference, a shear field of 10 (s−1).
strongly on the particle size and becomes more rapid as the size increases.
The kinetics of the destabilization process is illustrated in Figure 2.5, similar
to the processes illustrated for Brownian flocculation. It is clear that shear-
induced aggregation is different from Brownian and may lead to the forma-
tion of very large objects if emulsion starts to destabilize.
Gravity-induced aggregation. The creaming-induced movements of droplets
may also lead to collisions (McClements 2004a, Melik and Fogler 1988).
For small droplets, such as the micrometer-sized droplet in our example,
it may take about 20 s to reach the next droplet, whereas the Brownian
flocculation is much faster. However, when the particles are larger, gravity-
induced aggregation becomes more important. For 10 µm-sized particles,
the creaming time to reach an encounter is lower than the time for Brownian
motions leading to the same event.
The outcome of collision events depends on the strength of the collision
energy relative to the interactive forces between the particles. Thus, we will
discuss interparticle interactions between emulsion droplets in Section 2.8.
2.6 coalescence
Coalescence is the process when two separate droplets fuse to form one new large
droplet. This process can be described as being dependent on the stability of the film
separating two adjacent droplets (Kabalnov 1998, McClements 2004a, Walstra 1993).
Adjacent droplets could be two droplets flocculated together forming a doublet.

It could also be droplets in the cream layer being in permanent contact with each
other, or it could be a concentrated emulsion where the discontinuous phase has a
high volume fraction (above about 0.6–0.7). We may also observe coalescence as an
immediate consequence of a collision event.
The film represents a thermodynamically quite unstable system. Despite this,
there is no uniform theory describing the possible events that may lead to coales-
cence, and maybe this reflects the nonuniformity in the nature of these processes
(Chan et al. 2011, Kabalnov 1998). However, a few factors of relevance for emul-
sion coalescence and how they influence the film stability have been identified.
These factors are particle size, interfacial tension, surface interactions, interfa-
cial viscosity, emulsifier solubility, phase transitions, and presence of solid
particles.
Droplet size. Smaller droplets allow for a larger contact surface when counted
on the complete emulsion (Walstra 1993). If we expect that the probability
for coalescence to be proportional to random disturbances, it may be pro-
portional to the contact area, thereby becoming a more important desta-
bilizing mechanism for the finer dispersed emulsions. On the other hand,
smaller droplets have a higher internal pressure that may stabilize them
against disturbances.
Surface interaction. When droplets have aggregated together, the film separat-
ing them may drain. The drainage proceeds until an equilibrium thickness
has been obtained. In the Scheludko cells, a thickness of the semistable
film could be in the range 25–200 nm (Chan et al. 2011, Scheludko 1967).
The equilibrium thickness may correspond to the distance when we have
a balance between attractive and repulsive surface interactions. However,
this distance is in most systems quite short, and the actual films tend to be
thicker as a consequence of slow drainage.
Interfacial tension. A low interfacial tension allows disturbances to create
large protrusions or dents at the interface, which may lead to the formation
of holes or bridges, thereby causing a sudden collapse of the films (Chan
et al. 2011). A higher surface tension reduces disturbances and is therefore
expected to lead to more stable films. This consequence of the interfacial
tension is interesting as a very counter intuitive aspect of the emulsion
technology.
Interfacial viscosity. The molecules or particles forming an interfacial layer
usually appear more or less like a solid structure at the interface. This
solid-like character slows down the drainage, thereby ensuring slow drain-
age. A dense layer also creates surface pressure. The surface pressure will
rapidly heal sudden disturbances that may lead to the rupture of the film.
The flow of emulsifier contributes to the stabilizing action by dragging the
liquid into the film when moving along the surface. The relation between
surface rheological parameters and emulsion stability has recently been
reviewed by Pelipenko et al. (2012).
Emulsifier solubility. In the film, the emulsifier is exposed to an external

pressure, which may remove them from the interface. This dissolution hap-
pens quite readily if soluble in the dispersed phase. This effect is suggested
to be an important factor behind solubility rules such as the Bancroft rule,
according to which the phase in which the emulsifier is most soluble will
be the continuous phase (Bancroft 1913), and the HLB rules (Davies 1957).
Phase transitions in the emulsifier layer. The film is a planar structure. Certain
emulsifiers transform into nonplanar structures (liquid crystalline phases).
Such transformations can be imagined as a consequence of the stress in the
film and may lead to film fracture (Bergenstahl and Claesson 1997).
Presence of solid particles. Particles may influence the stability of the film.
Only particles that are weakly wetted by the dispersed phase present in the
film may prevent it from thinning and form a thick and, therefore, stable
layer (see Chapter 4). Particles wetted by the dispersed phase are expected
to destabilize the emulsion (Fredrick et al. 2010). We can expect them to
protrude into the film and give rise to destabilizing bridges. Fat crystal-
lization in an emulsion may lead to the formation of hydrophobic particles
inside the oil droplets that strongly contributes to the destabilization of such
oil-in-water emulsions (Boode and Walstra 1993).
2.7 ostwald rIpenInG

Due to the interfacial tension, the emulsion droplets display an internal overpressure,
termed the Laplace pressure.
4⋅γ
Π Laplace = (2.6)
d
where:
γ is the interfacial tension between oil and water
The Laplace pressure of our example emulsion is about 0.4 bar, which is a significant
pressure. The pressure leads to a dissolving force when acting on the molecules in
the internal phase. The energy released due to the pressure is about 4 J/mol that may
increase the solubility about 0.2%. The flux from drop to drop depends on the solu-
bility, the distance between the droplets (the flux over a distance is proportional to
1/d), the Laplace pressure (the effect is proportional to 1/d), and the total area (pro-
portional to 1/d) (Kabalnov and Shchukin 1992). Thus, the rate of the process will
be proportional to 1/d3 and therefore strongly depend on the size. Very fine emulsion
may be coarsened rapidly, although the process decay when the particle size reach
somewhere around 0.5–5 µm. The process depends on solubility of the dispersed
phase. Typically, low molecular oils (hydrocarbons with a molar mass below, maybe,
200 g/mol) are exposed to this process. The presence of a high-molecular component
in the system may delay the process and therefore stabilize the emulsion (Welin-
Berger and Bergenstahl 2000).
2.8 surface InteractIons In emulsIon systems

Colloidal surface interactions are, as pointed out above, important in emulsions
(Bergenstahl and Claesson 1997, McClements 2004b). The relative importance of
the different interactions depends on the properties of the surfaces involved and the
continuous phase.
2.8.1 Van der Waals InteractIons

Dipolar and induced-dipolar interactions cause this general and fundamental
attractive interaction. The interaction is comparable weak long range (maybe
100 nm or more). The interaction is in principal proportional to the refractive
index difference between the phases. The refractive index difference between oil
and water is large and only fairly high concentrations of dissolved material in the
water phase could reduce the van der Waals interactions significantly.
2.8.2 solVatIon InteractIons

That hydrophilic surfactants may stabilize lyophobic colloids is an old obser-
vation. However, the nature of the hydration interactions created by the hydro-
philicity of the surfactants is a controversial issue. Thus, we limit ourselves
to describe the interaction as an empirically identified short-range repulsive
interaction, which is operating between hydrophilic surfaces (Bergenstahl and
Claesson 1997, McClements 2004b). The strength of the interaction is larger than
the van der Waals interaction but the range of the interaction is very short, typi-
cally 2 nm.
The interactions between surfaces are usually assumed to depend on average
properties of the liquid between the surfaces. However, at very short distances, the
size of the molecule becomes comparable with the distance separating the particles.
The interaction turns oscillating (between attraction and repulsion) at short distances
due to packing constraints of the solvent molecules.
2.8.3 electrostatIc repulsIon

Electrostatic repulsion is generally considered as the classical source of stability in
typical colloidal systems (Lyklema 2000). The repulsive interaction is caused by
charge at the interfaces, and the range of the interaction is determined by the coun-
terion concentration.
Charges in colloidal systems mainly originate from adsorption of ionic emul-
sifiers, macromolecules, or particles. Adsorption of charged surfactants (for ins-
tance, sodium dodecyl sulfonate [SDS]) leads to a charge density, slightly depending
on the packing density of the adsorbed species, but for a soap with one hydro-
carbon chain, it is about 0.3 nm2 and may provide us with a charge density of
about 3 × 1018 unit charges per square meter, which is equal to about 0.5 C/m2.
Adsorption of charged macromolecules usually gives a somewhat lower charge
density.
The counterions have a strong affinity, electrostatic as well as by Van der Waals
forces, to such a charged surface and may adsorb into a tight layer. This first tight
layer is termed the Stern layer. A majority of the charges are compensated already
in the Stern layer.
Assuming that 90% of the charges are compensated within the Stern layer, we
obtain a potential at the Stern layer of about 70 mV.
Outside the Stern layer, the accumulation of counterions only depends on the bal-
ance between electrostatic interaction and the thermal diffusion. The charge of the
surface decays by the counterions layer.
To obtain the interaction strength, the surface charge is expressed as surface
potential (ς). The relation between surface charge and surface potential at the Stern
layer (e.g., the outer Helmholtz plane, approximately equal to the experimental zeta
potential) is given by the Grahame equation, which can be approximated to:
σ
ς= (2.7)
ε r ⋅ ε0 ⋅ κ
where:
σ is the surface charge density (C/m2)
εr and ε0 are the relative and vacuum dielectric constants, respectively (C/V m)
κ is the reciprocal Debye layer thickness
The key parameter here is the Debye layer. The attraction is counteracted by entropic
mobility of the ions and the result is a diffused layer of ions, usually termed the dou-
ble layer, schematically illustrated in Figure 2.6. The concentration of counterions
− +
−+ +
−+ + + −
−
− + + + +
−
− + + +
− + −
−
Cx
ψx
Stern layer
fIGure 2.6 The counterion accumulation at a charge surface and the counterion concentration
(Cx) and surface potential (ψx) as a function of the distance (x) (counted from the Stern layer).
in the double layer decays by increasing the distance from the surface with the decay
constant κ −1, the Debye length. The Debye length is obtained from the counterion
concentration and valence:
ε r ⋅ ε0 ⋅ RT
λ D = κ −1 = (2.8)
2F2 ∑ ci ⋅ zi2
where:
R is the universal gas constant (J/mol K)
F is Faraday constant (C/mol)
ci is the concentration of counterion i (mol/m3)
zi is the valence of counterion i
The Debye length varies from about 0.7 nm for physiological salt solution (0.9%
NaCl) to 5 nm in quite soft water-like normal tap water (0.15 ppm, 3°dH or 2° Clark).
The potential as well as the counterion concentration scale according to the Debye
length are now given:
ς x = ς 0 ⋅ e − ( x / λ0 ) (2.9)
ci ( x ) = ci (∞) ⋅ e −[( F / RT )⋅( ς x x )/( λD )] (2.10)
When two planar surfaces with double layers approach each other, the double layers
of both the surfaces overlap and an osmotic repulsion is created due to the excess
concentration in the overlap region. A repulsive pressure is generated (van’t Hoff
equation):
P(h) = RT ⋅ 
 ∑c (h) − ∑c (∞)
i i (2.11)
The following expression for the repulsive pressure is obtained as the surplus con-
centration is a function of the Debye layer and the surface charge:
1 2 − ( h / λD )
P(h) = 2ε0ε r ⋅ ⋅ ς0 ⋅ e (2.12)
λ D2
By comparing the repulsive electrostatic force with the attractive van der Waals
force, assuming additivity, Dejaugin, Landau, Vervey, and Overbeek in parallel were
able to create a theory (the DLVO theory) with quantitatively predictive abilities
(Derjaguin and Landau 1941, Verwey and Overbeek 1948).
The outcome of the theory was that at high charge and low ionic strength, the repul-
sion dominates and the system remains stable. At high ionic strength, the attraction
dominates and the system becomes destabilized. At low charge and/or, at intermedi-
ate ionic strength, a repulsive barrier is created that provides an activation energy that
aggregating particles need to overcome. This activation energy leads to slower aggrega-
tion rate as only a fraction of the collisions results in aggregation. The stability factor is
the ratio between slow aggregation and fast aggregation (the aggregation rate obtained
when every collision is assumed to lead to aggregation). A typical stability factor can
be 106 or comparable numbers depending on surface charge and ionic strength.
2.8.4 polymer-Induced InteractIons

Polymers in solutions are characterized by the low level of mixing entropy, which
makes the interaction terms much more important to the solution properties. A sec-
ond consequence is that the polymers may adsorb at surfaces with a quite limited
loss of entropy. Hence, we may observe adsorption with a limited reduction in the
interfacial tension when polymers are involved.
Polymers may create various interactions depending on the solubility properties
(Table 2.1). The interactions between the surfaces can be described as based on the
density of monomers close to the surface.
Interactions between surfaces due to nonadsorbed polymers. A surface

immersed in a solution of polymers with good solvent interactions is due
to volume exclusion effects creating a thin layer depleted of monomer units
close to the surface (Bibette et al. 1990, Claesson et al. 2001). If two such
layers overlap, the concentration of monomer units close to the surface will
be further lowered and an attractive osmotic interaction will be obtained.
The attractive force, depletion attraction, may lead to an aggregation
between the particles. Such aggregation is commonly observed in colloidal
systems with well-solubilized macromolecules.
The range of the interaction is comparable to the radius of gyration.
Hence, the depletion effect is reduced if the molecular dimensions are
small. McClements has, for instance, shown that the critical concentration
of added polymer to induce depletion flocculation strongly depends on the
molar mass of added maltodextrin (from 30% for 500 Da to 7% for 1800 Da).
(McClements 2000). Similarly, the effects are reduced if the polymers are
very polydisperse.
Interactions between adsorbed polymers fully covering the surface. A surface
is usually rapidly covered by an adsorbed layer of polymers when immersed
into a solution of polymers with less good solvent interactions or high sur-
face affinity. The surface activity of polymers is high even if the solubility
is decently good due to limited entropy of mixing in polymer systems.
table 2.1
surface Interactions caused by the presence of polymers
type of Interaction polymer surface Interaction other conditions
Bridging Adsorbing Low surface coverage
Steric repulsion Adsorbing High surface coverage and good solvent
conditions
Depletion attraction Nonadsorbing Intermediate polymer concentration
An adsorbed polymer layer can be described as a brush. The density of

the layer can be described by the concentration of monomers as a function
of distance from the surface. If two layers of adsorbed polymers overlap,
the concentration of monomers will be altered and an osmotic repulsive
effect is obtained according to the van’t Hoff equation (Claesson et al. 2001,
McClements 2004b).
The range of the interaction is comparable to the thickness of the
adsorbed layer, typically in the size range of the radius of gyration of the
adsorbed molecule. The repulsive interaction is only obtained if the poly-
mers are irreversibly adsorbed, well soluble, and completely covering the
interface.
Interactions between adsorbed polymers partially covering the surface. The sur-
face layer of adsorbed polymers includes empty patches, if we have insuffi-
cient polymer to fully cover the surface. Individual adsorbed molecules may
bridge between the surfaces when two layers with empty patches are pulled
together. The result is an attractive bridging force between the surfaces.
The range of the interaction is comparable to the size range of the
molecules. The attractive interaction is only obtained if the polymers are
partially covering the interface.
2.8.5 BrIdges due to a thIrd phase

Precipitating material between dispersed particles may cause material bridges
(Butt and Kappl 2009). The conditions include that the precipitating material wet
the particles (contact angle less than 90°) and that there is a surface tension between
the precipitating phase and the surrounding liquid. Bridge formation can be caused
by surplus emulsifiers being present as a third phase or phase segregating polymers
in the aqueous phase. It can be observed that a surplus of emulsifier may destabilize
dispersions. The explanation is that the precipitation of the liquid crystalline emulsi-
fier may create bridges between particles (Richardsson et al. 2004).
references
Bancroft, W.D. 1913. The theory of emulsification. V. Journal of Physical Chemistry 17,
501–519.
Bergenstahl, B. and Claesson, P. 1997. Surface forces in food emulsions, in Food Emulsions.
Friberg S. and Larsson K. (eds.) Marcel Dekker, New York, 57–110.
Bibette, J., Roux, D., and Nallet, F. 1990. Depletion interactions and fluid-solid equilibrium in
emulsions. Physical Review Letters 65, 2470.
Boode, K. and Walstra, P. 1993. Partial coalescence in oil water emulsions. 1. Nature of the aggre-
gation. Colloids and Surfaces A: Physicochemical and Engineering Aspects 81, 121–137.
Butt, H.J. and Kappl, M. 2009. Normal capillary forces. Advances in Colloid and Interface
Science 146, 48–60.
Chan, D., Klaseboer, E., and Manica, R. 2011. Film drainage and coalescence between
deformable drops and bubbles. Soft Matter 7, 2235–2264.
Chanamai, R. and McClements, D.J. 2000. Dependence of creaming and rheology of monodisperse
oil-in-water emulsions on droplet size and concentration. Colloids and Surfaces A:
Physicochemical and Engineering Aspects 172, 79–86.
Claesson, P.M., Blomberg, E., and Poptoshev, E. 2001. Surface forces and emulsion stability,
in Encyclopedic Handbook of Emulsion Technology. Sjoblom J. (ed.) Marcel Dekker,
New York, Chapter 13, pp. 305–327.
Davies, J.T. 1957. A quantitative kinetic theory of emulsion type, I. Physical chemistry of
the emulsifying agent, in Proceedings of the 2nd Congress on Surface Activity, Vol. 1,
Butterworths, London, 426–438.
Dawson, K.E. 2002. The glass paradigm for colloidal glasses, gels, and other arrested states driven
by attractive interactions. Current Opinion in Colloid and Interface Science 7, 218–227.
Derjaguin, B. and Landau, L. 1941. Theory of the stability of strongly charged lyophobic sols and
of adhesion of strongly charged particles in solutions of electrolytes. Acta Physicochimica
URSS 14, 663; Reprinted in: Progress in Surface Science 1993, 43, 30–59.
Fredrick, E., Walstra, P., and Dewettinck, K. 2010. Factors governing partial coalescence in
oil-in-water emulsions. Advances in Colloid and Interface Science 153, 30–42.
Hunter, R.J. 1986. Foundations of Colloid Science, Vol. 1, Oxford University Press, Oxford.
Kabalnov, A.S. 1998. Coalescence in emulsions, in Modern Aspects of Emulsion Science.
Binks B.P. (ed.). The Royal Society of Chemistry, Cambridge, Chapter 7, pp. 205–257.
Kabalnov, A.S. and Shchukin, E.D. 1992. Ostwald ripening theory: Applications to fluorocar-
bon emulsion stability. Advances in Colloid and Interface Science 38, 69.
Lopez-Montilla, J.C., Herrera-Morales, P.E., Pandey, S., and Shah, D.O. 2002. Spontaneous
emulsification: Mechanisms, physiochemical aspects, modelling and applications.
Journal of Dispersion Science and Technology 23, 219–268.
Lyklema, H. 2000. Electric double layer, in Fundamentals of Interface and Colloid Science,
Vol. 2, Liquid-Solid Interfaces. Academic Press, London, Chapter 3, pp. 3.1–3.232.
Maldonado-Valderrama, J., Martin-Rodriguez, A., Gálvez-Ruiz, M.J., Miller, R., Langevin, D., and
Cabrerizo-Vilchez, M.A. 2008. Foams and emulsions of β-casein examined by interfacial
rheology. Colloids and Surfaces A: Physicochemical and Engineering Aspects 323, 116–122.
McClements, D.J., 2000. Comments on viscosity enhancement and depletion flocculation by
polysaccharides. Food Hydrocolloids 14, 173.
McClements, D.J. (ed.) 2004a. Emulsion stability, in Food Emulsions: Principles, Practices,
and Techniques, 2nd ed., CRC Press, Boca Raton, FL, Chapter 7, pp. 267–339.
McClements, D.J. (ed.) 2004b. Emulsion stability, in Food Emulsions: Principles, Practices,
and Techniques, 2nd ed., CRC Press, Boca Raton, FL, Chapter 3, pp. 33–93.
Melik, D.H. and Fogler, H.S. 1988. Fundamentals of colloidal stability in quiescent media.
in Encyclopedia of Emulsion Technology, Vol. 3, Becher P. (ed.), Marcel Dekker,
Pelipenko, J., Kristl, J., Rosik, R., Baumgartner, S., and Kocbek, P. 2012. Interfacial rheology:
An overview of measuring techniques and its role in dispersions and electrospinning.
Acta Pharmaceutica 62, 123–140.
Richardsson, G., Bergenstahl, B., Langton, M., Stading, M., and Hermansson, A.M. 2004.
The function of alpha-crystalline emulsifiers on expanding foam surfaces. Food Hydro-
colloids 18, 655–663.
Scheludko, A. 1967. Thin liquid films. Advances in Colloid and Interface Science 1, 391.
Walstra, P. 1993. Principles of emulsion formation. Chemical Engineering Science 48, 333–349.
Walstra, P. (ed.) 2003. Changes in dispersity, in Physical Chemistry of Foods. Marcel Dekker,
Welin-Berger, K. and Bergenstahl, B. 2000. Inhibition of Ostwald ripening in local anesthetic
emulsions by using hydrophobic excipients in the disperse phase. International Journal
of Pharmaceutics 200, 249–260.
Verwey, E.J.W. and Overbeek, J.Th.G. 1948. Theory of the Stability of Lyophobic Colloids.
Elsevier, Amsterdam, the Netherlands.
3 Formulation of Emulsions
Marie Wahlgren, Björn Bergenståhl,
Lars Nilsson, and Marilyn Rayner
Contents
3.1 Introduction .................................................................................................... 52
3.2 Functionality that Ingredients Should Give to Emulsions .............................. 52
3.2.1 Nutrition and Health ........................................................................... 52
3.2.2 Texture and Flavor .............................................................................. 53
3.2.3 Shelf-Life Stability ............................................................................. 54
3.2.3.1 Emulsion Stability................................................................ 54
3.2.3.2 Chemical Stability ............................................................... 56
3.2.3.3 Microbiological Stability ..................................................... 57
3.2.3.4 Freeze–Thaw Stability ......................................................... 58
3.3 Issues to Consider When Choosing Ingredients for Emulsions...................... 59
3.4 Key Ingredients in Emulsions......................................................................... 62
3.4.1 Fats and Oils ....................................................................................... 62
3.4.2 Low Molar Mass Emulsifiers..............................................................64
3.4.3 Proteins ...............................................................................................66
3.4.4 Polysaccharides................................................................................... 70
3.4.5 Protein–Polysaccharide Complexes.................................................... 74
3.4.6 Particles .............................................................................................. 74
3.5 Evaluation of Emulsion Formulation and Ingredient Performance ................ 77
3.5.1 Emulsification Capacity ......................................................................80
3.5.2 Emulsion Stability Index .................................................................... 82
3.5.3 Assessing Gravitational Separation—Creaming Index...................... 82
3.5.4 Accelerated and Environmental Stress Tests ...................................... 86
3.5.5 Evaluation of Texture .......................................................................... 88
References ................................................................................................................90
ABSTRACT In this chapter, we describe some of the main concerns when it comes
to formulating emulsions. This includes the choice of ingredients, such as emulsifiers,
oils, preservatives, and thickeners. This is done with a focus on how these ingredients
can give the desired properties of the emulsions, such as texture, flavor, nutrition,
and stability. Commonly encountered thickeners and emulsifiers are described,
and the methods to characterize the key properties of emulsion and ingredient are
discussed.
51
3.1 IntroduCtIon
Almost all industrially processed emulsion-based food products are made up of a
wide variety of constituents, including fats and oils, emulsifiers, texture modifiers,
preservatives, antimicrobial agents, antioxidants, pH adjusters, sweeteners, salts,
coloring agents, flavors, and, of course, water. Each of these has been included in the
food product due to its intrinsic function or a combination of functions with other
compounds in the formulation. They are there to provide the overall quality of food
products such as nutritional value, flavor, texture, and shelf life. In this chapter, we
will discuss how the ingredients deliver these quality attributes to emulsions, and
we will also give a more general description of some of the key ingredients in emul-
sions, primarily oils, emulsifiers, and texture modifiers.
Food ingredients can be described on several levels:
1. Molecular (e.g., H2O, glucose, kappa casein, etc.)

2. Nutritional (e.g., proteins, lipids, carbohydrates, minerals, etc.)
3. Composite ingredients or recipe (e.g., milk, eggs, flour, salt, etc.)
4. Functional ingredients (e.g., emulsifiers, thickeners, preservatives, etc.)
Food manufacturers, product developers, and formulators are generally concerned

with the mass fraction of composite ingredients and functional ingredients because
they are normally purchased and used in this form.
3.2 FunCtIonalIty that IngredIents

should gIve to emulsIons
3.2.1 NutritioN aNd HealtH
A key function of any food emulsion is certainly its nutritional value. As emulsions
contain both lipophilic and hydrophilic regions, they have the capability to include
both water-soluble and oil-soluble components of high nutritional value. Emulsions
can increase the bioavailability of lipophilic nutrients such as vitamin E (Mayer,
Weiss, and McClements 2013, Yang and McClements 2013) or other beneficial com-
ponents, such as curcumin, that have low solubility in water (Ting et al. 2014). One of
the main nutritional concerns when it comes to emulsions is the composition of the oil
phase. Health benefits can be obtained, for example, by formulating products contain-
ing omega-3 oils (Berasategi et al. 2014, Moore et al. 2012). Another important health
aspect of food emulsions is the development of low-calorie products. In this case, one
often tries to manufacture products with low oil content, such as low-fat spreads, that
still has a texture similar to the original high-fat product (Chronakis 1997, Kasapis
2000). The aim is to formulate a product with low oil content that still has compa-
rable texture, flavor, mouthfeel, and visual aspects as its traditional high-fat product.
However, as the volume fraction of oil phase often is important for emulsion struc-
ture, this poses specific problems that need to be addressed; for example, the addition
of texturizing macromolecules (polysaccharides and proteins) will compensate the
lower oil fraction in maintaining the microstructure in low-calorie products.
Formulation of Emulsions 53
The effectiveness of emulsions as vehicles for the delivery of individual nutritional

compounds is affected by emulsion properties such as the surface area of the oil
droplets and the availability of the oil interface for digestive enzymes. Furthermore,
in the in vivo situation, the properties of the emulsion will also affect the gastric
emptying of the stomach, where emulsions prone to phase separation in the stomach
show a more rapid empting than emulsions that are stable (Golding and Wooster
2010). The structure of the fat used for the emulsion will also affect digestion; for
example, solid fat is digested more slowly than liquid fat (Michalski 2009).
3.2.2 texture aNd Flavor

In this section, a short overview of the area is given, and for a more thorough read-
ing, we recommend some recent reviews on the topic (Chung and McClements 2013,
Stokes, Boehm, and Baier 2013). The texture of emulsions is strongly dependent
on its rheological properties. Rheology of the emulsion is in turn dependent on the
volume fraction of the dispersed phase, the degree of flocculation of the dispersed
phase, and rheological properties of the continuous phase. In most cases, especially
if the drops are small, the rheological properties of the dispersed phase are less
important. Another factor that influences the mouthfeel and taste is how the emul-
sion may aggregate and coalesce in the mouth due to mixing with saliva, interactions
with the mucosa, the change in temperature, and the mechanical treatment while
eating (Benjamins et al. 2009).
In oil-in-water (O/W) emulsions, the rheological properties of the continuous
phase are often modified by the addition of polymers. As the degree of flocculation
of the oil droplets may also influence rheology, factors that affect flocculation such
as the type of emulsifier, pH, and salt should also be considered. The physical prop-
erties of dispersed oil phase could affect the rheology by the formation of crystal-
line bridges between different oil droplets leading to semicoalesced drops (Fredrick,
Walstra, and Dewettinck 2010). Thus, the melting temperature of the oil phase may
affect the texture. The fraction of oil also influences mouthfeel, where high-fat O/W
emulsions are usually perceived to have high creaminess, to be smooth and rich in
flavor (Chung and McClements 2013). The critical level of fat content to achieve the
mouthfeel related to fattiness seems to be around 15% (Malone, Appelqvist, and
Norton 2003). In low-fat products, increasing the viscosity in the continuous phase
can to some degree compensate the low-fat content and give products that have simi-
lar flavor and mouthfeel as high-fat products. However, the key parameter may not
be the rheology as a bulk property but rather the rheology of the film formed in the
mouth cavity upon eating the food product (Malone, Appelqvist, and Norton 2003).
In the case of water-in-oil (W/O) emulsions, such as spreads, the state of the fats
are important not only for mouthfeel but also for spreadability. The ratio between
liquid and solid fat will thus affect the rheology. Also, for these types of products,
the type of polymorphic form of the lipid crystals will influence the property of the
product, as a transition from β′ crystals (preferred in margarine-type products) to
β crystals is associated with larger crystals (greater than 20 µm), giving a gritty or
sandy mouthfeel, low spreadability, and oil–fat separation (Heertje 2014, Sato and
Ueno 2011). Mouthfeel of emulsions can also be altered by the presence of particles
or fat crystals. Large particles will give a sandy mouthfeel usually described as
tallowness (Watanabe et al. 1992).
When it comes to flavor, the release of flavoring components from the dispersed
phase is important. The release will be affected by how these molecules are trans-
ported out of the dispersed phase and thus by properties such as diffusion coefficient
of the component, droplet size of the dispersed phase, and interaction with other
ingredients in the emulsions (such as the emulsifier). The release will also be influ-
enced by partitioning of the flavoring ingredient into two phases and thus be affected
by the concentration of the dispersed phase. This is especially important for O/W
emulsions, as it is the concentration of aroma in the water phase and the head space
(gas phase above the emulsion) that influences its taste. Low-fat products can show a
burst of flavor due to the quick release of the oil-soluble components, whereas high-
fat products often display a more continuous release of components that partition
to the oil phase (Bayarri, Taylor, and Hort 2006). When designing and producing
low-fat products, the release profile of the oil-soluble components may have to be
modulated, for example, by encapsulation.
It has also been seen that in systems that have the same release of aroma com-
ponents into the gas phase, changes in the rheology of the emulsion still can affect
taste. This could be attributed to the difference in the release pattern between volatile
aroma compounds and more water-soluble taste compounds such as sugar (Bayarri
et al. 2006), where the latter is more sensitive to the rheology. When it comes to the
water-soluble components, they will predominately be in the water phase, and thus
O/W emulsions will have a quick influence on the taste. However, if taste masking
is desired, the water-soluble components can sometimes be encapsulated in double
emulsions.
3.2.3 SHelF-liFe Stability

The shelf-life stability of food emulsions is governed by factors that affect both
chemical and microbiological stability, in addition to issues that have to do with the
stability of the emulsion as such. There are also special issues, for instance, the sta-
bility of emulsions in frozen food, which are of technical and industrial importance.
3.2.3.1 emulsion stability

In Chapter 2, Bergenståhl describes factors that lead to the destabilization of emul-
sions in more detail. Destabilization of emulsions is mainly caused by creaming/
sedimentation, coalescence, and Ostwald ripening. For macroscopic emulsions,
creaming/sedimentation occurs due to the density difference between the oil and
water fraction; it can partially be reduced by decreasing the droplet size of the
emulsion, increasing the viscosity of the continuous phase, or by decreasing the dif-
ference in densities between the two phases (see Section 3.5.3). Coalescence leads
to the formation of larger oil droplets, which may eventually lead to a complete
phase separation of the emulsion. This can mainly be controlled by the adsorp-
tion of surface-active compounds to the interface that hinders drop–drop con-
tact through either a steric barrier or an electrostatic repulsion, thus resulting in
the process of coalescence. Increased viscosity of the continuous phase can also
decrease coalescence (as well as the rate of creaming/sedimentation) to some extent.
Another mechanism that drives the evolution of droplet size is caused by the pres-
sure difference between the inside and outside of a curved surface. This so-called
Laplace pressure is higher for a more curved surface, for example, small droplets;
this is the driving force Ostwald ripening, which leads to an increase in particle
size of the emulsions at the expense of smaller droplets. In this case, the solubility
of the dispersed phase in the continuous phase is of major importance; that is, a low
solubility slows down or prevents Ostwald ripening. Hence, Ostwald ripening is
typically not observed in triglyceride O/W emulsions but, for instance, can occur
for more soluble oils such as aromatic and essential oils. Ostwald ripening can also
be decreased by increasing the viscosity in the continuous phase (decreases diffu-
sion) and systems with low curvature. Pickering emulsions, for example, have been
suggested to decrease Ostwald ripening, as they might have a local zero curvature
(Tcholakova, Denkov, and Lips 2008).
Flocculation is the aggregation of droplets. Flocculated systems may have desired
properties for formulation such as beneficial rheology, but extensive flocculation
might lead to increased creaming and thus may lead to coalescence. Changes in the
degree of flocculation can also affect the rheology of the emulsions, changing prop-
erties such as mouthfeel.
The colloidal stability of the emulsion will be governed by the repulsive/attractive
forces between individual droplets of dispersed phase, the energy and rate of droplet
collisions, the viscoelastic properties of the interface between oil and water, and the
solubility of the dispersed phase in the continuous one. The choice of an emulsifier
could influence all of these, and a proper choice of viscosity modifier will influence
all kinetic factors such as collision of droplets and diffusion of dissolved molecules.
The most important repulsive and attractive forces between emulsions droplets
are summarized as follows:
Hydrophobic effect. This is the main reason for the instability of emulsions.
The hydrophobic interaction is based on the exclusion of nonpolar compo-
nents from water.
van der Waals attraction. These forces exist in all systems. Between small mol-
ecules, van der Walls forces are of short range and decay with increasing dis-
tance between the molecules proportional to the distance raised to the power
of minus six. However, in a colloidal system, they can be of a much more
long range, decaying with the reciprocal of distance. Together with the elec-
trostatic forces, it is the basis for the Derjaguin–Landau–Verwey–Overbeek
(DLVO) theory (Verwey and Overbeek 1948).
Electrostatic repulsion. This can be an important stabilizing force for food
emulsions. Both proteins and ionic emulsifiers can be charged, depend-
ing on the pH, and when adsorbed, at the droplet interface giving rise
to electrostatic repulsion. Emulsions stabilized by electrostatic repulsions
are sensitive to salt and, in many cases, sensitive to pH. This sensitivity
toward salt is due to the decay in the range of the electrostatic repulsion
caused by the presence of ions and the effect strongly increases with the
valance of the ions. Thus, it is the ionic strength that is the key issue when
it comes to stability in emulsions based on ionic emulsifiers. One should
be aware that the ionic strength of buffers changes with pH. In some cases,
there could also be specific ion interactions; for example, an interaction
between calcium ions and casein that leads to aggregation (Dickinson and
Davies 1999).
Depletion attraction and steric repulsion. These interactions are caused by
the presence of macromolecules in the continuous phase. Depletion attrac-
tion is due to the fact that macromolecules (proteins, polymers, and col-
loidal particles) having no affinity toward the interface will be excluded
in the space between two approaching emulsion drops; this will lead to
an osmotic pressure gradient, which then favors aggregation. Depletion
attraction is typically observed in emulsions containing dissolved neutral
polysaccharides. Thus, the addition of polysaccharides to alter the rheology
or to form complexes with emulsifying agents may lead to depletion aggre-
gation (Magnusson and Nilsson 2011). Steric repulsion is induced by mac-
romolecules adsorbed at the interface; this is mainly due to the excluded
volume effect, as adsorbed molecules come close together (Israelachvili
1985). Steric repulsion thus requires not only the affinity of the macromol-
ecule to the interface but also a high solubility of the macromolecule in the
continuous phase. The latter allows parts of the adsorbed macromolecule
to protrude into the continuous phase, giving rise to steric hindrance. Both
nonionic low molecular emulsifiers and polymers might stabilize the emul-
sion through steric repulsion. These systems are less sensitive to salt and pH
than electrostatic stabilized emulsions.
3.2.3.2 Chemical stability

One key issue for the chemical stability of emulsions is the oxidation of fats;
especially fats with a high degree of unsaturation are susceptible to this problem
(Moore et al. 2012, Waraho, McClements, and Decker 2011). The most common
cause of fat oxidation in O/W emulsions is the interaction between transition met-
als and lipid hydroperoxides located at the oil–water interface, which produces
highly reactive peroxyl and alkoxyl radicals (Frankel 1998, McClements and
Decker 2000).
One way to handle fat oxidation could be to add antioxidants such as vitamin E
or phenolic substances, but a proper choice of ingredients and ingredient quality can
also be of importance. For example, Charoen et al. (2012) showed that different bio-
polymers used as emulsifiers for rice oils differed in their capability to protect the
oil from oxidation. They speculate that this could partly be due to the degree of spe-
cific binding capacity of iron to the polymers, but they could not exclude that it was
caused by impurities such as heavy metal ions used in the polymers. This illustrates
that it is important to be aware of the oxidative impurities in the ingredients that are
used; this could be heavy metal ions as well as peroxides. Waraho, McClements,
and Decker (2011) review the oxidation of lipids in emulsions and point out that
there will be a difference in the oxidation process in pure oil when compared with
one in an emulsion. This is because the water phase in the emulsion may include
oxidative agents such as transition metals and iron and ingredients such as eth-
ylenediaminetetraacetic acid (EDTA) and iron-binding proteins (e.g., lactoferrin)
that may decrease oxidation in emulsions (Waraho, McClements, and Decker 2011).
Phenolic compounds have been seen to be pro-oxidatives, especially in the presence
of iron (Medina et al. 2012, Sørensen et al. 2008).
Antioxidants can, as mentioned, be added to the formulation, and the activity of
these antioxidants will depend on their location in the emulsion and solution condi-
tions such as pH. It has been shown that nonpolar antioxidants are more effective
in emulsions as compared to nonpolar oxidants, which are more effective in bulk
oils (Frankel 1998). This is called the polar paradox and is probably related to the
fact that the antioxidant has to be close to the lipids that it should protect. There is
a growing interest to use naturally occurring phenolic compounds such as caffeine,
coumaric acid, and rutin as antioxidants (Kikuzaki et al. 2002, Medina et al. 2012,
Sørensen et al. 2008). These compounds have, however, also been seen to, at some
conditions, be pro-oxidative (Sørensen et al. 2008). This phenomena of having both
anti-oxidative and pro-oxidative characteristics depending on the formulation further
highlights the importance to know the function of the specific additive at the condi-
tions used for each food product. Another problem with several phenolic compounds
is their low solubility (Löf, Schillén, and Nilsson 2011) and, in these cases, their
existence as dispersed particles in the continuous phase, which, of course, reduces
their antioxidative capacity.
3.2.3.3 microbiological stability

Microbiological stability is especially important for O/W emulsions as they often
have a high water activity. The shelf life from a microbiological viewpoint will be
dependent on packaging, processing (e.g., pasteurization), and the choice of ingredi-
ents. When it comes to ingredients, both their intrinsic microbiological load and their
ability to function as antimicrobiological ingredients are important. Furthermore,
from a formulation viewpoint, ingredients other than preservatives can give bacte-
ricidal effects. For example, components in essential oils are antimicrobial (Burt
2004) and so are some emulsifiers, for example, lysozyme–xanthan gum conjugates
(Hashemi, Aminlari, and Moosavinasab 2014) and monocaprylate (Hyldgaard et al.
2012). Some of the more common food preservatives such as ascorbic acids and its
salts (Lück 1990) are also used in emulsions. Other bactericides are the peptide nisin
(Castro et al. 2009) that is common in several food emulsions such as dairy prod-
ucts and sausages (Abee, Krockel, and Hill 1995). When choosing preservatives, it
is important to understand how the property of the emulsions such as pH and salt
content will affect the preservative. The choice of other ingredients and their concen-
tration might also affect the action of the preservative. Nisin, for example, has been
seen to be strongly affected by the composition of the emulsion such as oil content
and oil/surfactant ratio (Castro et al. 2009). The homogenization as such may also
affect the preservative, especially if it is sensitive to surface adsorption, heat, or shear
(Zapico et al. 1999).
In W/O emulsions, the microbiological growth is reduced due to the limited space
in the water droplets and the inability for the microorganisms to transport themselves
in between droplets; however, as pointed out by others, W/O emulsions such as mar-
garine and spreads also need to show how microbiological safety is obtained during
the shelf life (Charteris 1996, Delamarre and Batt 1999). For these products, spoilage
is often due to moulds and can be reduced by the addition of preservatives such as
sorbates and benzoates (Delamarre and Batt 1999).
3.2.3.4 Freeze–thaw stability

A special case of stability is the freeze–thaw stability of emulsions (Degner et al.
2014). It is often seen that frozen emulsions changes when thawed, for example,
manifested as a full phase separation or that the emulsion becomes grainy and
watery. There are several reasons for these effects; one is that the lipid crystals
can form and lead to partial coalescence of the semifrozen emulsion, which upon
reheating goes forward to full coalescence and phase separation. Crystallization
pattern of the lipids is one of the main factors that will determine if a freeze-stable
emulsion can be obtained. Magnusson, Rosén, and Nilsson (2011) have shown that
for high dispersed volume fraction O/W emulsions, oils contain high amounts
of unsaturated fatty acids, have a high percentage of crystallized triglycerides at
−25°C, and thus have a high rate of susceptibility for freeze–thaw instabilities. The
volume fraction of oil will also affect the freeze–thaw stability; in an unpublished
study, it was shown that the freeze–thaw stability of mayonnaise was increased by
decreasing the oil fraction. This is probably because of the reduced contact time
between oil droplets. Another mechanism that is seen for emulsions with oils that
do not crystallize before ice formation is that coalescence is triggered by increas-
ing the concentration of the dispersed phase when larger and larger volumes of the
water phase are removed due to ice formation. In this case, freeze–thaw instabili-
ties can be decreased by not only the right freezing conditions but also the right
choice of the product composition. The freeze–thaw stability of emulsions can be
increased by the addition of cryprotectants such as polyols (sucrose, glucose, fruc-
tose, trehalose, and maltose), antifreeze proteins, gelatin, and some carbohydrates
(Degner et al. 2014). These alter the crystallization of water and the morphology
of the ice crystals; however, some of them can also function by increasing the vis-
cosity, and thus decreasing the number of oil droplet collisions leading to coales-
cence. Addition of polysaccharides has also been seen to improve the freeze–thaw
stability. This could be due to several factors; however, an increased viscosity of
the nonfrozen phase and the capability of some polysaccharides to form protec-
tive layers around the dispersed phase hindering coalescence play a major role
(Degner et al. 2014). The emulsifier is critical when it comes to destabilization
due to increased concentration, but can also be important for lipid crystallization-
induced freeze–thaw instabilities. Emulsifiers are able to stabilize the emulsion
also at a high concentration, for example, some Pickering emulsions using quinoa
starch granules or egg yolk granules (Marefati et al. 2013, Rayner et al. 2014) pro-
teins such as caseins (Degner et al. 2014) and hydrophobic starch that give a thicker
interfacial coatings around the fat droplets are good in this sense.
In frozen and cold-stored foods, it is especially important to understand how
the emulsifier itself is affected by the decrease in temperature; for example, several
low-molecular emulsifiers lose their solubility below the so-called Kraft point and
thus the function of these emulsifiers will decrease.
3.3 Issues to ConsIder when ChoosIng

IngredIents For emulsIons
One of the key ingredients to choose for an emulsion formulation is the emulsifier.
The emulsifier lowers the surface energy between the two phases and thus affects the
size of the emulsion droplets. It should also create a barrier for coalescence and drop-
let growth during storage. Emulsions can be stabilized by low-molecular emulsifiers,
proteins and other polymers, and particles. Table 3.1 gives a comparison between
them and Figure 3.1 shows a schematic structure of the various general classes of
emulsifiers at the oil water interface. As a rule-of-thumb, low-molecular emulsifiers
lowers the surface tension more than surface-active macromolecules, but they adsorb
reversibly to the interface; they often form complexes with other ingredients in prod-
ucts such as proteins, and might be less good than high-molecular emulsifiers when
it comes to reducing coalescence during storage. When choosing an emulsifier, it is
important to consider its compatibility with other ingredients of the product; many
components, such as some preservatives, are surface active themselves and might
interact with the interface. Another issue is salt concentration and pH. Both charged
small-molecule emulsifiers and proteins are strongly affected by salt concentration
and proteins are especially strongly affected by pH.
When choosing ingredients for a multicomponent system as an emulsion, it is
very important to not only understand the solubility of components in the two phases
but also to partition the ingredients into each phase and to the interface between
the two phases. For example, whatever thickener is used, it should be partitioned
into the continuous phase of the formulation. One also has to be aware that the
partitioning of components might change the phase behavior of the ingredient. One
simple example of this is that the partitioning of small surface-active substances to
the oil–water interface will shift the apparent critical micelle concentration (CMC)
for these components to a higher concentration, which is dependent on the surface
area of the dispersed phase, and thus affected by the droplet size and the amount of
dispersed phase.
For shelf life, the purity of the ingredient is critical, especially components that
trigger oxidation of the oil, for example, heavy metal ions or peroxides. One should
also be aware of the concentration of surface-active substances such as fatty acids
in the oil. The latter could interact in different ways with the mechanisms of sta-
bilization of the oil droplets either by competing with the chosen emulsifier at the
oil–water interface or by interacting with it thus changing its properties. This is of
special importance for particles, but could also likely affect, for example, fatty acid-
binding proteins or biopolymers that form inclusion complex with the surface-active
components.
Also, the stability of the ingredients during homogenization has to be considered.
As discussed previously, the homogenization process negatively affects the preserva-
tive effect of nisin. Another example is the hydrophobically modified starch, which
60
taBle 3.1
Comparison of Functional Characteristics and Formulation attributes of various general Classes of emulsifiers
used in Food emulsions
general Class small molecular weight surfactants macromolecules Particles
Approx. size ~0.4 to 1 nm 2–200 nm 10 nm to 10 µm
Surface active Yes Yes Yes—via partial dual wettability
Amphiphilic Yes (head and tail) Yes (hydrophobic and hydrophilic No (unless Janus particles)
regions)
Adsorption kinetics Fast in dynamic equilibrium Medium partially irreversibly Slow but essentially irreversible
Desorption energy Low <10 kT High and increasing if Exceptionally high greater than
conformational changes occurs at several thousand to tens of
the interface, several thousand kT million kT, depending on particle
size and contact angle
Chemical types Nonionic HLB 12–16 Nonionic HLB Ionic Proteins Polysaccharides Colloidal solids Colloidal
7–10 θ < 90° solids θ > 90°
Food examples Polysorbates Monoglycerides Phospholipids Caseinates egg Modified starches, Modified starch Fat crystals
proteins celluloses in and cellulose
solution crystals/particles
Solubility/ Water Oil Water Water Water Water Oil
dispersibility
Emulsion type O/W W/O O/W O/W O/W O/W W/O
Usage level (g/goil) ~0.05 ~0.05 ~0.05 ~0.05 ~1 to 1.5 ~0.02 to 1 ~0.02 to 1
pH stability Good Good Depends on pKa Poor Good Variable Good
Salt stability Good Good Poor at I > CFC Poor at I > CFC Good Variable Good
Source: McClements, D.J. “Emulsion ingredients.” In Food Emulsions: Principles, Practices, and Techniques. CRC Press, Boca Raton, FL, 2005.
CFC, critical flocculation concentration; I, ionic strength.
Water OH
Oil
HO Monoglyceride
O O
Lecithin
O
O
O O
O
O O P O
O O
wO
O OH
O x
HO OH
O O
z y
Polysorbate Oil Water

(a) (b)
Water Water
Macromolecules θ Particle
θ < 90° oil in water emulsion Oil

(or if θ > 90° water in oil emulsion)
Oil
(c) (d)
FIgure 3.1 Schematic structure of the various general classes of emulsifiers at the oil–
water interface: (a) nonionic small molecular weight emulsifiers; (b) ionic small molecular
weight emulsifiers; (c) amphiphilic macromolecules with hydrophobic and hydrophilic regions
such as proteins and modified starches; (d) particles such as starch granules, fat crystals, and
whey protein microgels. Relative sizes and other attributes are provided in Table 3.1.
has been shown to decrease its molecular weight during homogenization (Modig
et al. 2006, Nilsson, Leeman et al. 2007). Soy proteins have shown disruption as
well as aggregation induced by high-pressure homogenization (Roesch and Corredig
2003). Several issues such as heat, shearing, and the adsorption into surfaces of the
equipment can affect the ingredients in the emulsion. Another issue could be that
the surface-active components might induce the leakage of components from gas-
kets and other plastic and rubber parts of the equipment. Thus, an incompatibility
between the process and the ingredients used has to be considered during the devel-
opment process.
In food products, a further complication is that many of the ingredients used are
often very complex mixtures, for example, mixture of proteins, polar lipids, and
polysaccharides. One good example here is egg yolk, which is used to stabilize
mayonnaise-type emulsions. In such cases, it may be difficult to know which ingredient
actually contributes to the emulsification and stabilization actions; to complicate things
further, the components at the interface can vary with solution properties such as pH
(Magnusson and Nilsson 2013, Nilsson et al. 2006, Nilsson, Osmark et al. 2007). The
interaction between ingredient components may enhance the stability of emulsions as
well as cause instability, and the effect of adding individual ingredients can be difficult
to predict.
Finally, one should be aware of the variation and inhomogeneity of the ingredi-
ents and at least have some knowledge if this might affect batch-to-batch variation
of products. Several commercial emulsifiers are mixtures that show a variation in
chain length of the hydrophobic tail (cf. sorbitan esters and ethoxylated sorbitan
esters) or variation in molecular weight (cf. all polymers) and even variation in the
composition, for example, lecithin and whey proteins. These variations can affect
the composition of the molecules at the interface and could influence issues such as
shelf-life stability, rheology, droplet size, and so on.
3.4 Key IngredIents In emulsIons

3.4.1 FatS aNd oilS
Oil is not only often the major source of energy in a food emulsion, but it also acts
as the phase where key nutritional components such as antioxidants and fat-soluble
vitamins will be dissolved. Thus, the choice of fats used in the emulsion process will
influence the nutritional value of the final product. Generally speaking, a high degree
of saturated lipids, especially omega-3 and omega-6 lipids, is considered to be linked
to health benefits, and a factor that is used to improve the nutritional value of some
foods (Berasategi et al. 2014, Moore et al. 2012, Sørensen et al. 2008)
The most common oils in food emulsions are triglycerides (see Figure 3.2). The
properties of the triglycerides are governed by the chain length and the degree of
saturation of the lipophilic part of the molecule (Larsson 1986). The longer the chain
length and the more saturated fatty acids are in a triglyceride, the higher will be the
melting point. Triglycerides are often classified depending on the chain length of the
fatty acid part into high- (>12), medium- (6–12), and low-chain (<6) triglycerides.
Table 3.2 shows some of the more common food oils and fats.
As described in Chapter 2, the character of the oil phase will influence emul-
sion stability when it comes to oxidation, coalescence, and Ostwald ripening.
Furthermore, the crystallinity of the oil affects both the stability and the organoleptic
properties of the emulsion. Thus, the key properties are the temperature at which the
α β′ β
Hexagonal Orthorhombic Triclinic

(a) (b) (c)
FIgure 3.2 (a–c) Schematic description of crystalline structures for triglycerides with
unsaturated fatty acid side chains.
taBle 3.2
Common Food oils/lipids
Production vitamin e saturated lipids monounsaturated Polyunsaturated
name [million tons (may 2014)]a (mg/100g)b (g/100g)b (g/100g)b (g/100g)b melting Pointc
Oil, Coconut 3.43 0.09 86 6 2 25

Oil, Cottonseed 4.99 35.3 26 18 52 −1
Oil, Olive 3.19 14.4 14 73 10 −6
Oil, Palm 62.35 15.9 43 37 9 35
Oil, Palm Kernel 7.23 3.81 81.5 11 2 24
Oil, Peanut 5.84 15.7 17 46 32 3
Oil, Rapeseed 26.02 −10
Oil, Soybean 46.34 8.1 16 23 58 −16
Oil, Sunflower 15.52 41.8 10 45 40 −17
Fish oil 1 30 27 34
Butter fat 2.8 60 28 4 32–35
Lard 0.6 40 45 11 41
Sources: a Agriculture, U.S.D.O., World Production, Markets, and Trade Reports: Oilseeds, United States Department of Agriculture, Washington, DC, 2014.
b United States Department of Agriculture—Agricultural Research Service, http://ndb.nal.usda.gov/ndb/foods.
c Engineering Tool Box, http://www.engineeringtoolbox.com/oil-melting-points-d_1088.html.

63
oil will crystallize and the type of polymorph that is formed during crystallization.
Triglycerides exist in several polymorphic forms, where β is the most stable one. Due
to the differences in treatment, for example, cooling rate, the crystals can be locked
into other less stable polymorphic forms, where α and β′ are the most common ones
for triglycerides; this is considered in several textbooks and in a recent review by Sato
et al. (2013). As discussed previously, β′ has more appealing organoleptic properties
than β. The different polymorphic forms also have different melting points, which
can be important for the stability of the emulsions. As a rule of thumb, the less stable
polymorphic forms, the lower the melting temperature. In more complex systems,
the crystallization will be dependent on the mixture of triglycerides and the addition
of other components. One example is the addition of diacylglycerol to blends of palm
super olein to increase the onset temperature for crystallization (Ng et al. 2014).
Apart from triglycerides, most oils also contain traces of numerous other com-
ponents such as diacylglycerol, monoacylglycerols, free fatty acids, phospholipids,
tocopherols, and minerals. As reviewed by Chen, McClements, and Decker (2011),
these trace compounds can affect the stability of the oil when it comes to oxidation
but also when it comes to emulsion stability. For example, free fatty acids, mono- and
diglycerides as well as phospholipids are surface active and can cause foaming upon
mixing, or destabilization of protein-stabilized emulsions. However, the effect on
lipid oxidation of these compounds, as described by Chen, McClements, and Decker
(2011), is not straightforward, although free fatty acids have been seen to accelerate
the oxidation of triacylglycerols. Tocopherols, on the other hand, have a beneficial
effect on reducing the oxidation as they work as antioxidants.
3.4.2 low Molar MaSS eMulSiFierS

Low-molecular emulsifiers are characterized by having regions of the molecule that
are more water loving (hydrophilic) and regions that are more water hating (hydro-
phobic). This is the main driving force for these molecules to adsorb into the inter-
face between oil and water, where they lower the surface tension between the two
phases. The amphiphilic character of these molecules also triggers the formation
of different self-assembled structures in solution (see Figure 3.3). The type of self-
assembled structures that are obtained depends on the concentration of the emulsifi-
ers, as well as is partly affected by the intrinsic properties of the molecules. The type
of structures formed ranges from micellar structures to reversed structures (reversed
micelles, L2 phases (is an inverse micellar solution), or reversed hexagonal phases).
Emulsifiers with an even balance between hydrophilic and hydrophobic parts often
form lamellar liquid crystals, which forms vesicles when dispersed in water. The
critical packing parameter (CPP) is a generalization of the self-assembling properties
of surfactants, describing the properties as a geometrical balance between the area
needed for the polar group relative and the area needed for the hydrophobic group
(Israelachvili, Mitchell, and Ninham 1976) (see Figure 3.3). This can be used to esti-
mate what type of structure an emulsifier will give. When CPP is equal to one, it will
favor lamellar structure, whereas if the CPP is lesser than 1/3, it will favor micelles;
a CPP of above 2 will favor reversed micelles. For more details on the liquid crystal-
line phases formed by common food emulsifiers, there is a review by Krog (1997).
CPP < 1/3 1/2 < CPP < 2 CPP > 2

(a) (b) (c)
V: volume of hydrophobic chain

CPP = V l: length of hydrophobic chain
al
a: area of head group
FIgure 3.3 Schematic description of some self-assembled lipid structures and an explana-
tion of packing parameter: (a) CPP < 1/3, (b) CPP > 1/2 and < 2, (c) CPP > 2.
Friberg and Wilton (1970) suggested that the presence of lamellar liquid crystalline
phases is a strong indication of a good emulsifier in simple systems.
The functionality of low-molecular emulsifiers is, in a wide interpretation, deter-
mined by their solution properties. Although the character of low-molecular emulsi-
fiers is such that they contain regions that are water soluble and those that are more
lipophillic, their overall character can make them more soluble in one of the two
phases. Thus, emulsifiers can be found in a range from highly soluble in the oil to
more soluble in the water phase. The effect of solubility on emulsion character was
first expressed in the Bancroft (1913, p. 501) rule, stating that “hydrophilic colloid
will tend to make water the dispersing phase while a hydrophobic colloid will tend
to make water the disperse phase.” To describe the degree of hydrophilicity con-
tra lipophilicity, it is very popular to use the hydrophilic–lipophilic balance system
according to Griffin (1954). The HLB number is expressed as a number based on the
molecular weight of hydrophobic components compared to the molecular weight of
the molecule. The HLB number can also be estimated from the chemical structure
according to molecular group contributions as stated by Davies (1957):
taBle 3.3
hlB values as Predictor for the use of emulsifiers
hlB value applications example of emulsifiersa
3.5–6 W/O emulsifier Glycerol monostearate
7–9 Wetting agent Sorbitan monolaurate
8–18 O/W emulsifier Tween 80
13–15 Detergent Tween 81
15–18 Solubilization Sodium Oleate
Source: Davies, J.T., Proceedings of 2nd International Congress Surface Activity,

a
Butterworths, London, 1957.
HLB = ∑ Hydrophilic group numbers − ∑ Hydrophobic group numbers + 7 (3.1)
The HLB value of an emulsifier is often used as a rule of thumb (see Table 3.3). However,
one should be aware of the fact that solution conditions might change the HLB balance
of a system; for example, the addition of salt can screen charge groups, making the
system appear less hydrophilic. Another factor of importance is the temperature. The
effective HLB value is strongly temperature dependent. For ethoxylated emulsifiers, the
emulsifier gets less hydrophilic with increasing temperature and finally becomes insoluble
in water at a temperature denoted as the cloud point. In an emulsion system, this can be
followed by the phase inversion temperature (PIT), which corresponds to the temperature
at which the effective HLB is about 6 (Shinoda and Sato 1969). Emulsions stored at a tem-
perature of 25°C–60°C below the PIT are usually more stable. However, in food applica-
tions, this is rarely used, as ethoxylated surfactants are uncommon for food applications.
Another important temperature to consider for emulsifiers is the Krafft point (Krafft and
Wiglow 1895). The Krafft point is the temperature below which the surfactant has low
solubility and, hence, cannot form micelles. Technical functionality (such as foaming
and emulsifying action) is only obtained above the Krafft temperature. High-melting fat
bases (fully hardened C18-dominated fats) or long paraffinic chains creates high-melting
emulsifiers with Krafft temperatures in the range of 40°C–60°C. Precipitating emulsifiers
may contribute to fat crystallization and solid emulsifier may have a textural functional-
ity; however, for most applications, such high melting points are unsuitable. Intermediate
melting fat bases (C14–C18 fats with some unsaturation) give emulsifiers with Krafft or
transition temperatures between 30°C and 50°C. These emulsifiers could be used to cre-
ate stable α-gels and usually display well-performing properties in baking applications.
Low-melting fat (highly unsaturated fat), branched hydrocarbons and inclusion of aro-
matic groups, gives low Krafft points, sometimes below 0°C. Table 3.4 summarizes some
examples and usages of common low-molecular weight emulsifiers.
3.4.3 ProteiNS
Proteins function both as emulsifiers and as rheological modifiers in the formula-
tion of food emulsions. The character of proteins in emulsions will be based on their
taBle 3.4
Common low-molecular weight emulsifiers
number
name eu/usa solubility uses Comment
Lecithin E322/184.1400 Dispersible but insoluble in water, Margarine, chocolate, breads and cakes, bubble gum, salad dressings, Mixture of phosphoric acid,
where it swells on hydration. and sauces choline, fatty acids, glycerol,
Soluble in oils and fats. glycolipids, triglycerides, and
phospholipids
Fatty acid salts E470/172.863 Sodium and potassium salts are Baked goods (e.g., bread and cakes), confectionery, dairy products, Charged at normal and low pH
soluble in water. Calcium salts are margarines, spreads, shortenings, salad dressings, and sauces
insoluble in water.
Sodium stearoyl E481/172.846 Dispersible in warm water and Fine bakery wares, emulsified liqueur, fat emulsions, desserts, Negatively charged
lactylate soluble in hot edible oils and fats. beverage whiteners, and minced and diced canned meat products
Citric acid esters E472/172.832 Dispersible in hot water, insoluble Fats for stabilizing, also as synergists for antioxidants, baking fat Negatively charged
of MG in cold water, and soluble in emulsions, bakery margarines and shortening for stabilizing, mar-
edible oils and fats. garine, mayonnaise, salad dressings, sauces, and in low-calorie foods
Mono and E471/184.1505 Oil Baked goods, confectionery (e.g., chewing gum, toffees, and caramels), Nonionic
diglycerides dairy products, creams, desserts, edible ices, margarines, shortenings
Polyglycerol esters E475/172.854 Water Cakes and icings, margarine, and salad oils Nonionic
of FA
Propylene glycol E477/172.856 Oil Whippable icing Nonionic
esters of FA
Polyoxyethylene E435/172.836 Water Fine bakery wares, fat emulsions for baking purposes, milk and cream Nonionic cloud point around
(20) sorbitan analogues, emulsified sauces, soups, dietary food supplements, carriers
monooleate and solvents for colors, fat-soluble antioxidants, and antifoaming agents
Polyoxyethylene E433/172.840 Water Fine bakery wares, fat emulsions for baking purposes, milk and cream Nonionic cloud point around
(80) sorbitan analogues, emulsified sauces, soups, dietary food supplements,
monostearate dietetic foods, carriers and solvents for colors, fat-soluble
antioxidants, and antifoaming agents
67
tertiary structure in the solution and at the interface, their size, the net charge and
charge distribution, their capability to form gels, and the distribution of hydrophilic
and hydrophobic groups. There are numerous proteins that are used in emulsions,
and the choice of protein emulsifier is often not only based on function but also based
on what food group the emulsion is related. There are several traditional or natural-
occurring emulsions that, at least, are partially stabilized by proteins, such as mayon-
naise, dairy products, and sausages. Table 3.5 presents some of the more common
protein emulsifiers. Most of these emulsifiers are mixtures of several different protein
species. Thus, depending on the production and formulation conditions, the actual
proteins at the interface may differ although the same protein emulsifier is used.
taBle 3.5
Common Commercial Protein emulsifiers and example
of Proteins that are Part of the emulsifier
emulsifier Key Proteins molecular weight IP
Whey proteina β-Lactoglobulin 18.6 5.3
α-Lactalbumin 14.2 4.8
Bovine serum albumin 66 5.1
Caseins b
α1-Casein 23 4.1
α2-Casein 25 5.3
β-Casein 24 5.1
κ-Casein 19 5.6
Egg whitec Ovalbumin 45 4.5
Ovotransferrin 77.7 6.0
Ovomucoid 28 4.1
Lysozyme 14.3 10.7
Egg yolkc Phosvitin 160–190
Low-density lipoproteins 16–135
Cobalamin-binding proteins 39
Riboflavin-binding proteins 37
Biotin-binding proteins 72
α- and β-Lipovitellins 400
Soy proteind α-Conglycinin 18–33
β-Conglycinin 104
σ-Conglycinin 141–171
Glycinin 317–360
Sources: a Kinsella, J.E. and Whitehead, D.M., Advances in Food and Nutrition
Research, Academic Press, San Diego, CA 1989.
b Swaisgood, H.E., J. Dairy Sc., 76, 10, 3054–3061, 1993.
c Awade, A.C., Z. Lebensm. Unters. For., 202, 1–14, 1996.
d Clarke, E.J. and Wiseman, J., J. Agr. Sci., 134, 111–124, 2000.
IP, isoelectric point.

Factors affecting the protein adsorption into the interface during competitive adsorp-
tion from solution are size, charge and hydrophobicity of the protein, the transport
conditions of proteins to the surface during emulsification, if adsorbed proteins can
be exchanged by proteins in solution, and the degree of conformational changes of the
protein at the interface (Nilsson et al. 2006, Nilsson, Osmark et al. 2007, Wahlgren and
Arnebrant 1991). Thus, it is a complex issue to understand what proteins are actually
adsorbed at the interface. Magnusson and Nilsson (2013) reviewed this recently for egg
yolk in high internal phase emulsions and discussed that the main property governing
adsorption was the hydrophobicity of the proteins and that there is a preference for
HDL and LDL proteins to adsorb at the interface. In the case of milk proteins, Surel
et al. (2014) have seen that in mixtures of casein micelles and whey proteins, casein
dominates at the interface when the fraction of casein in the solution is above 25%.
Proteins get their amphiphilic character from the mixture of hydrophilic and
hydrophobic amino acids. The amino acid sequence (secondary structure) also gives
the template for the three-dimensional structure of the protein (tertiary structure).
However, one should be aware that the tertiary structure will vary due to solution
conditions, and that proteins in solution have a well-defined tertiary structure, which
could be considerably changed and even lost when adsorbing at an interface. Proteins
are often divided into different categories based on their tertiary structure. The most
common structures are random coil (casein), globular proteins (whey proteins and
egg proteins), and rod-like structures (fibrinogen, collagen, and gelatin). In many
cases, the protein has a defined molecular weight but for some food proteins such as
gelatin, this is not the case. The distribution of hydrophobic groups within the poly-
mer is important. For globular proteins, the hydrophobic groups are mainly found
inside the core of the protein shielding them from water. Upon adsorption into the
oil–water interface, these hydrophobic groups could orient themselves toward the oil,
which might lead to conformational changes of the protein. A few proteins especially
κ-Casein has very distinctive hydrophilic and hydrophobic domains, which together
with its semirandom coil structure makes them especially suitable as emulsifiers.
The casein proteins α1-, α2-, β-, and κ-caseins form complex called casein micelles.
Although these proteins play a large biological and a technical role, the structure of
the casein micelles is still debated (Dalgleish 2011, Horne 2002).
The main difference between low-molecular weight emulsifiers and proteins is
that while the adsorption of the former is completely reversible, when the concentra-
tion is lowered, proteins have a tendency to adsorb irreversibly. This makes them less
sensitive to changes such as dilution. However, even if the adsorption is irreversible
toward the lowering of concentration, the protein could still be exchanged by other
species (proteins or low-molecular ones) that have higher driving force for adsorp-
tion, for example, a higher reduction of the surface tension at the oil–water interface.
The kinetics of these events and the conformational changes of the proteins can
be slow, on the timescale of hours to days, and can lead to postproduction changes
of the emulsion. Furthermore, proteins are sensitive to heat, enzymes, and solution
conditions such as pH and ionic strength, which lead to degradation, aggregation,
and other protein changes. These events can also lead to long-term change of emul-
sions stabilized by proteins. Another difference between low molar mass (or small)
emulsifiers and proteins is the rheology of the adsorbed layer. Proteins often form
thicker, more viscous layers than small emulsifiers (Bosa and van Vlieta 2001). This
is in most cases positive for the long-term stability of the emulsion. Proteins might
stabilize emulsions through electrostatic repulsion and, thus, several protein sys-
tems show tendencies to aggregate at pH close to the isoelectric point of the pro-
tein emulsifier. If such aggregation does not lead to coalescence, it could lead to an
increase in the viscoelasticity of the emulsion (Wu, Degner, and McClements 2013).
In systems where both small molecular emulsifiers and proteins are present, there
might be a competition between the components at the interface or there might be
a cooperative adsorption (Maldonado-Valderrama and Patino 2010, Nylander et al.
2008, Rodríguez, García, and Niño 2001, Waninge et al. 2005). The competitive
adsorption of proteins and small emulsifiers are strongly concentration dependent,
and at concentrations below the CMC of the emulsifiers, proteins often dominate at
the interface (Wahlgren and Arnebrant 1992). The order in which the components
reach the surface might also be important as small surface-active components can-
not always remove already adsorbed proteins (Karlsson, Wahlgren, and Trägårdh
1996, Wahlgren 1995). Cooperative adsorption may occur when the protein com-
plexes with the low-molecular emulsifier, which, for example, is common for many
ionic surfactants (Maldonado-Valderrama and Patino 2010). It is often seen that the
adsorption of low-molecular emulsifiers to protein-stabilized emulsions have a detri-
mental effect on the emulsion stability (Wilde et al. 2004). Furthermore, there could
be strong interactions between proteins and surfactants in solution, changing the
structure and behavior of the proteins (Nylander et al. 2008).
Proteins also have the capability to change the rheology of the emulsions, espe-
cially if they are triggered to aggregate and to form a gel. Gel formation is often
induced by heating and denaturation of the proteins but could also be an effect of
pH, for example, the change in the rheology between milk and yoghurt. For example,
increased viscosity through the addition of proteins is important in low-fat products
such as margarines, sausages, and spreads (Chronakis 1997). Common proteins used
to form gel structures are milk-based systems such as whey proteins (Chronakis 1997,
Youssef and Barbut 2011) and soy proteins (Youssef and Barbut 2011).
3.4.4 PolySaccHarideS
Polysaccharides primarily function as viscosity modifiers in emulsions. However,
hydrophobically modified polysaccharides are also used as emulsifiers. A thorough
description of polysaccharides is given in Food Polysaccharides and Their Applications
(Stephen, Phillips, and Williams 2006). The properties of a polysaccharide is given by
the structure of the smallest repeating saccharide units, the degree of branching of the
polymer, and its molecular size. Differing from proteins, polysaccharides typically
have a high degree of polydispersity when it comes to branching and molecular weight.
There can also be a large batch-to-batch variation, which might lead to variation in
performance. Table 3.6 presents some of the more common polysaccharide groups and
these will also be discussed subsequently.
Traditionally, exudate gums, such as gum arabic, have been used as emulsifiers
especially in flavored beverages (Dickinson 2003). These are natural polysaccha-
rides that are produced by plants as a protection against bacteria and dehydration.
taBle 3.6
some Key Polysaccharides
name molecular structure Function viscosity modification
Starch amylase Essentially linear (1 → 4)-α-d-glucan Stabilizers, thickeners, Starch gelatinization; the ordered crystalline regions
and hydrophobically undergo melting, permitting granule swelling. This

modified starches, can be followed by the recrystallization and
which function as formation of helix structure.
Starch amylopectin Clusters of short (1 → 4)-α-d-Glc chains attached by emulsifiers Modification can make the starch cold, swelling or
α-linkages of 0–6 of other chains nonswelling.
Modified starches Native starch both amylose and amylopectin-modified
by chemicals for example acetate, phosphate, and
sodium octenyl succinate
Carboxymethylcellulose HO2CCH2-groups at 0–6 of linear (1 → 4)-β-d-glucan Stabilizer, thickener, Semisoluble polymer with a wide range of viscosities,
and water retention viscosity decreases with temperature.
Chitosan (1 → 4)-2-acetamido-2-deoxy-β-d-glucose and Emulsifier and Positively charged, partly hydrophobized polymer that
2-amino-2-deoxy-β-d-glucose rheological modifier can gel depending on pH and the presence of
multivalent negative ions
Carrageenans Sulfated d-galactans, units of (1 → 3)-β-d-Gal and Stabilizer, thickener, Forms salt- or cold-setting reversible gels in an
(1 → 4)-3,6-anhydro-α-d-Gal alternating; pyruvate and gelation aqueous environment. Gelling ability is seen for
and Me groups Carrageenans that form helical structures.
(Continued)
71
72
taBle 3.6 (Continued )

some Key Polysaccharides
name molecular structure Function viscosity modification
Gum arabic Acidic l-arabino-, (1 → 3)- and (1 → 6)-β- d-galactan, Emulsifier, Low viscosity at high concentrations, less than 40%
highly branched with peripheral l-Rhap attached to encapsulating agent, rheology strongly affected by pH and electrolyte.
d-GlcA. Minor components of glycoproteins stabilizer, and
thickener
Pectins Linear and branched (1 → 4)-α-d-galacturonan (partly Gelation, thickener, High-molecular-weight pectins form gels at low pH
methyl esterified and acetylated); chains include and stabilizer (2.5–3.5) and in the presence of high sucrose
(1 → 2)-l-Rhap, and branches d-Galp, l-Araf, concentration (>55%), or other cosolutes (e.g.,
d-Xylp, d-GlcA sorbitol and ethylene glycol)
Xanthan gum Cellulosic structure, D-Manp (two) and GlcA- Stabilizer and thickener Solutions have a thixotropic behavior; gels are formed
containing side chains, acetylated and pyruvylated on at high concentration or in the presence of plant
Man galactomannans such as locust bean gum.
Source: Stephen, A.M. et al., Food Polysaccharides and Their Applications, CRC Press, Boca Raton, FL.
One has to be aware that there is a very high variation in the composition between
gums obtained from different species (Stephen, Phillips, and Williams 2006) and that
this variation might affect the emulsion produced. These contain a heterogeneous mix-
ture of highly branched polysaccharides and a small amount of proteins (2% for gum
arabic) covalent attached to the polysaccharides. Gum arabic is thought to have a water
blossom structure built up of an amino acid core of around 400 units onto which bulky
polysaccharide units of 250 kDa are grafted (Stephen, Phillips, and Williams 2006).
It is the protein moieties of the exudated gums that make them surface active. Alftrén
showed that for gum arabic and mesquite gum, the amount of protein in the polysac-
charide fraction increased with increasing molecular mass and that these high protein
content/high-molecular-weight fractions were preferentially adsorbed into emulsion
droplets (Alftrén et al. 2012, Evans, Ratcliffe, and Williams 2013). The emulsification
capacity of gum arabic is lost upon heating (Williams, Phillips, and Randall 1990).
Another group of polysaccharide that is dependent on a protein fraction for its emul-
sifying properties are modified pectins (Akhtar et al. 2002, Dickinson 2003). Although
pectin is mainly used as a rheological modifier in emulsions, if they are modified, they
might work as emulsifiers (Dickinson 2003). The modification is, in most cases, acety-
lation; however, depolymerization using acids has also been used (Dickinson 2003).
Akhtar et al. (2002) have shown that depolymerized citrus pectin of 70% esterifica-
tion gives good stable emulsions, although only 25% of the pectin is adsorbed into the
interface and that upon storage, there are some flocculation that increase particle size.
For example, hydrophobic modification of polysaccharides starch can also pro-
duce molecules that are surface active and can be used as emulsifiers. Nilsson and
Bergenståhl (2006, 2007) have done extensive studies of hydrophobically modified
starch and shown that they are good emulsifiers and that the surface load of OSA-
starch can be as high as 16 mg/m2. They have also shown that it is the high molar
mass components of the polymer that are selectively adsorbed to the emulsion drop-
lets (Nilsson, Leeman et al. 2007).
According to Dickinson (2013), xanthan gum, which has high viscosity at low shear,
is established as the first choice when it comes to using them as rheological modifiers
for stabilizing emulsions, but there are a large range of polysaccharides that are used for
improving texture in food emulsions. Polysaccharides can increase the viscosity of an
emulsion either by some gelation mechanism, such as those triggered by the addition of
calcium ions (alginate, pectins, and carrageenan) and the formation of double helices
and crystallization (starch), or by nonspecific chain–chain interactions determining
the viscosity. Nongelling polysaccharides, especially if they are linear and are good
solvents, often behave as random coil polymers. At low concentrations, such polymers
behave more or less as Newtonian liquids but as the concentration increases, the poly-
mer chains start to overlap and the rheological behavior of the polymer changes. Above
the so-called overlap concentration, the viscosity becomes non-Newtonian (shear thin-
ning) and the viscosity increases more steeply with polymer concentration. Gelling of
polysaccharide can also form continuous water-swollen networks at low concentra-
tions. To obtain such systems, the polysaccharides contain both regions that form the
physicochemical bonds between the polymers and the nonbinding regions that primar-
ily hold the water. The regions that form the bonds are usually well ordered, allowing
for helices, egg-box, ribbon–ribbon, and double helix–ribbon structures.
In complex food products, for example, the interaction with the polysaccharides
with other components of the product might lead to segregated networks when pro-
teins and polysaccharides phase segregate. The polysaccharides, especially starch,
can also form inclusion complex with small emulsifiers such as monoglycerides and
fatty acids (Eliasson 1986, Tufvesson, Wahlgren, and Eliasson 2003). These interac-
tions are often stronger than the tendency for the emulsifier to adsorb into interfaces
and thus it will lower the amount of the emulsifier available (Lundqvist, Eliasson,
and Olofsson 2002). The complexes as formed will also have additional properties
such as melting point than the double helices normally formed in starch.
3.4.5 ProteiN–PolySaccHaride coMPlexeS

Compatibility between proteins and polysaccharides becomes important during the
modification of textures in food products. Mixtures of proteins and polysaccharides
can lead to three major scenarios (Schuh et al. 2013):
1. A single homogenous phase is formed.

2. The polysaccharide and proteins phase segregates into different phases as
the mixtures are not thermodynamically compatible.
3. Protein and polysaccharides aggregate and form insoluble complex.
In complex food systems, these types of interactions might either lead to the stabili-
zation of the emulsion or, especially in the latter case, destabilization.
Lately, there has been an interest in using the protein–polysaccharide complex
as emulsifiers in food systems (Evans, Ratcliffe, and Williams 2013). As discussed
previously, some traditional polysaccharide emulsifiers are probably protein–
polysaccharide complexes; other methods to obtain such complexes could be the
formation of Maillard conjugates (Akhtar and Dickinson 2007, Zhang, Chi, and
Li 2013) or electrostatic complexes (Harnsilawat, Pongsawatmanit, and McClements
2006, Koupantsis, Pavlidou, and Paraskevopoulou 2014, Xu et al. 2014). Protein–
polysaccharide complexes are used in the encapsulation of emulsion droplets, lead-
ing to the protection of sensitive substances (Xu et al. 2014), or the encapsulation of
flavors (Koupantsis, Pavlidou, and Paraskevopoulou 2014). Protein–polysaccharide
complex stabilized emulsions have been shown to be more stable to stress, for exam-
ple, heat than emulsions only stabilized by the protein (Harnsilawat, Pongsawatmanit,
and McClements 2006). They have also been seen to be insensitive to pH and salt
concentration (Zhang, Chi, and Li 2013).
3.4.6 ParticleS
In addition to small molecular-weight surfactants and macromolecules, colloidal
particles can be utilized to stabilize emulsions. Particle-stabilized emulsions (com-
monly referred to as Pickering-type emulsions) are possible as a result of the proper-
ties of the particles, where a combination of size, form, and partial dual wettability
of both the oil and water phases confers Pickering particles several useful properties
and the ability to create emulsion droplets that are highly stable against coalescence
(Rayner et al. 2014). Particle-stabilized emulsions in general, and in food-based

particles in particular, have received an increasing interest in recent years and for
this reason, a separate chapter has been devoted to particle-stabilized emulsions in
this book (see Chapter 4). However, particle-stabilized emulsions are by no means a
recent discovery, being reported in the scientific literature during the early twentieth
century. One of the first particle-stabilized food products was mayonnaise, a popular
condiment based on an O/W emulsion, which was first formulated in 1756, where the
finely ground (and somewhat hydrophobic) mustard particles adsorb at the oil–water
interface and cover a fraction of the oil droplets preventing coalescence, in addition
to other surface-active components found in egg and other ingredients (Binks 2007).
The commonly used types of particles in fundamental studies (where there is an
abundance of literature to be found in the fields of soft matter and physical chemistry)
are often inorganic/synthetic such as clays, silica, alumina, titanium oxides, and latex-
based particles (Aveyard, Binks, and Clint 2003). However, there has been a recent
and an increasing trend toward developing suitable food-based particles, which are not
only edible but also maintain the consumer perception of being wholesome or natu-
ral. Three general approaches can be taken to obtain food-based particles. They can
be built up or synthesized from molecules extracted from food-based materials (e.g.,
aggregation, crystallization, cross-linking, precipitation, etc.); they can be obtained
by reducing the size of existing structures (e.g., milling, crushing, hydrolyzing, etc.);
and they can be isolated as with their innate biological structures intact. Also, in many
cases, a breaking-down step (i.e., to dissolve the working material) is a precursor for
synthesis. Examples of a synthesis approach for generating edible particles include
starch nanocrystals (Li, Sun, and Yang 2012), chemically modified starch nanospheres
(Li, Sun, and Yang 2012, Tan et al. 2012), flavonoid particles (Luo et al. 2011, 2012),
chitin nanocrystals (Tzoumaki et al. 2011, 2013), soy protein particles (Paunov et al.
2007), whey protein microgels (Destribats et al. 2014), insoluble corn protein (zein)
particles (De Folter, Van Ruijven, and Velikov 2012), solid lipid particles, and fat crys-
tals (Gupta and Rousseau 2012). Examples of particles formed from the breakdown of
larger structures include cellulose nanocrystals (Kalashnikova et al. 2011, 2013), cocoa
particles (Gould, Vieira, and Wolf 2013), ethyl cellulose particles (Jin et al. 2012), and
cryomilled fractured modified starch particles (Yusoff and Murray 2011). Examples of
particles directly isolated include lactoferrin nanoparticles (Shimoni et al. 2013), bacte-
ria chitosan networks (Wongkongkatep et al. 2012), natural spore particles (Binks et al.
2005, 2011), hydrophobic bacteria (Dorobantu et al. 2004), egg yolk granules (Ercelebi
and Ibanoglu 2010, Eriksson 2013, Laca et al. 2010, Rayner et al. 2014), and starch
granules isolated from a variety of botanical sources (Li et al. 2013, Rayner, Timgren
et al. 2012, Timgren et al. 2011), with or without hydrophobic modification (Rayner,
Sjöö et al. 2012, Song et al. 2014, Timgren et al. 2013).
Generating food-grade particles for stabilizing emulsions has been of a recent
interest because Pickering-type emulsions are generally more stable against coales-
cence and Ostwald ripening, and have the potential to enhance oxidative stability
(Aveyard, Binks, and Clint 2003, Binks 2002, Kargar et al. 2012). The reason for
their high stability is due to the particles preventing interfacial interaction by vol-
ume exclusion; that is, particles adsorbed at the oil–water interface create a physical
barrier preventing drop–drop contact. Because Pickering particles are often tens to
thousands of nanometres in diameter, this physical layer is quite large in comparison

to surfactant molecules (~1 nm) and protein molecules (~5 nm). As in most types
of emulsion formulations, the size of emulsion droplets in Pickering emulsions is
governed either by the amount of emulsifier relative to the dispersed phase or by the
intensity of the emulsification device (Chevalier and Bolzinger 2013, Tcholakova,
Denkov, and Lips 2008). Generally droplet size decreases with increasing particle
concentration (at fixed dispersed phase content) to a certain extent after which it
levels out, and excess particles begin to accumulate in one of the phases. Beyond this
level of particle-to-dispersed phase ratio, a further reduction in droplet size can only
be achieved by improving the emulsification conditions (Frelichowska, Bolzinger,
and Chevalier 2010). However, in Pickering emulsions, the size of the stabiliz-
ing particles ultimately limits the size of the emulsion drops that can be formed.
Generating small particles is a common objective of many studies, as it reduces the
amount required (milligram of particles per milliliter of dispersed phase) to stabi-
lize a given emulsion droplet interface, that is, emulsification capacity. Requiring a
high concentration of emulsifier for creating a stable emulsion is generally undesir-
able from a formulation viewpoint, as emulsifiers can be expensive, have a negative
impact on the overall taste, or have their concentration limited by regulatory aspects.
On the other hand, if the particles themselves are food components (i.e., starch gran-
ules, egg yolk granules, or fat crystals), this may not necessarily be the case, as the
particles themselves contribute to the nutritional profile, the perceived product qual-
ity, and/or the sensory properties of the formulation in a positive way. Furthermore,
for Pickering emulsions, achieving a small droplet size may not be as crucial as
in conventional emulsion formulations, where small droplet size is often correlated
with improved emulsion stability, as creaming is often a precursor to coalescence. In
contrast, Pickering emulsions droplets of large size (even on the millimeter scale) if
successfully formed, seem to be stable to coalescence over extended periods of time
(Aveyard, Binks, and Clint 2003, Binks 2007, Laredj-Bourezg et al. 2012, Marku
et al. 2012, Timgren et al. 2013). However, large droplets are also susceptible to
creaming, which is a major drawback of this type of emulsion. Some examples of
how gravimetric separation is reduced in Pickering formulations include the careful
choice of particle size and the amount to generate droplets that are of neutral buoy-
ancy (Rayner, Sjöö et al. 2012), or in cases where particle properties create weak
gel (Dickinson 2010, 2012), which is further improved in cases having high level of
dispersed phase (drops + particles), that is, space-filling conditions.
Particles as emulsion stabilizers have enabled formulators in the reduction or the
removal of low-molecular weight surfactants, which, in some cases, have a limit on
the amount that can be used in food emulsion recipes. Due to the relatively large
size of the stabilizing particles (in comparison to molecular surfactants and poly-
mers emulsifiers), they make up a significant volume fraction of the emulsion for-
mulation per se, and once adsorbed at the oil–water interface, they are essentially
thermodynamically trapped there (Aveyard, Binks, and Clint 2003). Furthermore,
if they possess a sufficient level hydrophobicity, it can also lead to particle aggrega-
tion. Because if it is energetically favorable for a particle to adsorb at the oil–water
interface, it is also favorable for the particles to adsorb to each other and tend to
exist in a state of weak aggregation in the continuous phase (Dickinson 2013), and
the particles may form a network or a gel-like structure with increased viscoelastic
moduli (Dickinson 2013). This can also result in the emulsion having a yield stress,
which, even if relatively small, will assist in preventing creaming, drop–drop con-
tact, and coalescence under quiescent storage conditions (Dickinson 2012). This may
also prove to be the reason why particle-stabilized emulsions, even when the surface
coverage of particles at the oil–water interface is much less than a closely packed
monolayer, can remain stable over several years of storage (Timgren et al. 2013).
Rheological properties resulting from particle–particle interactions may also have
the added benefit of reducing the need for additional thickeners and viscosity modi-
fiers in particle-stabilized formulations (Dickinson 2013).
3.5 evaluatIon oF emulsIon FormulatIon

and IngredIent PerFormanCe
The evaluation and characterization of emulsions and ingredient performance in food
formulations can be performed on several different time and length scales. The com-
plexity of these expensive type measurements or characterization methods can vary
from using simple visual observations of creaming in a glass container to elaborate
neutron scattering experiments. In any case, the method and time window of evalua-
tion should reflect the formulations’ fitness of intended use. For example, diary emul-
sions as refrigerated products need to be stable over the span of its best before date
(due to microbial spoilage), but not necessarily too much beyond that. Furthermore,
the emulsions should remain stable when exposed to likely environmental stresses
encountered during processing, packaging, storage, and consumption.
There are numerous methods to characterize emulsions and the ingredients
making up emulsion formulations, all of which cannot be described in detail
here. For a more thorough description of characterization method, we recommend
McClements’s (2005a) work and references therein. However, in Table 3.7, we sum-
marize some of the more important ingredient properties investigated and their pos-
sible characterization.
On a microstructural level, the emulsion droplet size distribution is perhaps the
most central quantifying measure in emulsion science, as emulsion characteristics
and performance are highly dependent on the droplet size distribution. There are
numerous methods to assess particle size distributions in emulsions, including direct
droplet measurement by microscopy (light, confocal, electron, etc.), automated
particle counters (i.e., Coulter counter), light scattering (i.e., Malvern Mastersizer),
dynamic light scattering, diffusional wave spectroscopy, nuclear magnetic reso-
nance, and sedimentation or centrifugation (Walstra 2005). These techniques vary
with respect to the range of sizes covered, measurement principles, degree of sample
preparation and dilution, as well as physical limitations of the methods. The inter-
ested reader is directed to comprehensive and critical reviews on emulsion charac-
terization techniques for more details (Dalgleish 2003, Dickinson 2013, McClements
2005b, 2007, Sherman 1995, Walstra 2005).
From a consumers’ perspective, aside from taste, the two most important emul-
sion properties are physical appearance (creaming, sedimentation, phase separa-
tion, graininess, etc.) and texture (mouthfeel, viscosity, etc.). These macroscopic
78
taBle 3.7
Characterization of emulsion Ingredient Properties
Key Properties (reviews) examples of methods Comments
Surface tension of biopolymers, surfactants, Wilhelm plate, drop volume, Pendent drop Pendent drop works well for soluble substances and for
and proteinsa oil–water interface. It is also good for measuring dynamic
change in surface tension. Wilhelm plate are easier to use
for nonsoluble materials
Interfacial rheology that is especially Interfacial rheology can be measured as dilatational The methods listed here are suitable for measurements
important for biopolymers and proteins.b deformation (oscillating increase and decrease of close to equilibrium
Presence of low-molecular emulsifiers often surface, e.g., a pendant drop) Langmuir trough gives both dilatation and shearing
decrease interfacial elasticityc Shearing deformation where the area is constant but the deformation
shape is changed (deep-channel surface viscometer)
Crystallinity and polymorphismd X-ray diffraction is the preferred method for identifying The use of synchrotron radiation enables analyses on a
crystalline structure but DSC and FTIR can be used as short timescale and thus kinetic phenomena can be
complementary techniques studied. X-ray can also be used to study self-associated
structures of lipids and synchrotron radiation can be used
to study structures at interfacese
Melting point and amount of solid materialf DSC, ultrasonics, and heat-controlled microscopy One should be aware that the scanning rate used in DSC
sometimes are too fast to have the sample in equilibrium
and this can affect the measured melting temperature
(Continued)
taBle 3.7 (Continued )
Characterization of emulsion Ingredient Properties
Key Properties (reviews) examples of methods Comments
Molecular weight/mass distribution is critical There are numerous methods for molecular weight Size exclusion and FFF can be linked to light-scattering
properties especially for proteins and determination; for high-molecular-weight molecules, detectors to get additional information such as shape. FFF
polysaccharides. Information on branching FFF,g size exclusion,h gel electrophoreses, rheology of is especially well suited for larger polymers and proteins
and shape can also be important to diluted solutions, and analytical ultracentrifugation can such as starch molecules. MALDI TOF is primarily used
understand biopolymers be used. Mass spectroscopy such as MALDI TOFi to get composition and structural information
Sources: a Drelich, J. et al., Encyclopedia of Surface and Colloid Science, Marcel Dekker, New York, 2002.
b Pelipenko, J. et al., Acta Pharm., 62, 2,123–140, 2012.
c Maldonado-Valderrama, J. and Patino, J.M.R., Curr. Opin. Colloid Interface Sci., 15, 4, 271–282, 2010.
d Sato, K., Chem. Eng. Sci., 56, 2255–2265, 2001.
e Cristofolini, L., Curr. Opin. Colloid Interface Sci., 19, 3, 228–241, 2014.
f McClements, D.J., Food Emulsions: Principles, Practices, and Techniques, CRC Press, Boca Raton, FL, 2005a.
g Nilsson, L., Food Hydrocolloids, 30, 1, 1–11, 2013.
h Hagel, L., Protein Purification, John Wiley & Sons, Inc., Hoboken, NJ, 2011.
i Harvey, D.J., Mass Spectrom. Rev., 30, 1:1–100, 2011.
DSC, differential scanning calorimetry; FFF, field flow fractionation; FTIR, Fourier transform infrared spectroscopy.
79
properties are a result of microstructure and molecular interactions within the emul-
sion formulation (Corredig and Alexander 2008) and are closely related to the ability
of the formulation to stabilize and maintain the stability of the emulsion droplets,
as well as the rheology of the resulting dispersion. For this reason, in the develop-
ment of food-based emulsions, the effectiveness of emulsifiers and the evaluation of
texture are often studied. From the material presented in the earlier sections of this
chapter, it is apparent that there is a large variety of emulsifiers that can be used in
formulating food-based emulsions. The fundamental performance of an emulsifier
can be described by the emulsifying capacity (EC) as the minimum amount required
to produce a stable emulsion and its ability to produce small drops during homogeni-
zation. The emulsion stability index (ESI) is a measure of the ability of an emulsifier
to prevent droplets from aggregating, flocculating, and coalescing over time.
3.5.1 eMulSiFicatioN caPacity

When formulating a food emulsion, it is useful to know the minimum amount of
emulsifier required to create a stable emulsion. The EC of a water-soluble emulsi-
fier is defined as the maximum amount of oil that can be dispersed in an aqueous
solution containing a specific amount of emulsifier without the emulsion breaking
down or inverting into a W/O emulsion (Sherman 1995). The EC of an oil-soluble
emulsifier is determined in a similar way, except that water is added to the oil phase
and it would invert into an O/W emulsion. This test is practically carried out by
placing the continuous phase containing the emulsifier into a vessel with a high-
speed mixer (e.g., ultra-turrax) and carefully titrating the dispersed phase into the
vessel. Phase inversion can be monitored via electrical conductivity (Allouche et al.
2004, Gu et al. 2000) or optically using a colorimeter or spectrophotometer in reflec-
tance mode (McClements 2002) or by adding a dye to one of the phases (Timgren
et al. 2013). The larger the volume of oil that can be added before phase inversion,
the higher will be the EC of the emulsifier. This test is widely used due to its relative
simplicity, but has several drawbacks that prevent its application as a standardized
procedure (Dalgleish 2003, McClements 2005b, 2007, Sherman 1995). The main
problem identified with the procedure is that the amount of emulsifier required to
stabilize an emulsion depends on the oil–water interfacial area rather than the oil-
volume fraction, thus EC depends on the size of the droplets produced during agita-
tion. This in turn is highly sensitive to the type of mixing/homogenization apparatus
used, its energy intensity (see Chapter 1), the volume of emulsion being processed,
the viscosity of the oil phase, and the temperature of processing. As such, EC should
be regarded as a qualitative index that depends on the specific setup and conditions
used to carry out the test that can be used to compare different emulsifiers tested
under the same conditions.
An alternative way of estimating the amount of emulsifier required to form an
emulsion that takes into account the amount of interfacial area generated is via the
surface load, Γs, which corresponds to the mass of emulsifier required to stabilize a
unit area of droplet surface (Dickinson 1992). This is determined by first generating
a stable emulsion by homogenizing a known amount of oil, water, and emulsifier.
Then the amount of emulsifier adsorbed at the oil–water interface is determined
via a mass balance of the emulsifier; that is, the amount adsorbed at the oil–water
interface, which is found by considering the initial concentration of emulsifier in the
continuous phase, Cini , minus the amount remaining in the continuous phase after
emulsification, Cend . This is carried out experimentally by carefully separating the
droplets from the continuous phase by centrifugation or filtration and determin-
ing the remaining concentration of the emulsifier (Tcholakova et al. 2002). The
interfacial area to which the emulsifiers are adsorbed is found by measuring the
specific surface area of the emulsion by either microscopy or an automated particle
size analyzer. The specific surface area, S, is determined from the surface mean
diameter, d32
d32 =
∑Nd i
3
i i
(3.2)
∑Nd i
2
i i
where:
Ni is the number of drops with diameter di
Because the specific surface area, S, is the sum of all the surface areas of all drops
divided by the sum of all their volumes (m2/m3), we can calculate S from d32:
4π(d32 /2)2 6
S= = (3.3)
(4/3)π(d32 /2)3 d32
Now, including this into the mass balance or the emulsifier over the continuous phase
and interface, we get
Vcts (Cini − Cend ) (1 − φ) d32

ΓS = = (Cini − Cend ) (3.4)
Vdisp S φ 6
Here, Vcts and Vdisp are the volumes of the continuous and dispersed phases,
respectively, and ϕ is the volume fraction of dispersed phase, that is,
Vdisp
φ= (3.5)
Vcts + Vdisp
Typically, the values of ΓS for molecular food emulsifiers is around a few milligram
per square meter, but is much larger for particles, hundreds to thousands milligram per
square meter, as ΓS is also directly related to the thickness of the interfacial layer. These
estimates of surface load values provide some knowledge with respect to the minimum
amount of emulsifier that is required to make an emulsion having droplets of a given
size and the dispersed phase fraction. However, in practice, an excess of emulsifier is
often used, as not all emulsifiers are ideally adsorbed into the oil–water interface during
homogenization due to kinetic limitations, as well as due to the partitioning equilibrium
conditions between the interface and the continuous phase. Furthermore, the surface
load of some types of emulsifiers is also sensitive to formulation conditions such as ionic
strength, pH, the concentration of macromolecules, temperature, and so on.
3.5.2 eMulSioN Stability iNdex

The emulsification capacity, presented in Section 3.5.1, gives information on the
ability to create an emulsion with a given formulation, but does not necessarily take
into account the evolution of emulsion stability over time. One expression of the
emulsion stability over time is the ESI, which is based on particle size measurements
performed at given time intervals and is defined as
d( 0 )t
ESI = (3.6)
d( t ) − d( 0 )
where:
d( 0 ) is the initial mean droplet diameter of the emulsion
d( t ) is the mean droplet diameter measured after a storage time, t (McClements
2005b)
Some of the main strengths of this method include that the mean droplet diameter
can be readily determined in analytical instruments, the evolution of particle size
microstructure is often a precursor to quality deterioration on the macrostructure
(creaming and phase separation, etc.) and can re-repeated over relevant timescale for
the shelf life of the product. A similar index is also sometimes used that compares
the specific surface area of the emulsions rather than just mean droplet size as a mea-
sure of how much coalescence has taken place. This surface coalescence index (SCI)
is more sensitive to the fate of smaller drops (as they have a relatively larger surface
are to volume ratio) and can be calculated by
S( 0 ) − S( t )
SCI = (3.7)
S( 0 )
where:
S( 0 ) is the initial specific surface area of the emulsion
S( t ) is new specific surface area of the emulsion measured after a storage time,
t (Anton, Beaumal, and Gandemer 2000)
S is calculated directly from 6/d32
It should be noted that there is no compelling evidence that a single index such as
EC, ESI, or SCI can be used to ultimately compare the effectiveness or the stability
of emulsifiers if they have been produced under different homogenization conditions.
Still, these indices are very useful when comparing a series of emulsifiers of emulsion
formulations produced under standardized conditions or in situations when the influ-
ence of specific changes are being made to the formulation, processing conditions, or
functionality of a specific ingredient that is being studied (McClements 2005b).
3.5.3 aSSeSSiNg gravitatioNal SeParatioN—creaMiNg iNdex

Gravitational separation of emulsions is one of the most common instability mecha-
nisms encountered in food and personal care products, thus formulators need to
know at what degree creaming or sedimentation is likely to occur over the shelf life
of a relevant product. Due to the fact that emulsion droplets in the context of food
emulsions typically never have the same density as the continuous phase and are
large enough for the buoyant forces to overcome viscous resistance and Brownian
motion, they allow gravitational separation to be observed on a relevant timescale.
As most edible oils at room temperature have a lower density than aqueous solu-
tions, oil droplets in O/W emulsions will tend to rise to the top of the container in
a process referred to as creaming, leaving the depleted layer by an emulsion drop at
the bottom of the container, often referred to as serum. These terms likely originate
from the prevalence of diary emulsions. For W/O emulsions, the sedimentation of
water droplets is observed, although generally at a much slower rate due to the higher
viscosity of the oil. However, the opposite can be observed, where the sedimentation
of oil droplet can occur if fat crystals or other weighing agents are added to the oil
phase, or in some cases, in particle-stabilized emulsions, where a higher density of
the stabilizing particle layer increases the overall density of the droplet causing them
to settle (Rayner, Timgren et al. 2012).
The rate at which a single spherical droplet or particle will cream (or settle) in a
Newtonian fluid can be predicted by the Stokes velocity:
2 gr 2 (ρ2 − ρ1 )
vStokes = − (3.8)
9η1
where:
g is acceleration due to gravity
r is the particle radius
ρ1 and ρ2 are the densities of the continuous and dispersed phases, respectively
η1 is the continuous phase viscosity
The settling or creaming rate of drops and particles indicated by Stokes equation is
somewhat idealized, as in reality, emulsions drops are not all the same size and will be
interacting during creaming or settling. Furthermore, Stokes law is mainly applicable
at low concentrations of the dispersed phase. However, Stokes equation does provide an
illustration of the factors that have the most impact on the gravitational tendency, specif-
ically the viscosity of the fluid surrounding the droplets, their relative density, and, to a
large degree, the droplet size due to the exponent. For example, an oil droplet creams at
a rate of 0.1 mm/day if its diameter is 0.1 µm, and will cream at a rate of 10 mm/day if the
diameter is 1 µm, if all other conditions remain constant. In many practical situations,
these conditions are not constant and are more complex; for example, there is often an
increase in the effective particle size during creaming due to coalescence, flocculation,
or Ostwald ripening (see Chapter 2), which results in Equation 3.8 under predicting the
rate of gravitational separation (McClements 2007). Therefore, it is often more practical
to directly quantify gravitational separation of the emulsions during storage.
The extent of creaming or sedimentation in an emulsion can be monitored by
visual observation. This method is cheap and straightforward, only requiring the
emulsions to be stored in an appropriate environment in clear glass vials or test
tubes. The layer formed by creamed emulsion droplets can be readily seen, and often
Vemuls Hrel oil Released oil layer
Vtotal Hemuls Creamed emulsion
Vserum Serum layer

Hserum
(a) (b)
Vemuls Hserum
EI = × 100% CI = × 100%
Vtotal Hemuls
FIgure 3.4 (a) Test tube showing creamed emulsions as defined in EI; (b) schematic test
tube showing layers as defined in CI.
the serum layer is transparent or optically distinct to such a degree that its height can
be determined, or its volume estimated. The emulsion index (EI) is a measure of the
volume of an emulsion layer formed relative to the total volume given by the follow-
ing equation, with the volumes defined in Figure 3.4:
Vemuls
EI = × 100% (3.9)
Vtotal
The relative heights of the creamy layer is also used to define the creaming index (CI):
H serum
CI = × 100% (3.10)
H emuls
Assuming that there is no significant coalescence or creation of an oil later (called

oiling-off, schematically illustrated in Figure 3.4b), the CI will start at zero and will
increase until all the emulsion drops are packed into the cream layer, after which the
CI index reached a final value. The height of the final cream layer depends on the
volume fraction of the oil, ϕ, and the maximum packing parameter droplets, which
is approximately P ~ 0.6, for random closed packing (McClements 2007).
 φ
CI final =  1 −  × 100% (3.11)
 P
Knowing the expected CI final can be relevant if emulsions with different dispersed
phase volumes are to be compared, as the more dispersed phases, the lesser will
be the serum. This concept had also been extended to adjust for the fact that
particles, and especially the larger food-based ones used in stabilizing Pickering
emulsions, also contribute to the amount of dispersed phase observed. The total
amount of nonseparated emulsion can be expressed as the relative occluding vol-
ume (ROV).
Vemuls
ROV = (3.12)
Vdisp + Vparticles
where:
Vemuls is the volume of the observed emulsion (i.e., the nonclear fraction) after
emulsification
Vdisp is the known volume of the added dispersed phase
Vparticles is the known volume occupied by the added particle stabilizers
In a completely phase-separated system, ROV equals to 1; that is, there is no increase

in the emulsion layer beyond that of its constituent phases. Figure 3.5 illustrates the
differences in EI and ROV for starch granule-stabilized emulsions with varying dis-
persed phase fraction (oil content 12.5%–33.2%), but at constant starch-to-oil ratio of
214 mg/mL oil. Here, we can see that although the EI increases as expected with oil
content, the ROV is higher for the less tightly packed systems. It should be noted that
in this system, there was no significant change in droplet size over time or between
formulations with different dispersed phase fractions.
Although the use of digital camera has greatly simplified the analysis of visual
observations of gravitational separations, there are several limitations to the visual
inspection of gravitational separation. It is often difficult to distinguish between the
layers in creaming/settling emulsions by the visual observation of a glass vial or a test
tube. One means to overcome this limitation is using an apparatus that scans the height
of the glass tube with a monochromatic beam of light near infrared part of the spec-
trum while monitoring the amount of scattered and transmitted light. This can give
an improved accuracy with respect to the boundaries between the creamed and serum
layers, as well as the possibility to quantify the concentration of emulsion drops as a
function of height. Some modern commercial instruments have also implemented mul-
tiple light-scattering techniques (e.g., Turbiscan Lab) that enable the measurement of
1.1 6
1.0 5
Emulsion index
0.9 4
ROV
0.8 3
0.7 2
0.6 1
0.5 0
0.1 1 10 0.1 1 10
Storage time (weeks) Storage time (weeks)
12.5% oil 16.6% oil 25% oil 33.2% oil
FIgure 3.5 EI and ROV of quinoa starch granule-stabilized oil-in-water emulsions. (Data
from Timgren, A. et al., Procedia Food Sci., 1, 95–103, 2011.)
concentrated emulsions and the measurement of both the particle size and phase thick-
ness continuously over time (Mengual et al. 1999), detecting the creaming long before
it is visible to the naked eye. The other major limitation of the gravitational separa-
tion analysis is that it takes a significantly long time to monitor instability that may
occur after weeks of storage and that the storage conditions in reality are not neces-
sarily those found in a controlled laboratory environment. This can be overcome in
two ways: (1) by increasing the gravitational field where the droplets cream/settle to
accelerate storage and (2) by exposing the emulsions to environmental stresses that
may trigger instability.
3.5.4 accelerated aNd eNviroNMeNtal StreSS teStS

The rate of gravitational separation and droplet coalescence can be accelerated
by the centrifugation of emulsions at a fixed speed for a certain length of time
(Sherman 1995). After which, the separation is monitored using the same means
as in a normal storage trial (visual observation, digital imaging, light scattering,
etc.). Alternatively, there are commercial analytical instruments that both scan and
centrifuge samples to gain further information (i.e., Lumisizer). However, precau-
tionary measures must be taken while using this approach, especially if there is a
difference in the rate of droplet size evolution between the normally stored and
accelerated samples. For example, if the droplet size is changing due to coales-
cence or Ostwald ripening, which in turn affects the rate of gravitational separation,
then these may or may not be reflected by merely increasing the gravitational field.
Furthermore, if the emulsion has a complex rheology (i.e., not just a Newtonian
continuous phase), increasing the gravitational field may over exaggerate the sepa-
ration. For example, if the emulsion has a weak gel structure (a finite yield stress)
that can be overcome in the centrifuge (but not under normal storage), the emulsion
will be forced to separate in conditions when it normally would not, if it is left to
settle over the normal time frame. For these reasons, it is imperative to compare the
results of accelerated creaming tests with those made using long-term normal stor-
age to validate the methods before routinely using an accelerated test for a given
type of formulation (McClements 2007).
In addition to gravitational separation, coalescence can also be accelerated using
centrifugal methods, as these methods essentially force droplets together. In this
case, the coalescence stability is determined by measuring the change in droplet
size distribution and/or the extent of oiling-off after the emulsion has been cen-
trifuged for a specific speed and time. Here, the coalescence stability is quanti-
fied in terms of the maximum centrifugation force that the emulsion can tolerate
before a change in the microstructure is observed (droplet size and or oiling-off).
The particle size distribution of the emulsion droplets can be measured before and
after centrifugation and the data can be represented as either the entire distribu-
tion or ESI. Alternatively, Tcholakova et al. (2002, 2006) have developed a cen-
trifugal method that can provide quantitative data about O/W emulsions stability
to coalescence. An emulsion is added to a transparent centrifuge tube and is loaded
into a centrifuge. This emulsion is then subjected to a centrifugal acceleration at
an appropriate intensity and time. The oils droplets will tend to move toward the
0 ζ0 ζ1 ζ2 ζ
ω
Cream
Oil Serum
Hrel HC
Z 0
FIgure 3.6 Schematic image of the thickness of the oil layer released during a forced
coalescence test.
axis of rotation (z direction in Figure 3.6) due to their relatively lower density. This
is the case for almost all food emulsions. Initially, the emulsion droplets form a
creamy layer in which they are forced into close proximity but keep their initial
form. As the centrifugal force is increased, they are pressed tighter and tighter
together and eventually the interfacial layer surrounding and stabilizing the drop-
lets will rupture, releasing a layer of oil (Hrel) on the top of the emulsion column in
the tube (see Figure 3.6).
The critical pressure that the emulsion can withstand before the oil is released
CR
when the film ruptures is described as a critical osmotic pressure, POSM . For a full
derivation, refer to Tcholakova et al. (2002, 2006) and references therein.
Hc
(Voil tot − Voil rel )
∫
= ∆ρgk φ( z )dz = ∆ρgk
CR
POSM (3.13)
ATT
0
where:
Δρ is the density difference between the oil and aqueous phases
gk is the centrifugal acceleration
φ(z) is the local volume fraction of oil along the z direction along the centrifugal
field
Voil tot is the total volume of oil in the emulsion
Voil rel is the volume of oil released
ATT is the interior cross-sectional area of the test tube containing the emulsion
After centrifugation, the height of the creamy layer, HC and oil released Hoil rel can
be easily measured, where Hoil rel = Voil rel / ATT . POSM
CR
may be readily calculated from
experimental data, if one assumes that the centrifugal field is homogenous through
the column of creamed emulsion, HC, and can be represented by the square of the
angular frequency, ω times the mean distance of the emulsion layer from the axis of
rotation, ζ:
ω2 (ζ1 + ζ 2 )
gk ≈ = constant (3.14)
2
Tcholakova et al. (2002, 2006) have proven that this is a reasonable assumption, as
more precise calculations that take into the account the spatial variation of the field
compared to using a mean distance gives a relatively small difference in the result,
and the imposed error is within the experimental accuracy of the measurements. This
method has been demonstrated to be particularly useful for monitoring the coales-
cence stability of different types of protein-stabilized emulsion with various composi-
tions (Denkov, Tcholakova, and Ivanov 2006, Tcholakova et al. 2002, 2003, 2006).
In addition to centrifugation, other types of accelerated coalescence tests include
subjecting the emulsions to other types stress such as mechanical forces (extended
homogenization, pumping, vibration, shearing, extruding, whipping, shaking, and
mixing) environmental stresses (freeze–thaw cycling, thermal processing, and heat
abuse), as well as compositional stresses (drying causing a change in solute com-
position, changes in pH and ionic strength, etc.). All of which with the purpose
of emulating some sort of typical event, environmental stress, or process that the
emulsion under consideration should withstand during processing, transport, shelf
life, and use. The formulation in general—and the performance of its emulsifier in
particular—is evaluated in a variety of conditions depending on its application to
establish a design space, in which a particular emulsifier is expected to successfully
function. Examples of test methods and experimental conditions/protocols can be
found in McClements (2007) and references therein.
3.5.5 evaluatioN oF texture

As pointed out previously, the rheology of an emulsion is important not only to
the taste, texture, and mouthfeel (Le Révérend et al. 2010) but also for the cream-
ing stability and coalescence (Tadros 2004). In this chapter, we only give a short
overview of the rheology of an emulsion; for a more extensive review, we recom-
mend Derkach (2009), Tabilo-Munizaga and Barbosa-Cánovas (2005), and Tadros
(2004). Texture can be evaluated using techniques such as rheological and texture
analyzers. Although texture analyzers can give a quick information and comparison
of systems, the information gained especially from oscillating rheology gives more
knowledge.
Emulsion droplets can, from a rheological viewpoint, in most cases be consid-
ered as hard spheres. The rheology of such systems are strongly dependent on the
concentration at low concentrations; where there are no droplet–droplet interactions,
the system will follow Einstein’s (1906) law for hard spheres. However, at moderate
concentrations, the interactions between the droplets will affect the rheology and in
these regimes, the rheology can be described by semiempirical equations such as the
Krieger–Dougherty equation (Krieger 1972).
−[( 5 / 2 ) φc ]
 φ 
η = η0  1 − e  (3.15)
 φc 
where:
η0 is the viscosity of the continuous phase
φc is the is the volume fraction at random close packing of spheres
φe is the volume fraction of oil in the emulsion.
Addition of rheological modifiers will further complicate the rheological proper-

ties of emulsions; for emulsions in the concentrated regime or emulsions containing
viscosity modifiers, the system will often become viscoelastic. For such systems,
measuring the rheology using oscillating measurements can further elucidate the
character of the emulsion. This allows for the use of small strains and stresses on
the material leading to measurements that does not destroy the structure of the sam-
ples. The sample is subjected to a sinusoidal shear deformation and the resultant
stress response is measured. The frequency and the strain/stress on the sample can
normally be varied and the response is divided into a viscous component G″ (loss
module) and an elastic component G′ (storage module). Such measurements can give
information of the viscoelastic character of the emulsion and, for example, describe
if the emulsion has a gel-like character or not. A characteristic for gels is that G′ is
higher than G″.
Oscillating measurements can, for example, be used to follow the buildup of a gel
during heating or cooling. In oscillating measurements, one can either change the
frequency of the oscillation or the strain/stress. Figure 3.7 shows typical frequency
measurements of Pickering emulsions for weak and strong gels and Figure 3.8 shows
a strain curves for the same samples. The frequency curve gives information on how
the emulsions react to stress during different time frames (time is proportional to
1/frequency). Stress or strain test are often used for gel-like emulsions and prefer-
ably measured at a frequency where the rheological properties of the gel are in a
100,000
10,000
1,000
G′ and G′′ (Pa)
100
10
0
0.1 1 10
Frequency, f (Hz)
FIgure 3.7 Elastic modulus (G′, in Pa) and viscous modulus (G″, in Pa) versus frequency
(f, in Hz) for Pickering emulsions with 19% Miglyol oil-in-water (O/W) emulsion stabilized
by chitosan G″ (closed squares), G′ (open squares) with 55% Miglyol O/W emulsion stabilized
by octenyl succinic anhydride (OSA)-modified quinoa starch G″ (closed circles), G′ (open
circles).
10,000
1,000
100
G′ and G′′ (Pa)
10
0.1
0.01
1.00E−01 1.00E+00
Complex shear strain
FIgure 3.8 Elastic modulus (G′, in Pa) and viscous modulus (G″, in Pa) versus complex
shear strain for Pickering emulsions with 19% Miglyol O/W emulsion stabilized by chitosan
G″ (closed squares), G′ (open squares) with 55% Miglyol O/W emulsion stabilized by OSA-
modified quinoa starch G″ (closed circles), G′ (open circles).
linear region. Typically, in strain tests, the strain is increased until the structure of the
emulsions is broken and the gel starts to flow, which is shown as a rapid decease of
G′. As can be seen in Figures 3.7 and 3.8, even if the gel is stiffer (high G′) and has a
smaller linear gel region (frequency), it might still flow at lower strains.
The rheology of emulsions is especially important to avoid creaming in nons-
pace filled systems. To arrest creaming of an emulsion drop the elastic modulus of
the surrounding media must exceed the stress exerted by the droplet on the fluid
due to buoyancy. The bouncy force on a droplet will be F = VΔρg (where V is the
volume of the drop 4/3 πR3, Δρ the density difference between the phases, and
g gravity). The resulting stress is the force applied over an area, in this case the sur-
face area of the drop (4πR2), thus the stress exerted by the droplet will be RΔρg/3.
The stress asserted by normal emulsions droplets will thus be normally below
0.1 Pa. Therefore, to arrest creaming, gelling polymers only have to withstand
the stress asserted by the droplet. Furthermore, this also means that to predict the
resistance of the system to creaming, the rheology has to be measured at a low
stress, which can be obtained by constant stress or creep measurements. However,
normally we would like the system to flow when handled; therefore, a good rheo-
logical modifier should yield at higher stresses. In this respect, stress curves are
more informative.
reFerenCes
Abee, T., L. Krockel, and C. Hill. 1995. “Bacteriocins: Modes of action and potentials in food
preservation and control of food poisoning.” International Journal of Food Microbiology
28:169–185.
Agriculture, U.S.D.O. 2014. World Production, Markets, and Trade Reports: Oilseeds.
Washington, DC: United States Department of Agriculture.
Akhtar, M., and E. Dickinson. 2007. “Whey protein–maltodextrin conjugates as emulsi-
fying agents: An alternative to gum arabic.” Food Hydrocolloids 21 (4):607–616.
doi:10.1016/j.foodhyd.2005.07.014.
Akhtar, M., E. Dickinson, J. Mazoyer, and V. Langendorff. 2002. “Emulsion stabilizing prop-
erties of depolymerized pectin.” Food Hydrocolloids 16 (3):249–256. doi:http://dx.doi
.org/10.1016/S0268-005X(01)00095-9.
Alftrén, J., J.M. Peñarrieta, B. Bergenståhl, and L. Nilsson. 2012. “Comparison of molecular and
emulsifying properties of gum arabic and mesquite gum using asymmetrical flow field-flow
fractionation.” Food Hydrocolloids 26 (1):54–62. doi:10.1016/ j.foodhyd.2011.04.008.
Allouche, J., E. Tyrode, V. Sadtler, L. Choplin, and J.L. Salager. 2004. “Simultaneous conductiv-
ity and viscosity measurements as a technique to track emulsion inversion by the phase-
inversion-temperature method.” Langmuir 20 (6):2134–2140. doi:10.1021/la035334r.
Anton, M., V. Beaumal, and G. Gandemer. 2000. “Adsorption at the oil–water interface and
emulsifying properties of native granules from egg yolk: Effect of aggregated state.”
Food Hydrocolloids 14 (4):327–335. doi:10.1016/s0268-005x(00)00009-6.
Aveyard, R., B.P. Binks, and J.H. Clint. 2003. “Emulsions stabilised solely by colloidal par-
ticles.” Advances in Colloid and Interface Science 100–102:503–546.
Awade, A.C. 1996. “On hen egg fractionation: Applications of liquid chromatographyt o the
isolation and the purification of hen egg white and egg yolk proteins.” Z lebensm Unters
Forsch 202:1–14.
Bancroft, W.D. 1913. “The theory of emulsification, V.” Journal of Physical Chemistry
17 (6):501–519.
Bayarri, S., A.J. Taylor, and J. Hort. 2006. “The role of fat in flavor perception: Effect of par-
tition and viscosity in model emulsions.” Journal of Agricultural and Food Chemistry
54:8862−8868.
Benjamins, J., M.H. Vingerhoeds, F.D. Zoet, E.H.A. de Hoog, and G.A. van Aken. 2009.
“Partial coalescence as a tool to control sensory perception of emulsions.” Food
Hydrocolloids 23 (1):102–115. doi:10.1016/j.foodhyd.2007.11.017.
Berasategi, I., M. Garcia-Iniguez de Ciriano, I. Navarro-Blasco, M.I. Calvo, R.Y. Cavero,
I. Astiasaran, and D. Ansorena. 2014. “Reduced-fat bologna sausages with improved
lipid fraction.” Journal of the Science of Food and Agriculture 94 (4):744–751.
doi:10.1002/jsfa.6409.
Binks, B.P. 2002. “Particles as surfactants—Similarities and differences.” Current Opinion in
Colloid & Interface Science 7 (1–2):21–41.
Binks, B.P. 2007. “Colloidal particles at liquid interfaces.” Physical Chemistry Chemical
Physics 9 (48):6298–6299. doi:10.1039/b716587k.
Binks, B.P., A.N. Boa, M.A. Kibble, G. MacKenzie, and A. Rocher. 2011. “Sporopollenin
capsules at fluid interfaces: Particle-stabilised emulsions and liquid marbles.” Soft Matter
7 (8):4017–4024.
Binks, B.P., J.H. Clint, G. Mackenzie, C. Simcock, and C.P. Whitby. 2005. “Naturally occurring
spore particles at planar fluid interfaces and in emulsions.” Langmuir 21 (18):8161–8167.
Bosa, M.A., and T. van Vlieta. 2001. “Interfacial rheological properties of adsorbed protein lay-
ers and surfactants: A review.” Advances in Colloid and Interface Science 91:437–471.
Burt, S. 2004. “Essential oils: Their antibacterial properties and potential applications in
foods—A review.” International Journal of Food Microbiology 94 (3):223–253.
doi:10.1016/j.ijfoodmicro.2004.03.022.
Castro, M.P., A.M. Rojas, C.A. Campos, and L.N. Gerschenson. 2009. “Effect of preservatives,
tween 20, oil content and emulsion structure on the survival of Lactobacillus fructiv-
orans in model salad dressings.” LWT—Food Science and Technology 42 (8):1428–1434.
doi:10.1016/j.lwt.2009.02.021.
Charoen, R., A. Jangchud, K. Jangchud, T. Harnsilawat, E.A. Decker, and D.J. McClements.
2012. “Influence of interfacial composition on oxidative stability of oil-in-water emul-
sions stabilized by biopolymer emulsifiers.” Food Chemistry 131 (4):1340–1346.
doi:10.1016/j.foodchem.2011.09.128.
Charteris, W.P. 1996. “Microbiological quality assurance of edible table spreads in new prod-
uct development.” Journal of the Society of Dairy Technology 49:87–98.
Chen, B., D.J. McClements, and E.A. Decker. 2011. “Minor components in food oils: A
critical review of their roles on lipid oxidation chemistry in bulk oils and emulsions.”
Critical Reviews in Food Science and Nutrition 51 (10):901–916. doi:10.1080/104083
98.2011.606379.
Chevalier, Y., and M.-A. Bolzinger. 2013. “Emulsions stabilized with solid nanoparticles:
Pickering emulsions.” Colloids and Surfaces A: Physicochemical and Engineering
Aspects 439:23–34. doi:10.1016/j.colsurfa.2013.02.054.
Chronakis, I.S. 1997. “Structural–functional and water-holding studies of biopolymers in low
fat content spreads.” Lebensmittel-Wissenschaft und -Technologie 30:36–44.
Chung, C., and D.J. McClements. 2013. “Structure–function relationships in food emul-
sions: Improving food quality and sensory perception.” Food Structure 1:106–126.
doi:10.1016/j.foostr.2013.11.002.
Clarke, E.J., and J. Wiseman. 2000. “Developments in plant breeding for improved nutritional
quality of soya beans I. Protein and amino acid content.” Journal of Agricultural Science
134:111–124.
Corredig, M., and M. Alexander. 2008. “Food emulsions studied by DWS: Recent advances.”
Trends in Food Science and Technology 19 (2):67–75.
Cristofolini, L. 2014. “Synchrotron X-ray techniques for the investigation of structures and
dynamics in interfacial systems.” Current Opinion in Colloid & Interface Science
19 (3):228–241. doi:10.1016/j.cocis.2014.03.006.
Dalgleish, D.G. 2003. Food Emulsions Their Structures and Properties. F.E. Stig, K. Larsson,
and J. Sjöblom, Chapter 1, pp. 1–44.
Dalgleish, D.G. 2011. “On the structural models of bovine casein micelles—Review and pos-
sible improvements.” Soft Matter 7 (6):2265. doi:10.1039/c0sm00806k.
Davies, J.T. 1957. “A quantitative kinetic theory of emulsion type. I. Physical chemistry of
the emulsifying agent.” Gas/liquid and liquid/liquid interfaces. Proceedings of the 2nd
International Congress Surface Activity, Butterworths, London.
De Folter, J.W.J., M.W.M. Van Ruijven, and K.P. Velikov. 2012. “Oil-in-water Pickering emul-
sions stabilized by colloidal particles from the water-insoluble protein zein.” Soft Matter
8 (25):2807–2815.
Degner, B.M., C. Chung, V. Schlegel, R. Hutkins, and D.J. McClements. 2014. “Factors influ-
encing the freeze-thaw stability of emulsion-based foods.” Comprehensive Reviews in
Food Science and Food Safety 13 (2):98–113. doi:10.1111/1541-4337.12050.
Delamarre, S., and C.A. Batt. 1999. “The microbiology and historical safety of margarine.”
Food Microbiology 16:327–333.
Denkov, N.D., S. Tcholakova, and I.B. Ivanov. 2006. “Globular proteins as emulsion
stabilizers—Similaries and differences with surfactants and solid particles.” 4th World
Congress on Emulsions, Lyon, France.
Derkach, S.R. 2009. “Rheology of emulsions.” Advances in Colloid and Interface Science
151 (1–2):1–23. doi:10.1016/j.cis.2009.07.001.
Destribats, M., M. Rouvet, C. Gehin-Delval, C. Schmitt, and B.P. Binks. 2014. “Emulsions
stabilised by whey protein microgel particles: Towards food-grade Pickering emul-
sions.” Soft Matter 10: 6941–6954. doi:10.1039/C4SM00179F.
Dickinson, E. 1992. Introduction to Food Colloids. Oxford: Oxford University Press.
Dickinson, E. 2003. “Hydrocolloids at interfaces and the influence on the properties of dis-
persed systems.” Food Hydrocolloids 17:25–39.
Dickinson, E. 2010. “Food emulsions and foams: Stabilization by particles.” Current Opinion
in Colloid & Interface Science 15 (1–2):40–49. doi:10.1016/j.cocis.2009.11.001.
Dickinson, E. 2012. “Use of nanoparticles and microparticles in the formation and stabilization
of food emulsions.” Trends in Food Science and Technology 24 (1):4–12. doi:10.1016/
j.tifs.2011.09.006.
Dickinson, E. 2013. “Stabilising emulsion-based colloidal structures with mixed food ingredients.”
Journal of the Science of Food and Agriculture 93 (4):710–721. doi:10.1002/jsfa.6013.
Dickinson, E., and E. Davies. 1999. “Influence of ionic calcium on stability of sodium casein-
ate emulsions.” Colloids and Surfaces B-Biointerfaces 12:203–212.
Dorobantu, L.S., A.K.C. Yeung, J.M. Foght, and M.R. Gray. 2004. “Stabilization of oil-water
emulsions by hydrophobic bacteria.” Applied and Environmental Microbiology 70
(10):6333–6336.
Drelich, J., Ch. Fang, and C.L. White. 2002. “Measurement of interfacial tension in fluid-fluid
systems.” In Encyclopedia of Surface and Colloid Science, 3152–3166. Marcel Dekker.
Einstein, A. 1906. “Eine neue Bestimmung der Moleküldimensionen.” Annalen der Physik
324 (2):289–306. doi:10.1002/andp.19063240204.
Eliasson, A.-C. 1986. “On effects of Surface active agents on the gelatinization of starch—a
calorimetric investigation.” Carbohydrate Polymers 6:463–476.
Ercelebi, E.A., and E. Ibanoglu. 2010. “Stability and rheological properties of egg yolk gran-
ule stabilized emulsions with pectin and guar gum.” International Journal of Food
Properties 13 (3):618–630. doi:10.1080/10942910902716984.
Eriksson, M. 2013. “Pickering emulsions based on food grade biological particles.” MSc,
Department of Food Technology, Engineering and Nutrition, Lund University (KLT
920, 2013).
Evans, M., I. Ratcliffe, and P.A. Williams. 2013. “Emulsion stabilisation using polysaccharide–
protein complexes.” Current Opinion in Colloid & Interface Science 18 (4):272–282.
doi:10.1016/j.cocis.2013.04.004.
Frankel, E.N. 1998. Lipid Oxidation. Dundee, Scotland: Oily Press.
Fredrick, E., P. Walstra, and K. Dewettinck. 2010. “Factors governing partial coalescence in
oil-in-water emulsions.” Advances in Colloid and Interface Science 153 (1–2):30–42.
doi:10.1016/j.cis.2009.10.003.
Frelichowska, J., M.A. Bolzinger, and Y. Chevalier. 2010. “Effects of solid particle content
on properties of o/w Pickering emulsions.” Journal of Colloid Interface and Science
351 (2):348–356. doi:10.1016/j.jcis.2010.08.019.
Friberg, S., and I. Wilton. 1970. “Liquid crystals—the formula for emulsions.” American
Perfumer and Cosmetics 85:27–30.
Golding, M., and T.J. Wooster. 2010. “The influence of emulsion structure and stability on
lipid digestion.” Current Opinion in Colloid & Interface Science 15 (1–2):90–101.
doi:10.1016/j.cocis.2009.11.006.
Gould, J., J. Vieira, and B. Wolf. 2013. “Cocoa particles for food emulsion stabilisation.” Food
& Function 4 (9):1369–1375. doi:10.1039/c3fo30181h.
Griffin, W.C. 1954. “Calculation of HLB values of non-ionic surfactants.” Journal of the
Society of Cosmetic Chemists 5 (4):249–256.
Gu, Y.X., Y.H. Huang, B. Liao, G.M. Cong, and M. Xu. 2000. “Studies on the characterization
of phase inversion during emulsification process and the particle sizes of water-borne
microemulsion of poly(phenylene oxide) ionomer.” Journal of Applied Polymer Science
76 (5):690–694. doi:10.1002/(sici)1097-4628(20000502)76:5<690::aid-app11>3.0.co;2-q.
Gupta, R., and D. Rousseau. 2012. “Surface-active solid lipid nanoparticles as Pickering
stabilizers for oil-in-water emulsions.” Food & Function 3 (3):302–311. doi:10.1039/
c2fo10203j.
Hagel, L. 2011. “Gel filtration: Size exclusion chromatography.” In Protein Purification,
51–91. John Wiley & Sons.
Harnsilawat, T., R. Pongsawatmanit, and D.J. McClements. 2006. “Stabilization of model

beverage cloud emulsions using protein−polysaccharide electrostatic complexes
formed at the oil−water interface.” Journal of Agricultural and Food Chemistry
54:5540−5547.
Harvey, D.J. 2011. “Analysis of carbohydrates and glycoconjugates by matrix-assisted laser
desorption/ionization mass spectrometry: An update for the period 2005–2006.” Mass
Spectrometry Reviews 30 (1):1–100. doi:10.1002/mas.20265.
Hashemi, M.M., M. Aminlari, and M. Moosavinasab. 2014. “Preparation of and studies on the
functional properties and bactericidal activity of the lysozyme–xanthan gum conjugate.”
LWT—Food Science and Technology 57 (2):594–602. doi:10.1016/j.lwt.2014.01.040.
Heertje, I. 2014. “Structure and function of food products: A review.” Food Structure 1 (1):
3–23. doi:10.1016/j.foostr.2013.06.001.
Horne, D.S. 2002. “Casein structure, self-assembly and gelation.” Current Opinion in
Colloid & Interface Science (7):456–461.
http://ndb.nal.usda.gov/ndb/foods.
http://www.engineeringtoolbox.com/oil-melting-points-d_1088.html.
Hyldgaard, M., D.S. Sutherland, M. Sundh, T. Mygind, and R.L. Meyer. 2012. “Antimicrobial
mechanism of monocaprylate.” Applied and Environmental Microbiology 78
(8):2957–2965. doi:10.1128/AEM.07224-11.
Israelachvili, J.N. 1985. Intermolecular and Surface Forces: With Applications to Colloidal
and Biological Systems. London: Academic Press.
Israelachvili, J., D.J. Mitchell, and B. Ninham. 1976. “Theory of selfassembly of hydrocar-
bon amphiphiles into micelles and bilayers.” Journal of the Chemical Society, Faraday
Transactions 2 72:1525–1568.
Jin, H., W. Zhou, J. Cao, S.D. Stoyanov, T.B.J. Blijdenstein, P.W.N. De Groot, L.N. Arnaudov,
and E.G. Pelan. 2012. “Super stable foams stabilized by colloidal ethyl cellulose par-
ticles.” Soft Matter 8 (7):2194–2205. doi:10.1039/c1sm06518a.
Kalashnikova, I., H. Bizot, P. Bertoncini, B. Cathala, and I. Capron. 2013. “Cellulosic nanorods
of various aspect ratios for oil in water Pickering emulsions.” Soft Matter 9 (3):952–959.
doi:10.1039/c2sm26472b.
Kalashnikova, I., H. Bizot, B. Cathala, and I. Capron. 2011. “New Pickering emulsions stabi-
lized by bacterial cellulose nanocrystals.” Langmuir 27 (12):7471–7479. doi:10.1021/
la200971f.
Kargar, M., K. Fayazmanesh, M. Alavi, F. Spyropoulos, and I.T. Norton. 2012. “Investigation
into the potential ability of Pickering emulsions (food-grade particles) to enhance the
oxidative stability of oil-in-water emulsions.” Journal of Colloid and Interface Science
366 (1):209–215. doi:10.1016/j.jcis.2011.09.073.
Karlsson, C.A.-C., M.C. Wahlgren, and A.C. Trägårdh. 1996. “Time and temperature aspects
of β-lactoglobulin removal from methylated silica surfaces by sodium dodecyl sulphate.”
Colloids and Surfaces B: Biointerfaces 6 (4):317–328.
Kasapis, S. 2000. “Novel uses of biopolymers in the development of low fat spreads and
soft cheeses.” In Novel Macromolecules in Food Systems, edited by G. Doxastakis and
V. Kiosseoglou. Elsevier Science, B.V.
Kikuzaki, H., M. Hisamoto, K. Hirose, K. Akiyama, and H. Taniguchi. 2002. “Antioxidant
properties of ferulic acid and its related compounds.” Journal of Agricultural and Food
Chemistry 50:2161−2168.
Kinsella, J.E., and D.M. Whitehead. 1989. “Proteins in whey: Chemical, physical, and func-
tional properties.” In Advances in Food and Nutrition Research, edited by E. Kinsella
John, 343–438. San Diego, CA: Academic Press.
Koupantsis, T., E. Pavlidou, and A. Paraskevopoulou. 2014. “Flavour encapsulation in
milk proteins—CMC coacervate-type complexes.” Food Hydrocolloids 37:134–142.
doi:10.1016/j.foodhyd.2013.10.031.
Krafft, F., and H. Wiglow. 1895. “Ueber das Verhalten der fettsauren Alkalien und der Seifen
in Gegenwart von Wasser, III. Die Seifen als Krystalloide.” Berichte der deutschen
chenischen Gesellschaft 28:2566–2573.
Krieger, I.M. 1972. “Rheology of monodisperse lattices.” Advances in Colloid and Interface
Science 3:111.
Krog, N. 1997. “Food emulsifiers.” In Food Emulsions, edited by S. Friberg and K. Larsson,
141–188. New York: Marcel Dekker.
Laca, A., M.C. Saenz, B. Paredes, and M. Diaz. 2010. “Rheological properties, stability and sensory
evaluation of low-cholesterol mayonnaises prepared using egg yolk granules as emulsifying
agent.” Journal of Food Engineering 97 (2):243–252. doi:10.1016/ j.jfoodeng.2009.10.017.
Laredj-Bourezg, F., Y. Chevalier, O. Boyron, and M.A. Bolzinger. 2012. “Emulsions sta-
bilized with organic solid particles.” Colloids and Surfaces A: Physicochemical and
Engineering Aspects 413:252–259. doi:10.1016/j.colsurfa.2011.12.064.
Larsson, K. 1986. “Physical properties—Structural and physical characteristics.” In The Lipid
Handbook, edited by F.D. Gunstone, J.L. Harwood, and F.B. Padley, 321–384. London:
Chapman & Hall.
Le Révérend, B.J.D., I.T. Norton, P.W. Cox, and F.Spyropoulos. 2010. “Colloidal aspects of
eating.” Current Opinion in Colloid & Interface Science 15 (1–2):84–89. doi:10.1016/
j.cocis.2009.11.009.
Li, C., Y. Li, P. Sun, and C. Yang. 2013. “Pickering emulsions stabilized by native starch gran-
ules.” Colloids and Surfaces A: Physicochemical and Engineering Aspects 431:142–149.
doi:10.1016/j.colsurfa.2013.04.025.
Li, C., P. Sun, and C. Yang. 2012. “Emulsion stabilized by starch nanocrystals.” Starch−Stärke
64 (6):497–502. doi:10.1002/star.201100178.
Löf, D., K. Schillén, and L. Nilsson. 2011. “Flavonoids: Precipitation kinetics and
interaction with surfactant micelles.” Journal of Food Science 76 (3):N35–N39.
doi:10.1111/j.1750-3841.2011.02103.x.
Lück, E. 1990. “Food applications of sorbic acid and its salts.” Food Additives & Contaminants
7 (5):711–715. doi:10.1080/02652039009373936.
Lundqvist, H., A.-C. Eliasson, and G. Olofsson. 2002. “Binding of hexadecyltrimethylam-
monium bromide to starch polysaccharides. Part I. Surface tension measurements.”
Carbohydrate Polymers 49 (1):43–55.
Luo, Z., B.S. Murray, A.L. Ross, M.J.W. Povey, M.R.A. Morgan, and A.J. Day. 2012. “Effects
of pH on the ability of flavonoids to act as Pickering emulsion stabilizers.” Colloids and
Surfaces B: Biointerfaces 92:84–90. doi:10.1016/j.colsurfb.2011.11.027.
Luo, Z.J., B.S. Murray, A. Yusoff, M.R.A. Morgan, M.J.W. Povey, and A.J. Day. 2011. “Particle-
stabilizing effects of flavonoids at the oil-water interface.” Journal of Agricultural and
Food Chemistry 59 (6):2636–2645. doi:10.1021/jf1041855.
Magnusson, E., and L. Nilsson. 2011. “Interactions between hydrophobically modified starch
and egg yolk proteins in solution and emulsions.” Food Hydrocolloids 25 (4):764–772.
doi:10.1016/j.foodhyd.2010.09.006.
Magnusson, E., and L. Nilsson. 2013. “Emulsifying properties of egg yolk.” In Eggs: Nutrition,
Consumption and Health, 69–85.
Magnusson, E., C. Rosén, and L. Nilsson. 2011. “Freeze–thaw stability of mayonnaise
type oil-in-water emulsions.” Food Hydrocolloids 25 (4):707–715. doi:http://dx.doi.
org/10.1016/j.foodhyd.2010.08.024.
Maldonado-Valderrama, J., and J.M.R. Patino. 2010. “Interfacial rheology of protein–surfactant
mixtures.” Current Opinion in Colloid & Interface Science 15 (4):271–282. doi:10.1016/
j.cocis.2009.12.004.
Malone, M.E., I.A.M. Appelqvist, and I.T. Norton. 2003. “Oral behaviour of food hydrocolloids
and emulsions. Part 1. Lubrication and deposition considerations.” Food Hydrocolloids
17 (6):763–773. doi:10.1016/s0268-005x(03)00097-3.
Marefati, A., M. Rayner, A. Timgren, P. Dejmek, and M. Sjöö. 2013. “Freezing and freeze-
drying of Pickering emulsions stabilized by starch granules.” Colloids and Surfaces A:
Physicochemical and Engineering Aspects 436:512–520. doi:http://dx.doi.org/10.1016/j.
colsurfa.2013.07.015.
Marku, D., M. Wahlgren, M. Rayner, M. Sjöö, and A. Timgren. 2012. “Characterization
of starch Pickering emulsions for potential applications in topical formulations.”
International Journal of Pharmaceutics 428:1–7.
Mayer, S., J. Weiss, and D.J. McClements. 2013. “Behavior of vitamin E acetate delivery sys-
tems under simulated gastrointestinal conditions: Lipid digestion and bioaccessibility
of low-energy nanoemulsions.” Journal of Colloid and Interface Science 404:215–222.
doi:10.1016/j.jcis.2013.04.048.
McClements, D.J. 2002. “Colloidal basis of emulsion color.” Current Opinion in Colloid &
Interface Science 7 (5–6):451–455. doi:10.1016/s1359-0294(02)00075-4.
McClements, D.J. 2005a. “Characterization of emulsions.” In Food Emulsions: Principles,
Practices, and Techniques. Boca Raton, FL: CRC Press.
McClements, D.J. 2005b. Food Emulsions: Principles, Practices, and Techniques, edited by F.M.
Clydesdale, CRC Press Series in Contemporary Food Science. Boca Raton, FL: CRC Press.
McClements, D.J. 2007. “Critical review of techniques and methodologies for characterization
of emulsion stability.” Critical Reviews in Food Science and Nutrition 47 (7):611–649.
doi:10.1080/10408390701289292.
McClements, D.J., and E.A. Decker. 2000. “Lipid oxidation in oil-in-water emulsions: Impact
of molecular environment on chemical reactions in heterogeneous food systems.”
Journal of Food Science 65 (8):1270–1282. doi:10.1111/j.1365-2621.2000.tb10596.x.
Medina, I., I. Undeland, K. Larsson, I. Storrø, T. Rustad, C. Jacobsen, V. Kristinová, and
J.M. Gallardo. 2012. “Activity of caffeic acid in different fish lipid matrices: A review.”
Food Chemistry 131 (3):730–740. doi:10.1016/j.foodchem.2011.09.032.
Mengual, O., G. Meunier, I. Cayré, K. Puech, and P. Snabre. 1999. “TURBISCAN MA
2000: Multiple light scattering measurement for concentrated emulsion and suspen-
sion instability analysis.” Talanta 50 (2):445–456. doi:http://dx.doi.org/10.1016/
S0039-9140(99)00129-0.
Michalski, M.C. 2009. “Specific molecular and colloidal structures of milk fat affecting
lipolysis, absorption and postprandial lipemia.” European Journal of Lipid Science and
Technology 111:413–431.
Modig, G., L. Nilsson, B. Bergenståhl, and K.G. Wahlund. 2006. “Homogenization-induced deg-
radation of hydrophobically modified starch determined by asymmetrical flow field-flow
fractionation and multi-angle light scattering.” Food Hydrocolloids 20 (7):1087–1095.
Moore, R.L., S.E. Duncan, A.S. Rasor, W.N. Eigel, and S.F. O’Keefe. 2012. “Oxidative
stability of an extended shelf-life dairy-based beverage system designed to contribute to
heart health.” Journal of Dairy Science 95 (11):6242–6251. doi:10.3168/jds.2012-5364.
Ng, S.P., O.M. Lai, F. Abas, H.K. Lim, B.K. Beh, T.C. Ling, and C.P. Tan. 2014. “Compositional
and thermal characteristics of palm olein-based diacylglycerol in blends with palm super
olein.” Food Research International 55:62–69. doi:10.1016/j.foodres.2013.10.035.
Nilsson, L. 2013. “Separation and characterization of food macromolecules using field-flow frac-
tionation: A review.” Food Hydrocolloids 30 (1):1–11. doi:10.1016/ j.foodhyd.2012.04.007.
Nilsson, L., and B. Bergenståhl. 2006. “Adsorption of hydrophobically modified starch at oil/
water interfaces during emulsification.” Langmuir 22:8770–8776.
Nilsson, L., and B. Bergenståhl. 2007. “Adsorption of hydrophobically modified anionic starch
at oppositely charged oil/water interfaces.” Journal of Colloid and Interface Science
308 (2):508–513. doi:10.1016/j.jcis.2007.01.024.
Nilsson, L., M. Leeman, K.G. Wahlund, and B. Bergenståhl. 2007. “Competitive adsorption
of a polydisperse polymer during emulsification: Experiments and modeling.” Langmuir
23:2346–2351.
Nilsson, L., P. Osmark, C. Fernandes, and B. Bergenståhl. 2006. “Competitive adsorption

of water soluble plasma proteins from egg yolk at the oil/water interface.” Journal of
Agricultural and Food Chemistry 54 (18):6881–6887.
Nilsson, L., P. Osmark, C. Fernandes, and B. Bergenståhl. 2007. “Competitive adsorption of
proteins from total hen egg yolk during emulsification.” Journal of Agricultural and
Food Chemistry 55:6746−6753.
Nylander, T., T. Arnebrant, M. Bos, and P. Wilde. 2008. “Protein/emulsifier interactions.” In
Food Emulsifiers and Their Applications, edited by G.L. Hasenhuettl and R.W. Hartel.
New York: Springer Science + Business Media.
Paunov, V.N., O.J. Cayre, P.F. Noble, S.D. Stoyanov, K.P. Velikov, and M. Golding. 2007.
“Emulsions stabilised by food colloid particles: Role of particle adsorption and wettabil-
ity at the liquid interface.” Journal of Colloid and Interface Science 312 (2):381–389.
doi:10.1016/j.jcis.2007.03.031.
Pelipenko, J., J. Kristl, R. Rosic, S. Baumgartner, and P. Kocbek. 2012. “Interfacial rheology:
An overview of measuring techniques and its role in dispersions and electrospinning.”
Acta Pharmaceutica 62 (2):123–140. doi:10.2478/v10007-012-0018-x.
Rayner, M., D. Marku, M. Eriksson, M. Sjöö, P. Dejmek, and M. Wahlgren. 2014. “Biomass-
based particles for the formulation of Pickering type emulsions in food and topical
applications.” Colloids and Surfaces A: Physicochemical and Engineering Aspects.
458:48–62. doi:http://dx.doi.org/10.1016/j.colsurfa.2014.03.053.
Rayner, M., M. Sjöö, A. Timgren, and P. Dejmek. 2012. “Quinoa starch granules as sta-
bilizing particles for production of Pickering emulsions.” Faraday Discussions
158:139–155.
Rayner, M., A. Timgren, M. Sjöö, and P. Dejmek. 2012. “Quinoa starch granules: A candidate
for stabilising food-grade Pickering emulsions.” Journal of the Science of Food and
Agriculture 92 (9):1841–1847. doi:10.1002/jsfa.5610.
Rodríguez, P.J.M., J.M.N. García, and M.R.R. Niño. 2001. “Protein–lipid interactions at the
oil–water interface.” Colloids and Surfaces B: Biointerfaces 21:207–216.
Roesch, R.R., and M. Corredig. 2003. “Texture and microstructure of emulsions prepared
with soy protein concentrate by high-pressure homogenization.” LWT—Food Science
and Technology 36 (1):113–124. doi:10.1016/s0023-6438(02)00208-6.
Sato, K. 2001. “Crystallization behaviour of fats and lipids: a review.” Chemical Engineering
Science 56:2255–2265.
Sato, K., L. Bayés-García, T. Calvet, M.À. Cuevas-Diarte, and S. Ueno. 2013. “External fac-
tors affecting polymorphic crystallization of lipids.” European Journal of Lipid Science
and Technology 115 (11):1224–1238. doi:10.1002/ejlt.201300049.
Sato, K., and S. Ueno. 2011. “Crystallization, transformation and microstructures of
polymorphic fats in colloidal dispersion states.” Current Opinion in Colloid & Interface
Science 16 (5):384–390. doi:10.1016/j.cocis.2011.06.004.
Schuh, V., K. Allard, K. Herrmann, M. Gibis, R. Kohlus, and J. Weiss. 2013. “Impact of
carboxymethyl cellulose (CMC) and microcrystalline cellulose (MCC) on functional
characteristics of emulsified sausages.” Meat Science 93 (2):240–247. doi:10.1016/
j.meatsci.2012.08.025.
Sherman, P. 1995. “A critique of some methods proposed for evaluating the emulsifying
capacity and emulsion stabilizing performance of vegetable proteins.” Italian Journal of
Food Science 7 (1):3–10.
Shimoni, G., C. Shani Levi, S. Levi Tal, and U. Lesmes. 2013. “Emulsions stabilization by
lactoferrin nano-particles under invitro digestion conditions.” Food Hydrocolloids
33 (2):264–272. doi:10.1016/j.foodhyd.2013.03.017.
Shinoda, K., and K. Sato. 1969. “The stability of O/W type emulsions as functions of tempera-
ture and the HLB of emulsifiers: The emulsification by PIT-method.” Journal of Colloid
and Interface Science 30:258–263.
Song, X., Y. Pei, W. Zhu, D. Fu, and H. Ren. 2014. “Particle-stabilizers modified from indica
rice starches differing in amylose content.” Food Chemistry 153:74–80. doi:http://
dx.doi.org/10.1016/j.foodchem.2013.12.046.
Sørensen, A.D.M., A.M. Haahr, E.M. Becker, L.H. Skibsted, B. Bergenståhl, L. Nilsson,
and C. Jacobsen. 2008. “Interactions between iron, phenolic compounds, emulsifiers,
and pH in omega-3-enriched oil-in-water emulsions.” Journal of Agricultural and Food
Chemistry 56 (5):1740–1750.
Stephen, A.M., G.O. Phillips, and P.A. Williams, eds. 2006. Food Polysaccharides and Their
Applications. Boca Raton, FL: CRC Press.
Stokes, J.R., M.W. Boehm, and S.K. Baier. 2013. “Oral processing, texture and mouthfeel:
From rheology to tribology and beyond.” Current Opinion in Colloid & Interface
Science 18 (4):349–359. doi:10.1016/j.cocis.2013.04.010.
Surel, C., J. Foucquier, N. Perrot, A. Mackie, C. Garnier, A. Riaublanc, and M. Anton. 2014.
“Composition and structure of interface impacts texture of O/W emulsions.” Food
Hydrocolloids 34:3–9. doi:10.1016/j.foodhyd.2013.06.016.
Swaisgood, H.E. 1993. “Review and update of casein chemistry.” Journal of Dairy Science 76
(10):3054–3061. doi:http://dx.doi.org/10.3168/jds.S0022-0302(93)77645-6.
Tabilo-Munizaga, G., and G.V. Barbosa-Cánovas. 2005. “Rheology for the food industry.”
Journal of Food Engineering 67 (1–2):147–156. doi:10.1016/j.jfoodeng.2004.05.062.
Tadros, T. 2004. “Application of rheology for assessment and prediction of the long-term
physical stability of emulsions.” Advances in Colloid and Interface Science 108–109:
227–258. doi:10.1016/j.cis.2003.10.025.
Tan, Y., K. Xu, C. Liu, Y. Li, C. Lu, and P. Wang. 2012. “Fabrication of starch-based nano-
spheres to stabilize Pickering emulsion.” Carbohydrate Polymers 88:1358–1363.
Tcholakova, S., N.D. Denkov, I.B. Ivanov, and B. Campbell. 2002. “Coalescence in
β-lactoglobulin-stabilized emulsions: Effects of protein adsorption and drop size.”
Langmuir 18 (23):8960–8971. doi:10.1021/la0258188.
Tcholakova, S., N.D. Denkov, I.B. Ivanov, and B. Campbell. 2006. “Coalescence stability
of emulsions containing globular milk proteins.” Advances in Colloid and Interface
Science 123–126:259–293. doi:10.1016/j.cis.2006.05.021.
Tcholakova, S., N.D. Denkov, and A. Lips. 2008. “Comparison of solid particles, globular
proteins and surfactants as emulsifiers.” Physical Chemistry Chemical Physics 10 (12):
1608–1627. doi:10.1039/b715933c.
Tcholakova, S., N.D. Denkov, D. Sidzhakova, I.B. Ivanov, and B. Campbell. 2003.
“Interrelation between drop size and protein adsorption at various emulsification condi-
tions.” Langmuir 19 (14):5640–5649. doi:10.1021/la034411f.
Timgren, A., M. Rayner, P. Dejmek, D. Marku, and M. Sjöö. 2013. “Emulsion stabilizing
capacity of intact starch granules modified by heat treatment or octenyl succinic anhy-
dride.” Food Science & Nutrition 1 (2):157–171. doi:10.1002/fsn3.17.
Timgren, A., M. Rayner, M. Sjöö, and P. Dejmek. 2011. “Starch particles for food based
Pickering emulsions.” Procedia Food Science 1:95–103.
Ting, Y., Y. Jiang, C.-T. Ho, and Q. Huang. 2014. “Common delivery systems for enhancing
in vivo bioavailability and biological efficacy of nutraceuticals.” Journal of Functional
Foods 7:112–128. doi:10.1016/j.jff.2013.12.010.
Tufvesson, F., M. Wahlgren, and A.C. Eliasson. 2003. “Formation of amylose–lipid complexes
and effects of temperature treatment. Part 1. Monoglycerides.” Starch - Stärke 55 (2):61–71.
Tzoumaki, M.V., T. Moschakis, V. Kiosseoglou, and C.G. Biliaderis. 2011. “Oil-in-water emul-
sions stabilized by chitin nanocrystal particles.” Food Hydrocolloids 25 (6):1521–1529.
Tzoumaki, M.V., T. Moschakis, E. Scholten, and C.G. Biliaderis. 2013. “In vitro lipid diges-
tion of chitin nanocrystal stabilized o/w emulsions.” Food & Function 4 (1):121–129.
Verwey, E.J.W., and J.Th.G. Overbeek. 1948. Theory of the Stability of Lyophobic Colloids.
Amsterdam, the Netherlands: Elsevier.
Wahlgren, M. 1995. “Removal of T4 lysozyme from silicon oxide surfaces by sodium dodecyl
sulfate.” Surface and Colloid Science Symposium, Lund, Sweden, October 18–19.
Wahlgren, M., and T. Arnebrant. 1991. “Protein adsorption to solid surfaces.” Trends in
Biotechnology 9 (1):201–208.
Wahlgren, M., and T. Arnebrant. 1992. “The concentration dependence of adsorption from a
mixture of b-lactoglobulin and sodium dodecyl sulfate onto methylated silica surfaces.”
Journal of Colloid and Interface Science 148:201–206.
Walstra, P. 2005. “8 Emulsions.” In Fundamentals of Interface and Colloid Science, edited by
J. Lyklema, 1–94. Academic Press.
Waninge, R., P. Walstra, J. Bastiaans, H. Nieuwenhuijse, T. Nylander, M. Paulsson, and
B. Bergenståhl. 2005. “Competitive adsorption between β-casein or β-lactoglobulin and
model milk membrane lipids at oil-water interface.” Journal of Agricultural and Food
Chemistry 53 (3):716–724.
Waraho, T., D.J. McClements, and EA. Decker. 2011. “Mechanisms of lipid oxidation in food disper-
sions.” Trends in Food Science and Technology 22 (1):3–13. doi:10.1016/ j.tifs.2010.11.003.
Watanabe, A., I. Tashima, V. Matsuzaki, J. Kurashige, and K. Sato. 1992. “On the formation
of granular crystals in fat blends containing palm oil.” Journal of the American Oil
Chemists’ Society 69:1077–1080.
Wilde, P., A. Mackie, F. Husband, P. Gunning, and V. Morris. 2004. “Proteins and emulsi-
fiers at liquid interfaces.” Advances in Colloid and Interface Science 108–109:63–71.
doi:10.1016/j.cis.2003.10.011.
Williams, P.A., G.O. Phillips, and R.C. Randall. 1990. “Structure-function relationships of
gum arabic.” In Gums and Stabilisers for the Food Industry, edited by G.O. Phillips,
D.J. Wedlock, and P.A. Williams, 25. Oxford: IRL.
Wongkongkatep, P., K. Manopwisedjaroen, P. Tiposoth, S. Archakunakorn, T. Pongtharangkul,
M. Suphantharika, K. Honda, I. Hamachi, and J. Wongkongkatep. 2012. “Bacteria
interface Pickering emulsions stabilized by self-assembled bacteria-chitosan network.”
Langmuir 28 (13):5729–5736. doi:10.1021/la300660x.
Wu, B., B. Degner, and D.J. McClements. 2013. “Microstructure & rheology of mixed colloidal
dispersions: Influence of pH-induced droplet aggregation on starch granule–fat droplet mix-
tures.” Journal of Food Engineering 116 (2):462–471. doi:10.1016/ j.jfoodeng.2012.12.020.
Xu, W., W. Jin, C. Zhang, Z. Li, L. Lin, Q. Huang, S. Ye, and B. Li. 2014. “Curcumin loaded
and protective system based on complex of κ-carrageenan and lysozyme.” Food
Research International 59:61–66. doi:10.1016/j.foodres.2014.01.059.
Yang, Y., and D.J. McClements. 2013. “Vitamin E bioaccessibility: Influence of carrier oil
type on digestion and release of emulsified alpha-tocopherol acetate.” Food Chemistry
141 (1):473–481. doi:10.1016/j.foodchem.2013.03.033.
Youssef, M.K., and S. Barbut. 2011. “Effects of two types of soy protein isolates, native and
preheated whey protein isolates on emulsified meat batters prepared at different protein
levels.” Meat Science 87 (1):54–60. doi:10.1016/j.meatsci.2010.09.002.
Yusoff, A., and B.S. Murray. 2011. “Modified starch granules as particle-stabilizers of oil-in-
water emulsions.” Food Hydrocolloids 25 (1):42–55. doi:10.1016/j.foodhyd.2010.05.004.
Zapico, P., M. de Paz, M. Medina, and M. Nuñez. 1999. “The effect of homogenization of
whole milk, skim milk and milk fat on nisin activity against Listeria innocua.” Inter-
national Journal of Food Microbiology 46: 151–157.
Zhang, B., Y.J. Chi, and B. Li. 2013. “Effect of ultrasound treatment on the wet heating
Maillard reaction between β-conglycinin and maltodextrin and on the emulsifying
properties of conjugates.” European Food Research and Technology 238 (1):129–138.
doi:10.1007/s00217-013-2082-y.
4 Particle-Stabilized
Emulsions
CONTeNTs
4.1 Introduction .................................................................................................. 101
4.2 Particles as Emulsifiers ................................................................................. 102
4.3 Stability of Particle-Stabilized Emulsions .................................................... 106
4.3.1 Coalescence ...................................................................................... 106
4.3.2 Ostwald Ripening ............................................................................. 108
4.3.3 Creaming or Sedimentation .............................................................. 109
4.4 Structure and Rheology of Pickering Emulsions ......................................... 109
4.5 Influence of Other Components in the System ............................................. 112
4.6 Food-Grade Particles for Emulsion Stabilization ......................................... 113
4.7 Additional Functional Properties of Particle-Stabilized Emulsions............. 118
References .............................................................................................................. 118
ABSTRACT This chapter describes particle-stabilized emulsions, the so-called

Pickering emulsions. The principles behind particle stabilization of emulsions are
described. The key properties of particle-stabilized emulsions such as stability, rhe-
ology, and the influences of other ingredients on the system, as well as some addi-
tional functional properties imparted by the stabilized particles in food emulsion
formulations, are highlighted.
4.1 INTRODUCTION
One specific type of emulsion that has received increasing attention, especially
since the late twentieth century, is the particle-stabilized emulsion. This inter-
facial phenomenon was initially described by Walter Ramsden (1903) in the first
known publication on the topic. However, this type of emulsion is generally referred
to as Pickering emulsion, named after the second scientist to publish on the topic,
S.U. Pickering (1907). Both Ramsden and Pickering independently observed that
solid particles were able to stabilize the interface between two immiscible phases.
More papers were published on Pickering emulsion during the first decade of the
twenty-first century than during the entire twentieth century. The renewed interest
in Pickering emulsions can be related to the fact that these emulsions have a very
high long-term stability toward coalescence and Ostwald ripening. The ability to
101
use different types of particles is another explanation. A wide variety of particles

have been used to stabilize emulsions, as reported in the literature. From the early
use of mainly colloidal particles of clay or copper sulfate to custom-made particles
of silica and latex with specific surface treatments, there has been increasing interest
in natural and food-grade particles (see Section 4.6). Rapid developments within the
field of nanotechnology for the manufacturing and characterization of nanoparticles
have also widened the type and the number of particles that can be used for Pickering
emulsions. The variety of particles to be used enables the formulation of emulsions
with tailor-made properties and functionalities.
4.2 PARTICLes As eMULsIFIeRs

Finkle, Draper, and Hildebrand (1923) postulated that particles can stabilize emul-
sions only if they are wetted by both liquids. The adsorbed particles will then form a
steric (and in relevant cases electrostatic) barrier between the oil droplets, stabilizing
the emulsion against coalescence. The adsorption of particles does not give rise to a
reduction in the interfacial tension as such; thus, this is not a stabilizing mechanism
for Pickering emulsions (Vignati, Piazza, and Lockhart 2003). The long-term stabil-
ity of emulsions will be discussed in Section 4.3.
The detachment energy (Equation 4.1) of a particle adsorbed at the oil–water
interface depends on the contact angle, θ; the oil–water interfacial tension; and the
size of the particle (Levine, Bowen, and Partridge 1989a, 1989b). Once adsorbed,
particles are held in a very deep energy minimum when the contact angle is close
to 90° or the particle is larger than a few tens of nanometers. Figure 4.1 shows the
particle size needed to obtain the energy of detachment for various contact angles.
As the radius has such a pronounced effect on the detachment energy, a radius in the
100,000
105 kT
103 kT
10,000 102 kT
Particle radius (nm)
1,000
100
10
1
0 30 60 90 120 150 180
Contact angle, θ (°)
FIGURe 4.1 Contour plot of equal detachment energy from a triglyceride–water interface
for various combinations of particle radii and contact angles.
Particle-Stabilized Emulsions 103
range of a couple of 100 nm is enough to give a detachment energy of 1000 kT at

contact angles as low as 10°. Thus, a particle does not have to be very hydrophobic
to adsorb strongly at the oil–water interface as long as it is in the size range above
100 nm. Equation 4.1 assumes that the local interface relative to the particle is flat.
The forces at the curved emulsion surface are, of course, more complex, as described
by Levine, Bowen, and Partridge (1989a, 1989b). However, Binks and Lumsdon
(2000) have shown experimentally that for nanoparticles in the range of 10–30 nm,
the stability of emulsions follows the expected pattern of having high stability at
intermediate hydrophobicity and low stability when they are either very hydrophilic
or very hydrophobic.
E = r 2πγ(1 − cos θ)2 (4.1)
Particles can stabilize both oil-in-water (O/W) and water-in-oil (W/O) emulsions
and there are also examples where particles have been used to stabilize at least one
of the phases in double emulsions (Matos et al. 2013; Spyropoulos, Frasch-Melnik,
and Norton 2011). As is the case with many emulsions systems, phase inversion may
occur when the fraction of the continuous phase is decreased in the system beyond
a certain point. For example, as shown by Binks and Lumsdon (2000), the fraction
of water where catastrophic phase inversion from W/O to O/W occurs is strongly
dependent on the hydrophobicity of the particles, and the fraction required for this
inversion increases with increasing particle hydrophobicity. This is in line with the
theoretical assumption that the contact angle and thus the hydrophobicity will be
important for what type of emulsion is formed at a water–particle contact angle
Water Oil
Oil Water
Water Oil
Oil Water
drop drop
(a) (b)
FIGURe 4.2 The location of a particle at the interface between oil and water is determined
by the contact angle, θ, measured through the water phase. This also determines the type of
emulsion formed, (a) θ < 90° will favor O/W emulsions, whereas (b) θ > 90° will favor W/O
emulsions. (Redrawn from Rayner, M. et al., J. Sci. Food Agric., 92, 9, 1841–1847, 2012.)
as measured through the water phase. A contact angle greater than 90° will favor
W/O emulsions, whereas a contact angle lesser than 90° will favor O/W emulsions,
as shown schematically in Figure 4.2. A high energy of desorption for larger par-
ticles means that particles far from a contact angle of 90° will also stabilize emul-
sions, providing a possibility to produce stable emulsions at a very high fraction of
the internal phase (Capron and Cathala 2013; Marku et al. 2012; Midmore 1998).
Such emulsions normally require a high amount of traditional surfactants, but, for
example, Capron and Cathala (2013) have shown that low amounts, less than 0.1% of
cellulose nanoparticles, could also stabilize O/W emulsions containing a dispersed
phase of 92%.
The droplet size of a Pickering emulsion will be governed by either the ratio
between the amount of particles and the dispersed phase volume or the energy avail-
able for emulsification. This means that above a critical concentration, where there
are enough particles to cover the dispersed phase, the droplet size of the emulsion
will decrease linearly with increasing particle-to-dispersed phase ratio, until a pla-
teau is reached where no further reduction in size can be obtained, due to the limita-
tion of the equipment used (Frelichowska, Bolzinger, and Chevalier 2010; Li, Sun,
and Yang 2012; Marku et al. 2012; Rayner, Sjöö et al. 2012; Timgren et al. 2011).
In the linear region, it is possible to theoretically estimate the amount of particles
needed to obtain a specific particle-stabilized emulsion droplet size. To do this, the
following assumptions are made: (1) that there is a strong preferential adsorption of
the particles into an ordered monolayer at the surface of the dispersed phase drop-
lets and (2) that the surface mean radius of the particle and droplet size are good
representatives of the system. These calculations were performed for quinoa starch
granules using a radius of 1.8 µm assuming a hexagonal close packing at the surface
using Equation 4.2 (see Figure 4.3).
250
200
Starch coverage (%)
150
100
50
0
0 200 400 600 800 1000 1200
Starch/oil ratio (mg/ml)
FIGURe 4.3 Starch coverage versus particle concentration for a Pickering emulsion stabi-
lized with quinoa starch particles of 1.8 µm and calculated assuming hexagonal close packing.
d32p
mp = 4Voρpϕ (4.2)
d32d
where:
mp is the mass of particles (kg)
Vo is the volume of the dispersed phase (m3)
ρp is the particle density (kg/m3)
φ is the packing density (i.e., the assumed hexagonal close packing φ ≈ 0.907)
d32p and d32d are the diameters of the particle and emulsion drop, respectively, (µm)
The theoretical maximum coverage, ΓM (mg/m2), could be calculated using Equation 4.3.
2
Γ M = ρp d pϕ ⋅ 106 (4.3)
3
Although illustrative, there is no perfect correlation between the measured particle
size and the theoretical plotted in Figure 4.4, as data points lie both above and below
the theoretical line. Furthermore, if one calculates the actual surface coverage of
the droplets, it decreases with decreasing particle/oil ratios and is well below what
would be expected for a closed packed layer for low ratios. There has been an ongo-
ing discussion on whether a closed packed layer is necessary to stabilize Pickering
emulsions. It has also been shown that a surface coverage well below close packing
gives rise to stable Pickering emulsions that do not change drop size either due to
storage (8 weeks to several years) or upon mild centrifugation (Marku et al. 2012).
Horozov and Binks (2006) have suggested that the emulsions could be stabilized
by bridging particle monolayers at concentrations below the full surface coverage.
1
Measured drop surface area (m2/ml oil)
0.1
Quinoa (2.25 μm)

0.01
Rice (4.46 μm)
Maize (14.9 μm)
Waxy barley (17.5 μm)
0.001
0.001 0.01 0.1 1
Estimated surface area (m2/ml oil)
FIGURe 4.4 The measured drop surface area versus estimated drop surface area for
starch particle-stabilized emulsions. Larger particles tend to produce larger drop area
than predicted, although in the case of starches, the particle shape may also have an
influence. (Open access image from Timgren, A. et al., Food Sci. Nutr., 1, 2, 157–171,
2013.)
Midmore (1998) showed that flocculation of silica particles could help the stabiliza-
tion at low surface coverage.
In a practical situation, one has to also consider to what extent the hydrophobicity
of particles will lead to particle–particle aggregation and thus several aggregated
particles act as single (larger) particles stabilizing an emulsion, thereby causing a
multilayer of particles, alternatively, a high degree of hydrophobicity may cause
aggregation of emulsion droplets. For example, Rayner, Sjöö et al. (2012) showed
that increasing the hydrophobicity of starch particles above a threshold value sta-
bilized O/W emulsions and led to an increase in droplet size; however, a decrease
in the amount of free starch not attached to the emulsion droplets was indicative of
particle aggregates adsorbed at the surface.
4.3 sTABILITY OF PARTICLe-sTABILIZeD eMULsIONs

Three main concerns regarding destabilization of emulsions are coalescence, Ostwald
ripening, and gravitational separation (creaming/sedimentation). Aggregation of
droplets may also occur, but this could be desired, as it might stabilize the system
from creaming, or nondesired, as it could give the emulsion a less favorable texture.
Previously (see Section 4.2), the high detachment energy for particles adsorbed to the
interface of Pickering emulsions was discussed. However, as pointed out by several
authors (Kruglyakov and Nushtayeva 2004; Tcholakova, Denkov, and Lips 2008),
this is not enough to completely describe stability in Pickering emulsions. The capil-
lary forces that work between the liquid phases and the adsorbed particles also affect
the stability. Theoretically, this has been thoroughly described by Kruglyakov and
Nushtayeva (2004). From a simplified theoretical point of view, the particles stabiliz-
ing the thin liquid film between the two droplets could be in either a monolayer or a
bilayer configuration. In reality, there could, of course, be several other conforma-
tions, including multilayers of particles, particle clusters, and different degrees of
imperfections, all of which influence the stability, most often decreasing it further,
compared to the idealized model.
The properties of the particle will influence the stability of the emulsions. As
discussed in Section 4.2, particle size and the interfacial contact angle will be the
properties influencing most; however, morphology and surface roughness of the
particles also influence the final stability of the emulsion. Particle roughness has
been suggested to both increase and decrease the stability of Pickering emulsions.
In a systematic study of particles with varying surface roughness, San-Miguel and
Behrens (2012) showed that increasing surface roughness of the particles increases
the emulsion stability to a point after which it decreases drastically. This point was in
the region when the particles where no longer homogenously wetted by the liquids.
4.3.1 CoalesCenCe
Both classical surfactants and particles can stabilize emulsions from coalescence
over long periods of time at optimal conditions. However, the stability of traditional
surfactants is highly dependent on small droplet sizes whereas Pickering emulsions
can be stable for months or even years for droplet sizes of greater than 100 µm
(Timgren et al. 2013).
FIGURe 4.5 The packing of particles between two particle-stabilized droplets. Left droplets
stabilized by monolayers, middle droplets stabilized by bridging particles, and right droplets
stabilized by individual particles and with aggregated particles in the continuous phase.
Coalescence occurs when a thin film of the continuous phase is drained from the
interface between two adjacent droplets. This leads to the nucleation of a pore, and
when this pore reaches a critical size, it starts growing rapidly and the two drops are
fused. Chen et al. (2013) have shown that the coalescence of Pickering emulsions can
occur via either the conventional route, where a liquid bridge grows continuously and
merges two droplets together, or via an oscillating mechanism, where bridges are
formed but fail to grow due to geometric restrictions until a stable bridge is formed
and the droplets merge. In Figure 4.5, the idealized packing of particles between two
droplets is described. If the energy of detachment for the particles is high, coalescence
will not occur due to a mechanism described by Denkov et al. (1992); Kruglyakov
and Nushtayeva (2004); and Tcholakova, Denkov, and Lips (2008). Contact between
the two discontinuous phases could occur through the menisci of the bare oil–water
interface. The stability of the film will, in this case, be the same as the maximum
capillary pressure drop, which can be resisted by the liquid menisci formed between
the adsorbed particles. The shape of the menisci depends on a number of factors
such as particle radius, antiparticle distance and configuration of the particle layer,
three-phase contact angle, oil–water interfacial tension, and the capillary pressure
across the fluid interface. This theory predicts that the maximum capillary pressure
that an emulsion can withstand prior to coalescence is higher (1) when particles have
a contact angle further away from 90°, (2) when the contact angle hysteresis is larger,
and (3) when the particles are small (Denkov et al. 1992). The size and contact angle
dependence is in the opposite direction from the factors that govern the detachment
energy from the surface, and thus theoretically there is an optimal particle size and
contact angle that will favor stable emulsions. To complicate things further, this will
be contingent on whether the particles at the surface form a mono or bilayer protec-
tion against coalescence (Denkov et al. 1992; Kruglyakov and Nushtayeva 2004;
Tcholakova, Denkov, and Lips 2008).
The ability to obtain the so-called partial coalescence is a specific character-
istic of particle-stabilized emulsions. In this case, two droplets start to merge but
instead of forming one larger droplet; the coalescence is arrested giving rise to
two merged droplets, sometimes obtaining peanut- or dumbbell-like shapes. This
occurs for emulsions that initially do not have the full surface coverage of the par-
ticles (Pawar et al. 2011), but when the droplets start to coalesce (thereby reducing
their total interfacial area), the particles do not detach. If the particle concentra-
tion is right, the merging of the drops will then lead to an increase in the surface
coverage. Given the right conditions, the particle concentration at the surface will
increase until the particles form a cohesive structure that resists the Laplace pres-
sure, hindering further coalescence (Pawar et al. 2011). Partial coalescence will
be important for the structure and rheology of the emulsion. This has been well
known in food systems and has been reviewed in detail by Fredrick, Walstra, and
Dewettinck (2010).
4.3.2 ostwald Ripening

Ostwald ripening occurs when the dispersed phase is partly soluble in the con-
tinuous phase. Due to Laplace pressure, the solubility will be higher at a curved
surface than at a flat surface, and thus small droplets will be dissolved and larger
droplets will grow. The kinetics of Ostwald ripening can be described using
LSW theory, originally mainly described by Lifshitz and Slyozov (1961) and by
Wagner (1961), as seen in Equation 4.4. Here rc is the size of the drop that is not
growing or decreasing, D is the diffusion constant of the dispersed phase in the
continuous phase, Vm is the molar volume of the dispersed phase, and Csol is the
solubility concentration. As can be seen from the equation, the rate of droplet
growth is dependent on the surface tension of the interface and the solubility of
the dispersed phase.
d 3 8γ DC V
dt
( )
rc = ow sol m
9 RT
(4.4)
Some authors have reported that Ostwald ripening is low for Pickering emul-
sions, at least at high particle coverage of the droplets (Ashby and Binks 2000;
Juarez and Whitby 2012). It has been shown, however, that Ostwald ripening
can be the driving force for instability at a low surface coverage of the droplets,
but as the surface coverage increases, Ostwald ripening stops (Ashby and Binks
2000; Juarez and Whitby 2012). Juarez and Whitby (2012) showed that at a low
surface coverage of toluene films, the Ostwald ripening was higher than that
predicted by the LSW theory. This was attributed to the aggregation of droplets
that was seen to be favored by low surface coverage. As the particle content
increased, they saw a drop in the ripening rate attributed to both higher surface
coverage and less aggregation. The surface tension between the oil and water is
an important factor and both proteins and especially surfactants lower the sur-
face tensions considerably, thereby decreasing Ostwald ripening. As described in
detail in a review by Tcholakova, Denkov, and Lips (2008), the surface tension of
a Pickering emulsion can be divided into the surface tension of the bare surface
γ b, which is always the same for each individual system and the macroscopic
apparent surface tension γa , which is dependent on the particle coverage, and that
is used in Equation 4.4. As discussed by Tcholakova, Denkov, and Lips (2008),
when the surface coverage becomes high, for example, due to initial shrinkage
of the drops, Ostwald ripening stops, as the apparent interfacial tension, γa , and
the drop capillary pressure, PC = 2γA/R, approach zero.
Another key component for Ostwald ripening is the solubility of the liquids
in each other. For example, triglyceride liquids have a rather low solubility in
water. However, the presence of micelles may increase the solubility of oil in
water. Drelich et al. (2012) showed that silica Pickering emulsions did not favor
transfer of oils between the emulsion droplets, but when a surfactant was added
in concentrations above the critical micelle concentration (CMC) of the surfac-
tant, oil was transferred between the emulsion droplets. Thus, one additional
advantage of Pickering emulsions compared to surfactant-stabilized emulsions
is that they do not change the apparent solubility of oil or an active substance in
the water phase.
4.3.3 CReaming oR sedimentation

Due to the size of the droplets and differences in density between the dispersed and
continuous phases, essentially all emulsions will tend to separate into one part con-
taining the continuous phase and another part containing a concentrated emulsion
phase. Under the right conditions, this separation will be slow or avoided as the
volume of the concentrated emulsion increases to fill the full void. Theoretically, this
occurs at the overlap concentration of the emulsion droplets.
Creaming and sedimentation arises because the droplets are large enough to
be affected by gravity. To counteract this, the size of the droplets can be reduced,
the viscosity of the continuous phase can be increased, or the density differences
between the continuous phase and the droplets can be altered. In the case of
Pickering emulsions, the particles adsorbed into the interface can be large enough
to affect the density of the drop and thus the surface coverage can affect either the
emulsion creams or sediments. As shown by Rayner, Timgren et al. (2012), the
surface coverage of particles can be matched in such a way that density match-
ing is achieved, thereby allowing the particles to float in the system; that is, a
neutral buoyancy system is achieved. As a result of weak aggregation of particles
and droplets, it is further possible to form a gel network structure in the systems,
which counteract sedimentation and creaming. In the event of an excess of par-
ticles, the particles themselves can form a stabilizing network. Binks and Lumsdon
(2000) investigated how adding more of the discontinuous phase affected cream-
ing and sedimentation and found that resistance toward creaming and sedimenta-
tion increased as the system approached the concentration where phase inversion
occurred.
4.4 sTRUCTURe AND RHeOLOGY OF PICKeRING eMULsIONs

The structural properties, often characterized in terms of its rheology, are highly
important for the properties of an emulsion. Rheological properties will have a large
impact on the sensory characteristics and eating qualities of food-based emulsions.
It may also set limitations for processing parameters in terms of, for example, mix-
ing, pumping, and packaging. Furthermore, properties such as viscosity and shear
thinning are also important for the consumer handling the product before using it,
for example, when being removed from its container.
All types of emulsions will become viscoelastic when they reach a certain level
of the dispersed phase where there is mechanical interaction between the droplets.
Mason, Bibette, and Weitz (1995) showed for surfactant-stabilized emulsions that
even very different emulsions systems follow the same dependence on the volume
fraction of the dispersed phase when the elastic modulus of the system is scaled to
γ/R. The scaled modulus have been seen to increase drastically at a volume fraction
around 0.63, which is equal to random close packing of solid spheres. Arditty et al.
(2004) showed that this behavior is not seen for Pickering emulsions. Instead the
emulsions follow the Mason curve if the scaling factor is the interfacial elasticity
and not the surface tension. This is due to the strong attachment of the particles to
the interface.
The excess particles can form a network in which the oil droplets are dis-
persed. In this case, the rheological properties of the gel will depend on both the
particle network as such, and on the dispersed phase droplets (Abend et al. 1998).
The dispersed phase will function as a filler in the particle network, giving rise to
a more elastic system. This behavior was shown to be formed by Aerosil 130 and
silicon oils (Sugita, Nomura, and Kawaguchi 2008) and by paraffin O/W emul-
sions stabilized by different bentonites and hectorites (Lagaly, Reese, and Abend
1999).
Pickering emulsions, where the adsorbed particles are able to form bridges
between the emulsion droplets, can have structures similar to flocculated particle
gels. Lee, Chan, and Mohraz (2012) showed that the rheology of these gels was
mainly influenced by the volume fraction of particles. In fact, systems with com-
pletely different oil-to-water ratios could have very similar rheological properties
as long as the particle ratio was the same. In contrast to surfactant-stabilized emul-
sions, these systems showed elastic properties even when the volume fraction of the
dispersed phase was below the random-close-packing limit for spheres. Therefore,
Pickering emulsions that are mechanically stable toward creaming at dispersed
phase volume fractions lower than the random-close-packing limit can be obtained.
Effects of particle aggregation causing droplet aggregation in Pickering emulsions
and increased elastic properties have also been observed both for starch particle-
stabilized emulsions (Rayner, Sjöö et al. 2012; Rayner, Timgren et al. 2012) and in
milk protein particles (Dickinson 2001).
Pickering emulsions stabilized by the same particles, that is, quinoa starch
granules, have been shown to have a weak gel behavior in a wide range of par-
ticle and oil concentrations (Figure 4.6). In a weak gel, the elastic modulus, G′,
is larger than the viscous modulus, G″; although being frequency dependent, the
gel strength can further be evaluated by the shear when the phase angle, δ, is
45° (Figure 4.7). As expected, the volume fraction of the dispersed phase mainly
determines the elastic properties and the stability of the gel (Timgren et al. 2011).
Marku et al. (2012) further showed that the viscoelastic behavior is influenced by
the properties of the dispersed phase, especially the melting point of the oil. Shear
can also be used in a controlled way to destabilize emulsions in specific applica-
tions (Whitby et al. 2011).
10,000
G′
1,000
G″
100
G′, G″ (Pa)
10
0.1
0.01
0 10 20 30 40 50 60
Oil concentration (%)
FIGURe 4.6 Viscoelastic properties in terms of elastic (G′) and viscous (G″) modulus for
quinoa starch-stabilized emulsions with buffer and medium-chain triglyceride oil. Particle con-
centration 214 mg/ml oil except for at 40% oil: 530 mg/ml oil. (Data combined from different
measurements Marku, D. et al., Int. J. Pharm., 428, 1–2, 1–7, 2012; Rayner, M. et al., Faraday
Discuss., 158, 139–155, 2012; Sjöö, M. et al. Submitted manuscript (under review J. Colloid.
Interf. Sci.), 2014; Timgren, A. et al., Procedia Food Sc., 1, 95–103, 2011.)
1,000
G′
G″
100
Elastic modulus, G′ (Pa)
10
0.1
0.00001 0.001 0.1 10
γ * (strain)
FIGURe 4.7 Elastic (G′) and viscous (G″) modulus as a function of complex strain of qui-
noa starch-stabilized emulsion.
4.5 INFLUeNCe OF OTHeR COMPONeNTs IN THe sYsTeM

Because the mechanism of stabilization is different in a Pickering emulsion com-
pared to a surfactant-based system, as described previously (see Section 4.2), parti-
cle-stabilized systems may behave differently in combination with other ingredients
present in the formulation. Surface-active components, especially when present at
high concentrations, tend to adsorb into the oil–water interface and may even dis-
place particles from the interface. When surface-active components are present, they
not only compete with the particles at the oil–water interface; they may also adsorb
into the surface of the particles themselves. The surface properties of the particle
may then affect the ability of the particles to act as a Pickering agent. Addition of
surfactants at concentrations below CMC has, for example, been shown to make
Pickering emulsions more stable (Binks, Rodrigues, and Frith 2007; Kawazoe and
Kawaguchi 2011). Addition of polymers may also lead to adsorption onto the surface
of particle, increasing the emulsion-stabilizing effect (Morishita and Kawaguchi
2009; Sugita, Nomura, and Kawaguchi 2008). The effect of adsorbed substances on
particle properties could also affect the degree of flocculation in the emulsion, alter-
ing the network formation and gel properties and/or also creaming or sedimentation
effects. This has been shown both for surfactants (Binks, Rodrigues, and Frith 2007)
and polymers (Sugita, Nomura, and Kawaguchi 2008). Phase inversion from an O/W
emulsion to a W/O emulsion could also be caused by surface-active impurities such
as fatty acid in triglycerol oils used in Pickering emulsions (Zhu et al. 2013), in this
case stabilized by CaCO3 nanoparticles.
In conclusion to the discussion above, surface-active ingredients are impor-
tant to consider in Pickering formulations as they can have substantial effect on
emulsion properties and stability. However, other ingredients may also influence
Pickering emulsion, although highly dependent on the system and particles used.
In a study of plate-like positively charged particles, Pickering emulsions were not
formed in the absence of salt, whereas the addition of salt led to particle adsorp-
tion and aggregation (Yang et al. 2006). It was then observed that the contact angle
was rather unaltered, whereas the zeta potential decreased with increasing salt
levels. Weak aggregation of droplets of different size has been seen in stabilized
emulsions without salt, whereas droplets were more homogenous in the presence
of NaCl for both starch (Timgren et al. 2013) and silica particles (Gautier et al.
2007). The salt concentration, however, had limited effect on the drop size dis-
tribution and rheological properties of the starch-stabilized emulsions (Rayner,
Sjoo et al. 2012). Also, combinations of salt and surfactant additions have been
investigated by rheological methods, showing the importance of the presence of
salt (Torres et al. 2007).
Depending on the system, and especially the particles used, pH is another prop-
erty that may have a large or no effect on the stabilization of Pickering emulsions,
and has been investigated in rather different ranges (Haase et al. 2010). Cosolvents
in the aqueous phase of emulsions can also have an effect on the interfacial tension
and particle interactions (Binks et al. 2013). For example, ethanol, glycerol, or simi-
lar ingredients may cause phase inversion, as shown for silica-stabilized emulsions
when glycerol concentrations are increased (Binks et al. 2013).
4.6 FOOD-GRADe PARTICLes FOR eMULsION sTABILIZATION

Particles of very different origin and composition can be used to stabilize Pickering emul-
sions. In model systems, silica particles have been extensively used as they can be varied
within a large range obtained through very specific particle characteristics in terms of
size and hydrophobicity. However, they are not suitable for use in food applications. For
food purposes, particles such as fat crystals, proteins, and starch have received increasing
interest, but there are numerous sources of potentially interesting particles (see Table 4.1).
TABLe 4.1
examples of Food-Based Particles Used as stabilizers for Pickering-Type
emulsions
Formulations/
Particle Type size and shape Applications References
Polysaccharides
Starch granules (quinoa, 0.5–20 µm in size Model systems of Dejmek et al. (2012);
rice, barley, and waxy and various O/W emulsions over Marefati et al. (2013);
maize) with various shapes (round, a wide range of oil Marku et al. (2012);
hydrophobic oblong, contents (5%–75%), Matos et al. (2013);
modifications (octenyl sharp-edged food emulsions, Rayner, Sjöö et al. (2012);
succinic anhydride polyhedral, etc.) salad dressings, Rayner, Timgren et al.
[OSA], or dry heating, depending on mayonnaise-type (2012); Rayner et al.
and in some cases botanical source emulsions, double (2014); Timgren et al.
without) emulsions, (2011, 2013)
encapsulation,
topical formulations,
cosmetics, and dried
emulsions (oil-filled
powders)
Starch nanocrystals Polygonal 50% paraffin O/W, Li et al. (2014);
structures sizes size of emulsions Miao et al. (2014)
40–100 nm drops containing
0.02–6.0 wt% starch
nanocrystals,
droplets were
33–13 µm,
respectively
Starch nanospheres Uniform O/W and W/O Tan et al. (2014)
spheres ~250 nm emulsions were
produced depending
on the water-to-oil
ratio. Using 1%
particles, emulsion
droplet sizes were
between 20–28 µm
(Continued)
TABLe 4.1 (Continued )

emulsions
Formulations/
Cyclodextrin (α-CD, Rhomboidal O/W to W/O Mathapa and Paunov (2013);
β-CD) prisms 2–20 µm depending on oil Moriyama, Saito, and
long depending phase fraction Bagchi (2013)
on type of 0.2–0.6 at 10 mm
cyclodextrin cyclodextrin α-CD
drops ~100 µm and
β-CD 15–30 µm
Chitin nanocrystals 240 × 20 nm Emulsion drop sizes Tzoumaki et al. (2011)
(from crab shells) between 10 and
100 µm depending on
the concentration of
chitin
Bacterial cellulose 189 × 13 × 6 nm 30% O/W (hexadecane) Capron and Cathala (2013);
nanocrystals (Nata de model system 0.2 and Kalashnikova et al.
coco cubes) 5 g/L cellulose (2011, 2012, 2013);
Cotton cellulose 855 × 17 × 7 nm particles in 50 mM Tasset et al. (2014)
nanocrystals NaCl. Emulsion drop
Algae cellulose 4000 × 20 × 15 nm size 4–10 µm, shorter
nanocrystals nanocrystals
promoted individual
droplets, and longer
nanocrystals
promoted networking-
like systems
Fat and Oils

Crystalline Submicron W/O emulsions for Frasch-Melnik, Norton,
monoglycerides as triglyceride controlled salt release, and Spyropoulos (2010);
seed particles for crystals droplet size Nadin, Rousseau,
triglyceride crystals 3.4–11 µm and Ghosh (2013);
depending on crystal Rousseau (2013)
composition
Proteins
Spray-dried soy protein Collapsed spheres O/W with oil phase Liu and Tang (2013);
isolate particles with (raisin shapes) fraction ranging Paunov et al. (2007)
CaP cores 2–10 µm from 0.1 to 0.9. At
6.4 wt% soy particles
emulsion drops
20–60 µm
(Continued)
TABLe 4.1 (Continued )

emulsions
Formulations/
Insoluble corn protein Spherical 50% O/W 0.2–2 wt% De Folter, Van Ruijven, and
zein 70–80 nm zein. Emulsion drops Velikov (2012)
50–100 µm diameter
depending on zein
concentration
WPM 100–300 nm pH-responsive soft Destribats et al. (2011,
depending on particles, O/W 2013, 2014)
pH and ionic emulsions, droplet
strength diameter 80–200 µm
at 0.1% WPM,
50% oil
Lactoferritin and ~10–20 nm O/W emulsions drop Meshulam and Lesmes
lactoferritin protein- sizes 0.5–10 µm (2014); Shimoni et al.
polysaccharide depending on the (2013)
complexes (alginate particle concentration
and carrageenan) and pH used. Affected
lipolysis rate/total
bioavailability
Miscellaneous
Egg yolk granules 0.3–2 µm Model O/W emulsions, Aluko and Mine (1997);
salad dressings-type Anton, Beaumal, and
emulsions, freeze- Gandemer (2000); Anton
thaw stable and Gandemer (1997);
mayonnaises, Ercelebi and Ibanoglu
cosmetics. Emulsion (2010); Eriksson (2013);
drop size of Laca, Paredes, and Diaz
10–70 µm depending (2012); Laca et al. (2010);
on the amount used Rayner et al. (2014)
Cocoa particles 2–5 µm 20% purified sunflower Gould, Vieira, and Wolf
oil, with 2%–10% (2013)
cocoa particles added.
Emulsion drops
~10 µm in diameter
Flavonoids (tiliroside, ~100 nm 20% oil-in-water in Luo et al. (2011, 2012)
rutin, and naringin) 1 mm flavonoids.
Emulsions drops
2–200 µm in
diameter depending
on the pH
WPM, Whey protein microgels.

Starch has received growing attention to be used as particles for the stabiliza-
tion of Pickering emulsions. In nature, a large variation in sizes and shapes of
starch granules can be found depending on the botanical origin. Most starches are
relatively large (greater than 15 µm or more) and thereby rather large and heavy
in order to stabilize Pickering emulsions (Timgren et al. 2013). However, there
are some plants producing extraordinarily small starch particles, which have been
shown to effectively stabilize emulsions (Timgren et al. 2011). Because starch is
hydrophilic, it is not likely to adsorb into the oil–water interface to a large extent,
although native starch has been used (Li, Li et al. 2013; Timgren et al. 2013). In
other studies, starch has been modified with octenyl succinic anhydride or ther-
mally treated using dry heat to obtain more hydrophobic character (Rayner, Sjöö
et al. 2012; Timgren et al. 2013; Yusoff and Murray 2011). Furthermore, the produc-
tion of starch-based nanospheres provides an additional source of starch particles
for particle stabilization (Tan et al. 2012). Modified starch has also previously been
used for a long time as an emulsifier, although in the molecular form, as reviewed
by Eliasson et al. (2013). In a Pickering emulsion, particles are mainly attached to
the drop surface, although free starch may be present in the system. If particles
are added to an emulsion already stabilized by surfactants, they are though less
likely to adsorb to the interface (see Figure 4.8). This would be the case also for
starch particles added to an emulsion stabilized with molecular starch. However,
an emulsion stabilized by starch particles can be heated, which then releases starch
molecules during the gelatinization process, although not destabilizing the emul-
sion (Marefati et al. 2013).
Fat crystals and solid lipid particles also have the potential of stabilizing
Pickering emulsion food systems such as whipped cream or spread. The mecha-
nisms of Pickering lipid crystal stabilization, fat crystal wetting, and the temperature
effects on fat crystal-stabilized emulsion have been previously reviewed (Ghosh and
Rousseau 2011; Rousseau 2013).
(a) (b) (c)
FIGURe 4.8 Particles at the drop interface and as free particles in the continuous phase
of O/W emulsions. (a) starch particle (0.5–15 µm) stabilized emulsion, (b) starch molecule
(40–400 nm) stabilized emulsion, (c) surfactant (1–10 nm) stabilized emulsion.
Proteins are a group of particles with important Pickering-stabilization ability.

The effect is contingent on the source of the protein (Hoffmann and Reger 2014; Li,
Xiao et al. 2013; Liu and Tang 2013; Shimoni et al. 2013). This includes particles
such as soy protein (Liu and Tang 2013; Paunov et al. 2007) and zein protein (De
Folter, Van Ruijven, and Velikov 2012). Milk and egg are traditional stabilizers in
food. Milk proteins are effective stabilizers, although their long-term mechanism as
Pickering stabilizers is debatable (Dickinson 2012). However, lactoferrin has been
isolated and used as a Pickering agent (Shimoni et al. 2013). High-density lipopro-
tein particles can be separated from egg yolk and can be used for emulsification (see
Figure 4.9), providing emulsions with less cholesterol as compared with the egg yolk
and high storage stability (Ercelebi and Ibanoglu 2010; Eriksson 2013; Laca et al.
2010; Rayner et al. 2014).
The increasing interest in particle stabilization has led to an investigation for
additional sources of particles, preferably with specific properties. Such alternatives
include the breaking down of cellulose into smaller particles (Kalashnikova et al.
2011, 2013) or producing chitin nanoparticles (Tzoumaki et al. 2011, 2013), with both
alternatives representing rod-like structures. Several common flavonoids have been
investigated with regard to their Pickering emulsification capacity and some were
found highly efficient (Luo et al. 2011, 2012).
(a) (b)
(c) (d)
FIGURe 4.9 Micrographs of emulsion droplets in mayonnaise stabilized with egg yolk
granules (a, b), quinoa starch granules (c), and liquid egg yolk (d), respectively. Scale bars
50 µm.
4.7 ADDITIONAL FUNCTIONAL PROPeRTIes OF

PARTICLe-sTABILIZeD eMULsIONs
In food product applications, Pickering particles can find their utility as a possible
alternative for emulsion stabilization. They may be used, for example, to alter the
product composition by adding other types of ingredients, or for reducing the fat
content. They may also provide additional functions to the system. One example is
to increase the freeze–thaw stability (Rayner et al. 2014) or the ability to use dry
emulsions as powders (Marefati et al. 2013). Pickering emulsions can further act as
vehicle for encapsulation and controlled release, and are being increasingly explored
for this purpose both in traditional and multiple emulsions such as water-in-oil-in-
water emulsions (Matos et al. 2013; Nadin, Rousseau, and Ghosh 2013; Marku et al.
2012; Shimoni et al. 2013).
ReFeReNCes
Abend, S., N. Bonnke, U. Gutschner, and G. Lagaly. 1998. “Stabilization of emulsions by het-
erocoagulation of clay minerals and layered double hydroxides.” Colloid and Polymer
Science 276 (8):730–737.
Aluko, R.E., and Y. Mine. 1997. “Competitive adsorption of hen’s egg yolk granule lipo-
proteins and phosvitin in oil-in-water emulsions.” Journal of Agricultural and Food
Chemistry 45 (12):4564–4570.
Anton, M., V. Beaumal, and G. Gandemer. 2000. “Adsorption at the oil–water interface and
emulsifying properties of native granules from egg yolk: Effect of aggregated state.”
Food Hydrocolloids 14 (4):327–335.
Anton, M., and G. Gandemer. 1997. “Composition, solubility and emulsifying properties of
granules and plasma of egg yolk.” Journal of Food Science 62 (3):484–487.
Arditty, S., V. Schmitt, J. Giermanska-Kahn, and F. Leal-Calderon. 2004. “Materials based on
solid-stabilized emulsions.” Journal of Colloid and Interface Science 275 (2):659–664.
Ashby, N.P., and B.P. Binks. 2000. “Pickering emulsions stabilised by Laponite clay par-
ticles.” Physical Chemistry Chemical Physics 2 (24):5640–5646.
Binks, B.P., P.D. Fletcher, M.A. Thompson, and R.P. Elliott. 2013. “Influence of propylene glycol
on aqueous silica dispersions and particle-stabilized emulsions.” Langmuir 29 (19):5723–33.
Binks, B.P., and S.O. Lumsdon. 2000. “Influence of particle wettability on the type and stabil-
ity of surfactant-free emulsions.” Langmuir 16:8622–8631.
Binks, B.P., J.A. Rodrigues, and W.J. Frith. 2007. “Synergistic interaction in emulsions
stabilized by a mixture of silica nanoparticles and cationic surfactant.” Langmuir 23
(7):3626–3636.
Capron, I., and B. Cathala. 2013. “Surfactant-free high internal phase emulsions stabilized by
cellulose nanocrystals.” Biomacromolecules 14 (2):291–296.
Chen, G., P. Tan, S. Chen, J. Huang, W. Wen, and L. Xu. 2013. “Coalescence of Pickering
emulsion droplets induced by an electric field.” Physical Review Letters 110 (6):064502.
De Folter, J.W.J., M.W.M. Van Ruijven, and K.P. Velikov. 2012. “Oil-in-water Pickering emul-
sions stabilized by colloidal particles from the water-insoluble protein zein.” Soft Matter
8 (25):2807–2815.
Dejmek, P., A. Timgren, M. Sjöö, and M. Rayner. 2012. New particle stabilized emulsions and
foams. WO Patent 2,012,082,065. December 15, 2011.
Denkov, N.D., I.B. Ivanov, P.A. Kralchevsky, and D.T. Wasan. 1992. “A possible mecha-
nism of stabilization of emulsions by solid particles.” Journal of Colloid and Interface
Science 150 (2):589–593.
Destribats, M., V. Lapeyre, M. Wolfs, E. Sellier, F. Leal-Calderon, V. Ravaine, and V. Schmitt.

2011. “Soft microgels as Pickering emulsion stabilisers: Role of particle deformability.”
Soft Matter 7 (17):7689–7698.
Destribats, M., M. Rouvet, C. Gehin-Delvat, C. Schmitt, and B.P. Binks. 2014. “Emulsions
stabilized by whey protein microgel particles: Towards food-grade Pickering emul-
sions.” Soft Matter 10 (36):6941–6954.
Destribats, M., M. Wolfs, F. Pinaud, V. Lapeyre, E. Sellier, V. Schmitt, and V. Ravaine.
2013. “Pickering emulsions stabilized by soft microgels: Influence of the emulsifica-
tion process on particle interfacial organization and emulsion properties.” Langmuir 29
(40):12367–12374.
Dickinson, E. 2001. “Milk protein interfacial layers and the relationship to emulsion stability
and rheology.” Colloids and Surfaces B: Biointerfaces 20 (3):197–210.
Dickinson, E. 2012. “Use of nanoparticles and microparticles in the formation and stabiliza-
tion of food emulsions.” Trends in Food Science and Technology 24 (1):4–12.
Drelich, A., J.L. Grossiord, F. Gomez, D. Clausse, and I. Pezron. 2012. “Mixed O/W emul-
sions stabilized by solid particles: A model system for controlled mass transfer triggered
by surfactant addition.” Journal of Colloid and Interface Science 386 (1):218–27.
Eliasson, A.C., B. Bergenstahl, L. Nilsson, and M. Sjoo. 2013. “From molecules to products—
Some aspects of structure-function relationships in cereal starches.” Cereal Chemistry
90 (4):326–334.
Ercelebi, E.A., and E. Ibanoglu. 2010. “Stability and rheological properties of egg yolk gran-
ule stabilized emulsions with pectin and guar gum.” International Journal of Food
Properties 13 (3):618–630.
Eriksson, M. 2013. “Pickering emulsions based on food grade biological particles.” MSc
thesis, Department of Food Technology, Engineering and Nutrition, Lund University,
Switzerland (KLT 920, 2013).
Finkle, P., H.D. Draper, and J.H. Hildebrand. 1923. “The theory of emulsification.” Journal of
the American Chemical Society 45 (12):2780–2788.
Frasch-Melnik, S., I.T. Norton, and F. Spyropoulos. 2010. “Fat-crystal stabilised w/o emul-
sions for controlled salt release.” Journal of Food Engineering 98 (4):437–442.
Fredrick, E., P. Walstra, and K. Dewettinck. 2010. “Factors governing partial coalescence in
oil-in-water emulsions.” Advances in Colloid and Interface Science 153 (1–2):30–42.
Frelichowska, J., M.A. Bolzinger, and Y. Chevalier. 2010. “Effects of solid particle content on
properties of o/w Pickering emulsions.” Journal of Colloid and Interface Science 351
(2):348–356.
Gautier, F., M. Destribats, R. Perrier-Cornet, J.-F. Dechézelles, J. Giermanska, V. Héroguez,
S. Ravaine, F. Leal-Calderon, and V. Schmitt. 2007. “Pickering emulsions with stimula-
ble particles: From highly- to weakly-covered interfaces.” Physical Chemistry Chemical
Physics 9:6455–6462.
Ghosh, S., and D. Rousseau. 2011. “Fat crystals and water-in-oil emulsion stability.” Current
Opinion in Colloid and Interface Science 16 (5):421–431.
Gould, J., J. Vieira, and B. Wolf. 2013. “Cocoa particles for food emulsion stabilisation.” Food
& Function 4 (9):1369–1375.
Haase, M.F., D. Grigoriev, H. Moehwald, B. Tiersch, and D.G. Shchukin. 2010. “Nanoparticle
modification by weak polyelectrolytes for pH-sensitive pickering emulsions.” Langmuir
27 (1):74–82.
Hoffmann, H., and M. Reger. 2014. “Emulsions with unique properties from proteins as emul-
sifiers.” Advances in Colloid and Interface Science 205:91–104.
Horozov, T.S., and B.P. Binks. 2006. “Particle-stabilized emulsions: A bilayer or a bridging
monolayer?” Angewandte Chemie 45 (5):773–776.
Juarez, J.A., and C.P. Whitby. 2012. “Oil-in-water Pickering emulsion destabilisation at low
particle concentrations.” Journal of Colloid and Interface Science 368 (1):319–325.
Kalashnikova, I., H. Bizot, P. Bertoncini, B. Cathala, and I. Capron. 2013. “Cellulosic nano-
rods of various aspect ratios for oil in water Pickering emulsions.” Soft Matter 9
(3):952–959.
Kalashnikova, I., H. Bizot, B. Cathala, and I. Capron. 2011. “New Pickering emulsions stabi-
lized by bacterial cellulose nanocrystals.” Langmuir 27 (12):7471–7479.
Kalashnikova, I., H. Bizot, B. Cathala, and I. Capron. 2012. “Modulation of cellulose nano-
crystals amphiphilic properties to stabilize oil/water interface.” Biomacromolecules 13
(1):267–275.
Kawazoe, A., and M. Kawaguchi. 2011. “Characterization of silicone oil emulsions stabilized
by TiO2 suspensions pre-adsorbed SDS.” Colloids and Surfaces A: Physicochemical and
Engineering Aspects 392 (1):283–287.
Kruglyakov, P., and A. Nushtayeva. 2004. “Emulsions stabilised by solid particles: The role of
capillary pressure in the emulsion films.” In Interface Science and Technology, edited by
D.N. Petsev, 641–676, Chapter 16. Elsevier, London.
Laca, A., B. Paredes, and M. Diaz. 2012. “Lipid-enriched egg yolk fraction as ingredient in
cosmetic emulsions.” Journal of Texture Studies 43 (1):12–28.
Laca, A., M.C. Saenz, B. Paredes, and M. Diaz. 2010. “Rheological properties, stability and
sensory evaluation of low-cholesterol mayonnaises prepared using egg yolk granules as
emulsifying agent.” Journal of Food Engineering 97 (2):243–252.
Lagaly, G., M. Reese, and S. Abend. 1999. “Smectites as colloidal stabilizers of emulsions.
II. Rheological properties of smectite-laden emulsions.” Applied Clay Science 14
(5–6):279–298.
Lee, M.N., H.K. Chan, and A. Mohraz. 2012. “Characteristics of Pickering emulsion gels
formed by droplet Bridging.” Langmuir 28 (6):3085–3091.
Levine, S., B.D. Bowen, and S.J. Partridge. 1989a. “Stabilization of emulsions by fine par-
ticles. I. Partitioning of particles between continuous phase and oil/water interface.”
Colloids and Surfaces 38 (2):325–343.
Levine, S., B.D. Bowen, and S.J. Partridge. 1989b. “Stabilization of emulsions by fine par-
ticles II. Capillary and van der Waals forces between particles.” Colloids and Surfaces
38 (2):345–364.
Li, C., Y. Li, P. Sun, and C. Yang. 2013. “Pickering emulsions stabilized by native starch
granules.” Colloids and Surfaces A: Physicochemical and Engineering Aspects
431:142–149.
Li, C., Y. Li, P. Sun, and C. Yang. 2014. “Starch nanocrystals as particle stabilisers of oil-in-
water emulsions.” Journal of the Science of Food and Agriculture 94 (9):1802–1807.
Li, C., P. Sun, and C. Yang. 2012. “Emulsion stabilized by starch nanocrystals.” Starch−Stärke
64 (6):479–502.
Li, Z., M. Xiao, J. Wang, and T. Ngai. 2013. “Pure protein scaffolds from pickering high inter-
nal phase emulsion template.” Macromolecular Rapid Communications 34 (2):169–174.
Lifshitz, I.M., and V.V. Slyozov. 1961. “The kinetics of precipitation from supersaturated solid
solutions.” Journal of Physics and Chemistry of Solids 19:35–50.
Liu, F., and C.H. Tang. 2013. “Soy protein nanoparticle aggregates as pickering stabilizers for
oil-in-water emulsions.” Journal of Agricultural and Food Chemistry 61 (37):8888–8898.
Luo, Z., B.S. Murray, A.L. Ross, M.J.W. Povey, M.R.A. Morgan, and A.J. Day. 2012. “Effects
of pH on the ability of flavonoids to act as Pickering emulsion stabilizers.” Colloids and
Surfaces B: Biointerfaces 92:84–90.
Luo, Z.J., B.S. Murray, A. Yusoff, M.R.A. Morgan, M.J.W. Povey, and A.J. Day. 2011. “Particle-
stabilizing effects of flavonoids at the oil-water interface.” Journal of Agricultural and
Food Chemistry 59 (6):2636–2645.
Marefati, A., M. Rayner, A. Timgren, P. Dejmek, and M. Sjöö. 2013. “Freezing and freeze-
drying of Pickering emulsions stabilized by starch granules.” Colloids and Surfaces A:
Physicochemical and Engineering Aspects 436 (5):512–520.
Marku, D., M. Wahlgren, M. Rayner, M. Sjöö, and A. Timgren. 2012. “Characterization

of starch Pickering emulsions for potential applications in topical formulations.”
International Journal of Pharmaceutics 428 (1–2):1–7.
Mason, T., J. Bibette, and D. Weitz. 1995. “Elasticity of compressed emulsions.” Physical
Review Letters 75 (10):2051–2054.
Mathapa, B.G., and V.N. Paunov. 2013. “Cyclodextrin stabilised emulsions and cyclodextrino-
somes.” Physical Chemistry Chemical Physics 15 (41):17903–17914.
Matos, M., A. Timgren, M. Sjöö, P. Dejmek, and M. Rayner. 2013. “Preparation and encapsu-
lation properties of double Pickering emulsions stabilized by quinoa starch granules.”
Colloids and Surfaces A: Physicochemical and Engineering Aspects 423:147–153.
Meshulam, D., and U. Lesmes. 2014. “Responsiveness of emulsions stabilized by lactoferrin
nano-particles to simulated intestinal conditions.” Food & Function 5 (1):65–73.
Miao, M., R. Li, B. Jiang, S.W. Cui, T. Zhang, and Z. Jin. 2014. “Structure and physicochemi-
cal properties of octenyl succinic esters of sugary maize soluble starch and waxy maize
starch.” Food Chemistry 151:154–160.
Midmore, B.R. 1998. “Preparation of a novel silica-stabilized oil/water emulsion.” Colloids
and Surfaces A: Physicochemical and Engineering Aspects 132 (2–3):257–265.
Morishita, C., and M. Kawaguchi. 2009. “Rheological and interfacial properties of Pickering
emulsions prepared by fumed silica suspensions pre-adsorbed poly(N-isopropylacryl-
amide).” Colloids and Surfaces A: Physicochemical and Engineering Aspects 335
(1–3):138–143.
Moriyama, H., Y. Saito, and D. Bagchi. 2013. “Characterization of cyclodextrin nanoparticles
as emulsifiers.” In Bio-Nanotechnology: A Revolution in Food, Biomedical and Health
Sciences, edited by M. Bagchi, H. Moriyama, and F. Shahidi, 476–486, Blackwell
Publishing Ltd., Oxford.
Nadin, M., D. Rousseau, and S. Ghosh. 2013. “Fat crystal-stabilized water-in-oil emulsions
as controlled release systems.” LWT—Food Science and Technology 56 (2):248–255.
Paunov, V.N., O.J. Cayre, P.F. Noble, S.D. Stoyanov, K.P. Velikov, and M. Golding. 2007.
“Emulsions stabilised by food colloid particles: Role of particle adsorption and wettabil-
ity at the liquid interface.” Journal of Colloid and Interface Science 312 (2):381–389.
Pawar, A.B., M. Caggioni, R. Ergun, R.W. Hartel, and P.T. Spicer. 2011. “Arrested coalescence
in Pickering emulsions.” Soft Matter 7 (17):7710.
Pickering, S.U. 1907. “Emulsions.” Journal of the Chemical Society 91:2001.
Ramsden, W. 1903. “Separation of solids in the surface-layers of solutions and ‘suspensions’
(observations on surface-membranes, bubbles, emulsions, and mechanical coagulation)—
Preliminary account.” Proceedings of the Royal Society of London 72:156–164.
Rayner, M., D. Marku, M. Eriksson, M. Sjöö, P. Dejmek, and M. Wahlgren. 2014. “Biomass-
based particles for the formulation of Pickering type emulsions in food and topical
applications.” Colloids and Surfaces A: Physicochemical and Engineering Aspects
458:46–62.
Rayner, M., M. Sjöö, A. Timgren, and P. Dejmek. 2012. “Quinoa starch granules as stabilizing
particles for production of Pickering emulsions.” Faraday Discussions 158:139–155.
Rayner, M., A. Timgren, M. Sjöö, and P. Dejmek. 2012. “Quinoa starch granules: A candidate
for stabilising food-grade Pickering emulsions.” Journal of the Science of Food and
Agriculture 92 (9):1841–1847.
Rousseau, D. 2013. “Trends in structuring edible emulsions with Pickering fat crystals.”
Current Opinion in Colloid and Interface Science 18 (4):283–291.
San-Miguel, A., and S.H. Behrens. 2012. “Influence of nanoscale particle roughness on the
stability of Pickering emulsions.” Langmuir 28 (33):12038–12043.
Shimoni, G., C. Shani Levi, S. Levi Tal, and U. Lesmes. 2013. “Emulsions stabilization by
lactoferrin nano-particles under in vitro digestion conditions.” Food Hydrocolloids 33
(2):264–272.
Sjöö, M., S. Cem Emek, T. Hall, M. Rayner, and M. Wahlgren. 2014. “Barrier properties of
heat treated starch Pickering emulsions.” Submitted manuscript (under review Journal
of Colloid and Interface Science). doi:10.1016/j.jcis.2015.03.004.
Spyropoulos, F., S. Frasch-Melnik, and I.T. Norton. 2011. “W/O/W emulsions stabilized by
fat crystals—Their formulation, stability and ability to retain salt.” In 11th International
Congress on Engineering and Food, edited by G. Saravacos, P. Taoukis, M. Krokida,
V. Karathanos, H. Lazarides, N. Stoforos, C. Tzia, and S. Yanniotis, 1700–1708.
Amsterdam, the Netherlands: Elsevier Science, B.V.
Sugita, N., S. Nomura, and M. Kawaguchi. 2008. “Rheological and interfacial properties of
silicone oil emulsions stabilized by silica particles.” Journal of Dispersion Science and
Technology 29 (7):931–936.
Tan, Y., K. Xu, C. Liu, Y. Li, C. Lu, and P. Wang. 2012. “Fabrication of starch-based nano-
spheres to stabilize pickering emulsion.” Carbohydrate Polymers 88 (4):1358–1363.
Tan, Y., K. Xu, C. Niu, C. Liu, Y. Li, P. Wang, and B.P. Binks. 2014. “Triglyceride-water emul-
sions stabilised by starch-based nanoparticles.” Food Hydrocolloids 36:70–75.
Tasset, S., B. Cathala, H. Bizot, and I. Capron. 2014. “Versatile cellular foams derived from
CNC-stabilized Pickering emulsions.” RSC Advances 4 (2):893–898.
Tcholakova, S., N.D. Denkov, and A. Lips. 2008. “Comparison of solid particles, globu-
lar proteins and surfactants as emulsifiers.” Physical Chemistry Chemical Physics 10
(12):1608–1627.
Timgren, A., M. Rayner, P. Dejmek, D. Marku, and M. Sjöö. 2013. “Emulsion stabilizing
capacity of intact starch granules modified by heat treatment or octenyl succinic anhy-
dride.” Food Science & Nutrition 1 (2):157–171.
Timgren, A., M. Rayner, M. Sjöö, and P. Dejmek. 2011. “Starch particles for food based
Pickering emulsions.” Procedia Food Science 1:95–103.
Torres, L.G., R. Iturbe, M.J. Snowden, B.Z. Chowdhry, and S.A. Leharne. 2007. “Preparation
of o/w emulsions stabilized by solid particles and their characterization by oscillatory
rheology.” Colloids and Surfaces A: Physicochemical and Engineering Aspects 302
(1–3):439–448.
Tzoumaki, M.V., T. Moschakis, V. Kiosseoglou, and C.G. Biliaderis. 2011. “Oil-in-water emul-
sions stabilized by chitin nanocrystal particles.” Food Hydrocolloids 25 (6):1521–1529.
Tzoumaki, M.V., T. Moschakis, E. Scholten, and C.G. Biliaderis. 2013. “In vitro lipid diges-
tion of chitin nanocrystal stabilized o/w emulsions.” Food & Function 4 (1):121–129.
Vignati, E., R. Piazza, and T.P. Lockhart. 2003. “Pickering emulsions: Interfacial tension, col-
loidal layer morphology, and trapped-particle motion.” Langmuir 19 (17):6650–6656.
Wagner, C. 1961. “Theorie der Alterung von Niederschlägen durch Umlösen (Ostwald-
Reifung).” Z. Elektrochem 65:581–591.
Whitby, C.P., F.E. Fischer, D. Fornasiero, and J. Ralston. 2011. “Shear-induced coalescence
of oil-in-water Pickering emulsions.” Journal of Colloid and Interface Science 361
(1):170–177.
Yang, F., S. Liu, J. Xu, Q. Lan, F. Wei, and D. Sun. 2006. “Pickering emulsions stabilized
solely by layered double hydroxides particles: The effect of salt on emulsion formation
and stability.” Journal of Colloid and Interface Science 302 (1):159–169.
Yusoff, A., and B.S. Murray. 2011. “Modified starch granules as particle-stabilizers of oil-in-
water emulsions.” Food Hydrocolloids 25 (1):42–55.
Zhu, Y., L.H. Lu, J. Gao, Z.G. Cui, and B.P. Binks. 2013. “Effect of trace impurities in triglyc-
eride oils on phase inversion of Pickering emulsions stabilized by CaCO3 nanoparticles.”
Colloids and Surfaces A: Physicochemical and Engineering Aspects 417:126–132.
Section II
High-Energy Processes
5 Droplet Breakup
in High-Pressure
Homogenizers
Andreas Håkansson
CONTeNTs
5.1 Introduction .................................................................................................. 125
5.2 The High-Pressure Homogenizer ................................................................. 126
5.3 An Historical Perspective on Fragmentation ................................................ 128
5.4 Fragmentation and Stabilization Mechanisms ............................................. 129
5.4.1 Fragmentation by Laminar Shear ..................................................... 129
5.4.2 Fragmentation by Turbulence ........................................................... 132
5.4.3 Fragmentation by Cavitation ............................................................ 136
5.4.4 Stabilizing Stresses and Comparisons .............................................. 137
5.5 Influence of Operating Parameters on Emulsification Result ...................... 139
5.5.1 Homogenizing Pressure .................................................................... 139
5.5.2 Second-Stage Effect and Thoma Number ........................................ 140
5.5.3 Viscosity of the Dispersed and Continuous Phases .......................... 141
5.5.4 Volume Fraction of Oil ..................................................................... 142
5.6 The Location of Drop Breakup .................................................................... 144
5.7 Conclusions on the Fragmentation Mechanisms in the HPH ....................... 145
References .............................................................................................................. 146
ABSTRACT In this chapter, a detailed review of the current understanding of the

physical processes of droplet breakup in a high-pressure homogenizer (HPH) is pre-
sented, covering breakup mechanisms by laminar shear in the gap inlet and its bound-
ary layers, by local turbulence in the gap exit jet, and by cavitation. Experimental
evidence of the effects of homogenization pressure, Thoma number, dispersed and
continuous phase viscosity, and dispersed phase volume fraction in relation to impli-
cations on dominant mechanisms of droplet breakup is also discussed.
5.1 INTRODUCTION
The HPH is used for size reduction or the disintegration of dispersed particles such
as cells, macromolecules, or emulsion drops. The by far largest application is the size
reduction of emulsion drops, which will be the focus of this chapter.
125
Information about the possibilities and limitations of HPHs as well as knowledge

about how drop size reduction is influenced by operating parameters and product
characteristics is of vital importance for the utilization of HPHs. Thus, understand-
ing the fragmentation process is of large importance for the efficient application of
homogenization processes as well as for the design of HPH valves.
Although breakup is a fundamental part of the process, it has been poorly under-
stood. A large number of theories have been put forward throughout the years,
many of which have been rejected later. There is still some uncertainty as to how
the remaining mechanisms interact to give rise to emulsification. However, today,
a consensus on what mechanism controls fragmentation is being established. Part
of the solution to the many contradictory views and widely differing experimental
results seen in many earlier studies has been found in understanding that difference
in design between different producers and models can lead to different dominating
mechanisms. Some of these differences will be discussed in this chapter.
It should also be noted that the result of any emulsification experiment (i.e., the
obtained drop size distribution) will be dependent on at least two simultaneously
occurring primary processes: fragmentation and coalescence of drops. For low-
volume fractions of oil and high concentration of emulsifiers (naturally present or
artificially added), the emulsification outcome will be dominated by fragmentation
(see further discussion in Section 5.5) and these systems will be the main interest of
this chapter.
This chapter will give an overview of the fragmentation mechanisms in the HPH,
both in terms of underlying physics and chemistry and how these are connected to
the experimentally obtained dependencies on operating parameters.
After a brief description of the basic function and hydrodynamics of an HPH
valve, different fragmentation mechanisms will be presented and discussed, includ-
ing a short section about older mechanisms, to give a historical background. The
large volume of experimental work on how the emulsification result varies with pro-
cess parameters is reviewed with special emphasis on what it implies for the active
mechanism of fragmentation. After a discussion of what is known about the loca-
tion of the drop breakup in the homogenizer valve, conclusions about fragmentation
mechanisms are drawn and discussed.
5.2 THe HIGH-PRessURe HOMOGeNIZeR

A schematic drawing of an HPH valve can be seen in Figure 5.1. Fluid enters the
valve from the bottom through a feed pipe. The forcer (upper part of the figure)
forces the flow radially through the narrow gap created between the forcer and the
seat. Often, the seat is inclined, giving rise to a narrowing region upstream of the
gap, referred to here as the inlet chamber. Downstream of the gap, the fluid exits into
a larger volume, referred to as the outlet chamber. Special impact rings are some-
times mounted on the valve in order to modify the outlet chamber geometry. The
design of an HPH valve is discussed in more detail in Chapter 6.
The gap height, h, can be varied by lowering or raising the position of the forcer.
Fluid-flow frictional forces increase with decreasing gap height, and thus a higher
pressure is required for a smaller gap height. In practice, the homogenizing pressure,
Droplet Breakup in High-Pressure Homogenizers 127
Impact
Gap ring
Forcer Gap
entrance exit
Outlet
re Inlet Gap chamber
ri chamber
Seat
FIGURe 5.1 Schematic, not to scale, representation of the high-pressure homogenizer valve.
(Note that the gap height is greatly exaggerated.)
ΔP, is set by adjusting the force applied on the forcer, which, in turn, sets the gap
height. Homogenization pressures are usually in the range of 5–40 MPa for food
applications, such as dairy processing of milk, but can be above 100 MPa for special
applications, such as cell breakage (Middelberg, 1995) or the disruption of macro-
molecules (Floury et al., 2002).
Because the gap height is small (order of magnitude 100 µm) and set implicitly
by the homogenizing pressure, it is generally unknown. The same is true for the
gap velocity, Ug, another important parameter for discussing fragmentation. Because
direct measurements are often not possible, gap height and velocity are often esti-
mated from semiempirical correlations such as (Phipps, 1975)
2 7/ 5 2
1  Q  5ρCν3C/ 5  Q   1 1  1 Q 
∆P = +  r 2 / 5 − r 2 / 5  + 2  2πr h  (5.1)
4  2πri h  h 3  2π   i e   e 
where:
ri and re are the gap inlet and exit radii, respectively (see Figure 5.1)
ρC is the continuous phase density
νC is the continuous phase kinematic viscosity
The homogenizing pressure is supplied through the displacement piston pumps

(often three or five) producing a near-constant volumetric flow rate, Q, as a function
of pressure drop. Furthermore, the volumetric flow rate is related to gap height and
gap (exit) velocity through
Q = 2πre hUg (5.2)
The hydrodynamic description in Equations 5.1 and 5.2 is vital for discussing frag-
mentation as it constitutes a link between the practically accessible parameters such
as Q and ΔP, and the more fundamental parameters h and Ug.
350 400
300 350
300
250
Ug (m/s)
250
h (μm)
200
200
150
150
100 100
50 50
20 40 60 80 100 20 40 60 80 100
(a) ΔP (MPa) (b) ΔP (MPa)
FIGURe 5.2 Gap height (a) and gap velocity (b) for a production-scale HPH as functions of
homogenizing pressure. See text for parameter values.
Gap height and gap velocity as a function of homogenizing pressure for a

production-scale homogenizer (Q = 10 m3/h, ri = 15 mm, re = 16 mm, ρC = 1000 kg/m3,
νC = 10−6 m2/s) can be seen in Figure 5.2. As can be seen in the figure, gap height
decreases rapidly with increasing pressure and gap velocity.
5.3 AN HIsTORICAL PeRsPeCTIVe ON FRAGMeNTATION

The first version of what should later evolve into an HPH was invented by the French
engineer Auguste Gaulin in the early 1900s. It was constructed for increasing the
product stability of milk by reducing the rate of cream separation due to fat glob-
ule size reduction. Dairy processing is still the largest application for high-pressure
homogenization, with approximately 270 Mt of dairy products treated with high-
pressure homogenization annually (IDF, 2010).
Gaulin was able to show an increase in stability against creaming but could only
speculate on the mechanism of drop size reduction:
The fat and the casein particles in being forced through the capillary tubes or orifice …
are either partially broken up or elongated into capillary filaments, and … are mashed
or squeezed, so far as to completely disintegrate them.
Gaulin, 1904: 2
Gaulin’s original explanation relies on emulsion drops interacting directly and

mechanically with the gap wall, much as we now know is the case in membrane
emulsification. This first explanation was refuted rather early when it was realized
that the gap height is substantially larger than the initial drop size; for example, in
milk, the initial mean fat globule size is 3–5 µm and the gap heights in production-
scale homogenizers are roughly 150 µm (see Figure 5.2a).
Alternative explanations followed. According to the explosion theory, drops were
compressed by the high pressures in the inlet chamber. It was thought that once the
pressures reach the outlet chamber and the pressure relaxes back to atmospheric,
the fast transition would cause a rapid expansion and subsequent explosion and
fragmentation of drops. This theory was disproved by noting that the compressibility
of liquids such as water and oil is very low. Even at a relatively high pressure differ-
ence such as 100 MPa, the volumetric change is just about 5%.
Fragmentation from impingement, that is, when drops of high velocity hit the
impact ring in the outlet chamber, has been another suggestion. It is now known
that a jet is created as the fluid exits the gap; however, the velocity declines rap-
idly as a function of distance to the gap exit, and drop velocities are relatively low
when reaching that far unless the impact ring distances are very short (Innings and
Trägårdh, 2007). Furthermore, fragmentation visualizations have shown breakup to
occur long before the impact ring (Innings, Fuchs, and Trägårdh, 2011). Thus, direct
impact is not responsible for the observed fragmentation.
5.4 FRAGMeNTATION AND sTABILIZATION MeCHANIsMs

When discussing fragmenting mechanisms, it must first be emphasized that there are
large geometrical differences between HPH models and that these may give rise to
differences in what mechanisms dominate fragmentation. Of special interest is that
of scale (laboratory, pilot, or production scale).
This section describes the characteristics, location, and fragmenting stress of
laminar shear, turbulence, and cavitation, the three mechanisms now attributed to
HPH fragmentation.
5.4.1 Fragmentation by Laminar Shear

Consider the spherical particle in Figure 5.3. The velocity gradient acting over the
particle is giving rise to an unevenly distributed force acting on the surface. If the
sphere is a solid particle, the velocity gradient will cause rotation and a lift force
∂vx
=0
∂x
∂vx
=G
∂y
FIGURe 5.3 Illustration of a spherical particle placed in a shear field and definition of
simple shear, G.
(Saffman, 1965). If the particle is liquid, on the other hand, three scenarios can occur
(Grace, 1982; Stone, 1994):
1. The drop deforms and elongates due to viscous drag on the interface.
2. The liquid inside the drop starts to rotate.
3. The drop in itself starts to rotate.
The degree of deformation increases with shear, G, but is countered by the interfacial
tension, as any deviations from the spherical shape will increase the surface pressure
and thus the free energy of the system. The extent of deformation is often described
in terms of a capillary number:
2Gµ C d
Ca = (5.3)
γ
where:
γ is interfacial tension
It will become increasingly favorable for the drop to break into two if it becomes
sufficiently elongated. One can show that this occurs approximately when the drop is
twice as long as it is thin, if only the surface energy is taken into consideration. Thus,
after it has been sufficiently elongated, it is expected to break.
Experimental studies of the deformation and breakup of fluid drops placed in
static shear fields were initiated by Taylor in the 1930s. An early conclusion from
this work was that the extent of deformation and point of breakage can be described
in terms of the capillary number. Furthermore, the critical value depends on the
dispersed-to-continuous phase viscosity ratio and the type of shear flow (Taylor,
1934). For a simple shear flow (i.e., a two-dimensional flow with a linear velocity
gradient in one direction), it has been shown experimentally that the critical capil-
lary number tends to infinity as the viscosity ratio tends to 4 (Grace, 1982), implying
that medium-to-high viscosity oils never break from laminar simple shear. However,
it should be noted that no such limiting viscosity ratio has been found for extensional
shear or the more general shear fields to which the HPH flow can be classified. Thus,
laminar shear could not be disregarded as a breakup mechanism per se.
A strong laminar shear can be found in the inlet chamber and boundary layers of
the gap in the HPH valve (Håkansson et al., 2011). It is the acceleration created from
the reducing flow through area when approaching the gap that produces high shear
in the inlet chamber. Experiments and hydrodynamic modeling has shown that the
maximum shear rate in the gap center can be estimated from (Håkansson et al., 2011,
2012; Innings and Trägårdh, 2007)
Ug
max(G) ≈ (5.4)
2h
The total pressure loss will be dominated by the expansion loss (i.e., third term in
Equation 5.1) for a production-scale HPH:
ρCUg2
∆P ≈ (5.5)
2
By substituting Equations 5.2 and 5.5 in Equation 5.4, the maximum shear experienced
by a drop in the inlet chamber can be shown to be proportional to the homogenizing
pressure:
re 2π
max(G) ≈ ∆P (5.6)
QρC
The fragmenting stress, σfrag, on a drop from a shear rate, G, can then be calculated from
σfrag = µ CG (5.7)
However, drops will only experience this high stress for a short period of time, just
before entering the gap. When taking into account the deformation timescale,
µD
τdef = (5.8)
σfrag
The deformation timescale is the minimal time required for deforming and break-
ing a drop. Walstra (1983) argued that only low viscosity drops could ever be frag-
mented by laminar shear, as it is reasonable to assume that the time spent in the high
shear region, τ, scales inversely with G and thus,
µD
τ > τdef => < C1 (5.9)
µC
Walstra (1983) suggested that C1 = 4 based on a very crude hydrodynamic model.

The conclusion is that medium and high viscosity drops cannot be fragmented by
laminar shear in the inlet chamber.
The second position of a high laminar shear is in the laminar boundary layers of
the gap. Here, it should be remembered that a large majority of the drops will pass
through the central parts of the gap as a consequence of the flow profile. Boundary
layers are therefore not expected to have a large influence on the emulsification result
if they do not extend into the central parts. Thus, it becomes interesting to know
under what conditions boundary layers merge in the gap. The flat plate approxima-
tion (Schlichting and Gersten, 2000) states that the boundary layer thickness, δ, as a
function of gap distance, x, can be approximated by
x νC
δ( x ) = 5 (5.10)
Ug x
The boundary layers will merge if the gap length, L g = re − ri, is long enough to give
δ(x = L g) = h/2, that is,
Q h
Lg ≥ (5.11)
2πreν C 100
200 0.035
0.03
150 0.025
Lg (mm)
Lg (mm)
0.02
100
0.015
50 0.01
0.005
0 0
5 10 15 20 5 10 15 20
(a) Q (m3/h) (b) Q (L/h)
FIGURe 5.4 Critical gap length, L g, for the laminar boundary layers in the gap to extend
into the center of the gap as a function of volumetric flow at a constant homogenizing pres-
sure (ΔP = 30 MPa). (a) Production-scale homogenizer (re = 16 mm) and (b) laboratory-scale
HPH (re = 3 mm).
Figure 5.4 illustrates the critical gap length needed for boundary layers to merge for a
production and a laboratory-scale homogenizer when fed with varying volumetric flow
rates, Q, while holding the homogenizing pressure constant at 30 MPa. For a production-
scale homogenizer (re = 16 mm), the gap would need to be approximately 50 mm long,
which is roughly 50 times longer than the actual gap length, whereas the boundary layers
would merge already after approximately 10 µm into the gap for a laboratory-scale HPH
(which is a short distance when compared to an actual length, close to 1 mm).
This analysis of boundary layers implies that laminar shear can influence emulsifica-
tion inside the gap for small-scale homogenizers, but not in the case of production-scale
HPHs. This could explain why experimental investigations often see different behavior
between production and laboratory homogenizers (e.g., Walstra and Smulders, 1998
and references therein).
5.4.2 Fragmentation by turbuLence

The linear fluid velocity in the gap is high (Figure 5.2b) and it has long been real-
ized that this might give rise to turbulence in the valve region. This has later been
supported by calculations of flow fields in the valve (e.g., Stevenson and Chen, 1997;
Kleinig and Middelberg, 1997) and more recently from velocity measurements in
carefully scaled models (Håkansson et al., 2011; Innings and Trägårdh, 2007).
A turbulent flow is not easily defined but can be characterized by a large degree
of random chaotic fluid motion. However, the fluid motion is not completely random
and there exists correlated fluid motions at different length-scales at each point in
time. These coherent structures are often referred to as turbulent eddies. The length
scales of these eddies, l, range from the geometrical scale of the flow, h, to a limiting
Kolmogorov length scale,
1/ 4
 ν3 
η= C  (5.12)
 ε 
where:
ε is the rate of dissipation of turbulent kinetic energy, which under static condi-
tions corresponds to the rate of energy input into the system
The rate of dissipation of turbulent kinetic energy can therefore be used as a measure
of the energy available for fragmentation in a turbulent flow.
Turbulent eddies of different length scales influence drops differently, which
makes it important to describe the relative amount of turbulent energy for eddies
of different sizes. This can be done with a spectrum of turbulent kinetic energy,
E(l), describing the contribution of eddies of length scale, l, to the total turbu-
lent kinetic energy. A standard Kolmogorov model spectrum (Pope, 2000) can
be seen in Figure 5.5 for a turbulent flow with ε = 109 m 2/s3, νC = 10−6 m 2/s, and
h = 150 µm (i.e., an oil-in-water emulsion in a production-scale HPH). Most of the
energy is contained in the larger scales close to l = h and eddies smaller than η
contain very little energy. Relevant drop sizes in a production-scale HPH are often
close to 1 µm; a reasonable interval of drop sizes has been inserted in Figure 5.5
as a comparison.
Fragmentation of a drop due to interactions with turbulent eddies was first
described theoretically by Kolmogorov (1949) and was then further developed by
Hinze (1955). The fragmentation is often described in terms of two limiting cases,
or more accurately as two mechanisms: turbulent inertial and turbulent viscous
fragmentation.
h η
E(l)
102 100
l (μm)
FIGURe 5.5 Turbulent spectra showing the turbulent kinetic energy as a function of eddy
length scales. Parameters chosen from the turbulent flow in the outlet chamber of a production-
scale HPH. Range of drop sizes has been inserted as comparison. (Note the logarithmic
scales and inverted horizontal axis.)
(a) (b)
FIGURe 5.6 Schematic illustration of interactions between drops and turbulent eddies
under turbulent inertial (a) and turbulent viscous (b) drop breakup. (Reprinted from J. Colloid
Interface Sci., 312, Vankova, N. et al., Emulsification in turbulent flow 1. Mean and maxi-
mum drop diameters in inertial and viscous regimes, 363–380, Copyright 2007, with permis-
sion from Elsevier.)
Inertial fragmentation originates from pressure fluctuations over the length scale
of the drop due to interactions with small turbulent eddies (i.e., l < d); see illustration
in Figure 5.6a. The stress on the drop can be estimated from
ρC uu
σfrag = d
(5.13)
2
where:
uu d is the average squared velocity fluctuation for eddies of length scales smaller
than the drop (l < d)
Emulsion drop sizes often have diameters of magnitude 1 µm when entering the
homogenizer. As seen in Figure 5.5, the drops will be in the far left region known
as the inertial subrange. The spectrum in this region can be approximated by (Pope,
2000)
E( κ) = C2 ⋅ ε2 / 3 ⋅ κ −5/ 3 (5.14a)
2π
κ= (5.14b)
l
and thus
∞
uu d
=
∫ E ( κ )d κ = 2 ε
2π/ d
2/3 2/3
d (5.15)
From the perspective of the drop, turbulent viscous deformation or fragmentation

from interactions with eddies much larger than the drop are indistinguishable from
laminar shear (c.f. Figures 5.3 and 5.6b). Thus, the stress on the drop under this
mechanism can be calculated from
σfrag = µ CGd (5.16)
where:
Gd is the turbulent shear created by eddies larger than the drop (l > d)
The turbulent shear is often estimated from
uu d
Gd = (5.17)
d
which is somewhat contradictory as uu d relates to eddies smaller than d and Gd to
eddies larger than d. However, as uu d increases with eddy size and Gd decreases
with eddy size, it could be argued that both are dominated by eddies of sizes close
to d. This can serve as a motivation for using Equation 5.17 as a first approximation
when a more refined model is unavailable.
Estimation of fragmenting stresses based on Equations 5.15 and 5.17 require an
estimation of the dissipation rate, ε. Here, it should be pointed out that the dissipation
rate varies considerably throughout the valve geometry. Because the full flow field
is rarely available, a mean efficient value, ε, is often calculated instead and seen as
characteristic of the process:
Ug3
ε = C3 (5.18)
h
with C3 = 1/20 (Mohr, 1987) or 1/80 (Innings and Trägårdh, 2007). Attempts have
been made in calculating local values of ε using computational fluid dynamics (e.g.,
Stevenson and Chen, 1997; Kleinig and Middelberg, 1997; Floury, Belletre et al.,
2004); however comparisons to experiments show that none of the proposed models
are able to describe the intense turbulent field accurately (Håkansson et al., 2012).
It should also be remembered that in addition to the spatial distribution of tur-
bulence, there is a stochastic component to the velocity fluctuations at each posi-
tion. The fragmenting stresses in Equations 5.13 and 5.16 are based on the average
velocity fluctuation. However, the stochastic nature of turbulence implies that much
more intense eddies exist temporarily. Because these eddies will give rise to faster
and more efficient breakup, the spread in the eddy energy should also be taken into
consideration, this was pointed out rather early by Kolmogorov (1949). However,
it is still not known if this has a significant effect on drop fragmentation in HPHs.
An important consequence of the Kolmogorov–Hinze theory is that an eddy of
length scale close to the drop diameter is most efficient in breaking the drop. This is
interesting in relation to Figure 5.5, which shows that most of the turbulent kinetic
energy is in eddies of significantly larger sizes. Thus, turbulent breakup of drops
does not occur where the total turbulent kinetic energy is highest but where the eddy
length scales are most efficiently distributed in comparison to drop sizes.
High-intensity turbulence is found in the outlet chamber of the HPH valve.

A high-velocity jet is created at the gap exit. The friction from the interaction
between the jet and low-velocity fluid in the outlet chamber gives rise to a developed
turbulent field. Experiments show that the highest levels of turbulent kinetic energy
for a drop traveling through the center of the gap is reached approximately 10 gap
heights into the outlet chamber. This distance is needed in order to transport energy
from the shear layers to the jet center. However, small-scale turbulent motion needs
approximately another 10 gap heights to obtain their maximum values as the trans-
port from large- to small-scale turbulent structures also requires time (Håkansson
et al., 2011; Innings and Trägårdh, 2007).
5.4.3 Fragmentation by cavitation

Due to narrow gap size in the homogenizer, the average fluid flow velocity is high
in the gap (see Figure 5.2b). Velocities are even higher in the early part of the gap
due to the development of a separation zone close to the seat (Håkansson et al., 2011;
Phipps, 1974).
From energy conservation considerations, it can be shown that these regions
of high velocity, and thus high dynamic pressure, must have a low static pressure.
The presence of low static-pressure zones have been shown experimentally in both
homogenizers (Loo and Carleton, 1953; Phipps, 1974) and other similar geometries
(Kawaguchi, 1971; Nakayama, 1964).
If the static pressure falls below the vapor pressure, vapor cavities can form and
expand. When transported into regions of higher local static pressure, these vapor
bubbles implode, sending out shock waves with extremely high local pressure and
temperature. The process is known as hydrodynamic boiling, or cavitation, and is
often characterized in terms of a cavitation number:
P∞ − PV (T )
NC = (5.19)
0.5ρCV∞2
where:
PV(T) is the vapor pressure at fluid temperature T
P∞ is the upstream static pressure
V∞ is a characteristic fluid velocity
Cavitation generally occurs when NC falls below a critical value and increase in
intensity with decreasing cavitation number (Brennen, 1995). Cavitation num-
ber decreases with increasing homogenizing pressures due to the increase in fluid
velocity.
There is ample experimental evidence showing cavitation taking place in the HPH.
Cavitation has been studied using a variety of techniques based on the different
properties of a cavitating flow, such as ultrasonic emissions (Håkansson et al., 2010;
Kurzhals, 1977), free radical formation (Floury, Legrand, and Desrumaux, 2004;
Shirgaonkar, Lothe, and Pandit, 1998), wear (Innings, Hultman et al., 2011; Phipps,
1985), and light scattering of vapor bubbles (Håkansson et al., 2010; McKillop
et al., 1955).
It is also known that cavitation can have an emulsifying effect as illustrated, for
example, in ultrasonic emulsification. This has led investigators to suggest cavita-
tion to be the dominant mechanism of drop fragmentation during high-pressure
homogenization (Kurzhals, 1977; Loo, Slatter, and Powell, 1950; McKillop, 1955);
see Section 5.5.2 for a more comprehensive discussion.
Due to the complexity of the cavitation process and interaction with the drop
interface, quantitative descriptions of the cavitation mechanisms, for example, in
terms of stresses, is not yet available.
Turning to the location of cavitation in the HPH valve, the gap inlet with the high
local velocity created close to the separation zone is a likely candidate (Phipps, 1974).
Cavitation visualizations agree with this and show cavitation bubbles in the beginning
or inside the gap (Håkansson et al., 2010; Phipps, 1974). These experiments indicate
the occurrence of bubble implosions before exiting into the outlet chamber, which is
also supported by comparisons of stability times of cavitation bubbles obtained from
bubble dynamics simulations with hold-up times (Håkansson et al., 2010).
An opposing view that cavitation extends into the outlet chamber has gained sup-
port by findings of intense wear in this region (Innings, Hultman et al., 2011).
5.4.4 StabiLizing StreSSeS and compariSonS

Emulsion drops are stabilized against fragmentation by two different stresses. First,
the deformation and fragmentation of a drop will be counteracted by the surface
forces; keeping the drop spherical minimizes the surface energy. The stabilizing
stress on a spherical drop due to the surface force is given by the Laplace pressure:
4γ
σstab,1 = (5.20)
d
Second, the deformation of the drop is countered by the viscous forces; more energy
is required to deform a more viscous drop. Hinze (1955) suggested
µD σfrag
σstab,2 = (5.21)
d ρD
where:
ρD is the dispersed phase density
µD is the dispersed phase kinematic viscosity
This formulation has gained experimental support from a large number of studies
(Calabrese, Chang, and Dang, 1986; Davies, 1985; Vankova, Tcholakova, Denkov,
Ivanov et al., 2007).
The total stabilizing stress can be obtained by the summation of Equations 5.20 and
5.21. A comparison of stresses can be seen in Figure 5.7. Fragmentation is assumed
to be due to the turbulent mechanisms with a dissipation rate relevant for production-
scale high-pressure homogenization (ε = 109 m2/s3). Figure 5.7a compares the total
stabilizing and fragmenting stresses for spherical drops of different diameters for a
1010
109
108
107
σ (Pa)
106
105
σstab = σstab,1 + σstab,2
104 σstab,1
σfrag
103 −2
10 10−1 100 101
(a) d (μm)
107
106
σ (Pa)
105
104
0 10 20 30 40 50
(b) μD (Pas)
FIGURe 5.7 (a) Illustration of fragmenting (turbulent inertial and viscous at ε = 109 m2/s3)
and stabilizing forces for µD = 30 mPas. (b) Stabilizing stress as a function of dispersed phase
viscosity for d = 20 nm (µC = 1 mPas, γ = 20 mN/m, ρD = 700 kg/m3).
viscous drop (µD = 30 mPas, µC = 1 mPas). The largest stable drop size is roughly
150 nm in diameter for this combination of parameters; that is, a drop size distribu-
tion being subjected to these hydrodynamic conditions for a sufficient time period
and without coalescence would be transformed toward a monodisperse size distribu-
tion with diameter 150 nm.
The Laplace pressure has also been included in the figure to show that it is
insufficient for stabilizing the drop by itself even if drops are relatively large.
Figure 5.7b displays the stabilizing stresses as functions of dispersed phase viscosity.
As seen in the figure, the resistance from viscous stabilizing stresses (Equation 5.21)
becomes important already at rather low dispersed phase viscosities.
5.5 INFLUeNCe OF OPeRATING PARAMeTeRs

ON eMULsIFICATION ResULT
The influence of operating parameters, such as homogenizing pressure, and product
characteristics, such as viscosity and oil content, is of essential importance for the
efficient utilization of HPHs. The experimental information on how drop sizes vary
with parameters can also help to shed some light on what mechanism actually gov-
erns fragmentation.
5.5.1 homogenizing preSSure

The increase in emulsion stability with increasing homogenizing pressures was
established early in HPH history and was attributed to a decrease in drop size.
A reduction in drop sizes with increasing pressures is also expected based on all
three of the suggested mechanisms in Section 5.4; see Equations 5.6, 5.18, and 5.19.
However, as light microscopy is not a suitable method for measuring submicron drop
sizes, it was not until special techniques for measuring average drop sizes and drop
size distributions were developed (Goulden, 1958; Walstra, 1968) that the relation
could be investigated quantitatively. A number of studies have now shown that an
average drop size (e.g., the Sauter mean diameter d32) scales as
d32 = C4 ⋅ ∆P q (5.22)
The constant C4 varies between homogenizers and the constant q is often found close
to −0.6 for pilot and production-scale homogenizers for low volume fractions of oil
(Phipps, 1985; Walstra and Smulders, 1998).
The experimentally obtained value of the constant q has been seen as an indi-
cation for turbulent fragmentation ever since Walstra (1969) pointed out that the
combination of Kolmogorov’s fragmentation theory with a reasonable scaling of
homogenizing pressure and dissipation rate of turbulent kinetic energy, ε, implies
that q = −0.6. However, as was pointed out by Phipps (1985), this is not a conclusive
evidence as it may very well be that other fragmentation mechanisms (e.g., cavita-
tion) could also act in accordance with q = −0.6. However, the higher q values often
found for laboratory-scale HPHs do imply that the fragmentations in these HPHs are
dominated by another mechanism (Walstra and Smulders, 1998).
In the last couple of years, it has also become possible to measure fragmenta-
tion rates directly (Vankova, Tcholakova, Denkov, Vulchev et al., 2007). The results
show that the experimentally obtained fragmentation rates are consistent in size and
scaling with turbulent breakup, as discussed in Section 5.4.2. This further supports
turbulence as the dominant mechanism of breakup.
5.5.2 Second-Stage eFFect and thoma number

The thermodynamic efficiency, that is, the ratio of the theoretically required energy
to that supplied to the HPH pump, can be a useful measure when comparing emul-
sification under different operating conditions. If the energy needed to increase sur-
face energy from that corresponding to a diameter of d0 to dend is the theoretical
minimum, the efficiency of homogenization is (Mohr, 1987).
6φD γ  1 1 
ηHPH =  −  (5.23)
∆Pp  dend d0 
where:
ϕD is the dispersed phase volume fraction
In other words, the efficiency is higher when obtaining smaller drops using the same
pressure or the same drop size when using a lower pressure difference. In discuss-
ing efficiency, it is important to properly define the term pressure difference. Many
HPHs are designed with two valves, or stages, coupled in series, as indicated in
Figure 5.8. The fluid enters the HPH at a pressure P0, which is often somewhat higher
than the atmospheric pressure due to feed pumps. The piston pump (or pumps) in the
HPH increases the pressure to a high pressure, P1, at which the fluid enters the first
valve. The back pressure to the first valve, P2, is set by adjusting the gap height in the
second valve. The fluid exits the HPH at pressure P3. It is important to note that it is
the pump pressure difference, ΔPp, that needs to be supplied to the system, whereas
it is the pressures over the gaps, ΔP1 and ΔP2, that control fragmentation. However,
if the upstream and downstream pressures of the HPH are assumed to be constant
(P3 = P0), then the pump homogenizing pressure is equal to the total pressure loss
over both stages. This explains why the pump pressure difference is often used to
describe valve pressure in discussing HPH droplet fragmentation.
The use of two consecutive valves is based on experiments reporting higher efficiency
for dual stages. The two-stage settings are often described using the Thoma number:
P2
Th = (5.24)
P1
Optimum efficiency is often reported in the region Th = 0.1 − 0.2 (Kurzhals, 1977;
Mohr, 1987; Pandolfe, 1982). This is interesting in relation to the fragmentation
mechanisms. Dividing the total pressure loss in two parts is expected to give lower
ΔPp = P1 − P0 ΔP1 = P1 − P2 ΔP2 = P2 − P3
P0 P1 P2 P3 ≈ Patm
Pump Stage 1 Stage 2
FIGURe 5.8 Schematic representation of pressures and pressure differences in the HPH.
laminar and turbulent stresses and thus larger drops and lower efficiency. Thus, if
laminar shear and/or turbulence dominated breakup, optimum efficiency would be
expected at Th = 0, corresponding to a single-stage configuration. However, using
a second valve also alters the absolute pressure in the outlet chamber. Although the
laminar and turbulent mechanisms only depend on the pressure difference over the
gap, cavitation is influenced by absolute pressure levels (see Equation 5.19). The
presence of an optimum efficiency as a function of Thoma number is, therefore, often
seen as evidence of cavitation controlling drop breakup. Studies of cavitation inten-
sity in HPHs show that the amplitude of a cavitation-induced ultrasound is affected
by the Thoma number (Håkansson et al., 2010; Kurzhals, 1977). Kurzhals (1977)
found that the Thoma number giving maximum efficiency also corresponded to the
most efficient emulsification, which supports this theory. A completely opposite view
is that the optimal Thoma number is obtained when suppressing cavitation in order
for it to not interact unfavorably with the turbulence (SPX, 2008). This is not an
implausible explanation as the presence of cavitation bubbles could influence the
structure and intensity of the turbulence (e.g., Iyer and Ceccio, 2002) and the sup-
pression of cavitation at sufficiently high Thoma numbers has been discussed theo-
retically and has been established experimentally (Håkansson et al., 2010; Kurzhals,
1977; Phipps, 1974).
A third explanation for the observed maximum, which does not rely on cavitation,
is based on observations of clustering in emulsions (Ogden, Walstra, and Morris,
1976). This has led investigators to propose that the second stage is needed in order
to break the clusters formed after the first stage. However, this fails to explain why
emulsions extracted after the first stage also seem to indicate optimal efficiency at
intermediate Thoma numbers (Loo and Carleton, 1953).
5.5.3 viScoSity oF the diSperSed and continuouS phaSeS

An increase in the dispersed phase viscosity is expected to stabilize the drop against
fragmentation during emulsification, as seen in Figure 5.7b. This has also been veri-
fied in a large number of experimental studies for HPHs (see Walstra and Smulders,
1998 and references therein) as well as for other emulsification equipment (e.g.,
Calabrese, Chang, and Dang, 1986).
A large number of these experiments have been summarized on the form
d32 = C5 ⋅µ Dp (5.25)
where:
p = 0.2 − 0.4 (Calabrese, Chang, and Dang, 1986; Kolb, 2001; Phipps, 1975).
Different suggestions have been made for explaining this value of p. Calabrese,
Chang, and Dang (1986) derived an expression for the limiting case when vis-
cous stresses dominate over surface forces in stabilizing a drop and obtained
p = 3/8. Walstra (1983), on the other hand, attributed the effect of the dispersed
phase viscosity to the timescale needed for drop deformation and breakup (c.f.
Equation 5.8). In this view, the fragmentation of more viscous drops is hin-
dered due to a lack of available time. Walstra and Smulders (1998) derived
p = 3/4 from a deformation timescale that limited drop fragmentation
Experimental evidence on the effect of the continuous phase viscos-

ity differs between investigations. Both Walstra (1974a) and Vankova,
Tcholakova, Denkov, Ivanov et al. (2007) observed a decrease in drop size as
a function of increasing continuous phase viscosity, whereas Pandolfe (1981)
observed a slight increase in drop size. The discrepancy can be understood by
realizing that the dispersed- and continuous-phase viscosity effects are not inde-
pendent of each other. Fragmentation from laminar shear is expected to depend
on the viscosity ratio, and by analogy, the same is true for turbulent breakup
from the viscous mechanism.
Experimental investigations often find a decrease in drop sizes with the
decreasing dispersed-to-continuous phase viscosity ratio (e.g., Wooster, Golding,
and Sanguansri, 2008; Lee and Norton, 2013). Studies often also report evidence
of a shift of mechanism occurring close to µD/µC = 4, which has been explained
in terms of a transition from fragmentation dominated by the turbulent inertial to
the turbulent viscous mechanism for decreasing viscosity ratios (Lee and Norton,
2013; Walstra and Smulders, 1998). Shift behavior for varying viscosity ratios
have also been reported in other studies. Emulsification experiments on simi-
lar homogenizer systems have shown that the number of drops formed in each
fragmentation event shifts from a small number (~4) when the dispersed phase
viscosity is low to a much larger number (~10) for higher viscosities (Tcholakova
et al., 2007). This would also be in line with the increased drop size distribution
width that has been reported for high viscosity oils in stirred tanks (Calabrese,
Chang, and Dang, 1986).
As discussed in Section 5.4.1, drops with high dispersed-to-continuous phase vis-
cosities (µD/µC > 4) cannot be broken by laminar shear in the inlet chamber due to
insignificant time available for deformation. Walstra (1983) argued that this applies
more generally (and thus also for turbulent viscous fragmentation in the outlet cham-
ber) from using a simplistic hydrodynamic model. This result would imply that lami-
nar shear (and the turbulent viscous mechanism by analogy) cannot fragment drops
of medium-to-high viscosities. However, the theoretical foundation for this general
case is not as solid as for the inlet chamber, and it should not be considered as an
established fact.
Continuous phase viscosity influences the global flow characteristics in addition
to direct effects discussed above. The Reynolds number is inversely proportional
to µC, which implies a turbulent reduction, which can give either a shift between
mechanisms or a reduction in fragmentation rate. High fluid viscosity also increases
the pressure losses and thus leads to lower gap velocities and velocity gradients
throughout the valve at a constant homogenizing pressure. This effect is expected
to decrease the efficiency of all three mechanisms with increasing continuous phase
viscosity.
5.5.4 voLume Fraction oF oiL

Experiments reveal an increase in drop size with increasing volume fraction of
oil (Phipps, 1985; Walstra, 1975). Several studies also show that q (Equation 5.22)
decreases with increasing volume fraction of oil (Phipps, 1985). Care must be taken
when interpreting these results. If the total amount of emulsifier is held constant
when increasing the volume fraction of the dispersed phase, this corresponds to a
decrease in the concentration of emulsifier available per volume of oil, and thus
there are two overlapping factors influencing the result: an increase in the amount of
dispersed phase and a decrease in the relative amount of emulsifier. The two factors
must be discussed separately.
Assuming a constant emulsifier-to-oil ratio, the volume fraction of oil is
expected to have two main effects on the fragmentation process. First, the number
of emulsion drops increases with increasing volume fraction of oil, which in turn
is expected to give rise to an increase in the rate of collisions and coalescence.
Coalescence can be described as a second-order process as it requires collisions
between pairs of drops, whereas fragmentation can be described as a first-order
process. Coalescence rate, therefore, increases faster than fragmentation rate with
increasing volume fraction of oil. Specific methods for measuring coalescence rate
during emulsification have been developed and confirm a fast increase in the rate
of coalescence with increasing volume fractions of oil (Lobo, Svereika, and Nair,
2002; Mohan and Narsimhan, 1997).
Second, increasing the volume fraction of the dispersed phase influences the flow
field in the homogenizer. Friction between the flow field and drops reduces the veloc-
ity gradients, which would imply decreased fragmentation rates. For turbulent flow,
the situation is more complex. Many theoretical studies on emulsification assume a
monotonic suppression of turbulent fluctuations, with an increase in volume fraction
of the dispersed phase (e.g., Coulaloglou and Tavlarides, 1977). However, addition
of dispersed phase drops can, depending on the properties of both flow field and
the drop, either enhance or attenuate on a turbulent flow (Gore and Crowe, 1989;
Poelma and Ooms, 2006). Prediction of the direction and size of the effect from
theory is presently not possible for volume fractions of technical relevance. However,
some experimental indications exist. Walstra (1974b) suggested that the smallest
eddy length scales were suppressed by the addition of polymers to the pre-emulsion
from observations of increasing drop sizes with polymer length scale. In a different
study, measurements of flow velocity in a scaled model indicated that the addition of
particles redistributes energy between different eddy length scales, which implies a
shift in the relative importance of the inertial to the viscous turbulent mechanisms
(Håkansson et al., 2013). In conclusion, experiments show a decreased efficiency
with an increased amount of the dispersed phase. However, when emulsifier con-
centration is high, the mechanisms involved are complex and not yet completely
understood.
Experiments show a decrease in resulting drop size as a function of increasing
emulsifier concentration as long as the concentration is low. For higher concentra-
tions, the effect is reduced in magnitude and the drop size levels out (see Walstra,
2005 and references therein). This has led investigators to suggest two different
regions: an emulsifier-poor region, where the surface chemistry of the drop emulsi-
fier system dominates, and an emulsifier-rich region, where the effects discussed
in Section 5.3 dominate (Tcholakova, Denkov, and Lips, 2008). The emulsifier
influences fragmentation directly through lowering of the interfacial tension, which

reduces the drop stability (see Equation 5.20). However, the major effect of emulsi-
fier concentration, and the origin of the two regimes, lies in how it influences the
coalescence behavior of the emulsion. For a more detailed discussion on the influ-
ence of emulsifier on coalescence behavior, the reader is referred to the reviews of
Tcholakova, Denkov, and Lips (2008), Bergenståhl and Claesson (1997).
5.6 THe LOCATION OF DROP BReAKUP

The location of regions of potentially fragmenting hydrodynamic conditions, that
is, a high laminar shear, sufficient turbulence of relevant length scale, and cavita-
tion in the HPH valve, is reasonably well known from hydrodynamic experiments.
These regions have been illustrated for a production-scale homogenizer shown
in Figure 5.9. As discussed in Section 5.4, laminar shear is found in the inlet and
intense turbulence is found in the outlet chambers at some distance from the gap
exit. Cavitation is most likely found only in the gap, but indications that it can extend
further have also been reported.
Knowing the location of drop breakup is interesting for finding the mechanism
responsible as the regions of intense potential fragmentation have a relatively low
overlap. Direct observations of fragmentation in HPHs have been carried out by
constructing valves with optical access (Innings and Trägårdh, 2005, 2007). Results
show breakup taking place in the outlet chamber, 10–20 gap heights downstream
of the gap exit. A comparison with visualizations of cavitation and measurements
of fluid flow show that the drops are broken up in a region where there is a high
intensity in small eddy length-scale turbulence (see Figure 5.9) (Håkansson et al.,
2011). Innings and Trägårdh (2005) also concluded that the morphology of breaking
Impact
ring
Cavitation
Turbulence
Laminar shear
Seat
Fragmentation
FIGURe 5.9 Illustration of the location of a high laminar shear, cavitation, and intense tur-
bulence according to hydrodynamic studies, compared to the region of drop breakup accord-
ing to visualization experiments.
drops was consistent with a combination of turbulent inertial and turbulent viscous
fragmentation.
5.7 CONCLUsIONs ON THe FRAGMeNTATION

MeCHANIsMs IN THe HPH
The experimentally observed scaling with different parameters, the location of
breakup, and hydrodynamics of the HPH valve all give information on the influence
of the different mechanisms on fragmentation.
In summary, laminar shear is unlikely to have a large effect on fragmentation
in pilot and production-scale homogenizers. Even if shear can be significant in the
inlet chamber, the time spent in these regions is insufficient for the breakup to occur.
Laminar shear will exist in the boundary layers of the gap; however, for pilot and
production-scale homogenizers, the boundary layer is thin and the majority of the
drops will, therefore, remain unaffected by laminar shear. It should be remembered
that with regard to small regions of high-velocity gradients, there is a considerable
difference between the HPH and other systems such as static mixers or impellers in
that the emulsion only passes through the valve once or at most a few times, whereas
mixers are often operated in batch mode until they have reached a steady-state drop
size. Regions giving high fragmenting rates while simultaneously only reaching a
small number of drops per passage should thus have a very limited effect on the
emulsification result in the HPH.
The boundary layers merge early in the gap, which can create high levels of
shear throughout the gap for laboratory-scale HPHs. This could give rise to lami-
nar breakup for these small machines. The difference in q value in Equation 5.22
between large- and small-scale HPHs could be interpreted as evidence for a differ-
ence in dominant mechanisms between HPH scales. Because both turbulence and
cavitation are expected to occur regardless of scale, it is reasonable to conclude that
laminar shear plays a major role for laboratory-scale homogenization.
Cavitation is present during high-pressure homogenization except when using a
high back pressure. Cavitation increases in amplitude with homogenizing pressure.
The observed optimum emulsification efficiency at an intermediate Thoma number
has been interpreted as evidence of cavitation-dominated fragmentation; however,
there are alternative explanations to this effect, which do not require cavitation-
driven breakup. Comparing the location of fragmentation and cavitation does not
support it being the dominant mechanism.
The scaling of homogenizing pressure, the dispersed phase viscosity, and
fragmentation rates are all consistent with the turbulent fragmentation theory.
Together with the observed match between the region of most intense small-scale
turbulence and the location of fragmentation, it serves as strong experimental
evidence for fragmentation by turbulent mechanisms. Inspection of the shape of
breaking drops together with the commonly observed shifts between mechanisms
with varying viscosity ratios indicate that both turbulent inertial and turbulent
viscous mechanisms are active to some extent and that the viscous mechanism
becomes more important for systems where the relative dispersed phase viscosity
is low.
ReFeReNCes
Bergenståhl, B.A., Claesson, P.M., 1997. Surface forces in emulsions. In: Food Emulsions,
Friberg, S.E., Larsson, K. (eds.), Marcel Dekker Inc., New York, 57–110.
Brennen, C.E., 1995. Cavitation and Bubble Dynamics. Oxford University Press, New York.
Calabrese, R.V., Chang, T.P.K., Dang, P.T. 1986. Drop breakup in turbulent stirred-tank con-
tactors. Part I: Effect of dispersed-phase viscosity. AIChE Journal 32(4), 657–666.
Coulaloglou, C.A., Tavlarides, L.L., 1977. Description of interaction processes in agitated
liquid-liquid dispersions. Chemical Engineering Science 32, 1289–1297.
Davies, J.T., 1985. Drop sizes of emulsions related to turbulent energy dissipation rates.
Chemical Engineering Science 40, 839–842.
Floury, J., Belletre, J., Legrand, J., Desrumaux, A., 2004. Analysis of a new type of high
pressure homogeniser. A study of the flow pattern. Chemical Engineering Science 59,
843–853.
Floury, J., Desrumaux, A., Axelos, M.A.V., Legrand, J., 2002. Degradation of methylcellulose
during ultra-high pressure homogenisation. Food Hydrocolloids 16, 47–53.
Floury, J., Legrand, J., Desrumaux, A., 2004. Analysis of a new type of high pressure
homogeniser. Part B. study of droplet break-up and recoalescence phenomena. Chemical
Engineering Science 59, 1285–1294.
Gaulin, A., 1904. Process of treating milk or similar liquids. US Patent No. 753,792, March 1.
Gore, R.A., Crowe, C.T., 1989. Effect of particle size on modulating turbulent intensity.
International Journal of Multiphase Flow 15, 279–285.
Goulden, J.D.S., 1958. Light transmission by dilute emulsions. Transactions of the Faraday
Society 54, 941–945.
Grace, H.P., 1982. Dispersion phenomena in high viscosity immiscible fluid systems and
application of static mixers as dispersion device in such systems. Chemical Engineering
Communications 14, 225–277.
Håkansson, A., Fuchs, L., Innings, F., Revstedt, J., Bergenståhl, B., Trägårdh, C. 2010. Visual
observation and acoustic measurement of cavitation in an experimental model of a high-
pressure homogenizer. Journal of Food Engineering 100(3), 504–513.
Håkansson, A., Fuchs, L., Innings, F., Revstedt, J., Trägårdh, C., Bergenståhl, B., 2011. High
resolution experimental measurement of turbulent flow field in a high pressure homoge-
nizer model and its implications on turbulent drop fragmentation. Chemical Engineering
Science 66(8), 1790–1801.
Håkansson, A., Fuchs, L., Innings, F., Revstedt, J., Trägårdh, C., Bergenståhl, B., 2012.
Experimental validation of k-ε RANS-CFD on a high-pressure homogenizer valve.
Chemical Engineering Science 71, 264–273.
Håkansson, A., Fuchs, L., Innings, F., Revstedt, J., Trägårdh, C., Bergenståhl, B., 2013.
Velocity measurements of turbulent two-phase flow in a high-pressure homogenizer
model. Chemical Engineering Communications 200, 93–114.
Hinze, J.O., 1955. Fundamentals of the hydrodynamic mechanism of splitting in dispersion
processes. AIChE Journal 1, 289–295.
IDF, 2010. The world dairy situation 2010. Bulletin of the International Dairy Federation 446.
Innings, F., Fuchs, L., Trägårdh, C., 2011. Theoretical and experimental analyses of drop
deformation and break-up in a scale model of a high-pressure homogenizer. Journal of
Food Engineering 103, 21–28.
Innings, F., Hultman, E., Forsberg, F., Prakash, B. 2011. Understanding and analysis of wear
in homogenizers for processing liquid food. Wear 271, 2588–2598.
Innings, F., Trägårdh, C. 2005. Visualization of the drop deformation and break-up process
in a high pressure homogenizer. Chemical Engineering & Technology 28(8), 882–891.
Innings, F., Trägårdh, C., 2007. Analysis of the flow field in a high-pressure homogenizer.
Experimental Thermal and Fluid Science 32, 345–354.
Iyer, C.O., Ceccio, S.L., 2002. The influence of developed cavitation on the flow of a turbulent
shear layer. Physics of Fluids 14, 3414–3431.
Kawaguchi, T., 1971. Entrance loss for turbulent flow without swirl between parallel discs.
Bulletin of the Japan Society of Mechanical Engineers 14, 355–363.
Kleinig, A.R., Middelberg, A.P.J., 1996. The correlation of cell disruption with homoge-
nizer valve pressure gradient determined by computational fluid dynamics. Chemical
Engineering Science 41(23), 5103–5110.
Kleinig, A.R., Middelberg, A.P.J., 1997. Numerical and experimental study of a homogenizer
impinging jet. AIChE Journal 43(4), 1100–1107.
Kolb, G., 2001. Zur Emulsionsherstellung in Blendensystemen. Doctoral Thesis, University
of Bremen, Germany.
Kolmogorov, A.N., 1949. On the breakage of drops in a turbulent flow. Doklady Akademii
Nauk SSSR 66, 825–828. (Originally in Russian. Reprinted and translated in Selected
Works of A.N. Kolmogorov, Volume 1: Mathematics and Mechanics, Tikhomirov, V.M.
[ed.], 1991, 339–343.)
Kurzhals, H.-A., 1977. Untersuchungen über die physikalisch-technischen Vorgänge beim
Homogenisieren von Milch in Hochdruck-Homogenisiermaschinen. Doctoral Thesis,
University of Hannover, Germany.
Lee, L., Norton, I. 2013. Comparing droplet breakup for a high-pressure valve homogenizer
and a microfluidizer for the potential production of food-grade nanoemulsions. Journal
of Food Engineering 114, 158–163.
Lobo, L., Svereika, A., Nair, M., 2002. Coalescence during emulsification. 1. Method develop-
ment. Journal of Colloid and Interface Science 253, 409–418.
Loo, C.C., Carleton, W.M., 1953. Further studies of cavitation in the homogenization of milk
products. Journal of Dairy Science 36, 64–75.
Loo, C.C., Slatter, W.L., Powell, R.W., 1950. A study of the cavitation effect in the homogeni-
zation of dairy products. Journal of Dairy Science 33, 692–702.
McKillop, A.A., Dunkley, W.L., Brockmeyer, R.L., Perry, R.L., 1955. The cavitation theory of
homogenization. Journal of Dairy Science 38, 273–283.
Middelberg, A.P.J. 1995. Process-scale disruption of microorganisms. Biotechnological
Advances 13(3), 491–551.
Mohan, S., Narsimhan, G., 1997. Coalescence of protein-stabilized emulsions in a high-
pressure homogenizer. Journal of Colloid and Interface Science 192, 1–15.
Mohr, K.-H., 1987. High-pressure homogenization. Parts II. The influence of cavitation on
liquid-liquid dispersion in turbulence fields of high energy density. Journal of Food
Engineering 6, 311–324.
Nakayama, Y., 1964. Action of the fluid in the air-micrometer. Bulletin of the Japan Society of
Mechanical Engineers 7, 698–707.
Ogden, L.V., Walstra, P., Morris, H.A., 1976. Homogenization-induced clustering of fat glob-
ules in cream and model systems. Journal of Dairy Science 59, 1727–1737.
Pandolfe, W.D., 1981. Effect of dispersed and continuous phase viscosity on droplet size of
emulsions generated by homogenization. Journal of Dispersion Science and Technology
2, 459–474.
Pandolfe, W.D., 1982. Development of the new gaulin micro-gap™ homogenizing valve.
Journal of Dairy Science, 65, 2035–2044.
Phipps, L.W., 1974. Cavitation and separated flow in a simple homogenizing valve and
their influence on the break-up of fat globules in milk. Journal of Dairy Research
41, 1–8.
Phipps, L.W., 1975. The fragmentation of oil drops in emulsion by a high-pressure homo-
genizer. Journal of Physics D: Applied Physics 8, 448–462.
Phipps, L.W., 1985. The High Pressure Dairy Homogenizer. The National Institute for
Research in Dairying, Reading.
Poelma, C., Ooms, G., 2006. Particle-turbulence interaction in a homogeneous, isotropic

turbulent suspension. Applied Mechanics Reviews 59, 78–89.
Pope, S.B., 2000. Turbulent Flows. Cambridge University Press, Cambridge.
Saffman, P.G. 1965. The lift on a small sphere in a slow shear flow. Journal of Fluid Mechanics
22(2), 385–400.
Schlichting, H., Gersten K., 2000. Boundary Layer Theory, 8th ed. Springer, Berlin, Germany.
Shirgaonkar, I.Z., Lothe, R.R., Pandit, A.B., 1998. Comments on the mechanism of microbial
cell disruption in high-pressure and high-speed devices. Biotechnology Progress 14,
657–660.
SPX, 2008. The effect of the second-stage homogenizing valve. Technical Bulletin 58. APV.
Stevenson, M.J., Chen, X.D., 1997. Visualization of the flow patterns in a high-pressure
homogenizing valve using a CFD package. Journal of Food Engineering 33, 151–165.
Stone, H.A. 1994. Dynamics of drop deformation and breakup in viscous fluids. Annual
Review of Fluid Mechanics 26, 65–102.
Taylor, G.I., 1934. The formation of emulsions in definable fields of flow. Proceedings of the
Royal Society A 146, 501–523.
Tcholakova, S., Denkov, N.D., Lips, A. 2008. Comparison of solid particles, globular pro-
teins and surfactants as emulsifiers. Physical Chemistry Chemical Physics 10(12),
1597–1712.
Tcholakova, S., Vanova, N., Denkov, N.D., Danner, T., 2007. Emulsification in turbulent
flow: 3. Daughter drop-size distribution. Journal of Colloid and Interface Science 310,
570–589.
Vankova, N., Tcholakova, S., Denkov, N.D., Ivanov, I.B., Vulchev, V.D., Danner, T. 2007.
Emulsification in turbulent flow 1. Mean and maximum drop diameters in inertial and
viscous regimes. Journal of Colloid and Interface Science 312, 363–380.
Vankova, N., Tcholakova, S., Denkov, N.D., Vulchev, V.D., Danner, T., 2007. Emulsification
in turbulent flow 2. Breakage rate constants. Journal of Colloid and Interface Science
313, 612–629.
Walstra, P. 1968. Estimating globule-size distributions of oil-in-water emulsions by spectro-
turbidimetry. Journal of Colloid and Interface Science 27(3), 493–500.
Walstra, P., 1969. Preliminary note on the mechanism of homogenization. Netherlands Milk
and Dairy Journal 23, 290–292.
Walstra, P. 1974a. Influence of rheological properties of both phases on droplet size O/W
emulsions obtained by homogenization and similar processes. Dechema Monographie
77, 87–94.
Walstra, P., 1974b. Turbulent depression by polymers and its effect on disruption of emulsion
droplets. Chemical Engineering Science 29, 882–885.
Walstra, P., 1975. Effect of homogenization on the fat globule size distribution in milk.
Netherlands Milk and Dairy Journal 29, 279–294.
Walstra, P., 1983. Formation of emulsions. In: Encyclopedia of Emulsion Technology Volume I:
Basic Theory, Becher, P. (ed.), Marcel Dekker Inc., New York, 57–127.
Walstra, P., 2005. Emulsions. In: Fundamentals of Interface and Colloid Science, Lyklema, J.
(ed.), Elsevier, Amsterdam, the Netherlands, 8.1–8.94.
Walstra, P., Smulders, P.E.A., 1998. Emulsion formation. In: Modern Aspects of Emulsion
Science, Binks, B.P. (ed.), Royal Society of Chemistry, Cambridge, 56–99.
Wooster, T.J., Golding, M., Sanguansri, P. 2008. Impact of oil type on nanoemulsion formation
and Ostwald ripening stability. Langmuir 24, 12758–12765.
6 High-Pressure
Homogenizer Design
Fredrik Innings
CONTENTS
6.1 History of the Homogenizer ......................................................................... 149
6.2 The Homogenizer ......................................................................................... 152
6.3 The Homogenization Device ........................................................................ 155
6.4 Wear in the Homogenization Gap ................................................................ 157
6.5 Scale Up and Gap Height.............................................................................. 161
6.6 Economic Aspects ........................................................................................ 164
6.7 The Homogenizer in a Processing Line ....................................................... 166
6.7.1 Split Homogenization ....................................................................... 166
6.7.2 Full Stream Homogenization ............................................................ 167
6.7.3 Partial Homogenization .................................................................... 167
References .............................................................................................................. 167
ABSTRACT Starting with the history of high-pressure homogenizers, the pros

and cons of design choices in modern commercial homogenizers are clarified, with
focus on efficiency, process line integration, wear, and investment and running costs.
6.1 HISTORY OF THE HOMOGENIZER

In 1899, Auguste Gaulin was granted a patent for a machine with the purpose of
stabilizing, or fixing, a fat emulsion against gravity separation, described in French
as “fixer la composition des liquides.” The machine and milk homogenized by it
was shown at the world fair in Paris in 1900. The first US patent, US756953 (Gaulin
1904) (Figure 6.1), was granted in 1904, and in 1909 Gaulin formed the Manton-
Gaulin Manufacturing Company and manufactured its first homogenizer.
The first Gaulin homogenizer was very similar to today’s machine with a three-
piston pump and a pump block with back pressure valves, but there was one impor-
tant exception. The homogenization device consisted of a bundle of capillary tubes
against one end of which a spring pressed a concave valve. The capillary tubes were
quite soon replaced with a single tube, and during the following 50 years, the homog-
enization gaps had the same basic wide gap design. In a wide gap design, the inner
diameter of the gap is in the order of a third outer diameter (Figure 6.2). This gives a
very wide, or long, gap resulting in much higher velocity at the inlet of the gap com-
pared to the outlet, and thus substantial friction pressure drop in the gap.
149
(a)
(b)
FIGURE 6.1 Patent sketch from the first US homogenization patent: (a) side view and
(b) top view. (Data from Gaulin, A., System for intimately mixing milk. US Patent US756953,
filed September 30, 1902, issued April 12, 1904.)
Di
Do
FIGURE 6.2 Principal drawing of an early wide gap design.

High-Pressure Homogenizer Design 151
In 1925, the two-stage homogenizer was patented by Gaulin, but it took quite
some time before it reached widespread use. For some products, the wide gap design
reduces the need for a second stage, as the outer part of the gap serves as a pressure
controller for the upstream part. This effect was optimized by Gaulin, and in 1955,
they patented the Liquid Whirling homogenizing valve, where the product is subject
to a number of rapid pressure and velocity changes as it flows through the gap.
During the following years, a lot of similar devices were invented, for example,
the perforated metal discs by Cherry-Burrell and the Multi-Flo compressed wire by
Crepaco. About 30 years ago, the knife-edge gap (Figure 6.3) was developed with a
large outer diameter in the order of 100 mm and a short gap in the order of 1 mm,
which is today’s standard design.
The last step in gap design was started by Gaulin patenting the MicroGap™ in
1983, US4383769. Here, the flow is distributed in six parallel gaps, making very high
production capacities possible (discussed further in Section 6.5).
Up to the 1960s, the homogenizers in the dairies were used to homogenize pas-
teurized milk with a shelf life of less than a week. To achieve this, a homogenization
pressure of 120–150 bar is required. First launched in Switzerland in 1961, asep-
tic milk with a shelf life of many months demanded a homogenization pressure of
220–250 bar, ensuring that the fat droplets are small enough to be mixed around by
the Brownian motion. To reduce the risk of recontamination of the milk, the homo-
genizer was placed upstream in the nonaseptic part of the process. In 1963, Alfa Laval
(now Tetra Pak) launched the VTIS™ aseptic process, where the milk was heated
by direct steam injection. The VTIS process demanded that the homogenizer be
placed downstream of the sterilizing section, thereby defining the need for an aseptic
homogenizer. There are two demands on an aseptic machine; first, it must be able to
handle the presterilization process, where the complete processing line is heated up
to 120°C–140°C for 20–40 minutes, killing any microorganisms remaining after the
cleaning; second, great care must be taken so that no microorganisms can enter the
sterile milk and recontaminate it. The presterilization is taken care of by upgrading
FIGURE 6.3 Modern knife-edge gap. (Courtesy of Tetra Pak.)

all seals and gaskets to qualities that can handle the high temperatures. The only
position microorganisms can contaminate the product in a homogenizer is along the
pistons. As the pistons move in and out they can bring microorganisms into the milk.
To counter this Alfa Laval prolonged the pistons and doubled the seals. Between the
seals, a sterilization zone was formed by flushing it with steam or hot condensate,
normally at 120°C.
6.2 THE HOMOGENIZER

The homogenizer is a large high-pressure pump with a homogenizing device (see
Figure 6.4). The piston pump is driven by a powerful electric motor (1) via belts
(2) and pulleys through a gearbox (5) to the crankshaft (3) and the connecting-rod
transmission, which converts the rotary motion of the motor to the reciprocating
motion of the pump pistons (4). The efficiency of the drive system is about 96% and
the access heat is cooled away with fans for the motor and a water cooler for the
crankcase.
The high-pressure pump of the homogenizer normally has three to five pistons
(4), running in cylinders in a high-pressure block (7). The pistons are made of highly
resistant materials and are sealed with double piston seals. Water is supplied to the
space between the seals to lubricate the pistons. A mixture of hot condensate and
steam can also be supplied to prevent reinfection when the homogenizer is placed
downstream in aseptic processes.
A piston pump is a positive pump and its capacity can only be adjusted by chang-
ing the speed of the motor or changing the size of the pistons. Each machine type
has a maximum crankshaft speed and a maximum crankshaft force. By choosing
the right diameter of the pistons, the needed capacity and pressure can be achieved;
1. Main drive motor

2. V-belt transmission
3. Crankshaft
4. Pistons
5. Gearbox
6. Hydraulic gap-setting system
7. Pump block
8. Homogenization device
FIGURE 6.4 The homogenizer machine. (Courtesy of Tetra Pak.)

(a) (b)
FIGURE 6.5 The wet end of the homogenizer: (a) Cut through the check valves seen from
the front, and (b) cut through a piston seen from the side. (Courtesy of Tetra Pak.)
therefore, a large piston diameter gives a high-capacity machine with a moderate

pressure, whereas a small piston diameter gives a high-pressure machine with a
moderate capacity. A larger machine type has a longer stroke length and/or more
pistons. In many cases, it can also produce a higher crankshaft force; that is, the
pistons can also have a larger diameter.
The wet end of the three-piston homogenizer can be seen in Figure 6.5. The flow
enters from the bottom right and passes through the first set of check valves. The
pistons are oriented into the image in Figure 6.5a but can be seen in Figure 6.5b. The
flow then passes the second set of check valves and continues to the homogeniza-
tion device. The pressure is measured just upstream of the homogenization gap. The
check valves used to be ball valves but nowadays a more advanced design is avail-
able, for example, the mushroom type that can be seen in Figure 6.5.
A piston pump will always generate a pulsating flow. Even though the motor
rotates with a constant rpm, the limited length of the connecting rod and the com-
pressibility of the product will result in the quite complex flow curve that can be seen
in Figure 6.6.
These flow variations will accelerate and decelerate the liquid creating a pulsat-
ing pressure in the pipes upstream and downstream of the homogenizer. The magni-
tude of the pulsations depends on the position of the homogenizer in the processing
line. The pulsations from a three-piston homogenizer in a typical line are shown in
Figure 6.7, where very high pressure peaks in the outlet line as well as negative pres-
sures in the inlet pipe can be seen.
These pressure variations will result in vibrations or even breakage of the pipes.
They will also disturb the control system of the processes upstream and downstream
of the homogenizer. But the most serious effect occurs when the inlet pressure drops
below the boiling point of the product. Then the product in the inlet valves will boil
and form cavitation bubbles that will implode and quite quickly wear down the valves.
To reduce the pressure peaks and to avoid cavitation, dampers are often used.
Figure 6.8 shows the simplest and the most common dampers (Bylund 2003).
Relative flow rate variation over one cycle

Relative flow variation in Relative flow variation out
15
10
5
0
0 0.25 0.5 0.75 1
−5
Percentage (%)
−10
−15
−20
−25
−30
−35
−40
Cycle
FIGURE 6.6 The flow variations from a three-piston homogenizer.
Pressure without dampers at inlet and outlet over one cycle

Pressure variations at inlet (bar) Pressure variations at outlet (bar)
14
12
10
8
Inlet (bar)
0
0 0.25 0.5 0.75 1
−2
Cycle
FIGURE 6.7 Pressure variations for a three-piston homogenizer in a typical process line.
The damper is just a closed pipe filled with air. If placed closed to the homogenizer,
they will work very well but have two main disadvantages. First, they need to be dis-
mantled and cleaned manually; second, the air will slowly dissolve into the product,
which limits the runtime of the dampers. There have been many attempts to solve
these two problems, for example, by adding cleaning in place (CIP) or compressed
air inlets to the dampers, but none has been totally successful in incorporating all the
requirements of the modern food industry.
FIGURE 6.8 Typical homogenization dampers. (Courtesy of Tetra Pak.)
As it is a positive pump, a homogenizer should not be operated in a series with

other positive pumps, unless there is a bypass—otherwise, the result can be extreme
pressure variations, which could damage the equipment. If the flow can be stopped
downstream of a high-pressure pump, a safety device must be installed that opens
before the pipe bursts.
6.3 THE HOMOGENIZATION DEVICE

In Figure 6.9, a typical two-stage homogenization device can be seen. The high-
pressure product enters from the pump block (9) and the homogenization pressure
is measured (8). The fluid is accelerated and passes through the first-stage gap (10).
The product then continues to the second homogenization stage (1), where it is
again accelerated as it passes through the gap between the seat (7) and the forcer
(6). The height of the gap and thus the homogenization pressure is adjusted by the
force the forcer is pressed against the seat. In old or in laboratory homogenizers
where the forcer is spring-loaded, this is often done by tightening a screw, but in
modern industrial homogenizers, it is done by a hydraulic or, in some machines,
1. Second homogenization stage

2. Hydraulic oil inlet
3. Hydraulic oil piston
4. Plunger
5. Product outlet
6. Forcer
7. Seat
8. Homogenization pressure transmitter
9. High pressure product from pump block
10. First homogenization stage
FIGURE 6.9 The homogenization device. (Courtesy of Tetra Pak.)
by a pneumatic system. By controlling the pressure of the incoming hydraulic oil

(2), the force developed by the hydraulic oil piston (3), affecting the forcer through
the plunger (4), is very accurately controlled. An additional benefit of the hydraulic
system is that it can be used to remotely control the gap position and, for example,
automatically open the gaps during CIP.
In most modern homogenizers, the pressure is measured before both the first and
the second stages, but traditionally the second stage pressure is not measured. To
find the correct hydraulic pressure for the second stage, the first-stage hydraulic pis-
ton is depressurized and the second-stage hydraulic piston is pressurized until the
desired second-stage pressure is achieved. Note also that the industrial definition of
the homogenization pressure is the total pressure measured on the pressure meter (8),
not the pressure drop over a gap, as can sometimes be seen in the scientific literature.
As the second stage is normally run with a much lower pressure, it would be pos-
sible to manufacture it with, for example, a smaller hydraulic oil piston and a weaker
housing, but for simplicity, the first and the second stages are normally interchangeable.
Among other factors, the second stage sets the pressure after the first gap and thus
controls the cavitation in and at the outlet in the first gap. The amount of cavitation
is dependent on the Thoma number (Th), in a homogenizer defined as the relation
between the second-stage pressure and the total pressure. It has been found that the
most efficient homogenization is achieved at a Thoma number of 0.15–0.2. Kurzhals
and Håkansson also found the highest levels of cavitation energy at these Thoma
numbers (Håkansson et al. 2010; Kurzhals 1997).
Almost all industrial homogenizers have the variable gap design, where it is the
force from the hydraulic piston or the screw that forces the gap to close. Some indus-
trial and many laboratory homogenizers, for example, the microfluidizer, use instead
fixed-round or slit-shaped orifices.
6.4 WEAR IN THE HOMOGENIZATION GAP

Due to the high velocities in the gap, wear can be quite severe, especially if the
product contains small and hard particles. Figure 6.10 shows an example of extreme
wear from a tomato paste production plant.
To counter wear, two methods are used. First, the seat and the forcer are made
of wear-resistant materials, from the hardest stainless steels (e.g., the device shown
in Figure 6.10) to tungsten alloys, ceramics, or even diamond-like carbon. Second,
the parts are made as simple and interchangeable as possible. Figure 6.11 shows an
example of a modern seat and forcer, where both parts can be turned over and are
made with a simple and cost-efficient design.
FIGURE 6.10 The wear pattern created on the forcer by tomato paste after 30 minutes of
homogenization. The material is Wallex 20. The wear pattern contains many grooves in the
radial direction, so that a wavy pattern is created in the circumferential direction. The typical
reflection is also seen as a wavy pattern in the radial direction, with the first crater being the
deepest. (Courtesy of Tetra Pak.)
FIGURE 6.11 A modern seat and a forcer that can be easily exchangeable and turnable.
(Courtesy of Tetra Pak.)
The amount of wear of the gap is totally dependent on the type of particles and
the amount of cavitation in the gap. Figure 6.12 shows the relative amount of wear for
a couple of particles/product types (Innings et al. 2010). Note that the scale is loga-
rithmic and the worst particles give a 5-log increase of the wear. Corundum particles
were added as a scientific reference, but the calcium particles are common in food
products in, for example, fortified milk.
The different products results in quite different wear patterns in the gap. Figure 6.13
shows a stainless steel seat that has been running on tap water for 600 hours at 600
bar and with a Thoma number of 0.2. The flow is from right to left, the length of the
undamaged gap was 0.5 mm, and the image is 0.7 mm wide. The seat had a surface
finish of Ra = 0.8 µm and on the unworn surfaces the machine groves can easily be
seen. The presence of the machined groves is a good measure, indicating that the
surface has not worn out. The wear pattern on the outlet is porous and rough, which
is typical for cavitation–erosion and is completely different compared to the case
where particles were added. Note that the inlet is totally undamaged, which shows
that a Thoma number of 0.2 suppresses the cavitation at the inlet-separation bubble.
Figure 6.14 shows a stainless steel seat that has been running on tomato paste for
2 hours at 280 bar and with a Thoma number of 0.2. A wide crater has been formed
Effect of particle type on wear

10,000 Corrundum, 10 μm
1,000 Corrundum, 3 μm
Total weight loss (mg/h)
100
Tomato
10
Calcium, 10 μm
1
0.1 Milk
0.01
0.001
0.0001
FIGURE 6.12 The weight loss for different particle types and with cavitation without added
particles as reference. Note that the scale is logarithmic.
FIGURE 6.13 Stainless steel seat that has been running for 600 hours at 600 bar on tap
water and with a Thoma number of 0.2.
FIGURE 6.14 Wear from tomato paste on a stainless steel seat with Th = 0.2. The surface
has a corrosive appearance.
FIGURE 6.15 Wear from calcium particles. Low cavitation (Th = 0.2). This wear has
removed all the machining marks in the gap region and has also caused a lot of deep craters on
the inlet and the outlet areas. The area between the craters has a matte appearance.
on both the forcer and the seat. The surface is smooth and has a matte look. No outlet
damages are appearing on either the forcer or the seat. The radial machine grooves
have been worn away also at the regions outside of the crater.
Figures 6.15 and 6.16 show the wear from calcium particles at the standard and a
low Thoma number, respectively. It shows that there is a great synergy between the
particles and the cavitation. The mechanism is likely to be that the imploding cavita-
tion bubbles accelerate the particles, so that they hit the gap surfaces at high velocities.
Figure 6.17 shows the profiles of the worn-out gaps from Figures 6.15 and 6.16.
The amount of wear is extensive in Figure 6.17b, but it should be noted that it was
still possible to keep the homogenization pressure, so it is not in any case worse than
a worn-out commercial valve.
This insensitivity to wear is one of the main advantages of the variable-gap type
homogenization devices, where it is the force from the hydraulic piston that keeps
the gap height low. As the gap is worn out, the force on the forcer closes the gap
slightly, making sure that the homogenization pressure is always constant. This is
of course only true if the wear is fairly even and not too severe. When the wear is
too severe, channels will be worn down into the seat and the forcer, and when all
the product flows in the channels, it will not be possible to close the gap anymore;
the machine cannot withstand the homogenization pressure, which indicates that it
FIGURE 6.16 Wear from calcium particles. High cavitation (Th = 0); the crater near the
inlet has increased considerably in depth and width and the reflection between the two has
started, see the crater in Figure 6.17b.
Forcer
Inlet
Seat
(a)
Forcer
Inlet
(b) Seat
FIGURE 6.17 (a, b) Profile of the worn-out gaps from calcium particles. (a) Standard cavita-
tion (Th = 0.2). (b) High cavitation (Th = 0). The length of the gap is always 0.5 mm and the
gap height is in the order of 100 µm.
is time to replace the parts. A fixed orifice does not have this automatic adjusting
function, so when a fixed orifice is worn, the pressure drop decreases and the flow
rate must be increased to keep the homogenization pressure and this is normally not
possible in an industrial processing line. The other advantage of the variable gap
is, of course, that the flow rate and the homogenization pressure can be controlled
independent of each other, making it possible to keep the homogenization pressure

when running the process line at reduced capacity.
It should be noted that an industrial gap will run in a more or less worn-out state
for most of its lifetime. During the first hours of its lifetime, it will be crisp with
sharp edges, but very soon these will be worn down and the gap will run with a more
rounded shape for thousands of hours. This must be taken into account when design-
ing and comparing gaps. The gap that is very efficient when testing it for a few hours
in the lab might lose its advantage after a few hours of production. This is also the
case why more extreme knife edge gaps, where the gap is made up of two razorblades
opposite each other are not possible in industrial homogenizers.
6.5 SCALE UP AND GAP HEIGHT

It has been found out that to achieve good homogenization efficiency, the gap height
cannot be too high. Figure 6.18 shows the homogenization efficiency presented with
the Sauter mean diameter as a function of the gap height. The graph is based on
many industrial homogenizers with different gaps and flow rates, where the effect of
only the gap height has been extracted. As can be seen, there is quite a large effect of
gap height: lower gap heights are more efficient.
At a constant pressure, the gap height is directly related to the flow rate and the
circumference of the gap; that is, to keep the gap height constant when doubling the
flow, the circumference and therefore the diameter of the gap need to be doubled as
well. Figure 6.19 shows the effect of capacity on the homogenization device. The dot-
ted line shows the calculated device diameter if the gap height is chosen to be 50 µm
as this was found to be highly efficient in Figure 6.18. The diameter for even a quite
moderate capacity of 30,000 L/h would need a 300 mm device. This works well in
theory but there are two reasons why it does not work in an industrial machine. First,
the force needed to close the gap would be extremely high, as most of the force from
the hydraulic piston is balancing the force created by the high-pressure product on the
1.0
Sauter mean diameter (μm)
0.8
0.6
0.4
0.2
0.0
0.000 0.050 0.100 0.150 0.200
Gap height (mm)
FIGURE 6.18 Sauter diameter as a function of gap height. Data gathered from a large num-
ber of industrial installations.
250
200
Device diameter (mm)
Gap height (μm)
150
D theoretical
D industrial
100
Gap industrial
50
0
0 5,000 10,000 15,000 20,000 25,000 30,000
Flow rate (l/h)
FIGURE 6.19 Theoretical device diameter as a function of capacity for a 50 µm gap (―);
industrial relevant device diameter restricting the device between 5 and 120 mm (- - -). Gap
height for an industrial homogenizer (•••).
(a) (b)
FIGURE 6.20 The force balance of the forcer for (a) a standard homogenization device and
(b) a balanced force-type device.
forcer, as can be seen in Figure 6.20a. Second, the weight of the seat and the forcer
would be in the order of 100 kg, and the weight of the housing needed to withstand
the pressure would be in the order of 1000 kg. As this is not industrially feasible,
the solution is to restrict the device diameter to, for example, 120 mm, shown by the
dashed line in Figure 6.19. The same is also true on the other end of the scale: it is not
possible to manufacture a device with a very small diameter. A 100 L/h homogenizer
would need a device with a diameter of 0.9 mm, and it is not possible to manufacture
if the gap shape should mimic that of an industrial machine. With the low-capacity
machines, the gap heights are therefore normally too small. This can be seen in the
black line in Figure 6.19, where the smallest device has been restricted to 5 mm. For
low capacities, gap height will be very small and for capacities larger than 13,000 L/h,
the gap will increase to more than 100 µm for a capacity of 30,000 L/h (Innings 2005).
Quite a few ideas have come up over the years to overcome this problem. The first
idea in commercial production was maybe the easiest one: to balance the forces by
restricting the force from the product on the forcer. This can be done by reducing the
area of the high-pressure product that is affecting the forcer by extending it and sealing
the center part of the forcer from the high-pressure product (Figure 6.20b). This is done
in the Soavi NanoVALVE™ and it can double the possible diameter and thus achieve a
high efficiency up to higher capacities. The disadvantage is that you get a complicated
high-pressure seal that will wear quickly and would need to be replaced regularly.
Another idea to increase the efficiency is to increase the circumference of the gap by
multiplying it. One way of doing it is shown in Figure 6.21, where two (Figure 6.21a)
24 26
1 2
20 17
23 21 16
8 25
7 6
12
1 4 8
2 3
A 18
4
22
1 A 21 23
9 8
14 3 24 19
15 5
25
12 16
7
h 20
13 7
6 10 9
4 6 8
11 12
10 16 11a 11c
5 13
(a) 2 (b) 15 11b
FIGURE 6.21 Homogenization device with (a) two gaps—one inside the other—and (b) three
gaps inside each other.
Hydraulic piston
Outlet
Gaps
Inlet
FIGURE 6.22 Five-parallel gap HD EnergyIQ™ concept from Tetra Pak. (Courtesy of
Tetra Pak.)
or three (Figure 6.21b) [Patents WO2000015327A1 (Innings and Malmberg 2000)

and WO1998047606A1 (Malmberg and Valencuk 2010)] gaps are positioned inside
each other. This, of course, increases the circumference of the gap, but it also makes
the design more complicated and the wear parts more expensive.
An alternative to placing the gaps inside each other is a parallel gap design, where
a number of gaps are placed above each other, each with its own seat and forcer.
This concept was introduced by Gaulin in 1983, in the system called Microgap; but
then as a fixed gap design even if the gap could be opened more during clean-
ing. Later, this design was developed further in the HD EnergyIQ™ concept
from Tetra Pak, where the parallel gaps work just as those in a one-gap industrial
machine (Malmberg and Valencuk 2010) (Figure 6.22). The gap height and thus
the homogenization pressure are controlled by the pressure from the hydraulic
piston.
6.6 ECONOMIC ASPECTS

In the homogenizer, electric energy is first transformed into rotation, and then into
pressure that is transformed into velocity that breaks up the drops before finally
being transformed into heat. In this process, the first two stages are very effective,
that is, the high-pressure pump of the piston transforms between 85% and 95% of
the electric energy into pressure. The drop break-up process is, on the other hand,
extremely ineffective in that less than 0.1% of the pressure energy is used to enlarge
the surface area of the drops. More than 99.9% of the energy is instead spent heating
up the product, giving it a heat increase of about 1°C for each 50 bar of homogeni-
zation pressure.
When looking into the economic aspects of a homogenizer, we have three main
cost types:
1. Investment cost
2. Electricity cost
3. Spare parts and maintenance cost
A homogenizer lasts for more than 50 years but is normally replaced after 30–40
years; this practice, however, is clearly not relevant to a calculation of yearly cost.
The cost for a standard industrial machine capable of a flow rate of 20,000 L/h
and a homogenization pressure of 200 bar is in the order of €120,000. Figure 6.23
shows the cost for the machine for the first year given that the machine runs
20 hours a day.
From Figure 6.23, it can be seen that electricity cost for one-year production
is about half of the cost of the brand new machine. The maintenance and spare
parts cost is a substantial €16,000 per year, but is dwarfed by the electricity cost.
It should be noted that this is only the case when running milk and other non-
abrasive products. If the machine is running an abrasive product such as tomato
juice or ketchup, the maintenance cost can be in the same order as the electric-
ity cost. The only cost not taken up in Figure 6.23 is the cooling and seal water
cost. Water is used to cool the crankcase and in nonaseptic machines also for
the piston seals. In aseptic machines, hot water is used and if it is not controlled
carefully, the cost for the water spent will be surprisingly high. In dairies, you
can sometimes see an open 6 mm pipe feeding water at 95°C to the seals. This
will give a fairly moderate flow of 15 L/min, but the yearly cost for this hot water
is about €15,000.
120,000
100,000
80,000
Cost/year (€)
60,000
40,000
20,000
0
Investment Electricity Maintenance
FIGURE 6.23 First year costs for an industrial homogenizer.

6.7 THE HOMOGENIZER IN A PROCESSING LINE

Normally, a homogenizer is used as a standalone unit operation. The homogenizer
is a quite noisy machine, so it is often placed in a soundproof room together with the
separators.
In general, the homogenizer is placed upstream, that is, before the final heating
section in a heat exchanger. In most pasteurization plants for consumption milk pro-
duction, the homogenizer is usually placed after the first regenerative section together
with the separator and the milk standardization system, as seen in Figure 6.24.
In the production of ultra high temperature (UHT) milk, the homogenizer is
generally placed upstream in indirect systems but always downstream in direct
systems, that is, on the aseptic side after UHT treatment. In the latter case, the
homogenizer is of an aseptic design with special piston seals, sterile steam con-
denser, and special aseptic dampers.
However, downstream location of the homogenizer is recommended for indirect
UHT systems, when milk products with a fat content higher than 6%–10% and/or
with increased protein content are going to be processed. The reason is that with
increased fat and protein contents, fat clusters and/or agglomerates (protein) form at
very high heat treatment temperatures. These clusters/agglomerates are broken up by
the aseptic homogenizer located downstream.
6.7.1 Split Homogenization

An aseptic homogenizer is more expensive to operate. In some cases, it is suffi-
cient if just the second stage is placed downstream. This arrangement is called split
homogenization.
Note that the whole section, including the heat exchanger, between the first and
the second stages in the homogenizer, has to withstand a fairly high pressure.
IW
Raw milk
Cream Standardized milk

Separator
Fat Homogenizer
standardization
Surplus
cream
Skim milk
FIGURE 6.24 The homogenizer in a pasteurization line. (Courtesy of Tetra Pak.)

FIGURE 6.25 The homogenizer in a partial homogenization pasteurization line. (Courtesy

of Tetra Pak.)
6.7.2 Full Stream Homogenization

Full stream or total homogenization is the most commonly used form of homogeni-
zation of UHT milk and milk intended for cultured milk products.
The fat content of the milk is standardized prior to homogenization, as is the
protein content in certain circumstances, for example, in yoghurt production.
6.7.3 partial Homogenization

Partial stream homogenization means that the main body of the skim milk is not
homogenized; only the cream together with a small proportion of the skim milk
is homogenized (Figure 6.25). This form of homogenization is mainly applied to
the pasteurized milk. The basic reason is to reduce operating costs. Total power
consumption is cut by some 80% because of the smaller volume passing through the
homogenizer. As sufficiently good homogenization can be reached when the product
contains at least 0.2 g casein/g fat, a maximum cream fat content of 18% is recom-
mended. The hourly capacity of a homogenizer used for partial homogenization is
about one-fifth of the full-stream capacity.
REFERENCES
Bylund, Gösta. 2003. Dairy Processing Handbook. Lund, Sweden: Tetra Pak Processing
Systems AB.
Gaulin, Auguste. 1904. System for intimately mixing milk. US Patent US756953, filed
September 30, 1902, issued April 12, 1904.
Håkansson, Andreas, Laszlo Fuchs, Fredrik Innings, Johan Revstedt, Björn Bergenståhl,
Christian Trägårdh. 2010. “Visual observations and acoustic measurements of cavitation
in an experimental model of a high-pressure homogenizer.” Journal of Food Engineering
100 (3): 504–513.
Hansson, Rikard, Rolf Malmberg. 1998. Homogenizer valve. European Patent
WO1998047606A1, filed April 21, 1998, issued October 29, 1998.
Innings, Fredrik. 2005. “Drop break-up in high-pressure homogenisers.” PhD Thesis, Lund
University, Sweden.
Innings, Fredrik, Erik Hultman, Fredrik Forsberg, Braham Prakash. 2010. “Understanding
and analysis of wear in homogenizers for processing liquid food.” Wear 271 (2011):
2588–2598.
Innings, Fredrik, Rolf Malmberg. 2000. A method of homogenization. European Patent
WO2000015327A1, filed September 13, 1999, issued March 23, 2000.
Kurzhals Hans-Albert. 1997. “Undersuchungen Uber die physikalisch-technichen Vorgänge beim
Homogenisiren von Milch in Hochdruck-Homogenisiermaschinen.” PhD Dissertation,
Technichen Universität Hannover, Germany.
Malmberg, Rolf, Jozo Valencuk. 2010. Homogenizer valve. US Patent, US 20140177382, filed
December 20, 2011, issued June 26, 2010.
7 High-Pressure
Homogenization with
Contents
7.1 Introduction .................................................................................................. 170
7.2 Technical Equipment .................................................................................... 170
7.2.1 Valves................................................................................................ 170
7.2.2 Orifices and Nozzles ......................................................................... 170
7.2.3 Flow Conditions ................................................................................ 172
7.3 Emulsification Mechanism ........................................................................... 173
7.3.1 Deformation ...................................................................................... 173
7.3.2 Breakup............................................................................................. 174
7.3.3 Disruption as Result of Breakup of Deformed Droplets .................. 174
7.3.4 Process Functions ............................................................................. 175
7.3.5 Droplet Stabilization ......................................................................... 178
7.4 Process and Material Parameters Influencing Emulsification ...................... 179
7.4.1 Geometry .......................................................................................... 179
7.4.2 Multiple Stage ................................................................................... 184
7.4.3 Multiple Passage ............................................................................... 185
7.4.4 Viscosity Ratio.................................................................................. 186
7.4.5 Stability ............................................................................................. 187
7.4.6 Scale-Up ........................................................................................... 189
7.5 Conclusion .................................................................................................... 189
References .............................................................................................................. 190
ABSTRACT The development of orifice-type high-pressure homogenization is

presented from both theoretical and technical perspectives, including patent litera-
ture. In addition to the approaches of Chapters 5 and 6, attention is paid to the role
of spatial variation of the flow field in different orifice geometries and the methods
available for its study, in particular modeling and simulation. Experimental results
from different orifice geometries are compared. A novel, more efficient homog-
enizer design suitable for partial homogenization is introduced, where remixing of
the excess continuous phase occurs within microseconds of droplet breakup.
169
7.1 IntRoDUCtIon
High-pressure homogenizers (HPHs) were developed approximately 100 years ago
(Gaulin, 1899) during the age of industrialization. The basic idea of combining a
high-pressure pump and a disruption system, such as a valve, was presented at the
1900 World Exposition in Paris and endures to this day. However, we still see ongo-
ing developments of pumps and disruption systems stemming from either daily
application problems or new product challenges.
Current techniques permit volume streams of up to 50,000 L/h and pressures
of up to 10,000 bar. However, homogenization pressures in industrial applications
today are in the range of 50–2000 bar. Piston pumps mainly serve as high-pressure
pumps. In bench-scale equipment, a single-piston pump commonly delivers the vol-
ume stream—or rather the pressure—but in production plants, up to eight piston
pumps are found. The disadvantage of the single-piston pump is that the pressure
and the volume stream can vary significantly over time, which results in a pulsation
of stresses on the product. Inhomogeneous stresses act on the droplets, thus mak-
ing product properties difficult to control. To reduce pulsation, several pistons are
combined in a phase-shifted manner. Valves are used to control the different pistons,
which are not usually influencing the quality of the emulsion.
Section 7.2 presents the current state of technical equipment. Section 7.3 dis-
cusses the mechanism of droplet breakup in these devices. Section 7.4 discusses the
influence of the main parameters on emulsification.
7.2 teCHnICAL eQUIPMent

The main part of an HPH is the disruption system. Here, the pressure built up by the
pump is expanded, resulting in specific flow conditions used for droplet disruption.
The disruption systems available on the market can be divided into two main groups:
valves and nozzles.
7.2.1 ValVes
Valves, also known as radial diffusors, are commonly used high-pressure disruption
systems. The fundamental idea is to reduce the flow diameter with a valve plunger,
which is pushed to a valve seat forming a small gap. These systems were introduced
and discussed in Chapters 5 and 6. Over the years, the geometry of flat valves has also
been developed. In this chapter, we will limit our discussion to the conventional old
flat valve and the new flat valve with a tapering inlet. The size range of the smallest
gap of the valves is generally 100 nm to several micrometers. Thus, valves are also
microstructured systems. In this chapter, however, we will use the definition of micro-
structured systems only for disruption systems with a fixed geometry, called orifices.
7.2.2 Orifices and nOzzles

The simplest technical solution of a microstructured system is a straightforward
round-shaped orifice, also called simple orifice (see Figure 7.1a) (EN ISO 5167-1,
2003). In contrast to valves, orifices are constructed without any movable parts,
High-Pressure Homogenization with Microstructured Systems 171
Ed a b
Bd2
Ed
(a) (b)
(c) (d)
FIGURe 7.1 Schematic drawing of a simple orifice (a) and modified types (b–d).
(Data as published by Stang, M., Zerkleinern und Stabilisieren von Tropfen beim mecha-
nischen Emulgieren, Dissertation, Universität Karlsruhe, 1998; Freudig, B., Herstellen von
Emulsionen und Homogenisieren von Milch in modifizierten Lochblenden, Dissertation,
Universität Karlsruhe, Germany, 3-8322-3147-1, 2004; Aguilar, F.A. et al., Chem. Ing. Tech.,
80, 5, 607–613, 2008.)
which is advantageous in the manufacturing. At constant viscosity of the emulsion,

the homogenizing pressure is adjusted by the volume stream or the orifice hole diam-
eter, respectively, to the cross-sectional area. Increasing the volume stream at target
pressure loss requires a numbering up of the orifices as realized in Bayer AG (1997,
2001). Numbering up is only limited by a minimum distance between the holes being
in the order of 6 (Aguilar et al., 2008). To ensure a constant homogenizing pressure
even for fluctuating volume flow rates, the number of orifices has to be automatically
adapted as well (Bayer MaterialScience AG, 2006).
The Bayer Company (now BTS) was among the first to patent and commercial-
ize orifices for high-pressure homogenization applications (Bayer AG, 1991). Basic
research on flow conditions in circular cross-sectional orifices and their effect on
droplet disruption was first reported by Stang and Schubert in the 1990s. An overview
of this work is published by Stang (1998). This was followed by intense research by
several groups, resulting in several patents (Cook and Lagace, 1985; Stone, Bentley,
and Leal, 1986; Muschiolik, Roeder, and Lengfeld, 1995; Stang, 1998; Penth, 2000;
Kolb, 2001; Tesch, 2002; Floury et al., 2004; Freudig, 2004; Aguilar et al., 2008).
Typical dimensions of the orifice are hole diameters of d = 0.1 to 1 mm and a thick-
ness of l = 0.4 to several millimeters.
Through modification of the orifice, nozzles were developed. In Figure 7.1b a
trench is shown used instead of a hole. The trench has the advantage that just the
smaller edge has an impact on the homogenization result. Thus, the larger one can
be used for an increase of the cross-sectional area and the volume flow rate (Aguilar
et al., 2008). This is limited only by the production accuracy of the smaller edge.
Impinging the free jet that develops in the orifice’s outlet section on a plate or a
second liquid jet (see Figure 7.1c) improves droplet breakup by inducing turbulent
disturbances (Aguilar et al., 2008). Similar effects are found for orifices with internal
steps deflecting the flow (Figure 7.1d) (Cook, 1985, 1990; Penth, 2000; Aguilar
et al., 2008).
7.2.3 flOw cOnditiOns

Depending on the high-pressure disruption system, the geometry different local flow
conditions are created. The resulting flow conditions also depend on the emulsion’s
material parameters such as the viscosity of the phases, or the viscosity ratio between
the droplets and the continuous phase (Grace, 1982; Bentley and Leal, 1986). Laminar
shear flow, elongational flow, as well as turbulent flow and cavitation-induced micro-
turbulences are usually found in industrial homogenization valves.
Generally, a laminar or relaminarizing flow is found after the high-pressure pump.
Due to the reduction of the cross-sectional area in front of the disruption system, the
stream is accelerated and elongated, which results in elongation and shear stresses.
From a critical homogenization pressure, the stream detaches on edges such as the
inlet edge and thus produces depression areas. In depression areas, cavitation may
occur. Furthermore, the detaching of the flow depicts instability in the stream and
may also induce a turbulent transition or a back flow area. Turbulence, as defined by
Kolmogorov, is an eddy cascade in which energy of the large eddies is divided into
small eddies that finally dissipate at the smallest eddy size (Kolmogorov scale) in
energy. In some cases, the inner core of the stream stays laminar but on the boundar-
ies first eddies can rise. Due to deflections in the orifice, local turbulences can be
enhanced. On the outlet, often the flow detaches from the wall again. Depending on
the outlet geometry, a free jet develops. At the boundaries of the free jet, transitional
or turbulent flow regions may develop and induce elongation and shear stresses.
Boundary effects in smaller outlet channels also influence local turbulences and
cavitation (Schlichting and Gersten, 2006).
Cavitation is referred to as the phenomena of genesis and collapse of bubbles
based on gas or vapor. In HPHs, the fluid must pass through a tapering, in which
the potential energy of the pressure is transformed into kinetic energy and thus into
the velocity of the fluid. Under several circumstances, the acceleration of the fluid
results in a pressure drop below the vapor pressure. As a result, the soluble gases and
the fluid itself start to produce nuclei, on which bubbles grow. After the junction,
the fluid velocity decreases and the pressure increases. The bubbles collapse and
produce strong fluid movement and heat. Microjets and turbulences are discussed
further.
Cavitation is a huge challenge for the service life of homogenization systems due
to the abrasion induced. Yet cavitation is also effective in disrupting emulsion drop-
lets. Through the collapsing of vapor or gas bubbles and induced microjets, pres-
sure fluctuations and microturbulences are produced, which result in the necessary
stresses for droplet disruption. Cavitation can be influenced significantly by a back
pressure, which can be produced by a second homogenization stage. With increase
of the back pressure, the overall cavitation is reduced, whereas the intensity of the
bubble implosion is increased (Bondy and Söllner, 1935; Silver, 1942). For this, a
simple counter pressure valve or a second disruption system can be applied.
Generally, in all microstructured devices, the same flow patterns arise as with
conventional flat valves. There are regions of laminar and turbulent flow, as well as
cavitation. They differ only in how long the different flow patterns exist and in the
height of tensions created.
7.3 eMULsIFICAtIon MeCHAnIsM

Due to capillary pressure, all premix emulsification processes, as typically used in
high-pressure homogenization, start with round droplets. The first process to produce
smaller droplets is to deform the droplet and form a filament. This can be done in two
ways: (1) by deforming the whole droplet to a long filament or (2) by deforming just
a part of the droplet surface to produce a bulge. Droplets are deformed and disrupted
by tensions, which result from different flow conditions and act on their interfaces.
The high-pressure disruption system creates the required local flow conditions. The
resulting flow conditions also depend on the emulsion’s material parameters, such as
viscosity of the phases, or the viscosity ratio between the droplets and the continuous
phase (Walstra, 1983; Armbruster, 1990).
The second process is the pinching of the fluid into droplets. Pinching may occur
due to forces introduced by surface tension and perturbations (e.g., Plateau-Rayleigh
instability) or recurring mechanical stresses.
In this chapter, we will use the nomenclature deformation for the first process,
and breakup for the second process; both processes taken together will be referred
to as disruption. After the deformation and breakup, a third process involved in the
stabilization of the droplets must be performed otherwise the droplet size will once
again increase (e.g., due to coalescence).
7.3.1 defOrmatiOn
Droplets are deformed by tensions that result from different flow conditions or vibra-
tions, and these have an effect on the droplet, especially the interfaces. The coun-
teracting tensions are the surface tension and the viscosity—or elastic effects—of
the surface and the dispersed phase (Arai et al., 1977). If the deforming tensions
exceed the counteracting tensions, the droplet starts to deform. To describe the pro-
cess of deformation, the three dimensionless numbers Weber (We), capillary (Ca),
and Ohnesorge (Oh) are established using the Reynolds number (Re):
v⋅x σ⋅ x We η ⋅ v We η
Re = , We = , Ca = = , Oh = = (7.1)
ν γ Re γ Re ρ⋅γ ⋅ x
Depending on the flow conditions and material parameters, the We, Ca, or Oh num-
ber can better describe the deformation. Besides the acting tensions, the duration of
the acting tensions is also important (Walstra, 1983).
Most work is done on the deformation of the whole droplet. Droplet deformation
due to laminar shear flow has been widely investigated (Walstra, 1983; Bentley and
Leal, 1985; Stone, Bentley, and Leal, 1986). However, it is restricted to a narrow
range of viscosity ratios between the dispersed and the continuous phases ηd/ηc for
single-droplet disruption, or between the dispersed phase and the emulsion ηd/ηe for
emulsions, respectively (Armbruster, 1990; Jansen, Agterof, and Mellema, 2001).
Laminar elongation flow is advantageous if highly viscous dispersed phases have
to be disrupted (Grace, 1982). It is usually found in the inlet of disruption systems.
Specific disruption systems are designed for increased elongation in the inlet flow.
Turbulence can be described as eddies of varying sizes (see flow conditions).
Depending on the relation of droplet size to eddy size, both the viscous and the
inertial regimes can be differentiated (Hinze, 1955). In the turbulent viscous regime,
in a Lagrange approach, the droplet is subject only to laminar shear and elongation
tensions due to the fact that the droplet is smaller than the eddies (Vankova et al.,
2007). Thus, the whole droplet is deformed as in laminar flow.
Two mechanisms are known to deform the interface of the droplet: capillary waves
or the turbulent inertial regime. Behrend (2002) showed that the smallest achievable
droplets by capillary waves are in the range of microns. This effect is thus negligible
for the production of submicron emulsions. The deformation of droplets in the tur-
bulent inertial regime is of great interest. This regime was first described by Hinze
(1955). In this regime, the droplets are in the same range as eddies or larger. This
results in a deformation of the surface due to several eddies acting on the surface.
7.3.2 Breakup
A breakup of droplet occurs, if the deformation reaches a critical value in deforma-
tion and time. The breakup of a formed filament can be explained by instabilities
or mechanical tensions (Tcholakova et al., 2011). The most often discussed case in
the literature is the Plateau–Rayleigh instability (Plateau, 1873; Eggers, 1997). This
effect is driven by a perturbation of the flow, which results in small difference in the
diameter of the filament. These differences produce differences in the local capillary
pressures and enforce the growth of large filament parts and the reduction of small
parts until the breakup occurs. Instead of perturbations, the filament surrounding
flow can also produce differences in the diameter of the filament.
7.3.3 disruptiOn as result Of Breakup Of defOrmed drOplets

Due to the fact that the most devices are not transparent—thus making a separate
investigation of the deformation and breakup impossible—most investigations
involve the complete disruption by examining the homogenization result offline.
Some authors used transparent devices, but investigated the complete process of dis-
ruption as well.
The basic idea in describing the complete disruption process is that a deformed
droplet breaks in any way. Here, the first approach is to examine the flow conditions
that disrupt the droplet at sufficient time. This value was called critical capillary or
Weber number. This approach only works if the flow conditions and the tensions are
known. Taylor (1932) completed the first remarkable work. Bentley and Leal (1986)
demonstrated the influence of the critical number on the flow pattern and the viscos-
ity ratio. This was expanded for the superposition of flow pattern by Grace (1982). Hu
showed that with this approach, the droplet size distribution also can be predicted.
Droplets are disrupted if they are deformed over a period of time, tdef, which
is longer than a critical deformation time, tdef,cr, and, if the deformation exceeds a
critical value. Both criteria must be fulfilled. The necessary deformation as well as
the time needed for droplet breakup depends on the external tension, σ; the droplet
diameter, x; the interfacial tension, γ; and the viscosities of the dispersed phase, ηd;
and the continuous phase, ηc. The last three parameters are material parameters,
which today can be measured offline using established methods (Walstra, 1993).
ηd
tdef,crit = (7.2)
σ − pk
The second approach is to measure the size of the remaining droplet, which has
resisted against the external forces and thus can be found in the final emulsion. Using
this approach, it is possible to measure the tensions appearing in the process by
inserting it into the Weber or capillary number.
7.3.4 prOcess functiOns

Because tensions are in most cases not known, several process functions were estab-
lished to estimate the resulting tensions and thus the droplet size. Several approaches
are available today in order to predict the droplet size. In general, these can be
divided into measuring, modeling, and simulation techniques.
The oldest approach is to measure the energy consumption of the machine.
However, this approach is problematic in that the efficiency or general losses of the
machine are not considered. Over the years, different empirical equations to cal-
culate energy dissipation—and thus tension—were established by taking process
parameters such as pressure, volume stream, flow type, and diameters of the disrup-
tion system as well as substance parameters into account (Vankova et al., 2007; see
also Chapter 5).
A simple alternative method is measuring the static pressure drop over the dis-
ruption device, Δp. The static pressure correlates to the mean tension over the whole
volume in diameter and time. It also correlates with the specific energy density, EV
(see Equation 7.3) (Karbstein, 1994). Mean tensions can be estimated as the energy
dissipating in a volume. In the general definition, as given in Equation 7.3, E and
P are the energy and power, respectively, being supplied by any emulsification
machine, V is the emulsion volume in the disruption zone, V is the emulsion volume
throughput, and tres is the residence time within the zone of disruptive stresses. In
the specific disruption energy, EV, the applied power is related to the volume stream;
both parameters are measurable in industrial processes.
E P ∆p ⋅V (7.3)
EV = PV ⋅ t res = = =  = ∆p
V V V
When passing the emulsion several (n) times through the disruption system, this
must be multiplied by the number of passes, n. This can be assumed as long as
the disruption is going on and the duration of tensions is the limiting factor, which
is often the case in HPHs. Once all droplets that cannot survive the stresses are
disrupted, this concept does not work anymore. An equilibrium droplet diameter is
found, which is determined by the maximum volume related power density, PV, also
called ε in Kolmogorov’s, Hinze’s, and Davies’s theory of droplet breakup in turbu-
lent flow (Kolmogorov, 1949; Hinze, 1955; Davies, 1972).
With the specific disruption energy, EV, a mean droplet diameter such as the
Sauter mean diameter x32 can be calculated in case of well-defined flow conditions
by the following process functions:
Laminar shear flow (Grace, 1982) is given by the following equation:
η  η 
x32 ∝ EV−1 ⋅ f  d  or x32 ∝ (n ⋅ ∆p)−1 ⋅ f  d  (7.4)
 ηe   ηe 
Laminar elongational flow (Walstra, 1983; Bentley and Leal, 1986; Chesters, 1991)
is given by the following equation:
x32 ∝ EV−1 or x32 ∝ (n ⋅ ∆p)−1 (7.5)
Isentropic turbulent flow (Arai et al., 1977) and microturbulences in the cavitational
flow (Bechtel, Gilbert, and Wagner, 1999, 2000; Behrend, Ax, and Schubert, 2000;
Behrend, 2002) are given by the following equation:
x32 ∝ EV−0.250.4 ⋅ η0d0.75 or x32 ∝ (n ⋅ ∆p)−0.250.4 ⋅ η0d0.75 (7.6)
In Equation 7.6, the exponent of the viscosity of the dispersed phase, ηd, is equivalent
to 0 for lowly viscous dispersed phase (ηd < 10 mPa s), whereas it is 0.75 for highly
viscous dispersed phases.
In defined flow conditions, the exponent of the disruption energy is either −1 or
between −0.25 and −0.4. For industrial homogenization valves, exponents around −0.6
have been published, for example by Walstra (1983), depicting the mix of flow condi-
tions found in homogenization valves. The overall equation describing the process
functions for high-pressure homogenization processes thus is:
x32 ∝ EV− b ⋅ ηcd or x32 ∝ (n ⋅ ∆p)− b ⋅ ηcd (7.7)
where:
b is in the range of 0.25–1
c is lesser than 0.75
This means that using Equation 7.7, we can predict the droplet size of an emulsion of
a specific recipe produced in a homogenizer with a specific valve.
In all approaches a change in the valve geometry is often a challenge, due to the fact
that changes in flow conditions are difficult to predict, as it stands today. In different
geometries, the tensions are distributed differently in location, time, type, and intensity
(Equation 7.7) using a mean tension over time, type, and location, thereby limited to
explaining local mechanisms or predicting the efficiency of a geometric modification.
A measuring technique such as Micro Particle Image Velocimetry (µ-PIF) or laser

doppler anemometry (LDA), recently significantly improved, allows for analyzing
the resulting tensions with high time and local resolution. Currently, these techniques
are used in basic research (Meinhart, Wereley, and Santiago, 1999; Sheng, Meng, and
Fox, 2000; Mielnik and Saetran, 2004), but much too expensive for standard applica-
tion in production. Analyzing typical flow parameters, such as discharge coefficient
(cD) and Re numbers, however, allows for characterizing the flow conditions found in
a valve outlet in general (Johansen, 1930; Wolf et al., 2012).
With the rapid development of computers, simulation methods have become
increasingly reliable in the description of the emulsification process in specific
geometries, such as those used in HPHs. Today, several works have been published
on one phase flow simulated by computational fluid dynamics (CFD) (Kleinig and
Middelberg, 1997; Stevenson and Chen, 1997; Aguilar, Freudig, and Schuchmann,
2004; Floury et al., 2004; Freudig, 2004; Innings, 2005; Steiner et al., 2006; Köhler
et al., 2007; Casoli, Vacca, and Berta, 2010; Köhler, 2010). Here, the continuous phase
flow is simulated, although typically the influence of the dispersed phase is neglected.
The approaches used in these publications differ in the geometry used and the way tur-
bulence is modeled. Also, data derived by direct numerical simulation are published
(Kissling, Schütz, and Piesche, 2011), which enable us to simulate the flow without
any additional turbulence model. Knowing the flow field is the basis for analyzing
tensions that are the important for droplet deformation and breakup. The authors sug-
gest using different tensions derived from simulation data: (1) the maximum tension
found in the simulated volume as total, (2) the maximum tensions found in the main
flow field, and (3) mean tensions over a volume of interest. Using these tensions, drop-
let sizes can be calculated using the models described (see Chapters 5 and 6). For
example, Steiner et al. (2006) used mean tensions over a volume of interest and calcu-
lated a mean tension over the slit volume. In this way, they demonstrated that it is possible
to describe a change in droplet size by parameters such as the volume stream or mate-
rial parameters such as the surface tension. Köhler (2010) calculated mean tensions in a
volume of fluid method (VOF) mesh for representative droplets by particle tracking for
the average time of the Reynolds-averaged Navier–Stokes equations (RANS) assump-
tions. The value derived thus was called mesoenergy density (see Figure 7.2). He could
thus demonstrate that effects on droplet sizes resulting from a change in flow type or
processing time can be explained qualitatively. Using the same method on nanoparticle
dispersions flowing through high-pressure valves, Wengeler (2007) also demonstrated
that distributions of the achieved particle size could be explained. However, the effect of
cavitation was not considered in all of these investigations.
Ongoing research focuses on two- or three-phase simulations of the continu-
ous phases filled with one or several droplets or the occurrence of cavitation and
sometimes on both. Baldyga et al. (2007) investigated the effect of cavitation in
HPHs equipped with orifices used for deagglomerating nanoparticle agglomerates.
Kissling, Schütz, and Piesche (2011) published a method to handle the well-known
problem of VOF models by sharpening the blurred interfaces.
Overall, the future challenges describing emulsification by simulation will be the
high spatial and temporal resolution of the flow and interfaces required, as well as
the interactions with molecules such as the emulsifier found in emulsions.
> mm μm nm
Macro Micro
Mesoenergy
energy density density tension
Static pressure Cell in mesh Molecule
FIGURe 7.2 Graphical explanation of the mesoenergy density. (Data from Köhler, K.,
Simultanes Emulgieren und Mischen, Logos Verlag, Berlin, Germany, 2010.)
In conclusion, models describing specific devices and flow conditions are avail-
able, and investigations are conducted to predict droplet size distributions resulting
from a high-pressure homogenization process. However, slight changes in the geom-
etry or other parameters often cause the models to fail. This is explainable by the fact
that most investigations in disruption are of single flow types, with homogenous flows
in space and enough time to break the droplet. As a result, most models are based on
these experiments. Yet, in most HPHs, the different flow patterns are often superposed
and inhomogeneous in time and space. This means that the flow conditions are often
transient and the flow on the walls differ completely from the main stream. In addi-
tion, droplet–droplet interactions resulting, for example, in their coalescence and thus
in a change of droplet sizes are usually not considered at all. As a result, most models
used to describe droplet size fail to calculate the final droplet size, in particular the
droplet size distribution especially arising through modifications.
7.3.5 drOplet staBilizatiOn

In breaking up droplets, a large surface area is created, which has to be stabilized
against sticking (agglomeration, flocculation) and coalescence (merger of droplets).
In HPHs, this occurs in short periods of time (usually within a microsecond). This
is why fast stabilization of the droplets has to be realized. In many cases, emulsi-
fier molecules are not able to do this job as efficiently as required (Walstra, 1983).
To estimate the influence of coalescence and agglomeration on resulting droplet
sizes, agglomeration or coalescence rates can be applied. The coalescence rate, Ω,
expresses the quantity of droplet coalescence incidents per volume and time. It can
be calculated by multiplying the collision frequency, C, with the coalescence prob-
ability, pcoal (Chesters, 1991) (see Equation 7.8).
Ω = pcoal ⋅ C (7.8)
The coalescence probability mainly depends on material parameters such as the sur-
face tension and elasticity, as well as droplet–droplet interaction forces of electro-
static, hydrodynamic, or steric nature. These values change over time through the
homogenization process due to the newly produced and emulsifier-reduced surfaces,
which are again covered with other emulsifier molecules adsorbing at them. This,
however, is a process that takes some time. This time depends on the emulsifier
molecule structure and chemical nature of the phases (Miller, 1990). Therefore, the
kinetics of the emulsifier(s) also influence the probability of coalescence and thus
the coalescence rate (Stang, Karbstein, and Schubert, 1994; Vankova et al., 2007).
The collision frequency, C, depends on the local flow conditions and surface area
per volume. In an isentropic turbulent flow, it can be calculated using Equation 7.9.
C = K ⋅ u ⋅ x 2 ⋅ n2 (7.9)
where:
K is the collision coefficient
u the velocity of the droplet (being dependent on the droplet’s diameter, x)
n is the number of droplets
As can be easily derived from Equation 7.9, the collision frequency increases with the
number of droplets (n2) and their size (x2). Even when the droplet size is reduced in
emulsification, the surface area and the number of droplets increase significantly and
have a strong influence on the collision frequency. Also, the flow has an important
impact on the droplet coalescence, as it generates the necessary acceleration of the drop-
lets. Both parameters increase with the specific energy input. Coalescence rates found
for typical emulsification processes can be mathematically described using first-order
kinetics. Kinetic rate constants derived from these can be used to describe the influence
of process and material characteristics on resulting droplet sizes. Detailed information
regarding droplet coalescence is found in Danner (2001) and Vankova et al. (2007).
7.4 PRoCess AnD MAteRIAL PARAMeteRs

InFLUenCInG eMULsIFICAtIon
This section shows high-pressure homogenization results as a function of the energy
density applied in processing. We further demonstrate the possibilities and limita-
tions of this approach. Karbstein (1994, p. 102) showed that a disruption-dominated
emulsification process results in a linear plot in a double logarithmic diagram, as
predicted by Equation 7.7.
7.4.1 GeOmetry
In Section 7.2, we introduced several disruption devices. In this section, we discuss
the influence of geometry on the emulsification result. In Figure 7.3 disruption sys-
tems are compared by producing corn oil droplets in water. As the droplet fraction
4.0 O/W
CE = 5% SDS
3.8 φ = 0.25 vol%
Δp = 100 bar
3.6 Orifice one stage
Droplet diameter, x903 (μm)
3.4
Bd1 = 1.5 mm
3.2 Ed = 2 mm
Ed = 4 mm
3.0 Ed = 8 mm
2.8 Bd1 = 4 mm
Ed = 2 mm
2.6 Ed = 8 mm
2.4
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Orifice diameter, Bd2 (mm)
FIGURe 7.3 Droplet diameter, x903; plotted over the valve diameter, Bd2; for different inlet,
Bd1; and outlet diameters, Ed2, by a constant homogenization pressure, Δp = 100 bar.
is rather low (1%), we can ensure that droplet–droplet coalescence rates are low, and
we have a disruption-dominated emulsification process.
In all cases, the droplet diameter decreases linearly with increasing specific
energy input (corresponding to the pressure difference applied), as predicted by
Equation 7.7. However, the position and the slope of the resulting curves differ. To
understand the differences, we have to discuss the geometries in detail.
The new-generation flat-valve geometry allows for droplet elongation in the inlet
as well as higher turbulence and cavitation in the outlet. Elongation of droplets to
filaments results in decreased mean Sauter diameters, as disruption of elongated fila-
ments is easier compared to spherical droplets. This results in decreased Sauter mean
diameters and a decreased slope of the curve in double logarithmic diagrams.
Results found for the simple orifice geometry are similar, indicating similar
local flow conditions. Regarding the geometry of the valve (see Figure 7.1) and flow
conditions already published by simulation, droplet elongation followed by a breakup
in local turbulences is a realistic scenario to be assumed. The Microfluidizer® geom-
etry equipped with the double-stage disruption system—thus applying back pressure—
allows for a further improvement of homogenization results. Comparable slopes of the
curves indicate that comparable flow conditions are responsible for droplet breakup.
Similar results are also found for double valves, which also apply back pressure to
the first stage (Kolb, 2001; Freudig, 2004; Karasch and Kulozik, 2008). All authors
hypothesize that the back pressure is responsible for an improved breakup. The reason
for this has to be investigated in more detail in future.
This shows that the geometry of microstructured valves is a factor of main influ-
ence on the homogenization result. We thus concentrate on this using the simple
basic geometry, as shown in Figure 7.1. We changed the three design parameters: the
inlet, outlet, and valve diameter. We produced emulsions and ensured that coales-
cence did not appear and lead to a misinterpretation of disruption results. For this,
we used sodium dodecyl sulfate (SDS) in excess, known to be a very fast stabilizing
emulsifier, and a low dispersed phase fraction of φ = 0.25 vol%. In Figure 7.3, the
maximum droplet diameter of the volume collective x903 is plotted over the valve
diameter, Bd2, for a different inlet, Bd1, and outlet diameter, Ed, at a constant homog-
enization pressure, Δp = 100 bar.
Remarkable is the fact that the droplet diameter can be modified by a factor of 2
in changing the geometry. The largest droplets are achieved by a valve diameter Bd1
around the 0.4 mm level. This effect can be seen in the maximum droplet size x903, as
well as in the mean Sauter droplet diameter x32 (see Köhler, 2010).
Furthermore, we see that the combination of large inlet and outlet diameters leads
to smaller droplets. However, this correlation is not significant and, therefore, cannot
be used for further optimization. Yet this is consistent with results reported by Stang
(1998, S. 123–124). He found that the homogenization result is not significantly influ-
enced by the ratio of inlet diameter to valve diameter.
We simulated the flow for the different modifications of the geometry and
compared the resulting tensions in each cell of the VOF-mesh. Laminar tensions
(or mesoenergy densities EV, m), were achieved by multiplying the elongation
and shear rate with the corresponding viscosity. The turbulence was modeled
using the RNG-k-ε-model. Turbulent tensions were calculated by the following
equation:
EV, m (ε) = ρk ⋅ (ε ⋅ υk )1/ 2 (7.10)
The tensions calculated were compared on the symmetry axis of the orifice. A discus-
sion of the local distribution of the tension can be found in Köhler (2010). As vegeta-
ble oil (viscosity: 60 mPa s) was used as the dispersed phase and water (viscosity of
1 mPa s) as the continuous phase, the breakup due to laminar shear can be neglected.
At this viscosity, ratio droplets will only rotate in pure laminar shear flow, but not
deform (Stone, Bentley, and Leal, 1986). Elongation in y-direction is negligible due
to low mesoenergy densities. Thus, only turbulent dissipation ε and elongation in the
x-direction are considered (see Figure 7.4). In the following diagram, the mesoen-
ergy densities resulting from the simulations are plotted over time for the symmetry
axis of orifices with a diameter Bd2 of 0.1 and 0.8 mm. The time is set to zero at the
moment when a fluid element is at the entrance of the orifice.
Both droplet devices are subjected to stresses prior to entering the orifice, as they
are accelerated and elongated. The tensions increase to maximum values in the first
microseconds after entering the orifice. The resulting tensions are comparable, and only
the absolute value and the duration change. With a decrease in the orifice diameter, Bd2,
elongation is increased. The main difference between the orifices is in the duration of the
tensions, especially those resulting from turbulence. This shows that droplet deforma-
tion and breakup kinetics may not be neglected in homogenization. The time between
One-phase simulation
104 water
RNG-k-ε turbo model
Δp = 100 bar
Bd1_Bd2_Ed
1,5_Bd2_2
103
Ev,m (kg/ms2)
Mesoenergy density
102 Bd2 (mm)
0.1 0.8
ε
dvx/dx
101
−20 0 20 40 60
Time, t (μs)
FIGURe 7.4 Comparison of the mesoenergy densities resulting from turbulence (ε) and
elongation in x-direction (dvx /dx) in the orifices of diameters Bd2 = 0.1 and 0.8 mm plotted
over time on the symmetry axis of the orifice (t = 0: entrance of the orifice).
the maximum tension by elongation and turbulence decreases from around 30–10 µs.
Thus, a droplet deformed by elongation before entering the orifice has less time to relax
before being subjected to the tensions resulting from turbulence.
Droplets are disrupted in the laminar elongational flow as soon as the Weber
number exceeds the value of 0.1 and the deformation time reaches a critical value
(Walstra, 1993). We calculated the droplet diameter at which the tension on the sym-
metry axis exceeds the critical Weber number. This calculated maximum droplet
diameter is indicated in the following diagram with elongation (see Figure 7.5). The
droplet diameter at which the tension persists long enough to break up the droplets
is indicated by time.
Only at small orifice diameters, the elongational tensions are high enough to
deform droplets up to their breakup. The duration of the deformation, however, is
not long enough to achieve droplets in a size of the experiment.
To describe the droplet disruption in turbulence, models working with a turbulent
Weber number are also available (Hinze, 1955; Arai et al., 1977). But so far, no equa-
tion for the critical turbulent Weber number is known. Thus, we used the approach of
Kolmogorov and calculated the maximum droplet diameter surviving the dissipation
rate. This maximum droplet diameter decreases slowly with increasing orifice diam-
eter, Bd2. For an orifice diameter of 0.8 mm, the model predicts the droplet size well,
although differences are found for the small diameters. Also, models of other authors
(Hinze, 1955; Davies, 1985; Vankova et al., 2007) are not able to predict the droplet
size found in our experiments. For small orifice diameters, elongation seems to play
a pronounced role in the breakup. In summary, this confirms that models describing
250.0
One-phase simulation
water
100.0 Δp = 100 bar
RNG-k-ε turbo model
Bd1_Bd2_E
Droplet diameter, x (μm)
1,5_Bd2_2
10.0
Calculated:
Elongation
Time
3.0 Turbulence
Experiment:
x903
1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Orifice diameter, Bd2 (mm)
FIGURe 7.5 Calculated maximum droplet diameters surviving elongation (calculated for a
dominating effect of the elongation rate or elongation time) and turbulent energy dissipation
compared to the maximum droplet diameter achieved in the experiments.
droplet breakup in homogenization orifices have taken interactions of the different

flow regimes into their account. Simulation of the local flow conditions and resulting
tensions will help in quantifying resulting droplet sizes and explain the differences
in homogenization devices.
In the next section, we show results of different homogenization devices used for
the homogenization of full-fat milk (φ = 3.5 vol% of fat). We used microstructured
orifices, a round and a slit orifice as well as a deflection orifice (U) and a deflection ori-
fice with an additional elongation zone in the deflection (UD) (see Figure 7.6). These
were produced in the Karlsruhe Institute of Technology (KIT) workshop. In addition,
commercial microstructured orifices (Microfluidics® Y- and Z-chamber) were used.
The Z-chamber is a deflection orifice, whereas the two orifice-leaving jets are shot on
each other in the Y-chamber.
As for the rapeseed oil-in-water (O/W) emulsion, we see a linear trend for the
correlation between the maximum size of the milk fat globule and the homogeni-
zation pressure (specific energy) for all orifices used. Significant differences were
found between the deflecting (U, UD Y- and Z-chamber) and nondeflecting orifices
(slit orifice, Orifice). Both the nondeflecting orifice and the round and slit orifice
deliver comparable homogenization results. A deflection results in an improved
breakup. A reason for this can only be found when analyzing the flow within the
orifice. This indicates an increased turbulent energy dissipation in the moment of
deflection.
Elongation or the shooting of two jets on each other, however, did not result in
an improved breakup (see Figure 7.7), as found for rapeseed oil. This indicates that
4
Milk
φ = 3.5 vol%
3
ϑ = 60–65°C
Full-stream
One stage
2
1
Slit orifice
Orifice
U
UD
Y-chamber
Z-chamber
0.4
45 100 500 1000
Homogenization pressure, Δp (bar)
FIGURe 7.6 Maximum volume-rated milk fat globule droplet sizes x903 after homogeni-
zation with different orifices at homogenization pressures up to 1000 bar. Homogenization
temperature 65°C.
3.3
3 Milk
φ = 3.5 vol%
ϑ = 65°C
2 Full-stream
Two stage
YZ
Y
Z
ZR YZ
0.3
45 100 500 1000
FIGURe 7.7 Maximum volume-rated milk fat globule droplet sizes x903 after homogeni-
zation with different orifices at homogenization pressures up to 1000 bar. Homogenization
temperature 65°C.
besides local stresses, inner stresses in the droplets have to be considered when
analyzing droplet breakup. Milk fat globules, for example, are known to resist elon-
gation due to their specific multilayer membrane composition (Kessler, 2002) and
will thus react differently to external stresses than rapeseed oil droplets.
Future research, therefore, has to concentrate not only on the continuous phase
flow in microstructured orifices but also on droplet deformation behavior.
7.4.2 multiple staGe

A multiple-stage process means interconnection of several disruption units in a
series. The use of not only one disruption device is nowadays well established for
commercial homogenization valves (see Chapters 5 and 6). Two to several stages
are reported in the literature (Kolb, 2001; Kessler, 2002; Freudig, 2004; Karasch
and Kulozik, 2008). As for commercial homogenization valves, this can be real-
ized for microstructured orifices. An example is given in Figure 7.7, where full-fat
milk is homogenized with a deflecting orifice and a nondeflecting orifice in a single-
stage or two-stage setup. The impact of the second stage on the resulting size of the
milk fat globule is significant and dominates one of the geometry of the first stage
(see Figure 7.7).
As the main benefits of a multiple-stage homogenizer, a smaller mean droplet
diameter and narrower droplet size distribution are discussed in the literature. These
effects are explained by modified cavitation (see Section 7.2.3.), modified turbulence
(Kolb, 2001), or a second breakup of droplets or agglomerates formed after the first
stage (Walstra, 1975).
As small droplets are more difficult to break and tensions are usually lowered in the sec-
ond stage, a second breakup scenario is unlikely. Only the big droplets in the collective that
survived the first stage may be broken. Narrow droplet size distributions, as usually found
in the two-stage homogenization, confirm this hypothesis. A breakup of agglomerates is
discussed especially in dairy homogenization. These agglomerates form when caseins are
adsorbed at several fat globules after the breakup in the first stage (Walstra, 1975). In any
case, a second stage will provide counter pressure in the outlet of the first stage and thus
modify the local flow conditions. Future research should focus on this effect.
7.4.3 multiple passaGe

Multiple passages (i.e., the emulsion passes the homogenization device several
times) are often used, if very small droplets, near droplet size distributions, or clearly
defined upper droplet sizes are required. The latter is often found in pharmaceutical
applications for intravenous emulsions. Contrary to the multiple-stage processing,
the pressure falls in between the homogenization steps to ambient pressure, and thus
the local flow conditions are different.
Figure 7.8 gives an example: Two emulsions, of 1% and 30% dispersed phase frac-
tion, respectively, were produced in a microstructured valve of a round shape in 1 and
8 passages each. By multiple passaging the number of large droplets is reduced in the
emulsion of 1% dispersed phase fraction. At the same time, many small droplets are pro-
duced. In the case of a higher dispersed phase fraction (φ = 30%), however, additional
effects are found: some large droplets also disappear after multiple passaging. However,
1.2
Orifice
Veg. oil-in-water
Volume distribution, q3 (μm−1)
φ = 1% O/W emulsifier: Tween 80
0.8
Number of passages n
φ = 30%
1
φ = 1%
0.4 8
1
φ = 30%
8
0
10−2 10−1 100 101
Droplet diameter, d (μm)
FIGURe 7.8 Multiple passage: droplet size distribution depending on the dispersed
phase content and the number of passages. (Data from Tesch, S., Charakterisieren mecha-
nischer Emulgierverfahren: Herstellen und Stabilisieren von Tropfen als Teilschritte beim
Formulieren von Emulsionen, Dissertation, Universität Karlsruhe, Germany, 2002.)
we do not see an increase in small droplets. As coalescence plays a more pronounced

role, small droplets cannot be stabilized and coalesced. Even so, the distribution nar-
rows, and very big droplets disappear.
7.4.4 ViscOsity ratiO

As emulsification mechanisms do not differ for commercial or microstructured valves,
the influence of material parameters such as the viscosity ratio between the dispersed
phase and the emulsion, ηd/ηe, is similar to the known knowledge. Usually, smallest drop-
lets are achieved at a constant energy density with a viscosity ratio between 0.1 and 1.
This can be demonstrated using the example of O/W and water-in-oil (W/O) emulsions.
In general, both W/O and O/W emulsions can be produced in HPHs. Differences
are found, however, in the homogenization result (here, Sauter mean diameter of the
droplets), when comparing results at constant specific energy (or homogenization pres-
sure). In Figure 7.9, the Sauter diameter of the emulsions produced by simple orifices
with a diameter of 0.5 mm is dependant on the energy density, the emulsion type, and
the dispersed phase fraction is plotted. In both emulsions, a fast stabilizing emulsifier
was used. As we see a negative and constant slope, and as no significant differences can
be seen between the different dispersed phase fractions, the homogenization process
is dominated by disruption. Also, the linear trend has a slope of 1, which indicates that
both elongation and shear flow play a dominant role (see process function).
Comparing both emulsion types, nearly the same range of homogenization results
can be achieved. Sauter droplet diameter sizes between 400 nm and 2 µm are pro-
duced. However, homogenization pressures required differ: although pressures of
100–400 bar are required for O/W emulsions, pressures up to 1000 bar are needed
2
Orifice
Water, veg. oil
O/W emulsifier: Tween 80
Sauter mean diameter, x12 (μm)
1
W/O emulsifier: Triodan
Dispersed phase fraction
φ (vol%):
W/O O/W
30% 30%
50% 60%
70%
0.1
106 107 2.107
Energy density, Ev (J/m3)
FIGURe 7.9 Influence of the dispersed phase fraction on the Sauter mean diameter is depen-
dant on the energy density for different emulsion types. (Data from Tesch, S., Charakterisieren
mechanischer Emulgierverfahren: Herstellen und Stabilisieren von Tropfen als Teilschritte
beim Formulieren von Emulsionen, Dissertation, Universität Karlsruhe, Germany, 2002.)
for W/O emulsions. At the same energy density (or homogenization pressure), the
droplets of the W/O emulsions are smaller by a factor of 2. This can be explained by
the breakup mechanism. Breakup is improved at the viscosity ratio ηd/ηe between
0.1 and 1, as is found in O/W emulsions. In W/O emulsions, the viscosity ratio ηd/ηe
is <<0.1, resulting in higher critical capillary numbers.
7.4.5 staBility
As discussed earlier, homogenization in HPHs is a fast process. Droplets deform and
break within milliseconds and less. Thus, the homogenization is only crowned by
success, if the emulsifier is fast enough to stabilize the newly formed surfaces.
One example for a critical homogenization in terms of stabilization is the homog-
enization of concentrated milk, also called cream. Dairy processes are one of the old-
est industrial high-pressure homogenization processes and up to now, the ones with
the biggest volume streams. In conventional processing, raw milk is separated prior to
homogenization into a low-fat phase (0.03–0.3 vol% fat, skim milk) and a fat-enriched
phase (13–42 vol% fat, cream) using a separator (Kessler, 2002). In the conventional
full stream homogenization process, the milk is first standardized to the final product fat
content by mixing these two phases. The product of this process is then homogenized at
pressures around 100 bar. Also, conventionally applied are partial stream homogeniza-
tion processes. Here, the cream is diluted with skim milk to a fat content of 13–17 vol%,
then homogenized, and afterward standardized again to the target fat concentration of,
for example, 3.5 vol% in full-cream milk. This reduces the required energy as less con-
tinuous phase has to be compressed to nearly the same homogenization pressure.
This two-step remixing process interrupted by high-pressure homogenization is

required as aggregation of fat globules (casein bridging) is found in homogenized cream
with more than 13 vol% of fat (Köhler, Karasch et al., 2008). The negative impact of
aggregation can be compensated by an increased homogenizing pressure up to fat con-
tents of 17 vol%. At fat contents higher than that, the process is controlled by coalescence
and mainly aggregation of the fat globules, leading to dissatisfying homogenization
results and product quality. Coalescence of the newly formed fat droplets is found until
adsorbing dairy proteins have stabilized the droplets. In the stabilization of milk fat
globules, a secondary droplet membrane is built up by adsorbing casein micelles and
submicelles, as well as lactalbumins and lactoglobulins (Walstra and Oortwijn, 1982;
Dalgleish, Tosh, and West, 1996). As the adsorbed casein micelles tend to adsorb into
more than one fat globule at the same time and also strongly interact, bridges between
fat globules are formed at increased fat content resulting in high aggregation rates.
These fat globule aggregates can be partially destroyed in the second homogenizing
stage (Darling and Butcher, 1978), as it is realized in conventional technical processes.
With increasing fat globule concentration, however, the coalescence and aggregation
rates increase as well (Ogden, Walstra, and Morris, 1976; Walstra, 1999; Köhler et al.,
2007), limiting partial homogenization to 17 vol% of fat. Thus, the energy reduction
potential cannot be fully exploited. The standardization process dilutes the product and
adds additional surface active proteins to it. However, in the common dairy process, the
standardization (which in fact is done by mixing) is located several meters behind the
homogenizer, and thus, happens too late. Coalescence and casein bridging has already
happened. When the mixing of the skim milk into the cream happens directly in the
moment, when the fat globules are disrupted, coalescence and casein bridging can be
avoided. This is realized in a specific microstructured homogenization valve, called
simultaneous emulsifying and mixing (SEM) valve (Köhler et al., 2007). Here, a micro-
structured mixing unit is integrated into a homogenization orifice. Mixing of skim milk
is realized within few microseconds at a high local mixing efficiency.
SEM valves thus enable (1) to combine the homogenization and standardization
step into one process unit, (2) to dilute the fat globules directly after their production,
and (3) to add additionally emulsifier molecules (here, dairy proteins) in the moment
of their need (droplet breakup) with high mixing intensity. Cream is pumped through
the orifice of the SEM valve at a fat content of 32–42 vol%, and the skim milk is
mixed into the cream in the first millimeters after the orifice outlet at pressures of
0.1%–20% of the homogenizing pressure.
Comparing the SEM partial homogenization results at 32 and 42 vol% fat, respec-
tively, to those of conventional full-stream homogenization at 3.5 vol% fat, the product
quality is fully maintained by the same homogenization pressure [with a slight, but
negligible improvement at 32 vol% and a slight loss of 42 vol% fat (see Figure 7.10)].
However, the new SEM process requires only 20% of the energy input com-
pared to the full-stream process, and only 60% of the energy applied in conven-
tional partial homogenization processing at comparable process parameters. This
results in considerable energy and cost savings in dairy processing without any
loss in product quality. In addition, two mixing units can be eliminated from the
process line resulting in less investment, cleaning, and maintenance costs (process
intensification).
Full-stream homogenization
φmilk = 3.5 vol%
6 φmilk = 32 vol%
SEM
φcream = 42 vol%
4 φcream = 32 vol%
φSM = 0.3 vol%
Distance, l = 3 mm
Pressure ratio
2 Cream/SM = 10/1.5
Volume ratio
Cream/SM = 1/10
Homogenization temperature:
ϑ = 65°C
0 One stage
40 50 100 150 200 250 300 350
FIGURe 7.10 Influence of the fat content of homogenized cream and the homogenization
pressure on the characteristic maximum droplet diameter, x903, for full-stream and partial
homogenization processing. When SEM valves are used, cream of 32 and 42 vol% fat were
mixed with skim milk (SM, φ = 0.3 vol%) within the valve. In the conventional full-stream
homogenization milk (volume fat content φ = 3.5 vol%) and cream (φ = 32 vol%) were homoge-
nized as full stream. (Data from Köhler, K. et al., Chem. Eng. Technol., 31, 12, 1863–1868, 2008.)
7.4.6 scale-up
A scale-up of disruption units can generally be realized by a geometric magnifica-
tion (scaling up) or by adding several units at same scale (numbering up). A scale-up
by a geometric magnification of orifices is usually limited as the local flow condi-
tions responsible for droplet breakup are changed. It was shown that, for example, on
simple orifices, a geometric magnification of the orifice diameter is only possible up
to a diameter d > 0.8 mm (Aguilar et al., 2008). A scale-up by increasing the number
of orifices (numbering up) is possible and technically realized, for example, in the
jet disperser® of Bayer Technology Services (BTS, Leverkusen, D) (Bayer AG, 1997,
2001). Important in this context is the distance between the different orifices, thus
the maximum number of holes per unit area. Aguilar et al. (2008) showed that a ratio
of the distance between two holes and the hole diameter has to be over 6.
Today, HPHs with flat valves are available with flow rates up to 50,000 L/h, for
example, in the dairy industry. In orifice systems, flow rates up to 5,000 L/h are
commercially available.
7.5 ConCLUsIon
HPHs are broadly used today in homogenization process. Besides valves, micro-
structured devices such as orifices of simple geometry (slit or hole) or one with
deflecting streams (e.g., Microfluidizer) are well established. Laminar, turbulent
and cavitational flow is found in those orifices producing shear, elongational, and
Reynolds tensions, which deform and break droplets in the timescale of microsec-
onds. Today, many models are available to describe the homogenization result for
specific homogenization devices qualitatively. However, a detailed understanding of
the homogenization process and the influence of especially geometric orifice param-
eters on droplet breakup is missing. A better understanding of local flow conditions
and a focus of ongoing research are required.
For large-scale production (e.g., the dairy process) commercial homogeniza-
tion valves are still state of the art. However, especially when specific solutions, for
example, improved breakup or stabilization, are required, microstructured devices
offer real alternatives. SEM-type valves offer the chance to change local conditions
during the moment of droplet breakup and stabilization such as temperature, pH,
chemical composition, or droplet concentration.
ReFeRenCes
Aguilar, F.A., Freudig, B., Schuchmann, H.P.: Herstellen von Emulsionen in Hoch-
druckhomogenisatoren mit modifizierten Lochblenden, Chemie Ingenieur Technik,
76 (4), 396–399, 2004.
Aguilar, F.A., Köhler, K., Schubert, H., Schuchmann, H.P.: Herstellen von Emulsionen in ein-
fachen und modifizierten Lochblenden: Einfluss der Geometrie auf die Effizienz der
Zerkleinerung und Folgen für die Maßstabsvergrößerung, Chemie Ingenieur Technik,
80 (5), 607–613, 2008.
Arai, K., Konno, M., Matinaga, Y., Saito, S.J.: Effect of dispersed-phase viscosity on the maxi-
mum stable drop size for breakup in turbulent flow, Chemical Engineering of Japan,
10 (4), 325–330, 1977.
Armbruster, H.: Untersuchungen zum kontinuierlichen Emulgierprozeß in Kolloidmühlen unter
Berücksichtigung spezifischer Emulgatoreigenschaften und der Strömungsverhältnisse
im Dispergierspalt, Dissertation, Universität Karlsruhe (TH), Germany, 1990.
Baldyga, J., Orciuch, W., Makowski, L., Malski-Brodzicki, M., Malik, K.: Break up of nano-
particle clusters in high-shear devices, Chemical Engineering and Processing, 46 (9),
851–861, 2007.
Bayer AG, Patentnr.: US4996004, Preparation of pharmaceutical or cosmetic dispersions,
February 26, 1991.
Bayer AG, Verfahren und Vorrichtung zur Herstellung einer parenteralen Arzneistoffzubereitung,
1997.
Bayer AG, Dispersion nozzle with variable throughput, July 4, 2001.
Bayer MaterialScience AG, Homogenizing nozzle and method for the production of an aque-
ous two-component polyurethane lacquer emulsion, 2006.
Bechtel, S., Gilbert, N., Wagner, H.G.: Grundlagenuntersuchungen zur Herstellung von
Emulsionen im Ultraschallfeld, Chemie Ingenieur Technik, 71 (8), 810–817, 1999.
Bechtel, S., Gilbert, N., Wagner, H.G.: Grundlagenuntersuchungen zur Herstellung von
Emulsionen im Ultraschallfeld Teil 2, Chemie Ingenieur Technik, 72 (5), 450–459,
2000.
Behrend, O.: Mechanisches Emulgieren mit Ultraschall., Dissertation, Universität Karlsruhe
(TH), Germany, 2002.
Behrend, O., Ax, K., Schubert, H.: Influence of continuous phase viscosity on emulsification
by ultrasound, Ultrasonics Sonochemistry, 7 (2), 77–85, 2000.
Bentley, B.J., Leal, L.G.: An experimental investigation of drop deformation and breakup
in steady, two-dimensional linear flows, Chemical Engineering Department, CIT,
December 21, 1985.
Bentley, B.J., Leal, L.G.: An experimental investigation of drop deformation and breakup
in steady two-dimensional linear flows, Journal of Fluid Mechanics, 167, 241–283,
1986.
Bondy, C., Söllner, K.: On the mechanism of emulsification by ultrasonic waves, Journal of
the Chemical Society Transactions of the Faraday Society, 31, 835–843, 1935.
Casoli, P., Vacca, A., Berta, G.L.: A numerical procedure for predicting the performance of
high pressure homogenizing valves, Simulation Modelling Practice and Theory, 18 (2),
125–138, 2010.
Chesters, A.K.: The modelling of coalescence processes in fluid- liquid dispersions: A review
of current understanding, Chemical Engineering Research and Design, 69 (4), 259–270,
1991.
Cook, E.J.: Patentnr.: 4533254, Microfluidizer (Teil I), August 6, 1985.
Cook, E.J.: Patentnr.: 4908154, Microfluidizer (Teil II), March 13, 1990.
Cook, E.J., Lagace, A.P.: Patentnr.: 4533254, Apparatus for forming emulsions, Biotechnology
Development, 1985.
Dalgleish, D.G., Tosh, S.M., West, S.: Beyond homogenization: The formation of very small
emulsion droplets during the processing of milk by a Microfluidizer, Netherlands Milk
and Dairy Journal, 50 (2), 135–148, 1996.
Danner, T.: Tropfenkoaleszenz in Emulsionen, Dissertation, Universität Karlsruhe (TH), 2001.
Darling, D.F., Butcher, D.W.: Milk-fat globule membrane in homogenized cream, Journal of
Dairy Research, 45 (2), 197–208, 1978.
Davies, J.T.: Turbulence phenomena at free surfaces, American Institute of Chemical Engineers
Journal, 18 (1), 169–173, 1972.
Davies, J.T.: Drop sizes of emulsions related to turbulent energy-dissipation rates, Chemical
Engineering Science, 40 (5), 839–842, 1985.
Eggers, J.: Nonlinear dynamics and breakup of free-surface flows, Reviews of Modern Physics,
69 (3), 865–930, 1997.
EN ISO 5167-1: Durchflussmessung von Fluiden mit Drosselgeräten in voll durchströmten
Leitungen mit Kreisquerschnitt Teil 1: Allgemeine Grundlagen und Anforderungen
(ISO 5167-1:2003); Deutsche Fassung EN ISO 5167-1:2003, 2003.
Floury, J., Bellettre, J., Legrand, J., Desrumaux, A.: Analysis of a new type of high pres-
sure homogeniser. A study of the flow pattern, Chemical Engineering Science, 59 (4),
843–853, 2004.
Freudig, B.: Herstellen von Emulsionen und Homogenisieren von Milch in modifizierten
Lochblenden, Dissertation, Universität Karlsruhe, Germany, 3-8322-3147-1, 2004.
Gaulin, A.: Patentnr.: Brecet nr. 295596, US625497, Appareil et Procédé pour la Stabilisation
du Lait, May 23, 1899.
Grace, H.P.: Dispersion phenomena in high-viscosity immiscible fluid systems and appli-
cation of static mixers as dispersion devices in such systems, Chemical Engineering
Communications, 14 (3–6), 225–277, 1982.
Hinze, J.O.: Fundamentals of the hydrodynamic mechanism of splitting in dispersion pro-
cesses, American Institute of Chemical Engineers Journal, 1 (3), 289–295, 1955.
Innings, F.: Drop Break-up in High-Pressure Homogenisers, Dissertation, Lund University,
Faculty of Engineering, 91-628-6646-X, 2005.
Jansen, K.M.B., Agterof, W.G.M., Mellema, J.: Droplet breakup in concentrated emulsions,
Journal of Rheology, 45, 227–236, 2001.
Johansen, F.C.: Flow through pipe orifices at low Reynolds Numbers, Proceedings of the
Royal Society of London. Series A, 126 (801), 231–245, 1930.
Karasch, S., Kulozik, U.: Hochdruckhomogenisierung von Milch mit modifizierten Lochblenden
im Vergleich zu konventionellen Flachventilen, Chemie Ingenieur Technik, 80 (8), 1117–
1124, 10.1002/cite.200800079, 2008.
Karbstein, H.: Untersuchungen zum Herstellen und Stabilisieren von Öl-in-Wasser-
Emulsionen, Dissertation, Universität Karlsruhe, 1994.
Kessler, H.G.: Food and Bio Process Engineering—Dairy Technology, Verlag A. Kessler,
München, Germany, 2002.
Kissling, K., Schütz, S., Piesche, M.: Numerical investigation on the deformation of drop-
lets in high-pressure homogenizers. in High Performance Computing in Science and
Engineering ‘10 (Editors: Nagel, Wolfgang E., Kröner, Dietmar B., and Resch, Michael
M.), 287–294, Springer-Verlag, Berlin, Germany, 978-3-642-15748-6, 2011.
Kleinig, A.R., Middelberg, A.P.J.: Numerical and experimental study of a homogenizer
impinging jet, American Institute of Chemical Engineers Journal, 43 (4), 1100–1107,
1997.
Köhler, K.: Simultanes Emulgieren und Mischen, Logos Verlag, Berlin, Germany, 978-3-8325-
2716-7, 2010.
Köhler, K., Aguilar, F.A., Hensel, A., Schubert, K., Schubert, H., Schuchmann, H.P.: Design
of a microstructured system for homogenization of dairy products with high fat content,
Chemical Engineering & Technology, 30 (11), 1590–1595, 10.1002/ceat.200700266,
2007.
Köhler, K., Aguilar, F.A., Hensel, A., Schubert, K., Schubert, H., Schuchmann, H.P.: Design of
a microstructured system for the homogenization of dairy products at high fat content.
Part II: Influence of process parameters, Chemical Engineering & Technology, 31 (12),
1863–1868, 2008.
Köhler, K., Karasch, S., Schuchmann, H.P., Kulozik, U.: Energiesparende Homogenisierung
von Milch mit etablierten sowie neuartigen Verfahren, Chemie Ingenieur Technik,
80 (8), 1107–1116, 10.1002/cite.200800070, 2008.
Kolb, G.E.: Zur Emulsionsherstellung in Blendensystemen, Dissertation, Universität Bremen,
Germany, 3-8265-9204-2, 2001.
Kolmogorov, A.N.: O Droblenii Kapel V Turbulentnom Potoke, Doklady Akademii Nauk
SSSR, 66 (5), 825–828, 1949.
Meinhart, C.D., Wereley, S.T., Santiago, J.G.: PIV measurements of a microchannel flow,
Experiments in Fluids, 27 (5), 414–419, 1999.
Mielnik, M.M., Saetran, L.R.: Micro particle image velocimetry—An overview, International
Workshop on Size Effects in Microfluidics and Heat Transfer—Fundamental and
Practical Aspects, 10, 83–90, September 16, 2004.
Miller, R.: Adsorption kinetics of surfactants at fluid interfaces: Experimental conditions and
practice of application of theoretical models, Colloids and Surfaces, 46, 75–83, 1990.
Muschiolik, G., Roeder, R.-T., Lengfeld, K.: Patentnr.: DE 19530247, Druckhomogenisator,
August 17, 1995.
Ogden, L.V., Walstra, P., Morris, H.A.: Homogenization-induced clustering of fat globules in
cream and model systems, Journal of Dairy Science, 59 (10), 1727–1737, 1976.
Penth, B., Patentnr.: WO/2000/061275, Method and device for carrying out chemical and
physical processes, April 7, 2000.
Plateau, J.A.F.: Statique expérimentale et théorique des liquides soumis aux seules forces
moléculaires, Acad. Sci. Brux. Mem., 2, 1873.
Schlichting, H., Gersten, K.: Grenzschicht-Theorie, 10th ed. (Editors: Schlichting, Hermann
and Gersten, Klaus), Online-Ressource, Springer-Verlag, Berlin, Germany, 978-3-540-
32985-5, 2006.
Sheng, J., Meng, H., Fox, R.O.: A large eddy PIV method for turbulence dissipation rate estimation,
Chemical Engineering Science, 55 (20), 4423–4434, 2000.
Silver, R.S.: The theories of stress due to collapse of vapour bubbles in a liquid, F. Inst. P.
Engineering (London), 154, 501ff., December 25, 1942.
Stang, M.: Zerkleinern und Stabilisieren von Tropfen beim mechanischen Emulgieren,
Dissertation, Universität Karlsruhe, Germany, 1998.
Stang, M., Karbstein, H., Schubert, H.: Influence of emulsifier adsorption kinetics and emulsi-
fication machine construction on disparity of oil-in-water emulsions in Food Colloids—
Proteins, Lipids and Polysaccharides (Editors: Dickinson, E. and Bergenstähl, B.),
382–392, 1994.
Steiner, H., Teppner, R., Brenn, G., Vankova, N., Tcholakova, S., Denkov, N.: Numerical simu-
lation and experimental study of emulsification in a narrow-gap homogenizer, Chemical
Engineering Science, 61 (17), 5841–5855, 2006.
Stevenson, M.J., Chen, X.D.: Visualization of the flow patterns in a high-pressure homogeniz-
ing valve using a CFD package, Journal of Food Engineering, 33 (1–2), 151–165, 1997.
Stone, H.A., Bentley, B.J., Leal, L.G.: An experimental-study of transient effects in the
breakup of viscous drops, Journal of Fluid Mechanics, 173, 131–158, 1986.
Taylor, G.I.: The viscosity of a fluid containing small drops of another fluid, Proceedings of the
Royal Society of London. Series A, 138 (834), 41–48, 1932.
Tcholakova, S., Lesov, I., Golemanov, K., Denkov, N.D., Judat, S., Engel, R., Danner, T.:
Efficient emulsification of viscous oils at high drop volume fraction, Langmuir, 27 (24),
14783–14796, 2011.
Tesch, S.: Charakterisieren mechanischer Emulgierverfahren: Herstellen und Stabilisieren
von Tropfen als Teilschritte beim Formulieren von Emulsionen, Dissertation, Universität
Karlsruhe (TH), Germany, 2002.
Vankova, N., Tcholakova, S., Denkov, N.D., Ivanov, I.B., Vulchev, V.D., Danner, T.:
Emulsification in turbulent flow, 1. Mean and maximum drop diameters in inertial and
viscous regimes, Journal of Colloid and Interface Science, 312, 363–380, 2007.
Walstra, P.: Effect of homogenization on the fat globule size distribution in milk, Netherlands
Milk and Dairy Journal, 29 (2–3), 279–294, 1975.
Walstra, P.: Formation of emulsions. in Encyclopedia of Emulsion Technology, Vol. 1 (Editor:
Becher, P.), 57–128, Marcel Dekker Inc., New York, 1983.
Walstra, P.: Principles of emulsion formation, Chemical Engineering Science, 48 (2), 333–349,
1993.
Walstra, P.: Casein sub-micelles: Do they exist?, International Dairy Journal, 9 (3–6), 189–192,
1999.
Walstra, P., Oortwijn, H.: The membranes of recombined fat globules. 3. Mode of formation,
Netherlands Milk and Dairy Journal, 36 (2), 103–113, 1982.
Wengeler, R.: Hydrodynamic Stress Induced Dispersion of Nanoscale Agglomerates by a High
Pressure Process, Cuvillier Verlag, Göttingen, Germany, 978-3-86727-182-0, 2007.
Wolf, F., Schuch, A., Köhler, K., Schuchmann, H.P.: Ansatz zur Beschreibung der zerkleiner-
ungsrelevanten Strömungsverhältnisse beim Emulgieren von W/O-Emulsionen mit
Lochblenden, Chemie Ingenieur Technik, 84 (12), 2215–2220, 10.1002/cite.201100065,
2012.
8 Rotor–Stator Devices
contents
8.1 Introduction .................................................................................................. 195
8.2 Technical Equipment .................................................................................... 196
8.2.1 Stirred Vessel .................................................................................... 196
8.2.2 Colloid Mill ...................................................................................... 197
8.2.3 Toothed-Rim Dispersing Machines .................................................. 198
8.2.4 Extruder ............................................................................................ 198
8.2.5 Rotor–Rotor Devices ........................................................................ 199
8.2.6 Pumps ............................................................................................... 199
8.3 Emulsification Mechanism ........................................................................... 199
8.4 Emulsification Parameters ............................................................................200
8.4.1 Droplet Disruption: Processing Parameters and Geometrical
Design of RSDs ................................................................................200
8.4.2 Emulsion Recipe: Viscosity Ratio and Dispersed Phase Fraction....200
8.4.3 Stabilization of Droplets against Coalescence .................................202
8.5 Scale-Up .......................................................................................................204
8.6 Conclusion ....................................................................................................204
References ..............................................................................................................205
ABSTRACT In this chapter, emulsification in pumps, stirred vessels, colloid mills,

toothed-rim dispersers, double rotor dispersers, and extruders are described. The uni-
fying parameter of energy density is described, and the positive consequences of
batch processing, particularly for emulsions, where the droplets slowly stabilize, are
highlighted.
8.1 IntroductIon
Rotor–stator devices (RSDs) are probably the most widely used emulsifying sys-
tem. They consist of a minimum of one rotating part. The complexity of these com-
ponents ranges from simple stirrer systems, such as propeller stirrers rotating in a
vessel as a stator, to rotor–rotor systems with two rotating parts, but with no stator.
Section 8.2 explains in detail about their structural design and operation.
One of the main benefits of RSDs is the fact that they can be run in batch, semi-
batch, and alternating as well as in a continuous mode, each having its respective
merits (Figure 8.1). The batch mode offers the advantage of realizing many process
operation steps in parallel. Thus, products are mixed, pasteurized, homogenized,
195
Continuous
Batch
Semibatch
(a) (b) Alternating
FIGure 8.1 Operation modes in the homogenization process realized in RSDs: (a) batch
and semibatch modes and (b) continuous and alternating methods.
and cooled in one vessel, which is used for mayonnaise-type products or sauces
in food industry. However, it cannot be ensured that the product volume in total is
processed on equal terms, which presents the main disadvantage in batch process-
ing. Especially the broad distribution of stresses acting on emulsion droplets and
residence time results in a broad distribution of droplet sizes and development of
by-products, often unwanted. Therefore, extreme process conditions as required; for
example, emulsions with droplets in the submicron-size range have to be realized by
continuous process.
By the movement of the rotor, the fluid in the device is accelerated, thereby form-
ing certain flow patterns. These flow patterns lead to tensions that deform and eventu-
ally break up emulsion droplets. Flow patterns typically found in RSDs are discussed
in Section 8.3.
In Section 8.4, influences of various processes and material parameters on the
emulsification result are described. These influences are discussed and illustrated
by way of resulting droplet size distributions in the manufacture of both oil-in-water
(O/W) and water-in-oil (W/O) emulsions.
8.2 technIcal equIpment

All rotor–stator and rotor–rotor devices have a rotating part in common.
Depending on the shape of the rotor, four groups of devices can be differentiated
(see Figure 8.2).
8.2.1 Stirred VeSSel

Stirred vessels are rotor–stator systems and are very often used in the industry. It
is a simple construction and thus often seems to be the cheapest solution. A motor
rotates the stirrer in a vessel. The stirrer has to fulfill two aspects. The main goal of
rotating the stirrer is to disrupt the droplets. In most cases, the area of high tensions
being responsible for the breakup of droplets is concentrated to the edges of the
stirrer plate. Therefore, the stirrer has to fulfil a second goal, to move all droplets to
areas of high tensions. The second aspect is a mixing problem, which is explained in
detail in Zlokarnik (1999). To optimize both aspects independently, one opportunity
Rotor–Stator Devices 197
(a) (b)
(c) (d)
FIGure 8.2 Rotating part in various rotor–stator and rotor–rotor devices: (a) stirred vessel,
(b) colloid mill, (c) toothed-rim dispersing machine, and (d) extruder.
is the use of several stirred tanks. The second one is to use a single stirred tank with
different stirrers (e.g., anchor and propeller stirrer), or a stirred tank combined with
a colloid mill (see Figure 8.1). An alternative is to use different stirrer plates on a
single stirrer rod.
Dispersion discs are proposed especially for high-viscous emulsions (viscosity
around 10 Pa s) to be produced with slowly adsorbing emulsifiers (Urban et al.,
2006). Geometric modifications start with simple teeth and end with complex geom-
etries (see Figure 8.3).
The broad distribution of tensions acting on droplets and residence time is a main
disadvantage of stirred vessels. The following RSDs have the benefit of a controlled
volume in which the droplets are disrupted.
8.2.2 Colloid Mill

In colloid mills, emulsification takes place in a conical gap between a rotor and a sta-
tor (see Figure 8.2b). The rotor as well as the stator can have a plain or a structured
surface. Depending on the viscosity of the fluid and the surface structures of the rotor
and the stator, colloid mills may also act as a pump. The tensions required for droplet
breakup depend mainly on the rotational speed, the rotor diameter, and the gap width.
The latter is easily modified through a vertical movement of the rotor. Often, a stirrer
is additionally mounted on the top of the rotor to adjust the volume stream. This has
(a) (b)
(c) (d)
FIGure 8.3 Geometric modifications of various stirred vessels: (a) propeller stirrer, (b) dis-
persion disc, (c) anchor stirrer, and (d) helical ribbon agitators. (a, b: Courtesy of EKATO®; c, d:
Courtesy of Turbo-Rührwerke.)
the benefit of using one motor for both applications; however, this poses a problem:
the emulsification and pumping effect cannot be adjusted separately.
8.2.3 toothed-riM diSperSing MaChineS

Toothed-rim or gear-rim dispersing machines consist mainly of rings with teeth
gearing with each other. Typically, the rotor is a plate with teeth, which is adjusted in
the stator. The emulsification effect can be modified by the diameter of the rings, the
number of rings, the number of teeth per ring, the gap in between the teeth, and the
number of stages, which means several rotor plates and stators are stacked on each
other. As colloid mills, toothed-rim dispersing machines may act as pump, with the
effect depending on the teeth and ring geometry.
8.2.4 extruder
An extruder consists of a housing and a minimum of one screw and one die. Typical
extruders are single-screw or twin-screw extruders, equipped with one or two screws,
respectively. If two screws are mounted, they intermesh and run in a counter or coro-
tating mode (Riaz, 2000). In general, extruders of up to eight screws are fabricated,
but often the benefit of, for example, a higher and a more constant volume stream is
not worth the problems and costs (Guy, 2001).The length of the extruder screw can
vary from some centimeters to several meters. Extruders can only be operated in a
continuous mode. Depending on geometric parameters and screw speed, volume flow
rates are between several tens of kilogram per hour and tons per hour. Extruders are
usually the device of choice if very high viscous phases, such as polymers or waxes,
have to be mixed and emulsified. The extruder screw is equipped with elements of
different design. As some elements are designed to essentially pump the liquid to the
die, others are adapted for improved mixing or emulsification effects. Simple extrud-
ers (also called mono pumps) have screws designed only for forward pumping a fluid.
Higher tensions, as required for emulsification, are produced through backward or
kneading elements, the latter existing in neutral, forward, and backward flow direc-
tion. In backward elements, the fluid is mostly pressed back through the gap of the
screw and the housing, which also increases the residence time of the fluid in the
extruder. At the end of the extruder barrels, a die is often mounted. This die is often
designed as either a simple hole or an orifice of spherical shape, reducing the cross-
sectional area by a high degree. If a die is mounted at the end of the extruder, both the
fluid residence time and the tensions acting on the fluid increase. The gaps between
the screw elements and the housing are in the size of some micrometer to some mil-
limeter. Screw rotational speeds usually are in the range of some hundred revolutions
per minute (rpm), with high-throughput extruders running at up to 1800 rpm.
8.2.5 rotor–rotor deViCeS

Nearly 10 years ago, a rotor–rotor device was placed on the market (Symex
GmbH&Co KG, 2000), with both rotors being toothed rims. If both rotors are coun-
terrotated, high tensions can be achieved, whereas lower tensions but higher volume
streams can be achieved if both turn in the same direction. This advantage is paid
by a second motor and gearbox. To increase the volume stream, it is often easier to
add a separate pump. Increasing tensions help when emulsions with droplets in the
submicron-size range are required. Some applications are established till date.
8.2.6 puMpS
The centrifugal pump is an emulsification device that is broadly distributed. The
blades of the pumps are comparable with the plates of a stirrer and have the same
effect. However, these pumps are most often adjusted to hauling a fluid, which results
in low tensions for emulsification. In some cases, these tensions are high enough for
the required emulsification effect. In this chapter, however, emulsification in or by
pumps is not further discussed, as in most cases, the pumping effect is dominant and
the emulsification effect is negligible.
8.3 emulsIFIcatIon mechanIsm

In all RSDs, droplet breakup is achieved by tensions produced in the flowing con-
tinuous medium, with the flow induced by the rotating device part. In general, the
shear flow is induced in the gap between the rotor and the stator, with the shear
rate depending on the velocity difference, the gap width, and the fluid viscosity.
Depending on the geometry, as well as circumferential and axial speeds (Taylor and
axial Reynolds number), Taylor vortices are found or flow turns turbulent (Taylor,
1923; Wimmer, 1988; Karbstein, 1994; Atiemo-Obeng, Penney, and Armenante,
2004). Some RSDs are designed for inducing additional elongational flow. Cavitation
is also found in some devices, however, often unwanted, and therefore neglected in
this chapter. Laminar and turbulent flow result in elongational or shear tensions act-
ing on droplets. Emulsification mechanisms resulting from these are explained in
detail in Section 7.3.
8.4 emulsIFIcatIon parameters

8.4.1 droplet diSruption: proCeSSing paraMeterS
and geoMetriCal deSign of rSdS
As discussed in Chapter 7, local tensions acting on droplets determine the resulting

droplet size. In all RSDs, an increase in the rotational speed as well as the diameter of
the device results in an increase of local tensions, thereby resulting in the formation of
smaller droplets. Same is true for an increase in the periods for which the tensions act
on the droplets. Due to the fact that these tensions as well as time cannot be analyzed,
in most RSDs, the specific energy density, Ev —summarizing both aspects—can be
used as a representative parameter, even if this value only accounts for mean tensions
and mean periods of deformation. Specific energy density values can easily be deter-
mined in RSDs by measuring, for example, the temperature shift directly before and
after the emulsification device, or by measuring the processing time and the electric
power uptake of the rotor (subtracting the idling drive power). The energy density in
RSDs is increased by different processes and material parameters. Examples include
a decrease in the gap size between the rotor and the stator, an increase in the rotor
diameter, in the number of toothed-rim devices, and in emulsion viscosity. Empirical
results gained by working with a wide range of continuous working RSDs colloid
mills and toothed-rim devices at different process parameters and emulsion recipes
prove that emulsions of comparable mean droplet sizes result from comparable abso-
lute values of the energy density, which are independent of the parameter set chosen
(Karbstein, 1994). Figure 8.4 gives an example for colloid mills. Here, the surface
geometry of the rotor is changed and the resulting energy density is measured.
Differences are found for stirred-vessel applications. Here, the effect of insuf-
ficient mixing dominates. With increasing processing time, more and more droplets
pass the areas of droplet disruption. However, once all droplets are disrupted to the
size corresponding to the maximum tensions found in this system, an increase in
processing time does not result any more in a decrease in droplet size. This is true
for not only stirred vessels but also all RSDs that run in batch mode.
8.4.2 eMulSion reCipe: ViSCoSity ratio and diSperSed phaSe fraCtion

Basic studies in a four-roll apparatus enable us to study the effect of single param-
eters in pure and quasi-stationary laminar shear or elongational flow (Grace, 1982;
Bentley and Leal, 1986). In a single droplet view, the ratio between the viscosities
10
Collid mill, 30% O/W
emulsifier: LEO-10
Simple teeth, 30°, with/without sawtooth

Cross-toothed with/without sawtooth
Simple teeth, 0°
Sawtooth
1
106 107 108
FIGure 8.4 Linear trend between the energy density and Sauter mean diameter indepen-
dent of the rotor geometry of a colloid mill. Process parameters varied: rotor speed and gap
width. (Data from Karbstein, H., Untersuchungen zum Herstellen und Stabilisieren von Öl-in-
Wasser-Emulsionen, Dissertation, Universität Karlsruhe, Germany, 1994.)
of both the dispersed and the continuous phases has a major effect on droplet
break-up. As long as the dispersed phase content is low, the single-droplet scenario
can also be assumed. However, at higher droplet numbers, as found in typical
emulsions, the viscosity ratio of the dispersed phase to the emulsion is relevant
(Armbruster, 1990; Jansen, Agterof, and Mellema, 2001). As RSDs often produce
flow patters similar to those of a four-roll apparatus, this basic research result
also applies. Figure 8.5 depicts the impact of the viscosity ratio on the Sauter
mean droplet size for a toothed-rim RSD. Emulsions of O/W and W/O types were
produced with varying process parameters (rotor speed and toothed-rim design).
The dispersed phase fraction was set to 30% and a fast adsorbing emulsifier was
chosen to ensure that droplet breakup dominates coalescence. The parameter given
in the diagram for the different dots is the specific energy, Ev, that resulted from
the different processing parameters and the emulsion viscosity. For its definition,
see Chapter 7.
Independent of the specific energy input and the emulsion type, the small drop-
lets are achieved at a viscosity ratio λ in the range 0.1–1, as found in the basic
studies (Bentley and Leal, 1986). As the mean droplet size increase strongly with
the increasing viscosity ratio, we can hypothesize that droplet breakup is mainly
induced by laminar shear tensions. Elongational tensions seem to play a second-
ary role.
With this knowledge, many effects of recipe or process parameters published
in the literature for RSDs can be explained. O/W emulsions, as often found for
life science applications, are characterized by λ values well above 1. An increase
in the viscosity of the continuous phase (and thus in the viscosity of the emul-
sion, respectively) is in most cases beneficial, as λ values decrease and are closer
20
Toothed rim dispersing machine
10
30% O/W emulsifier: LEO-10

Ev (J/m3): 1⋅107 5⋅107
2⋅107 7⋅107
3⋅107 8⋅107
4⋅107 1⋅108
1
30% W/O emulsifier: PGPR-90
Ev (J/m3): 1⋅106
2⋅106
5⋅106
1⋅107
0.1
0.01 0.1 1 10 100
Viscosity ratio, λ = ηd/ηe (−)
FIGURE 8.5 Influence of the viscosity ratio between the dispersed phase and the emulsion
on the Sauter mean droplet diameter of O/W and W/O emulsions produced by a toothed-rim
dispersing machine. (–) means no dimension. (Data taken from Karbstein, H., Untersuchungen
zum Herstellen und Stabilisieren von Öl-in-Wasser-Emulsionen, Dissertation, Universität
Karlsruhe, Germany, 1994; Tesch, S., Charakterisieren mechanischer Emulgierverfahren:
Herstellen und Stabilisieren von Tropfen als Teilschritte beim Formulieren von Emulsionen,
Dissertation, Universität Karlsruhe, Germany, 2002.)
to the optimal 0.1–1 range. Same is true for increasing the dispersed phase frac-
tion (see Figure 8.6), as the emulsion viscosity increases (see Figure 8.5). Both an
increase in the dispersed phase content and the continuous phase viscosity result in
an increase in specific energy density at constant process parameters (gap width and
circumferential speed). Therefore, emulsification in RSD is improved by these two
parameters.
8.4.3 Stabilization of DropletS againSt CoaleSCenCe

As for every emulsification device, droplets being disrupted in RSDs may also
coalesce as long as they are not stabilized by emulsifier molecules adsorbed at their
new interfaces.
In order to avoid this, droplets may be disrupted several times at capillary num-
bers just above the critical one. This results in some droplets being a little bit smaller
in size, which can be stabilized by adsorbing emulsifier molecules before they are
disrupted the second time. This is typically realized in RSDs. Thus, these devices
are highly recommended for the production of emulsions that are characterized by
high coalescence rates (see Chapter 7). Especially mayonnaise-like food emulsions,
that is, emulsions of a high dispersed phase fraction (such as 60–80 vol%), that have
to be stabilized with slow adsorbing emulsifiers (such as egg yolk) can only be pro-
duced in colloid mills, whereas high-pressure homogenization results in the breakup
of emulsions (see Figure 8.7).
100
Collid mill
O/W emulsifier: LEO-10
W/O emulsifier: PGPR-90
10
Dispersed phase fraction
O/W:
Open symbol φ = 10%–40%
Filled symbol φ = 50%–80%
W/O:
Filled symbol φ = 1%–70%
1 4
10 105 106 107 108
FIGure 8.6 Sauter mean droplet size of emulsions as function of the energy density: parame-
ters changed are the dispersed phase content φ (1%–80%) and the emulsion type (W/O and O/W).
Emulsions were produced on a colloid mill. (Data taken from Karbstein, H., Untersuchungen
Karlsruhe, Germany, 1994; Tesch, S., Charakterisieren mechanischer Emulgierverfahren:
Herstellen und Stabilisieren von Tropfen als Teilschritte beim Formulieren von Emulsionen,
Dissertation, Universität Karlsruhe, Germany, 2002.)
100
O/W emulsion
Dispersed phase content of 30%
10
LEO-10
Egg yolk
1
RSD
HPH, flat valve
0.1
104 105 106 107 108 109
FIGure 8.7 Sauter mean droplet size of emulsions as function of the energy density:
changed parameters include the production device and the emulsifier type at the constant
dispersed phase content and emulsion type. (Data from Karbstein, H., Untersuchungen
Karlsruhe, Germany, 1994.)
8.5 scale-up
Scale-up of emulsification processes in RSDs today is well established. Nearly
all devices are available from laboratory scale up to production scale. Stirrers are
usually scaled geometrically for vessels of some milliliter to some thousand liters.
Adapting the residence time distribution is a major challenge. Solutions are summa-
rized in Zlokarnik (1999).
For scaling up continuous emulsification processes in colloid mills and toothed-
rim dispersing machines, the rotor and stator diameters are scaled accordingly.
Process parameters are adapted until the specific energy density is in the same range
(Karbstein, 1994). Figure 8.8 gives an example for colloid mills.
8.6 conclusIon
RSDs are often used in industrial emulsification, for example, for the production
of mayonnaise, sauces, creams, and lotions. They are available from laboratory
to production scale. In terms of geometry, four different devices can be differen-
tiated. Depending on production defaults and product viscosity, stirred vessels,
colloid mills, toothed-rim dispersing machines, and extruders are used. Stirred
vessels as such or in combination with a continuous working RSDs such as a col-
loid mill or a toothed-rim device is the method of choice whenever batch process-
ing is required or emulsification is only one of many processing steps in emulsion
production.
The main parameters responsible for droplet breakup are the rotor diameter, rota-
tional speed, gap width, and the geometry of the RSD, as well as emulsion viscosity.
100 50% rapeseed O/W

emulsifier: salted egg yolk
10
Manufacturer 1
Rotor diameter
dR = 50 mm
Manufacturer 2
Rotor diameter
dR = 110 mm
1
106 107 108
Specific energy, Ev (J/m3)
FIGure 8.8 Sauter mean diameter as function of specific energy for colloid mills of differ-
ent rotor diameter and thus volume throughput. Diameter 1: pilot scale; diameter 2: produc-
tion scale. (Data from Karbstein, H., Untersuchungen zum Herstellen und Stabilisieren von
Öl-in-Wasser-Emulsionen, Dissertation, Universität Karlsruhe, Germany, 1994.)
All parameters influence the volume-specific energy applied, which in turn is a

measure for mean tensions acting on droplets over time. Disruption is efficient in
particular for a viscosity ratio between droplets and emulsion around 1.
Droplets not only have to be disrupted but also have to be stabilized in emulsifica-
tion processes. RSDs offer the chance of disrupting droplets in a gentle manner; that
is, droplets are disrupted several times and emulsifier molecules are given time to
stabilize them in between two disruption steps. Therefore, RSDs are the emulsifica-
tion machines of choice whenever stabilization of droplets is a challenge; that is, if
products have a high dispersed phase fraction and emulsifiers used are long-chained
and are of a complex molecular structure. The latter is found in foods or cosmetics
where consumers prefer natural ingredients.
reFerences
Armbruster, H.: Untersuchungen zum kontinuierlichen Emulgierprozeß in Kolloidmühlen unter
Berücksichtigung spezifischer Emulgatoreigenschaften und der Strömungsverhältnisse
im Dispergierspalt, Dissertation, Universität Karlsruhe (TH), Germany, 1990.
Atiemo-Obeng, V.A., Penney, W.R., Armenante, P.: Rotor-stator mixing devices. in Handbook of
Industrial Mixing (Editor: Paul, E.L., Atiemo-Obeng, V.A., and Kresta, S.M.), 479–505,
John Wiley & Sons, Hoboken, NJ, 2004.
Bentley, B.J., Leal, L.G.: An experimental investigation of drop deformation and breakup in
steady two-dimensional linear flows, Journal of Fluid Mechanics, 167, 241–283, 1986.
Grace, H.P.: Dispersion phenomena in high viscosity immiscible fluid systems and appli-
cation of static mixers as dispersion devices in such systems. Chemical Engineering
Communications, 14, 225–277, 1982.
Guy, R.: Extrusion Cooking: Technologies and Applications, CRC Press, Boca Raton, FL,
9781855735590, 2001.
Jansen, K.M.B., Agterof, W.G.M., Mellema, J.: Droplet breakup in concentrated emulsions,
Journal of Rheology, 45 (1), 227–236, 2001.
Karbstein, H.: Untersuchungen zum Herstellen und Stabilisieren von Öl-in-Wasser-Emulsionen,
Dissertation, Universität Karlsruhe (TH), Germany, 1994.
Riaz, M.N.: Extruders in Food Applications, Technomic Publishing Co., Inc., CRC Press,
Miami Shores, FL, 2000.
Symex GmbH&Co KG, Patent nr.: EP1825907A1, Homogenisator (“co-twister”), Veröffentli-
chungsdatum, August 29, 2007; Eingetragen, February 16, 2001; Prioritätsdatum,
February 18, 2000.
Taylor, G.I.: Stability of a viscous liquid contained between two rotating cylinders, Philosophical
Transactions of the Royal Society of London. Series A, 223 (605–615), 289–343, 1923.
Tesch, S.: Charakterisieren mechanischer Emulgierverfahren: Herstellen und Stabilisieren
von Tropfen als Teilschritte beim Formulieren von Emulsionen, Dissertation, Universität
Karlsruhe (TH), Germany, 2002.
Urban, K., Wagner, G., Schaffner, D., Röglin, D., Ulrich, J.: Rotor-stator and disc systems for
emulsification processes, Chemical Engineering & Technology, 29 (1), 24–30, 2006.
Wimmer, M.: Viscous flows and instabilities near rotating bodies, Progress in Aerospace
Sciences, 25 (1), 43–103, 1988.
Zlokarnik, M.: Rührtechnik—Theorie und Praxis, Springer-Verlag, Berlin, Germany, 3-540-
64639-6, 1999.
Section III
Low-Energy Processes
9 Microchannel
Emulsification
Aspects of Droplet
Generation, Channel Materials,
Operating Conditions, and
Scaling-Up Strategies
Isao Kobayashi, Marcos A. Neves,
and Mitsutoshi Nakajima
Contents
9.1 Introduction .................................................................................................. 210
9.2 Principle of Droplet Generation for MC Emulsification............................... 212
9.2.1 Droplet-Generation Mechanism ....................................................... 212
9.2.2 Effect of the Flow of the Dispersed and Continuous Phases............ 214
9.3 Effect of Channel Structure and Dimensions on Emulsification .................. 216
9.3.1 Grooved MC Arrays ......................................................................... 216
9.3.2 Straight-Through MC Arrays............................................................ 217
9.3.3 Effect of Channel Dimensions .......................................................... 218
9.4 Effect of Channel Materials and Their Surface Properties on
Emulsification ............................................................................................... 220
9.4.1 Silicon MC Arrays ............................................................................ 220
9.4.2 Polymer MC Arrays .......................................................................... 222
9.4.3 Stainless Steel MC Arrays ................................................................ 223
9.5 Effect of Process Factors on Emulsification ................................................. 223
9.5.1 Temperature ...................................................................................... 223
9.5.2 Viscosities of Dispersed and Continuous Phases ............................. 226
9.5.3 Electrolyte Concentration ................................................................. 226
9.6 Scaling-Up Strategies ................................................................................... 228
9.6.1 Grooved MC Array Chip .................................................................. 228
9.6.2 Asymmetric Straight-Through MC Array Chip ............................... 229
9.7 Concluding Remarks .................................................................................... 231
References .............................................................................................................. 232
209
ABSTRACT This chapter gives an overview of some of the important aspects of

microchannel (MC) emulsification that include droplet generation via MC arrays,
chip materials, operating conditions, and scaling-up strategies. General introduc-
tion that relates to MC emulsification is described in Section 9.1. Section 9.2 first
explains the unique and very mild droplet-generation mechanism on MC emulsi-
fication that is driven by interfacial tension dominant on a micron scale. It also
focuses on the effect of flow of the dispersed and continuous phases. Section 9.3
mainly introduces major microstructures used for MC emulsification: grooved MC
arrays and asymmetric straight-through MC arrays. The effect of channel dimen-
sions on the droplet size, droplet size distribution, and droplet productivity is also
briefly reviewed. Section 9.4 introduces MC emulsification chips made of differ-
ent materials and how their surface properties affect droplet-generation behavior.
Section 9.5 focuses on some important process factors in MC emulsification such
as temperature, liquid viscosity, and electrolyte concentration. Section 9.6 intro-
duces our recent scaling-up strategies on MC emulsification chips, revealing the
potential throughput capacity of greater than 10 t of monodisperse emulsions per
year. Possible future approaches for further scaling-up MC emulsification chips
are also discussed.
9.1 IntRoDUCtIon
Emulsions are immiscible liquid-liquid systems that consist of numerous droplets
dispersed in a continuous phase. Emulsions form the basis for many kinds of tradi-
tional foods, for example, milk, cream, beverages, dressings, dips, sauces, butters,
and desserts (Dickinson and McClements, 1996), and have been increasingly used in
nutritional beverages designed for infants, elderly people, and athletes, among oth-
ers (McClements, 2005). Many emulsion-based products are used in our daily life
in the form of foods (milk, butter, mayonnaise, etc.), cosmetics (facial creams, body
lotions, hair products, etc.), pharmaceuticals (drug-delivery systems, vitamin drops,
etc.), and chemicals (paints, sprays, lubricants, etc.). Among others, appearance, rhe-
ology, and stability depend on the size of emulsion droplets, as well as on their size
distributions. Traditional emulsification techniques generally require intense energy
input, resulting in polydisperse emulsions, whereas most of the energy applied is lost
as heat, which may be detrimental to heat-sensitive bioactive compounds, generally
loaded into emulsions for targeted delivery.
The physical stability of emulsions and their underlying interfacial aspects have
been extensively investigated (Walstra, 1996). Emulsions become kinetically stable
against droplet coalescence for a finite period when surfactants are added prior to
emulsification. During storage, emulsions can separate into their original constitu-
ents, oil, and aqueous phases, either due to coalescence of the droplets or due to
creaming/sedimentation resulting from density differences between the two phases.
Coalescence can be prevented by adding a surface-active compound, which reduces
coalescence by adsorbing to the interface. In addition, creaming or flocculation can
be reduced by controlling physicochemical properties, such as decreasing the droplet
size, increasing the continuous phase viscosity, or matching the densities of both the
continuous and dispersed phases.
Microchannel Emulsification 211
(a) 100 μm (b) 100 μm
FIGURe 9.1 Optical micrographs of protein-stabilized O/W emulsion droplets: (a) mono-
disperse emulsion prepared using microchannel emulsification and (b) polydisperse emul-
sion prepared using a rotor–stator homogenizer. (Reproduced from Food Hydrocolloids,
20, Saito, M. et al., Comparison of stability of bovine serum albumin-stabilized emulsions
prepared by microchannel emulsification and homogenization, 1020–1028, Copyright 2006,
with permission from Elsevier.)
Monodisperse emulsions with uniform droplets are useful in both scientific and
industrial fields, as they have advantages such as higher stability and offer simpler
understanding and control of physical properties (McClements, 2004). Monodisperse
emulsions are also promising templates for preparing monodisperse microparticles
and microcapsules (Vladisavljević and Williams, 2005). Figure 9.1 illustrates poly-
disperse and monodisperse emulsion droplets prepared using different techniques
(Saito et al., 2006). The average size and size distribution of an emulsion affect the
stability and properties of emulsion-based products. For instance, fine emulsions
were reported to have higher viscosities as compared to coarse emulsions due to
their narrower size distribution, decreased droplet deformability, and stronger col-
loidal interactions (Nisisako, Torii, and Higuchi, 2004). The visual appearance of an
emulsion depends on the droplet size as well; for example, an emulsion with droplets
greater than 10 μm appears turbid, and the turbidity increases with decreasing drop-
let size (Steegmans, Schroën, and Boom, 2009), until the droplet size is less than
the wavelength of visible light; the emulsion then becomes transparent (Steegmans
et al., 2010). Further, a wide size distribution in polydisperse emulsions promotes
coalescence (Nie et al., 2008) and Ostwald ripening.
Microfabrication technology, which originated from semiconductor technology,
makes it possible to precisely fabricate micrometer-sized channels on a microfluidic
chip. Microfluidic devices enable the formulation of emulsions with well-controlled
droplet sizes and at high energy efficiency, which improves product properties and
shelf life, but also reduces the total energy expenditure. Not only single emulsions
but also multiple emulsions can be successfully produced, although their produc-
tion scale is still limited. Microfluidic emulsification of immiscible fluids is of inter-
est to several different application areas because of the high monodispersity of the
obtained emulsions. The microfluidic channels have very narrow size distribution
(typically lesser than 1%), which is a necessary condition for the direct production
of monodisperse emulsions. In contrast to the traditional emulsification devices (e.g.,
colloid mills and high-pressure homogenizers), which use intense force fields to
disrupt large, premixed droplets into smaller and relatively polydisperse droplets
(McClements, 2004), over the last two decades, novel techniques have been devel-
oped for the gentle and controlled formation of droplets from two immiscible liquids,
without preemulsification. These techniques involve the injection of a pure dispersed
phase through microchannels (MCs) or membrane pores into a continuous phase,
so that enabling the production of uniformly sized droplets of controllable average
size, due to the possibility of regulating the droplet size according to the size of the
pores or channels, rather than turbulent flow or cavitation areas, which often occur in
traditional emulsification devices (Vladisavljević, Kobayashi, and Nakajima, 2012).
Among those novel emulsification techniques, MC emulsification, which was ini-
tially proposed by Kawakatsu, Kikuchi, and Nakajima (1997), is a very mild emulsi-
fication process driven by interfacial tension. This process can produce monodisperse
emulsions using a microfluidic channel array microfabricated on a silicon chip. The
MC array devices can afford the integration of hundreds of thousands of channels on
a single chip (Kobayashi, Mukataka, and Nakajima, 2005a), and the diameter of the
generated droplets can range from 1 μm (Kobayashi, Uemura, and Nakajima, 2007)
to several millimeter (Kobayashi, Wada et al., 2008). MC arrays can be fabricated
parallel to the MC plate surface as open microgrooves (Kawakatsu, Kikuchi, and
Nakajima, 1997) or normal to the plate surface as straight-through holes (Kobayashi
et al., 2002). Hereinafter, we will describe major features and outline recent develop-
ments on MC emulsification technique.
9.2 PRInCIPLe oF DRoPLet GeneRAtIon

FoR MC eMULsIFICAtIon
9.2.1 Droplet-Generation MechanisM
Kawakatsu, Kikuchi, and Nakajima (1997) discovered a unique droplet-generation
mechanism based on spontaneous transformation of the dispersed phase that has passed
through channels with terraces outside them. Sugiura et al. (2001) proposed a droplet-
generation mechanism for MC emulsification that is driven by interfacial tension domi-
nant on a micron scale. The droplet-generation process of MC emulsification has been
analyzed by experimental approach using high-speed video microscopy (Sugiura et al.,
2001) and numerical approaches using computational fluid dynamics (CFD) and Lattice
Boltzmann methods (Kobayashi, Mukataka, and Nakajima, 2004a; Kobayashi et al.,
2011; van Dijke, Schroën, and Boom, 2008; van der Zwan, Schroën, and Boom, 2009).
Figure 9.2 schematically depicts the droplet-generation mechanism on MC emul-
sification (Kobayashi et al., 2011). The droplet-generation process commences when
the dispersed phase of a discoid shape restarts to expand in the slot (terrace for
grooved MC arrays) after generating the last droplet (Figure 9.2a). This expansion
process ends at the moment the tip of this expanding dispersed phase reaches the slot
outlet (terrace outlet for grooved MC arrays) (Figure 9.2b). This moment is also the
starting point of the detachment process. In the initial stage of the detachment pro-
cess, the dispersed phase slowly expands over the slot outlet, as the internal pressure
of the dispersed phase over the slot outlet is higher than that in the slot (Figures 9.2c
and 9.3b, I). This internal pressure is an indicator of the Laplace pressure differ-
ence between the two phases. The difference in their internal pressures gradually
Slot Oil–water
interface
Channel
Flow direction 20 μm
(dispersed phase)
(a) (b) (c)
Droplet
Neck
20 μm
(d) (e) (f)
FIGURe 9.2 Visualized CFD result for droplet generation from an asymmetric straight-
through MC (tdet is the detachment time): (a) tdet: −40.0 ms, (b) tdet: 0.0 ms, (c) tdet: 35.3 ms,
(d) tdet: 41.8 ms, (e) tdet: 48.7 ms, and (f) tdet: 51.9 ms. (Data from Kobayashi, I. et al., Chem.
Eng. Sci., 66, 5556–5565, 2011.)
becomes smaller, as the dispersed phase increases over the slot outlet, and then their
pressure balance becomes opposite (Figure 9.3b, II). Rapid flow of the dispersed
phase toward the domain over the slot outlet was triggered by the change in the
internal pressure balance, forming a neck in the slot (Figure 9.2d and e). The internal
pressure at this neck sharply increased soon after the pressure balance became oppo-
site, as shown in Figure 9.3b, III. In this stage, the dispersed-phase flux at the neck
exceeds than that present in the front of the neck (van Dijke, Shroën, Boom, 2008),
leading to the instantaneous pinch-off of the neck and the completion of the detach-
ment process (Figure 9.2e and f). Such droplet-generation process, which does not
require a cross-flowing continuous phase, is periodically repeated to produce mono-
disperse emulsions by MC emulsification. Droplet generation for MC emulsification
is also a highly energy-efficient process with a typical energy input of 103–104 J/m3
(Sugiura et al., 2001; Kobayashi, Takano et al., 2008).
ΔPd,slot,over
ΔPd,slot,in
ΔPd,MC
(a)
ΔPd,MC
12
ΔPd,slot,in
ΔPd,slot,over
ΔPd (kPa)
I II III
0
20 30 40 50 60
(b) tdet (ms)
FIGURe 9.3 (a) Three-dimensional snapshot of the oil–water interface in the detachment
process. Here, ΔPd,MC is the internal pressure of the dispersed phase at the channel outlet,
ΔPd,slot,in is the internal pressure of the dispersed phase in the slot, and ΔPd,slot,over is the inter-
nal pressure over the slot. (b) Variation of ΔPd,MC, ΔPd,slot,in, and ΔPd,slot,over as a function of
detachment time (tdet). (Data from Kobayashi, I. et al., Chem. Eng. Sci., 66, 5556–5565, 2011.)
9.2.2 effect of the flow of the DisperseD anD continuous phases

Appropriate control of the flow of the dispersed phase is very important to achieve
the production of monodisperse emulsions at higher droplet throughputs by MC
emulsification. A practically important advantage of MC emulsification includes the
droplet size insensitive to the flow rate of the dispersed phase (or the pressure applied
to the dispersed phase) below its critical value (Figure 9.4). This feature indicates
that one can perform a successful MC emulsification without very precise control
of the flow of the dispersed phase. The monodispersity of the resulting emulsions
remains almost constant in a certain range of the flow rate of the dispersed phase
(Sugiura, Nakajima et al., 2002; Kobayashi, Nakajima, and Mukataka, 2003). This
feature is different from droplet generation driven by shear stress and elongation
(Vladisavljević, Kobayashi, and Nakajima, 2012). When a successful MC emulsifica-
tion is carried out, the droplet production rate is basically proportional to the flow rate
of the dispersed phase, achieving a maximum value at a critical value (Figure 9.4).
Critical value
Droplet production rate

Droplet size
Flow rate of the dispersed phase

(Pressure applied to the dispersed phase)
FIGURe 9.4 Variations of the droplet size and the droplet production rate for MC emulsifi-
cation as a function of the flow rate of the dispersed phase (pressure applied to the dispersed
phase).
Above the critical value, the resulting droplet size is sensitively varied by the flow
rate of the dispersed phase, and the droplet production rate becomes lower due to the
increase in the droplet size (Figure 9.4).
The flow transition of the dispersed phase during MC emulsification can be well
explained by a dimensionless number called the capillary number that expresses the
balance between the viscous and interfacial forces (Sugiura, Nakajima et al., 2002).
In MC emulsification, capillary number of the dispersed phase that flows in the chan-
nel (Cad,MC) is used to analyze the flow state of the dispersed phase during droplet
generation. Previous studies (Sugiura, Nakajima et al., 2002; Kobayashi, Mukataka,
and Nakajima, 2005a; Kobayashi et al., 2011) reported the critical Cad,MC values of
lesser than 0.05, below which droplets of uniform size were stably generated from
the channels. The interfacial tension is considered to be the dominant force in the
range below these critical Cad,MC values. The critical Cad,MC depends hardly on the
channel diameter, but depends greatly on the liquid viscosity, surface properties, and
channel length, as well as the operating temperature (Sugiura, Nakajima, and Seki,
2002b; Butron Fujiu, Kobayashi, Uemura et al., 2011). Above the critical Cad,MC, the
influence of the viscous force cannot be neglected, stabilizing the neck. This results
in the delayed pinch-off of the neck and the generation of nonuniformly sized large
droplets (Kobayashi et al., 2011). The internal pressure balance in and around the
neck is assumed to be a key factor for determining the droplet generation.
The influence of the flow rate of the cross-flowing continuous phase on the drop-
let size in MC emulsification has been investigated by Kawakatsu et al. (1999) and
Kobayashi, Hori et al. (2010). Kawakatsu et al. (1999) used grooved MC array
plates, and Kobayashi, Hori et al. (2010) used the asymmetric straight-through MC
array plates. The maximum Reynolds numbers of the cross-flowing continuous phase
were 15 for the grooved MC arrays and 28 for the asymmetric straight-through MC
arrays, suggesting its laminar state. For grooved MC arrays, the resulting droplet size
is not influenced by the cross-flowing continuous phase in the applied range, except in
the case of generating droplets whose size is similar to the well depth. The droplets
as large as the well depth become smaller as the flow rate of the continuous phase
increases. For asymmetric straight-through MC arrays, the resulting droplet size that
ranged between 75 and 180 μm was much smaller than the continuous-phase domain
over the slot outlets (1000 μm height). The size of droplets smaller than 100 μm was inde-
pendent of the flow rate of the cross-flowing continuous phase (Qc) applied in this study.
On the contrary, the size of droplets larger than 100 μm became sensitive to Qc in its range
over a critical value. An analysis of the force balance during droplet generation suggests
that the decrease in the resultant droplet size above a critical Qc may be attributed to the
fact that the buoyant force becomes effective for promoting the pinch-off of the neck and
droplet detachment. The critical Qc values were sufficiently high to appropriately sweep
away the droplets from the outlet of an MC array; therefore, large droplets of uniform size
can be collected from the module by applying Qc below the critical value.
9.3 eFFeCt oF CHAnneL stRUCtURe AnD

DIMensIons on eMULsIFICAtIon
9.3.1 GrooveD Mc arrays
Microstructure around the grooved MC array is schematically depicted in Figure 9.5a.
Each MC array consists of parallel channels, terraces outside them, and deep wells.
This microstructure was originally developed for measuring blood flow rate and for
characterizing blood flow phenomena in capillary, as one of the earliest microfluidic
Dispersed phase Grooved MC array

15 mm
10 mm
Channel
Terrace
Droplet Well filled with the

(a) continuous phase (b) Inlet hole for DP
Grooved MC array
Inlet hole
for CP
8 mm
Outlet hole
(c) 22.5 mm for CP
FIGURe 9.5 Grooved MC array chips of the dead-end and cross-flow types. (a) Three-
dimensional drawing of droplet generation from part of a grooved MC array. (b) Schematic view
of a grooved MC array chip of the dead-end type. DP stands for the dispersed phase. (c) Schematic
view of a simple grooved MC array chip of the cross-flow type. CP stands for the continuous phase;
DP stands for dispersed phase.
devices in the world (Kikuchi et al., 1992). At least two grooved MC arrays are posi-
tioned on each MC emulsification chip. Grooved MC arrays are sealed by physically
attaching them to a flat glass plate in a module during MC emulsification. Currently
available grooved MC arrays can produce monodisperse emulsions with droplet sizes
of 1–550 μm (Kobayashi, Uemura, Nakajima, 2007; Kobayashi, Hori et al., 2012).
Modules equipped with the grooved MC array chips can be classified as both dead-
end and cross-flow types (Figure 9.5b and c). Grooved MC array chips of the dead-end
type were initially used for MC emulsification. Four grooved MC arrays were arranged
on all four sides of a chip (Kawakatsu, Kikuchi, and Nakajima, 1997). A major advan-
tage of these dead-end grooved MC array chips includes an easy understanding of the
droplet-generation behavior from the channels. However, it is usually not straightfor-
ward to collect the emulsions produced when using such chips. To make a continu-
ous operation of MC emulsification possible, Kawakatsu et al. (1999) developed the
grooved MC array chips of the cross-flow type. Cross-flow grooved MC array chips
enable the collection of the generated droplets driven by the cross-flow of a continu-
ous phase in a well between two parallel-grooved MC arrays. On a simplest cross-flow
chip, two grooved MC arrays were arranged at both longitudinal sides of the central
well used as the cross-flow channel. The purpose of cross-flow was to collect droplets
from the module and not to control the droplet size. Zhang et al. (2013) demonstrated
a successful long-term production of monodisperse oil-in-water (O/W) emulsions for
seven days using a cross-flow grooved MC array chip. Cross-flow grooved MC array
chips are more suitable for higher droplet production rates, as many parallel MC arrays
can be incorporated on a chip (Kobayashi, Hori et al., 2010). However, the cross-flow
type chips still have a low throughput of monodisperse emulsions.
Kobayashi, Hori et al. (2012) proposed emulsification using grooved nanochannel
(NC) array chips of the dead-end type for producing monodisperse submicron emul-
sions. Monodisperse submicron emulsions with droplets as small as 0.5 μm were
produced by NC emulsification.
9.3.2 straiGht-throuGh Mc arrays

Over the last decade, Kobayashi and coworkers (Kobayashi et al., 2002) have developed
straight-through MC array chips that allow a much better utilization of the chip surface
leading to significantly higher throughputs. They initially designed symmetric straight-
through MC arrays with numerous circular and oblong channels that are compactly
arranged vertical to a chip. As depicted in Figure 9.6a, the dispersed phase that flows
the upstream side space enters circular channels, and the continuous phase that flows in
downstream side space to collect the generated droplets. For MC emulsification chips
consisting of 10-μm channels, straight-through MC arrays can position channels at
least 100 folds greater than grooved MC arrays. Symmetric straight-through MC arrays
with deep rectangular channels exceeding a slot aspect ratio of about 3 are capable of
generating uniform droplets (Kobayashi, Mukataka, and Nakajima, 2004b). However,
a major drawback of rectangular channels is unstable droplet generation when using a
dispersed phase of very low viscosity (e.g., decane and water).
Kobayashi, Mukataka, and Nakajima (2005b) developed asymmetric straight-
through MC arrays consisting of numerous pairs of circular channels on the upstream
Droplet
24 mm
Continuous
phase 10 mm
A A′
A−A′ cross section

Slot
Dispersed phase Asymmetric
straight-through
MC array
(a) (b) Channel
FIGURe 9.6 (a) Three-dimensional drawing of a droplet generation from part of an asym-
metric straight-through MC array. (b) Schematic view of an asymmetric straight-through MC
array chip of standard size.
(bottom) side and microslots on the downstream (top) side (Figure 9.6). The use of
these arrays enabled the stable production of monodisperse O/W and water-in-oil
(W/O) emulsions using two phases with viscosities of lesser than 1 mPa s (Kobayashi,
Mukataka, and Nakajima, 2005b; Kobayashi et al., 2009). Asymmetric straight-
through MC arrays are currently available for producing monodisperse emulsions
with droplet sizes of 10–180 μm (Kobayashi, Hori et al., 2010). When an asymmet-
ric straight-through MC array chip (Figure 9.6b) was used, the monodisperse emul-
sion droplets of n-tetradecane were successfully produced at a maximum flux of the
dispersed phase of 2700 L/(m2 h) (Vladisavljević, Kobayashi, and Nakajima, 2011).
A maximum droplet throughput per unit area of an asymmetric straight-through MC
array is independent of the channel diameter, as determined by CFD simulations
(Kobayashi et al., 2011). Asymmetric straight-through MC arrays are one of the most
commonly used microstructures for current MC emulsification researches.
9.3.3 effect of channel DiMensions

In MC emulsification, the resulting droplet size is mainly affected by channel and terrace/
slot dimensions. Channel cross section is usually trapezoidal, appearing close to square for
the grooved MC arrays or circle for the straight-through MC arrays in order to enhance
pressure drop inside the channel. If all terrace/slot dimensions are proportional to channel
diameter, the resulting droplet diameter depends proportionally on the channel diameter
(Figure 9.7) (Sugiura, Nakajima, Kumazawa et al., 2002; Kobayashi et al., 2011).
For the grooved MC arrays, the droplet size is tunable to some extent by the chan-
nel height and terrace length (Sugiura, Nakajima, Kumazawa et al., 2002; Sugiura,
Nakajima, and Seki, 2002a). Figure 9.8a shows the variation of the droplet size as
a function of the terrace length under a certain channel diameter. When the ratio of
terrace length to channel diameter is low (e.g., lesser than 2), droplet generation is
10
dMC Critical Ud,MC
8 5 μm
10 μm
6 30 μm
ddrop (−)
50 μm
100 μm
4
(a)
0
dMC
8 150 μm
200 μm
6 300 μm
ddrop (−)
400 μm
4
(b)
0 2 4 6 8 10 12
Ud,MC (mm/s)
FIGURe 9.7 Variation of dimensionless droplet diameter (ddrop) obtained from the
CFD simulations as a function of dispersed-phase velocity (Ud,MC) that flows in a
channel; effect of channel size. Here dMC is the channel diameter. (a) dMC of 5 to 100 μm.
(b) dMC of 150 to 400 μm. − in the graphs stands for the lines fitted the plots. (Data
from Kobayashi, I. et al., Chem. Eng. Sci., 66, 5556–5565, 2011.)
Maximum droplet-generation rate
Droplet size
(a) Terrace length/slot depth (b) Channel length
FIGURe 9.8 Variations of the droplet size as functions of terrace length (slot depth) (a) and
channel length (b).
unstable, and the droplet size decreases with the increasing terrace length. The small-
est droplet size can be obtained by a ratio of terrace length to channel diameter of
about 2. When the terrace length exceeded the channel diameter by more than 2 folds,
the droplet size increased with the increasing terrace length. This increasing slope
became lower in a range of very high ratios of terrace length to channel diameter. It
should be stated that the terrace length is determined by the mask design and cannot be
varied during chip fabrication. In contrast, the channel height can be readily tuned by
etching time in the case of fabricating the silicon-grooved MC array chips. The droplet
size gradually increases with the increasing channel height, if the terrace length is suf-
ficiently larger than the channel size. There is a maximum channel height for a given
channel width due to wet anisotropic etching described in Section 9.4.1. Analytical
models for predicting the size of the droplets generated from the grooved MC arrays
have been proposed by Sugiura, Nakajima, and Seki (2002a, 2004) and van der Zwan,
Shroën, and Boom (2009). The size of the droplets generated using the asymmetric
straight-through MC arrays is somewhat tunable by the slot depth, which is normally
controlled by the etching period. The droplet size varies with the slot depth, as shown
in Figure 9.8a. The diameter of circular channels is defined by the mask design, sug-
gesting that it is difficult to change the channel diameter during chip fabrication.
The channel length, which does not affect the droplet size, is a major param-
eter affecting the maximum droplet-generation rate from each channel. Figure 9.8b
shows the variation of the maximum droplet-generation rate as a function of the
channel length under a certain channel diameter. For a given channel diameter, lon-
ger channels lead to the production of monodisperse emulsions at higher droplet-
generation rates (Sugiura, Nakajima, and Seki, 2002b). Appropriately, narrow and
long channels are preferable for MC emulsification, as they have a greater pressure
drop of the dispersed phase, which is given by Fanning’s equation:
 ρ U 2  l 
∆Pd,MC = 4 f  d d   MC  (9.1)
 2   dMC 
where:
ΔPd,MC is the pressure drop of the dispersed phase in a channel
f is Fanning’s friction factor
ρd is the density of the dispersed phase
Ud is the velocity of the dispersed phase
lMC is the channel length
dMC is the channel diameter
9.4 eFFeCt oF CHAnneL MAteRIALs AnD tHeIR

sURFACe PRoPeRtIes on eMULsIFICAtIon
9.4.1 silicon Mc arrays
MC emulsification chips have been normally made of single-crystal silicon and
silicon-on-insulator substrates (Figure 9.9a). Single-crystal silicon and silicon-
on-insulator are used as materials in both grooved MC arrays and asymmetric
(a) (b) (c)
FIGURe 9.9 Photographs of MC emulsification chips made of silicon (a), PMMA (b), and
stainless steel (c). Top and bottom photographs in each frame are grooved MC array chips and
(asymmetric) straight-through MC array chips, respectively.
tABLe 9.1
Materials and typical techniques Used for Fabrication of Microchannel
Array Chips
Inherent surface
Materials Fabrication Methods Affinity Reference
Surface- Anisotropic wet Hydrophilic Kawakatsu, Kikuchi, and
oxidized etching, chemical dry Nakajima (1997); Kobayashi,
silicon etching Mukataka, and Nakajima (2005b)
Poly(methyl Injection molding, Hydrophobic Liu et al. (2005a); Kobayashi,
methacrylate) synchrotron Hirose et al. (2008)
lithography
Stainless steel Mechanical dicing, end Hydrophilic Tong et al. (2000b); Kobayashi,
milling Hori et al. (2012)
straight-through MC arrays, respectively. Silicon-etching techniques enable precise fab-

rication of the channels with a size over 1 μm. Grooved submicron channel arrays can
also be fabricated using these techniques (Kobayashi, Uemura, and Nakajima, 2007).
Materials, fabrication, and surface properties of MC arrays are listed in Table 9.1.
Grooved MC arrays are usually obtained by the first photolithography and aniso-
tropic wet etching for fabricating channels and the second photolithography and
anisotropic wet etching for fabricating terraces and wells (Kikuchi et al., 1992).
A potassium hydroxide solution is used for anisotropic wet etching. Anisotropic
etching is greatly preferable for MC emulsification, as the fabricated channels have
smooth, sharp, and well-defined edges. Isotropic wet etching is undesirable due to the
undercutting of mask features. Asymmetric straight-through MC arrays are made by
the first photolithography and dry etching, called “deep reactive ion etching” (DRIE),
for fabricating slots, the second photolithography and dry etching for fabricating cir-
cular channels, and the third photolithography and DRIE for penetrating the channels
to the inlet side of the chip (Kobayashi, Mukataka, and Nakajima, 2005b). During the
DRIE process, deep channels can be fabricated by including passivation with polymer
that prevents lateral etching of the sidewalls. The fabrication procedures of these MC
arrays are reviewed in detail in our previous publication (Vladisavljević et al., 2013).
Although silicon is an inherently hydrophobic material, its surface can be easily
modified with hydrophilic and hydrophobic treatments (Kobayashi, Takano et al.,
2008). The surface of the silicon MC arrays becomes hydrophilic after plasma-
oxidation to form a thin silicon dioxide layer. Hydrophobic MC arrays can be obtained
by surface oxidization and subsequent treatment with a silane coupling reagent.
In MC emulsification, the surface property of MC arrays has to be appropriately con-
trolled to achieve a successful production of monodisperse emulsions. A key factor
for a successful MC emulsification is to prevent wetting of the dispersed phase to the
MC array surface; that is, hydrophilic and hydrophobic channels are commonly used
to produce O/W and W/O emulsions, respectively. The contact angle of the dispersed
phase to the channel wall is an indicator that is useful for understanding the interaction
between the oil–water interface stabilized by surfactant molecules and the MC array
surface. Butron Fujiu, Kobayashi, Neves et al., (2011) demonstrated that droplets of
uniform size are stably generated via MC arrays when the interface that advances in
the channel has a high contact angle of the dispersed phase and a symmetrical shape.
Silicon MC arrays have a negatively charged surface in the water after hydrophilic
treatment, which is due to the presence of silanol group (–SiOH). To avoid the attrac-
tive interface/channel interaction, anionic and selected ionic surfactants (or proteins)
have to be used for producing monodisperse O/W emulsions by MC emulsification
(Tong et al., 2000a; Kobayashi, Nakajima, and Mukataka, 2003; Yin et al., 2005).
Silicon MC arrays after hydrophobic treatment have an uncharged surface in organic
solutions. There is a nonstrong repulsive interaction between the interface covered by
hydrophobic surfactant molecules and hydrophobic channel surface, although mono-
disperse W/O emulsions can be produced by appropriately selecting emulsion compo-
sitions and operating temperature (Kobayashi et al., 2009; Butron Fujiu et al., 2012).
9.4.2 polyMer Mc arrays

MC emulsification chips made of polymer have been developed for producing mono-
disperse W/O emulsions (Figure 9.9b). Poly(methyl methacrylate) (PMMA) was used
as a polymeric material for both grooved MC arrays and symmetric straight-through
MC arrays. PMMA is an inherently hydrophobic, antiacid, and antialkali polymer,
but weak in strong organic solvents. Grooved MC arrays made of PMMA were fabri-
cated by single-step injection molding using a metallic mold insert formed by nickel
electroplating (Liu et al., 2005a). Symmetric straight-through MC arrays made of
PMMA were fabricated by synchrotron radiation lithography and subsequent pat-
tern development that etches deep vertical channels (Kobayashi, Hirose et al., 2008).
These fabrication techniques enable a precise fabrication of the channels with a size
of over 1 μm on a PMMA chip. MC emulsification chips made of PMMA have
been successfully used for producing monodisperse W/O emulsions with average
droplet diameters of 20–50 μm without any surface modification. Liu et al. (2005b)
produced monodisperse O/W emulsions using PMMA-grooved MC arrays that were
coated by a hydrophilic polymer. Kobayashi et al. (2013) also demonstrated that

monodisperse O/W emulsions stabilized by an anionic surfactant can be produced
using a PMMA-straight-through MC array with a negative surface charge in water.
9.4.3 stainless steel Mc arrays

Metallic MC emulsification chips have also been developed for producing monodis-
perse O/W emulsions (Figure 9.9c). A stainless steel was used as a metallic material
for both grooved MC arrays and asymmetric straight-through MC arrays. Stainless
steel is a hydrophilic metal with acid, alkali, organic solvent, and shock-proof proper-
ties. Grooved MC arrays made of stainless steel were fabricated using both mechani-
cal dicing with diamond saw (Tong et al., 2000b) and end mills (Kobayashi, Wada
et al., 2012). The grooved MC arrays fabricated by mechanical dicing consisted of
only parallel channels (i.e., no terraces), whereas those fabricated using end mills
had both parallel channels and terraces, which was suitable for MC emulsification.
Asymmetric straight-through MC arrays made of stainless steel were obtained by
pilling up symmetric straight-through MC array chips, each consisting of slots or
circular channels (Kobayashi, Wada et al., 2008). The channel size of stainless steel
MC arrays is usually larger than 10 μm owing to the fabrication precision of the pre-
ceding processes. Monodisperse O/W emulsions with average droplet diameters of
30–1400 μm have been successfully produced using stainless steel MC emulsifica-
tion chips after surface oxidation. Stainless steel is among the most common materi-
als for manufacturing and handling of food and pharmaceutical products, whereas
technological development is needed for fabricating stainless steel MC arrays avail-
able for the practical production of emulsion samples.
9.5 eFFeCt oF PRoCess FACtoRs on eMULsIFICAtIon

9.5.1 teMperature
A variety of operating and environmental conditions may be responsible for emulsion
properties, such as the chemical and physical stabilities. For instance, different physi-
cal mechanisms may be responsible for the instability of food emulsions; among them,
creaming, flocculation, and coalescence are the most common (McClements, 2000).
In this section, we discuss features of MC emulsification, as affected by the operat-
ing temperature. Temperature-controlled MC emulsification has been applied to pre-
pare monodisperse food-grade microdispersions. For example, Sugiura et al. (2000)
prepared monodisperse O/W emulsions by high-temperature MC emulsification, aim-
ing to obtain solid lipid microparticles of uniform size. Iwamoto et al. (2002) applied
high-temperature MC emulsification at 40°C to prepare monodisperse W/O emulsions
used as templates for uniform gel microparticles of natural polymers. Neves, Ribeiro,
Kobayashi et al. (2008) also prepared monodisperse O/W emulsions containing func-
tional lipids with high melting points by conducting MC emulsification at 40°C. Aside
from the above-mentioned studies, where a specific temperature higher than the melting
point of the two phases was used during MC emulsification, Butron Fujiu, Kobayashi,
Neves et al. (2011) investigated the effect of temperature on MC emulsification using
different food-grade emulsifiers: sodium oleate, pentaglycerol monolaurate, and poly-

oxyethylene sorbitan monolaurate (Tween 20). They focused on MC emulsification char-
acteristics such as droplet-generation behavior, droplet size, droplet size distribution, and
droplet productivity, as affected by a wide-range operation temperature, between 10°C
and 70°C. They also evaluated the physical and physicochemical properties of the
two-phase systems used for producing O/W emulsions. As depicted in Figure 9.10,
Sodium oleate (ηc) = 1.7 e−0.023Tc

Tween 20 (ηc) = 1.6 e−0.021Tc
PGM (ηc) = 1.4 e−0.017Tc
Soybean oil (ηd) = 9 × 106e−0.040Tc
Soybean oil
100 Sodium oleate
Tween 20
PGM
Viscosity, ηd,ηc (mPa s)
10
(a)
0.1
100
Viscosity ratio, ξ (−)
10
Sodium oleate
Tween 20
PGM
(b)
0.1
0 10 20 30 40 50 60 70 80
Temperature (°C)
FIGURe 9.10 (a) Viscosity of the dispersed phase (soybean oil) and various continuous
phases (sodium oleate, Tween 20 or pentaglycerol monolaurate; 1 wt% in Milli-Q water),
measured at various temperatures. (b) Variation of the viscosity ratio as a function of tem-
perature. ηd is the viscosity of the dispersed phase, ηc is the viscosity of the continuous phase,
and ξ is the viscosity ratio, defined as ηd/ηc. Lines in (a) and (b) are fitted based on the expo-
nential equations presented over the graphs.
the viscosity of the continuous phase (ηc) at each temperature was independent of the
type of emulsifier. ηc values ranged between 0.3 and 1.4 mPa s and decreased exponen-
tially as the operating temperature increased from 10°C to 70°C. The viscosity ratio
(ξ), defined as ηd/ηc, where ηd is the viscosity of the dispersed phase, also decreased
exponentially as the operating temperature was increased, within the range applied in
this study. Figure 9.11 depicts the oil–water interface stabilized by sodium oleate, and
by Tween 20 in a rectangular channel at temperature of 70°C. The contact angle of the
dispersed phase to the channel wall (θd) was measured using a new MC array method
proposed by Butron Fujiu, Kobayashi, Neves et al. (2011), where the contact angle is
obtained by averaging the two contact angles (θd1 and θd2) indicated in Figures 9.11a
and b. The contact angle of the dispersed phase on the channel wall exceeded 130° and
decreased gradually as the temperature increased. Monodisperse O/W emulsions were
produced independent of the temperature applied and the type of emulsifier used. The
resultant droplet diameters, which ranged between 30 and 36 μm, decreased gradually
as the temperature increased. They also reported that the critical flow velocity increased
with the increasing operating temperature. Butron Fujiu et al. (2012) also evaluated the
influence of temperature on the production characteristics of W/O emulsions by MC
emulsification, using a hydrophobic MC array chip that was controlled between 10°C
and 55°C. In that case, the continuous phase included decane oil containing 5 wt% tetra-
glycerin monolaurate condensed ricinoleic acid ester, a hydrophobic emulsifier, and the
dispersed phase consisted of an aqueous solution containing 1 wt% of sodium chloride
and 5 wt% of polyethylene glycol. They reported that, at each operating temperature,
the resultant droplet diameter was also almost constant below a critical flow velocity of
the dispersed phase. Moreover, the maximum droplet-generation rate from a channel
gradually increased with the increasing operating temperature due to the decrease in
viscosity of both phases.
θd1 ≈ θd2
θd1
θd2
20 μm
(a)
θd1 > θd2
θd1
θd2
20 μm
(b)
FIGURe 9.11 Optical micrographs of the oil–water interface in a rectangular channel at

temperature of 70°C, stabilized by sodium oleate (a), and by Tween 20 (b). θd represents the
contact angle of the dispersed phase to the channel wall.
9.5.2 viscosities of DisperseD anD continuous phases

Taking into account that spontaneous droplet formation devices enable the combination
of good monodispersity, relatively high throughputs (if parallelized), and robustness,
the influence of many of their geometric parameters has been explored. Nevertheless,
the effect of viscosity of both phases on the droplet generation is not understood in full
yet. During the pinch-off of a droplet, the dispersed phase flowing out of the channel
has to be replaced by an equal amount of continuous phase; otherwise, the drainage
of the dispersed phase out of the nozzle will be hindered and the droplet will not cut
off. On this concern, van Dijke et al. (2010) evaluated the effect of the viscosities of
the phases by performing emulsification experiments with several dispersed phases
into various polyethylene glycol (PEG)–water–sodium dodecyl sulfate (SDS) mix-
tures as the continuous phase. Moreover, those authors also performed fluid dynamics
calculations of a comparable MC system with a single channel aiming to validate
their experimental observations. A range of dispersed phases (hexadecane, soybean
oil, and silicon oil), and continuous phases containing different amounts (from 0 to
20 wt%) of polyethylene glycol in Milli-Q® water were used, so that varying broadly
the ratio between the viscosities of both phases, and O/W emulsions were prepared
using MC emulsification. They found that considering the ratio of the dispersed phase
viscosity over the continuous phase viscosity, at high viscosity ratio the droplet size
was constant; the inflow of the continuous phase is fast compared to the outflow of
the dispersed phase. On the other hand, at lower viscosity ratio, the droplet diameter
increases until a minimal ratio is reached, where droplet formation is no longer pos-
sible. In addition, the limiting viscosity ratio value seems to be a function of the MC
design, and this should be adapted to the viscosity of fluids of both the dispersed and
the continuous phases. This study also indicated that long channels could be beneficial
for droplet formation at low viscosity ratio. A longer channel means a higher hydrody-
namic flow resistance, which slows down the flow from channel to the terrace during
droplet formation, therewith keeping the inflow of dispersed phase through the MC
more constant, and through this also the entire droplet formation process.
Vladisavljević, Kobayashi, and Nakajima (2011) also investigated the effect of
dispersed phase viscosity on the maximum flux and maximum droplet-generation
frequency in asymmetric straight-through MC emulsification, by experimental obser-
vation and through CFD simulation. They used soybean oil, medium-chain fatty-acid
triglyceride oil or n-tetradecane as the dispersed phase, with a viscosity of 50, 20,
and 2.7 mPa s, respectively, at 25°C. This study indicated that the maximum droplet
throughput in MC emulsification is strongly affected by the viscosity of the dispersed
phase. They also found that, based on the dispersed phase flux, droplets were formed
under three different regimes: the stable-size regime, the expanding-size regime, and
the blow-up regime, whereas the viscosity of the dispersed phase had a significant
effect on the critical dispersed phase flux for stable droplet formation.
9.5.3 electrolyte concentration

Besides the emulsification process itself, one major challenge concerning emul-
sions is the formulation of easy-to-disperse emulsions that are suitable for
various applications. Optimization of processing conditions, choice of emulsi-

fier, and other ingredients are the most important variables to achieve the desir-
able droplet size as well as suitable stability for each application (Neves, Ribeiro,
Fujiu et al., 2008). In particular, sodium chloride (NaCl) is well known as a
major additive in food industry used to preserve products by avoiding micro-
bial growth, and as a flavor enhancer in many canned vegetables, smoked and
cured meats, pickles, and cheeses, among other products. Moreover, salts play
an important role in governing the structure and texture of emulsions contain-
ing lipids, such as sauces, or salad dressings, due to electrostatic interactions
between the salt molecules and lipids. In order to elucidate the effect of electro-
lytes on the formation characteristics of O/W emulsion droplets, Kobayashi et al.
(2014) investigated the interfacial tension between the oil and water phases, zeta
potential of emulsion droplets as well as the electrostatic interaction between
the oil droplets and MC chip surfaces. The dispersed phase consisted of refined
soybean oil, and the continuous phase contained 1.0 wt% of a surfactant [either
surfactant, polyoxyethylene (20) sorbitan monolaurate (Tween 20) or SDS], and
NaCl as a model electrolyte (from 0 to 1.0 mol/L). As depicted in Figure 9.12a
and b, monodisperse O/W emulsions with average droplet diameter of 26 μm
and coefficient of variation values below 5% were produced from the asym-
metric through-holes when the NaCl concentration was lower than a threshold
level, which is dependent on the type of differently charged surfactants. For
instance, in the case of SDS-stabilized droplets, the threshold NaCl concentra-
tion was 0.3 mol/L, whereas in the case of Tween 20, the threshold concentration
was 0.5 mol/L, as indicated in Figures 9.12a and b, respectively. At NaCl concen-
trations above 0.2 mol/L, droplet formation via the asymmetric through-holes
became somewhat less stable and slower, whereas the oil droplets formed imme-
diately moved away from the slot outlets (Figure 9.13a, iii). They also reported
that a further increase in the NaCl concentration resulted in the unstable forma-
tion of remarkably larger droplets driven by shear stress due to the cross-flowing
80 60 50 50
Coefficient of variation (%)
Coefficient of variation (%)
70 50 40 40
Droplet diameter (μm)
40 40
30 30
30 30
20 20
20 20
10 10
10 10
0 0 0 0
0 0.2 0.4 0.6 0.8 1.0 1.2 0 0.2 0.4 0.6 0.8 1.0 1.2
(a) NaCl concentration (mol/L) (b) NaCl concentration (mol/L)
FIGURe 9.12 Influence of NaCl concentration on the average droplet diameter and coef-
ficient of variation of O/W emulsions produced using MC emulsification (WMS1-3 chip),
stabilized by SDS (a) and Tween 20 (b).
(i) 0.1 mol/L (ii) 0.3 mol/L (iii) 0.5 mol/L
(a) 100 μm
(i) 0.3 mol/L (ii) 0.75 mol/L (iii) 1.0 mol/L
(b) 100 μm
FIGURe 9.13 Optical micrographs of oil droplets formation from the slot outlets at differ-
ent NaCl concentrations (between 0.1 and 1.0 mol/L). (a) SDS-stabilized droplets. (b) Tween
20-stabilized droplets.
continuous phase or push-off force due to the presence of adjacent dispersed

phase expanding from the slot outlet.
9.6 sCALInG-UP stRAteGIes

9.6.1 GrooveD Mc array chip
MC emulsification chips produce emulsion droplets simultaneously from at least more
than hundred channels, enabling higher total droplet production rates. Kobayashi,
Wada et al. (2010) and Kobayashi, Neves et al. (2012) have scaled up grooved MC
array chips and asymmetric straight-through MC array chips for mass-producing
monodisperse emulsion droplets.
Cross-flow grooved MC array chips are advantageous for continuously produc-
ing monodisperse emulsions with droplet sizes smaller than about 10 μm. However,
a simplest cross-flow grooved MC array chip (Figure 9.5c) has only one cross-flow
channel and two holes at both the ends for the introduction and withdrawal of the
continuous phase. A maximum droplet production rate of this chip is normally lower
than 0.1 mL/h (Kawakatsu et al., 1999). The maximum droplet production rate is the-
oretically proportional to the channel size for a given size of grooved MC array chips.
Through- Through-
holes holes
(inlet) (outlet)
Grooved
MC array
50 μm
(a) (b)
FIGURe 9.14 (a) Photograph of a large grooved MC array chip of the cross-flow type. (Data
from Kobayashi, I. et al., Microfluid. Nanofluidics, 8, 252–262, 2010.) (b) Optical micrograph
showing the generation of soybean oil droplets of uniform size via part of a grooved MC
array. The continuous phase was Milli-Q water containing 1.0 wt% SDS. The flow rate of the
dispersed phase was 1.5 mL/h.
A large silicon-grooved MC array chip (Figure 9.14a) consists of seven cross-flow

channels for flowing the continuous phase and 14 parallel MC arrays with a total of
11,900 channels (Kobayashi, Wada et al., 2010). The use of this large MC emulsi-
fication chip can mass-produce soybean oil droplets of uniform size with average
diameters of approximately 10 μm at a maximum flow rate of the dispersed phase
of 1.5 mL/h (Figure 9.14b). It should be noted that the maximum production rate of
monodisperse emulsion droplets is almost inversely proportional to the dispersed
phase viscosity.
The large MC emulsification chip (Figure 9.14a) is available for mass-producing
uniform droplets on a laboratory scale, although further scaling up of this chip is
needed to attain practical-scale production. For instance, 32 MC arrays and 5.8 × 104
channels can be integrated on a 100 × 100 mm chip while maintaining the dimen-
sions of the MC arrays except for their length. Uniform attachment of the large chip
onto a glass plate and precise control of the flow of each phase inside a module would
be key points of successful emulsification if the 100 × 100 mm MC emulsification
chip is available. Another possible approach for further scale-up is to pile up grooved
MC array chips; however, it is difficult to monitor droplet generation via MC arrays
in each piled-up chip. It would be preferable to parallelize grooved MC array chips
as a further scaling-up approach.
9.6.2 asyMMetric straiGht-throuGh Mc array chip

Asymmetric straight-through MC array chips are advantageous for continuously
producing monodisperse emulsions with droplet sizes of greater than 10 μm at
higher droplet production rates. Only one asymmetric straight-through MC array
is fabricated on an initially designed 24 × 24 mm chip (Kobayashi, Mukataka,
and Nakajima, 2005b). For instance, more than ten thousands of vertical chan-
nels with a 10 μm diameter can be positioned within the 10 × 10 mm asymmetric
straight-through MC array. This asymmetric straight-through MC array chip are
Asymmetric
straight-
40 mm
through
MC
200 μm
40 mm
(a) (b)
FIGURE 9.15 (a) Photograph of a large asymmetric straight-through MC array chip.

(b) Optical micrograph showing the generation of tetradecane oil droplets of uniform size
via part of an asymmetric straight-through MC array. The continuous phase was Milli-Q
water containing 2.0 wt% Tween 20. (Data from Kobayashi, I. et al., Chem. Eng. Technol.,
35, 1865–1871, 2012.)
capable of producing monodisperse O/W and W/O emulsions at a maximum flow

rate of the dispersed phase of higher than 0.1 L/h when using two phases of low
viscosity (Kobayashi et al., 2009; Vladisavljević, Kobayashi, and Nakajima, 2011).
Previous CFD simulation results suggest that a maximum droplet throughput per
unit area of an asymmetric straight-through MC array chip is independent of the
channel diameter (Kobayashi et al., 2011). Kobayashi, Wada et al. (2012) developed
a large silicon-MC emulsification chip (Figure 9.15a) consists of four asymmetric
straight-through MC arrays with a total area of 484 mm 2 (Kobayashi, Wada et al.,
2010). Figure 9.15 depicts a large asymmetric straight-through MC emulsification
system, aiming for mass-producing droplets of uniform size on a liter per hour
scale. Monodisperse O/W emulsion droplets of tetradecane were successfully mass-
produced at a maximum flow rate of the dispersed phase of 1.4 L/h (Figure 9.15b).
It should be noted that the maximum throughput of monodisperse emulsions was
11.2 L/h.
The preceding maximum droplet productivity corresponds to a droplet through-
put capacity of 12.2 tons/year. Actual droplet productivity of the large asym-
metric straight-through MC array chip may become somewhat lower owing to
the necessity of regularly cleaning the emulsification module. An annual drop-
let throughput capacity of greater than 10 tons/year is considered to permit a
minimum industrial-scale production of monodisperse emulsions. A straightfor-
ward approach to increase droplet productivity is parallelization of the modules,
equipping each with the large asymmetric straight-through MC array chip. For
the large MC emulsification system depicted in Figure 9.16, both liquid phases
are delivered from vessels, not from syringes, thereby indicating that only one
vessel enables delivering a liquid phase to multiple modules over a long period of
time. Further scaling up the large asymmetric MC array chip is also a promising
approach for increasing droplet productivity. The fabrication of six inch asym-
metric MC array wafers could lead to the production of droplets of uniform size at
5-L tank
(dispersed phase)
Emulsification
module
FIGURe 9.16 Photograph of a large MC emulsification system for mass-producing mono-

disperse emulsions at a liter/hour scale. A large asymmetric straight-through MC array chip
is installed in the module.
a flow rate of the dispersed phase over 10 L/h and monodisperse emulsions with
a throughput over 100 L/h.
9.7 ConCLUDInG ReMARKs

This chapter has outlined recent technologies for the direct production of mono-
disperse emulsions using MC emulsification technology, whereas micron-scale
or submicron devices with well-defined channels enable successful production of
monodisperse emulsions with an average droplet diameter ranging from a single
micrometer to several hundreds of micrometer by exploiting the flow characteristics
of the two liquid phases and the force balance in the channels.
The effect of channel materials, as well as their surface properties on MC emul-
sification has also been incorporated in this chapter, outlining features of MC arrays
microfabricated on materials such single-crystal silicon and silicon-on-insulator
substrates, aside from presenting the latest developments on MC arrays fabricated
on polymer and stainless steel, which is among the most common materials used for
handling of food and pharmaceutical products.
The effect of different process factors has been also discussed, focusing especially
on how the operation temperature affects MC emulsification characteristics such as
droplet-generation behavior, droplet size, droplet size distribution, and droplet pro-
ductivity, whereas the resultant droplet diameter was almost constant below a critical
flow velocity of the dispersed phase, at each operating temperature. Moreover, the
maximum droplet-generation rate from a channel gradually increased with increasing
operating temperature due to the decrease in the viscosity of both phases. Furthermore,
it has been shown that the addition of NaCl, normally used as preservative in food
industry, in concentrations below a threshold level (from 0.3 to 0.5 mol/L, depending
on the type of differently charged emulsifier used), resulted in the successful formu-
lation of monodisperse O/W emulsions by MC emulsification. Further increases in
NaCl concentration lead to the unstable formation of remarkably larger droplets.
ReFeRenCes
Butron Fujiu, K., Kobayashi, I., Neves, M.A., Uemura, K., Nakajima, M. 2011. Effect of
temperature on production of soybean oil-in-water emulsions by microchannel emulsi-
fication using different emulsifiers. Food Science and Technology Research, 17, 77–86.
Butron Fujiu, K., Kobayashi, I., Uemura, K., Nakajima, M. 2011. Temperature effect on
microchannel oil-in-water emulsification. Microfluidics and Nanofluidics, 10, 773–783.
Butron Fujiu, K., Kobayashi, I., Uemura, K., Nakajima, M. 2012. Influence of temperature
on production of water-in-oil emulsions by microchannel emulsification. Colloids and
Surfaces A: Physicochemical and Engineering Aspects, 10, 773–783.
Dickinson, E., McClements, D.J. 1996. Advances in Food Colloids, Chapman & Hall, London,
p. 333.
Iwamoto, S., Nakagawa, K., Sugiura, S., Nakajima, M. 2002. Preparation of gelatin micro-
beads with a narrow size distribution using microchannel emulsification. AAPS
Pharmaceutical Science and Technology, 3, 25.
Kawakatsu, T., Kikuchi, Y., Nakajima, M. 1997. Regular-sized cell creation in microchan-
nel emulsification by visual microprocessing method. Journal of the American Oil
Chemists’ Society, 74, 317–321.
Kawakatsu, T., Komori, H., Nakajima, M., Kikuchi, Y., Yomemoto, T. 1999. Production of
monodisperse oil-in-water emulsion using cross-flow type silicon microchannel plate.
Journal of Chemical Engineering of Japan, 32, 241–244.
Kikuchi, Y., Sato, K., Ohki, H., Kaneko, T. 1992. Optically accessible microchannels formed in
a single-crystal silicon substrate for studies of blood rheology. Microvascular Research,
44, 226–240.
Kobayashi, I., Hirose, S., Katoh, T., Zhang, Y., Uemura, K., Nakajima, M. 2008. High-aspect-
ratio through-hole array microfabricated in a PMMA plate for monodisperse emulsion
production. Microsystem Technologies, 14, 1349–1357.
Kobayashi, I., Hori, Y., Neves, M.A., Uemura, K., Nakajima, M. 2012. Controlled produc-
tion of monodisperse submicron emulsions by nanochannel emulsification. 12th
International Conference on Microreaction Technology, February 20–22, Lyon, France.
Kobayashi, I., Hori, Y., Uemura, K., Nakajima, M. 2010. Production characteristics of large
soybean oil droplets by microchannel emulsification using asymmetric through-holes.
Japan Journal of Food Engineering, 11, 34–48.
Kobayashi, I., Mukataka, S., Nakajima, M. 2004a. CFD simulation and analysis of emulsion
droplet formation from straight-through microchannels. Langmuir, 20, 9868–9877.
Kobayashi, I., Mukataka, S., Nakajima, M. 2004b. Effect of slot aspect ratio on droplet for-
mation from silicon straight-through microchannels. Journal of Colloid and Interface
Science, 279, 277–280.
Kobayashi, I., Mukataka, S., Nakajima, M. 2005a. Production of monodisperse oil-in-
water emulsions using a large silicon straight-through microchannel plate. Industrial
Engineering & Chemistry Research, 44, 5852–5856.
Kobayashi, I., Mukataka, S., Nakajima, M. 2005b. Novel asymmetric through-hole array
microfabricated on a silicon plate for formulating monodisperse emulsions. Langmuir,
21, 7629–7632.
Kobayashi, I., Murayama, Y., Kuroiwa, T., Uemura, K., Nakajima, M. 2009. Production of
monodisperse water-in-oil emulsions consisting of highly uniform droplets using asym-
metric straight-through microchannel arrays. Microfluidics and Nanofluidics, 7, 107–119.
Kobayashi, I., Nakajima, M., Chun, K., Kikuchi, Y., Fujita, H. 2002. Silicon array of elongated
through-holes for monodisperse emulsion droplets. AIChE Journal, 48, 1639–1644.
Kobayashi, I., Nakajima, M., Mukataka, S. 2003. Preparation characteristics of oil-in-water
emulsions using differently charged surfactants in straight-through microchannel emulsi-
fication. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 229, 33–41.
Kobayashi, I., Neves, M.A., Wada, Y., Uemura, K., Nakajima, M. 2012. Microchannel emul-
sification using stainless-steel chips: Oil droplet generation characteristics. Green
Processing & Synthesis, 1, 353–362.
Kobayashi, I., Takano, T., Maeda, R., Wada, Y., Uemura, K., Nakajima, M. 2008. Straight-
through microchannel devices for generating monodisperse emulsion droplets several
microns in size. Microfluidics and Nanofluidics, 4, 167–177.
Kobayashi, I., Uemura, K., Nakajima, M. 2007. Formulation of monodisperse emulsions using
submicron-channel arrays. Colloids and Surfaces A: Physicochemical and Engineering
Aspects, 296, 285–289.
Kobayashi, I., Vladisavljević, G.T., Uemura, K., Nakajima, M. 2011. CFD analysis of micro-
channel emulsification: Droplet generation process and size effect of asymmetric
straight flow-through microchannels. Chemical Engineering Science, 66, 5556–5565.
Kobayashi, I., Wada, Y., Hori, Y., Neves, M.A., Uemura, K., Nakajima, M. 2012. Microchannel
emulsification using stainless-steel chips: Oil droplet generation characteristics.
Chemical Engineering & Technology, 35, 1865–1871.
Kobayashi, I., Wada, Y., Uemura, K., Nakajima, M. 2008. Generation of uniform drops
via through-hole arrays micromachined in stainless-steel plates. Microfluidics and
Nanofluidics, 5, 677–687.
Kobayashi, I., Wada, Y., Uemura, K., Nakajima, M. 2010. Microchannel emulsification for
mass production of uniform fine droplets: Integration of microchannel arrays on a chip.
Microfluidics and Nanofluidics, 8, 252–262.
Kobayashi, I., Zhang, Y., Neves, M.A., Hori, Y., Uemura, K., Nakajima, M. 2014. Influence of elec-
trolyte concentration on microchannel oil-in-water emulsification using differently charged
surfactants. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 440, 79–86.
Kobayashi, I., Zhang, Y., Neves, M.A., Wada, Y., Uemura, K., Nakajima, M. 2013. Production
of monodisperse O/W and W/O emulsions using PMMA straight-through microchan-
nel arrays. 3rd Asia-Pacific Chemical and Biological Microfluidics Conference, August
19–22, Seoul, Korea.
Liu, H., Nakajima, M., Nishi, T., Kimura, T. 2005a. Effect of channel structure on prepara-
tion of a water-in-oil emulsion by polymer microchannels. European Journal of Lipid
Science and Technology, 107, 481–487.
Liu, H., Nakajima, M., Nishi, T., Kimura, T. 2005b. Hydrophilic modification of polymer
microchannel for preparation of oil-in-water emulsion. Nippon Shokuhin Kagaku
Kogaku Kaishi, 52, 599–604.
McClements, D.J. 2005. Food Emulsions: Principles, Practice, and Techniques. 2nd ed. CRC
Press, Boca Raton, FL, p. 301.
Neves, M.A., Ribeiro, H.S., Fujiu, K.B., Kobayashi, I., Nakajima, M. 2008. Formulation of con-
trolled size PUFA-loaded oil-in-water emulsions by microchannel emulsification using
β-carotene-rich palm oil. Industrial Engineering & Chemistry Research, 47, 6405–6411.
Neves, M.A., Ribeiro, H.S., Kobayashi, I., Nakajima, M. 2008. Encapsulation of lipophilic
bioactive molecules by microchannel emulsification. Food Biophysics, 3, 126–131.
Nie, Z., Seo, M., Xu, S., Lewis, P., Mok, M., Kumacheva, E., Whitesides, G., Garstecki, P.,
and Stone, H. 2008. Emulsification in a microfluidic flow-focusing device: Effect of the
viscosities of the liquids. Microfluidics and Nanofluidics, 5, 585–594.
Nisisako, T., Torii, T., Higuchi, T. 2004. Novel microreactors for functional polymer beads.
Chemical Engineering Journal, 2004, 101, 23–29.
Saito, M., Yin, L.-J., Kobayashi, I., Nakajima, M. 2005. Preparation characteristics of mono-
disperse oil-in-water emulsions with large particles stabilized by proteins in straight-
through microchannel emulsification. Food Hydrocolloids, 19, 745–751.
Saito, M., Yin, L.-J., Kobayashi, I., Nakajima, M. 2006. Comparison of stability of bovine
serum albumin-stabilized emulsions prepared by microchannel emulsification and
homogenization. Food Hydrocolloids, 20, 1020–1028.
Steegmans, M.L.J., Ruiter, J. De, Schroën, K.G.P.H., Boom, R.M. 2010. A descriptive force-
balance model for droplet formation at microfluidic Y-junctions. AIChE Journal, 56,
2641–2649.
Steegmans, M.L.J., Schroën, K.G.P.H., Boom, R.M. 2009. Characterization of emulsification
at flat microchannel Y junctions. Langmuir, 25, 3396–3401.
Sugiura, S., Nakajima, M., Iwamoto, S., Seki, M. 2001. Interfacial tension driven monodis-
persed droplet formation from microfabricated channel array. Langmuir, 17, 5562–5566.
Sugiura, S., Nakajima, M., Kumazawa, N., Iwamoto, S., Seki, M. 2002. Characterization of
spontaneous-transformation-based droplet formation during microchannel emulsifica-
tion. Journal of Physical Chemistry B, 106, 9405–9409.
Sugiura, S., Nakajima, M., Seki, M. 2002a. Effect of channel structure on microchannel emul-
sification. Langmuir, 18, 5708–5712.
Sugiura, S., Nakajima, M., Seki, M. 2002b. Prediction of droplet diameter for microchannel
emulsification. Langmuir, 18, 3854–3859.
Sugiura, S., Nakajima, M., Seki, M. 2004. Prediction of droplet diameter for microchannel
emulsification: Prediction model for complicated microchannel geometries. Industrial
Engineering & Chemistry Research, 43, 8233–8238.
Sugiura, S., Nakajima, M., Tong, J., Nabetani, H., Seki, M. 2000. Preparation of monodis-
persed solid lipid microspheres using microchannel emulsification technique. Journal
of Colloid and Interface Science, 227, 95–103.
Tong, J., Nakajima, M., Nabetani, H., Kikuchi, Y. 2000a. Surfactant effect on production
of monodispersed microspheres by microchannel emulsification method. Journal of
Surfactants and Detergents, 3, 285–293.
Tong, J., Nakajima, M., Nabetani, H., Kikuchi, Y. 2000b. Production of oil-in-water micro-
spheres using a stainless steel microchannel. Journal of Colloid and Interface Science,
237, 239–248.
van der Zwan, E., Schroën, K., Boom, R. 2009. A geometry model for the dynamics of micro-
channel emulsification. Langmuir, 25, 7320–7327.
van Dijke, K.C., Isao Kobayashi, Karin Schroën, K., Uemura, K., Nakajima, M., Boom, R.
2010. Effect of viscosities of dispersed and continuous phases in microchannel oil-in-
water emulsification. Microfluidics and Nanofluidics, 9, 77–85.
van Dijke, K.C., Schroën, K.C.P.G.H., Boom, R.M. 2008. Microchannel emulsification:
From computational fluid dynamics to predictive analytical model. Langmuir, 24,
10107–10115.
Vladisavljević, G.T., Khalid, N., Neves, M.A., Kuroiwa, T., Nakajima, M., Uemura, K.,
Sato, S., Ichikawa, S., Kobayashi, I. 2013. Industrial lab-on-a-chip: Design, applica-
tions, and scale-up for drug discovery and delivery. Advanced Drug Delivery Reviews,
65, 1626–1663.
Vladisavljević, G.T., Kobayashi, I., Nakajima, M. 2011. Effect of dispersed phase viscosity on
maximum droplet generation frequency in microchannel emulsification using asymmet-
ric straight-through channels. Microfluidics and Nanofluidics, 10, 1199–1209.
Vladisavljević, G.T., Kobayashi, I., Nakajima, M. 2012. Production of uniform droplets using
membrane, microchannel and microfluidic emulsification devices. Microfluidics and
Nanofluidics, 13, 151–178.
Vladisavljević, G.T., Williams, R.A.T. 2005. Recent developments in manufacturing emul-
sions and particulate products using membranes. Advances in Colloid and Interface
Science, 113, 1–20.
Walstra, P. 1996. Emulsion stability. In: Becher, P. (Ed.) Encyclopedia of Emulsion Technology,
Vol. 4, Marcel Dekker, New York, pp. 1–62.
Zhang, Y., Kobayashi, I., Neves, M.A., Uemura, K., Nakajima, M. 2013. Long-term continu-
ous production of soybean oil-in-water emulsions by microchannel emulsification. Food
Science and Technology Research, 19, 995–1001.
10 Emulsification with
Microsieves and Other
Well-Defined
Contents
10.1 Introduction ................................................................................................. 236
10.2 Microsieves ................................................................................................. 238
10.2.1 Microsieve Structure and Pore Activation .................................... 238
10.2.2 Process Parameters and Droplet Formation Mechanism .............. 241
10.2.3 Toward Practical Application? ...................................................... 242
10.2.4 Metal Sieves .................................................................................. 243
10.2.5 Metal Sieves in Combination with Packed Beds ..........................246
10.3 Comparison of Emulsification with Microstructured Systems ...................248
10.4 Concluding Remarks................................................................................... 250
References .............................................................................................................. 250
ABSTRACT Membrane emulsification has been around for a couple of decennia,

and various aspects have been covered as described in Chapters 2 through 4. One of
the main issues is control of the droplet size, and quite often polydispersity is linked
to the polydispersity of the pores of the membrane. Therefore, the use of membranes
with equally sized pores is seen as a way to better control the eventual size of the
produced emulsion.
In this chapter, microsieves, and other devices with monodispersed pores, are
described, and their performance in cross-flow emulsification is compared to regular
membranes and more classic emulsification devices based on, among others, energy
density. Besides this, the structure of the microsieves and how this influences droplet
formation will be elaborated on, leading to suggestions for improved design. Further,
premix emulsification with metal sieves with uniform pores, and the combination of
metal sieves with glass bead beds, will also be described and compared to cross-flow
membrane emulsification and other emulsification techniques in the final part of this
chapter based on, among others, droplet size distribution.
235
10.1 IntRoDUCtIon
Since Nakashima and Shimizu filed their first Japanese patent in 1986, various
designs for and products made by membrane emulsification have been reported.
Mostly, cross-flow emulsification and premix emulsification have been used, and
various good reviews are available in literature. The interested reader is referred to
Joscelyne and Trägårdh (2000; general review); Charcosset, Limayem, and Fessi
(2004; general review); Vladisavljević and Williams (2005; general overview with
many products); van der Graaf, Schroën, and Boom (2005; double emulsions);
Charcosset (2009; specific for food); Nazir, Schroën, and Boom (2010; specific for
premix emulsification); Maan, Schroën, and Boom (2011; for spontaneous droplet
formation systems); and Vladisavljević, Kobayashi, and Nakajima (2012; for an
extensive comparison also with microfluidic devices).
Membrane emulsification has been used in various applications, and, especially,
the lower energy intensity needed for production of emulsions has been celebrated as
one of the benefits of this technology, as illustrated in Figure 10.1, which is adapted
after a figure from the PhD thesis of Schröder from the Technical University of
Karlsruhe in Germany, into which data from the research group of Professor
Schubert—Karbstein and Schubert (1995), Schröder (1999), Behrend and Schubert
(2001), and Lambrich and Schubert (2005)—have been added by the last author.
The energy required in membrane emulsification using a vegetable oil at 30 vol%
with a typical viscosity of 60 mPa s can be orders of magnitude lower than that of
more classic emulsification devices such as high pressure homogenizers and colloid
mills. What also distinguishes membrane emulsification from the other technologies
is that the energy density is a function of the volume fraction of oil that is required
100
Toothed colloid mill
Cross-flow membrane Ultrasound
30 vol% oil
emulsification (ceramic High-pressure homogenizers:
Mean droplet size, x (μm)
membrane) Standard valve

10
Sharpened edge valve
Microfluidizer®
Orifice valve
1 vol% 10 vol% 50 vol%

0.1 3
10 104 105 106 107 108
3
Energy density, Ev (J/m )
FIGURe 10.1 Energy density for various emulsification devices. The emulsion comprises
of 30% vegetable oil with a viscosity of 60 mPa s. (Based on Schröder, V., Herstellen van
Öl-in-Wasser Emulsionen mit Microporösen Membranen. PhD thesis, Technische Hochschule
Karlsruhe, Germany, 1999 and extended by Lambrich, U. and Schubert, H., J. Membr. Sci., 257,
76–84, 2005.)
Emulsification with Microsieves 237
in the emulsion. In classic emulsification devices, the entire emulsion is pressurized,

requiring the same amount of energy irrespective of volume fraction. The applied
energy is to a large extent lost as heat, while in membrane emulsification the two constitu-
ent phases are pressurized individually allowing for effective use of the applied energy.
In spite of this, membrane emulsification has not been adopted widely, and that is
caused by the throughput of the membranes that is too low to be cost-effective. A com-
parison between the required area for two of the most reported membranes, Shirazu
porous glass (SPG) and ceramics, is shown in Figure 10.2 for the production of an
emulsion that contains 30% oil at a rate of 20 m3/h (Gijsbertsen-Abrahamse, 2003). It is
clear that the required areas are very large for ceramic membranes (>1000 m2), and
while SPG membranes are more efficient and reduce the required area with roughly
a factor of 10, the areas are still far beyond what is cost effective for a bulk product.
Besides this, droplet size as well is still an issue. With cross-flow membrane emul-
sification, typically droplets can be produced that are 2–20 times the pore size (van der
Graaf et al., 2004). For stability issues, the produced droplets should be below 1 µm,
and this implies that the pores need to be accordingly smaller, leading to even larger
challenges regarding throughput of the membranes, that is, larger required surface areas.
Further, the droplet size distribution of emulsions made with membranes (that have a
pore size distribution) is still up for improvement, and for this it was suggested that
membranes with uniform pores could be the key, as described in the following section.
10,000
1,000
Required area (m2)
100
Ceramic membrane; 2% SDS [1]

10
Idem; 0.1% whey protein [1]
SPG membrane [2]
Idem [3]
Microsieve [4]
1
0 2 4 6
Transmembrane pressure (bar)
FIGURe 10.2 Required membrane surface area for the production of 30% oil emulsion at
20 m3/h. (Data were collected by Gijsbertsen-Abrahamse, A.J. et al., J. Membr. Sci., 230, 149–159,
2004, from: [1] Schröder, V. et al., J. Colloid Interface Sci., 202, 334–340, 1998; [2] Nakashima,
T. et al., Key Eng. Mater., 61–62, 513–516, 1991; [3] Fuchigami, T. et al., J. Sol-Gel Sci. Technol.,
19, 337–341, 2000; [5] Abrahamse, A.J. et al., J. Membr. Sci., 204, 125–137, 2002.)
10.2 MICROSIEVES
Through photolithographic techniques used in the microchip industry, it is possible
to make pores of uniform size in silicon wafers. For an extensive description of the
technology and various applications, please consult the book by van Rijn (2004) or
the papers written by van Rijn and his coworkers (van Rijn and Elwenspoek, 1995;
van Rijn et al., 1997). The geometry, spacing, and size of the pores can be designed at
will (see Figure 10.3). These sieves are known for their very low resistance because
the actual layer in which the pores are placed is extremely thin (1 µm, as is also vis-
ible in the images in Figure 10.3), which results in high fluxes. Both aspects were
expected to be beneficial for emulsification, since the required surface area could be
much smaller as for other membranes; therefore, these sieves have been tested exten-
sively in our lab, within the PhD project of Anneke Gijsbertsen-Abrahamse (2003).
10.2.1 Microsieve structure and Pore activation

The sieves were placed in a holder with a viewing area to allow microscopic obser-
vation of the droplet formation process. The oil phase was pressurized through the
sieve, and the droplet size was observed, as shown in the snapshots in Figure 10.4a.
What was immediately clear was that the droplets grew to sizes that were much
larger as the pores (the small black dots in the background of the images in
Figure 10.4a), that the droplets influenced neighboring droplets inducing polydis-
persity, and that not all pores were actually generating droplets. Besides, it was
found that the number of active pores was a function of the applied pressure; at
increasing pressure more pores become active (Abrahamse et al., 2002), but still
the percentage of active pores did not exceed 16% for the conditions and micro-
sieve design that were tested here, as illustrated in Figure 10.4b. This in itself does
not make microsieves unique; even lower numbers for active pores were reported
by Vladisavljević et al. (2007), who polished an SPG membrane to enable visual-
ization but concluded that although the membrane had a porosity of approximately
55%, the actual percentage of droplet forming pores was only 1% irrespective of
the applied transmembrane pressure.
The behavior of the microsieves was analyzed further and compared with that
of other membranes by Gijsbertsen-Abrahamse, van der Padt, and Boom (2004).
(a) (b) (c)
FIGURE 10.3 (a–c) Various designs of microsieves. (Pictures courtesy of Aquamarijn

Microfiltration.)
Cross-flow
t=0 t = 0.03 s t = 0.22 s

(a)
16 Cross-flow
velocity (m/s)
Fraction active pores (%)
12 0.011
0.017
0.028
8
0.039
0
3 6 9 12 15
(b) Transmembrane pressure (kPa)
FIGURe 10.4 (a) Microscopic observation of droplet formation on microsieves (top); three
snapshots taken in fast succession showing droplets growing to sizes much larger than the
underlying pores that push neighboring droplets of the pore inducing polydispersity, and
that are not formed from all pores. (b) Active pores as function of applied pressure on a
hydrophilic microsieve with round pores of 7 μm and thickness 5 μm (bottom). (Data from
Abrahamse, A.J. et al., J. Membr. Sci., 204, 125–137, 2002.)
The layout of the microsieve is very typical and shown in Figure 10.5, with a sub-
structure onto which the actual sieve is placed.
The resistances of the substructure and sieve are indicated by Rs and Rp(pore),
and the various pressures in the system are the pressure in the continuous phase
(p0), in the substructure (ps), and the applied pressure at the feed side of the to-
be-dispersed phase (p1). Besides, there is an activation pressure that needs to be
exceeded in order to make droplets, and this pressure is related to the Laplace
pressure of the pore. The flow of the to-be-dispersed phase through the sieve is
determined by the pressures in the system and the resistance of both substructure
and sieve, which in turn is determined by the number of active pores (Nact), as
shown in Equation 10.1:
Rp  p1 − p0 
N act =  − 1 (10.1)
Rs  ps − p0 
p0
ps − pact
Φp = Nact Rp
η Rp
ps
p1 − ps Rs
Φs =
η Rs
p1
FIGURe 10.5 Schematic representation of the microsieve structure with its typical sub-
structure onto which the actual sieve is placed (right), and an overview of the various pressure
gradients that are relevant in the microsieve (left). (After Gijsbertsen-Abrahamse, A.J. et al.,
J. Membr. Sci., 230: 149–159, 2004.)
The resistance of the substructure can be determined straightforwardly from the

dimensions, but for the pores, matters are more complex. This resistance is deter-
mined by both the resistance against flow through the pores 128 L / πd p4 and the ( )
( )
entrance effects that occur 24 / d p3 with these very short “channels” and is a func-
tion of the number of active pores (1 – Σ term), as also shown in Equation 10.2.
 128Lp 24   ∞

Rp = 
 πdp
4
+ 3  1 −
dp   ∑a ε
i =1
i
i +(1/ 2 )



(10.2)
The line in Figure 10.4b is a result of this model, leading to a typical activation
pressure of just over 3 bar and an increasing number of active pores at increasing
pressure. The resistance of the substructure and the pores are in the same order
of magnitude, and this also implies that if a pore starts forming droplets, locally
the pressure under that pore will be lower leading to preferential flow toward that
pore, which also does not allow neighboring pores to become active unless the
applied pressure is increased dramatically. Similar effects also play a role in regu-
lar membrane emulsification, where active pores will keep neighboring pores from
becoming active due to these pressure differences. If the pore resistance were to
be increased, the applied pressure as a whole needs to be higher to activate pores
because of increased flow resistance, but pore activation would be much more
gradual. This was also tested in the work of van der Graaf et al. (2004), and they
found that if the sieves were much thicker, all pores became active producing simi-
lar sized droplets.
Obviously, the flux of the microsieves is reduced if the longer pores are used
because of the additional resistance. As a matter of fact, the sieves used by van der
Graaf et al. (2004) are on the border of the limiting aspect ratio (width vs. length
of the pore) that technically can be made. The overall process for these thick sieves
is stable at 100% pore activation. Based on these data, and allowing for sufficient
spacing between the pores (typical porosity is 1%), the required area of the micro-
sieve was calculated for the same system that is described in Figure 10.2, and it is
clear that the required surface area is much lower, that is, in the order of 1–10 m2
(Gijsbertsen-Abrahamse, van der Padt, and Boom, 2004). Whether this is the deci-
sive step toward practical application is described in the respective section.
10.2.2 Process ParaMeters and droPlet ForMation MechanisM

For the design of actual emulsification processes, the droplet formation mechanism
needs to be understood. The first ones to derive equations for regular cross-flow
membrane emulsification were Peng and Williams (1998), who started from the
interfacial tension force, which keeps the droplet connected to the membrane pore,
and the shear force that tries to remove the droplet. The interfacial tension and shear
forces were summarized in a capillary number defined as:
ηcν c
Ca c = (10.3)
γ ow
In this capillary number defined for the continuous phase, ηc is the dynamic viscos-
ity of the continuous phase, vc is the characteristic average velocity of the continuous
phase, and γow is the interfacial tension between oil and water phase. At high Ca,
either the high shear (which may be a result of high continuous phase velocity or
high viscosity or both) and/or low interfacial tension leads to small droplets. Various
others have followed this approach, and van Rijn (2004) used it for microsieves.
Rayner and Trägårdh (2002) also used the capillary number to predict the droplet
size, and found reasonable agreement, but at the same time mentioned that there
was room for improvement in order to be able to capture the operating parameters
better. Also, Abrahamse et al. (2002) started from the force balance when analyz-
ing droplet formation with microsieves and identified various factors that influenced
the droplet size beyond what was expected from the basic force balance. This was
confirmed by Lepercq-Bost et al. (2008), who used a ceramic membrane, and found
similar effects.
Real breakthroughs in understanding of the droplet formation mechanism resulted
from extensive modeling studies and investigations with microfluidic devices. As
early as 2001, Abrahamse et al. reported computational fluid dynamics results that
showed that interfacial tension and surface properties, as reflected in the contact
angle, are very relevant for the droplet size. Rayner et al. (2004) used the surface
evolver under conditions for which the force balance does not hold, and they could
predict droplet size with an average error of 8%. When combining modeling with
observation with microfluidic T-junctions, which are a model for cross-flow emulsi-
fication, van der Graaf et al. (2004) and van der Graaf, Steegmans et al. (2005) came
to the conclusion that the flow of both the continuous and to-be-dispersed phases
determines the size of the droplets that are formed. In Figure 10.6, a comparison
is shown of the microscopic observation and the result obtained through Lattice
Boltzmann simulation (van der Graaf et al., 2004; van der Graaf, Steegmans et al.,
2005; van der Graaf et al., 2006).
van der Graaf and colleagues (2005) found that droplet formation takes place in
two parts. First, the droplet needs to obtain a certain size inside the continuous phase
channel, after which the still attached droplet starts detaching. However, during this
FIGURe 10.6 Comparison of Lattice Boltzmann simulations and microfluidic observations with
T-shaped junctions that are used as a model for cross-flow emulsification. In the bottom images,
the to-be-dispersed phase is pressurized through the vertical channel, and in the middle images this
phase is pushed toward the exit, while in the top images the droplet is snapped off from the feed.
(Data from van der Graaf, S. et al., Langmuir, 22, 4144–4152, 2006.)
phase, the droplet will still receive additional liquid from the feed channel, which
makes it grow larger. Because of this, the droplet size is not only a function of the
applied shear rate but also of the applied oil flow rate (transmembrane pressure),
which makes droplet size much less simple to predict than originally assumed, and
also makes process design more complex, since the flow of both phases needs to be
controlled accurately, unless the detachment time does not significantly contribute to
the total volume of the droplet.
To make matters even more complex, under the process conditions at which drop-
let formation takes place, dynamic interfacial tension effects are also expected to
take place, which implies that the actual value of the interfacial tension can be any-
thing between the equilibrium value that represents a surfactant saturated surface
and that of an empty interface. This was confirmed as early as 1998 by Schröder,
Behrend, and Schubert through the bursting membrane technique. van der Graaf
et al. (2004) have used droplet volume tensiometry to extrapolate interfacial tension
values to the conditions applied for droplet formation on a microsieve, and they also
showed that dynamic interfacial tension effects play a role there. Measuring dynamic
interfacial tensions under the conditions used in emulsification is far from trivial, but
also here there seems to be a breakthrough thanks to microfluidic investigations with
Y-shaped junctions as described by Steegmans et al. (2009).
10.2.3 toward Practical aPPlication?

In spite of the positive results for microsieves that allow for considerable reduction
of required areas as shown in Figure 10.2, there are still some factors to consider,
such as the chemical integrity of the sieves. The top layer of the microsieve is made
of silicon nitride, which is in principle inert, but various components, especially food
ingredients, are known to be able to adsorb onto this surface and influence its wet-
tability. If that occurs, droplet formation can be influenced to a very large extent or
even cease to occur as illustrated in the simulation work of Abrahamse et al. (2001).
Therefore, surface modification of the sieves to prevent adhesion of especially pro-
teins has been investigated extensively in the works of Arafat et al. (2004, 2007) and
Rosso et al. (2009) in which they show that adsorption of bovine serum albumin
and fibrinogen could be prevented by mildly attaching ethylene oxide chains to the
surface.
The price of the low porosity, high resistance microsieves that are required for
stable operation during emulsification will be higher as for regular microsieves and
much higher as for other membranes. But given the reduction in required surface
area, they still may be interesting, although in our opinion it is more likely that
application will not take place for a bulk product but for a niche market. Further
reduction in price may take place simply through the scale of production, but alter-
natively also production of polymeric microsieves, as proposed by Vogelaar et al.
(2005), is an interesting development. They used a template that is made by the
same techniques that are used to make the SiN microsieves, but now they make a
negative image from which repeatedly polymeric casts can be made. This consid-
erably reduces the costs of the microsieves; if, for example, 100 casts can be made
from one template, the cost would roughly reduce to about 1% of that of the SiN
microsieve.
Besides sieve design, different process designs have also been published,
mostly focusing on application of shear through more than just the flowing liquid.
For example, Stillwell and coworkers (2007) investigated a stirred cell with a
membrane at the bottom resulting in emulsions that were rather polydisperse due
to the differences in shear rates across the membrane. Rotating systems using
(metal) membranes with uniform pores were presented by Verena (Eisner)-
Schadler (2006), Verena (Eisner)-Schadler (2007), Aryantia et al. (2006), and
Yuan and coworkers (2008). This resulted in better control over droplet size; how-
ever, the droplets were rather large due to the fact that pores in metal sieves can-
not be made as small as in other microstructured systems.
10.2.4 Metal sieves

Although the metal sieves mentioned in the previous section could not produce small
droplets in a cross-flow emulsification setting, they have been tested extensively in
premix emulsification by Nazir and coworkers (2011, 2013a). Examples of tested metal
sieves with uniform pore size are shown in Figure 10.7.
During premix emulsification, a coarse emulsion is pushed through the sieve to
further refine the size of the droplets. It was found that in this case the droplets can
become considerably smaller than the smallest dimension of the pore as will be
discussed later, and this also distinguishes this technique from cross-flow emulsifi-
cation in which the droplet size is considerably larger as the pore size. In the experi-
ments carried out by Nazir and colleagues (2011, 2013a), various process conditions
were tested, mostly for 5% hexadecane in water emulsions stabilized by 0.5 vol%
Tween 20. The size of the premix emulsion was around 27 µm, and the emulsion was
pushed repeatedly through the sieves, with either rectangular or squared pores; an
illustrative result is shown in Figure 10.8.
Sieve specification Front view Back view
Pore size: 4 μm × 4 μm
Thickness: 60 μm
Porosity: 2.65%
Supporting mesh on back side
Pore size: 7.1 μm × 413.2 μm

Thickness: 200 μm
Porosity: 1.53%
Supporting structure on back side
FIGURe 10.7 Examples of metal sieves (Courtesy of Stork Veco B.V., Eerbeek, the
Netherlands) with squared and rectangular pores that were used in premix emulsification by
Nazir et al. (Data from Nazir, A. et al., J. Membr. Sci., 383, 1–2, 116–123, 2011; Nazir, A.
et al., Chem. Eng. Sci., 93, 173–180, 2013a.)
35
30
Droplet diameter, d32 (μm)
25
20
15
10
0
0 1 2 3 4 5 6
No. of passes, N (−)
FIGURe 10.8 Droplet size as function of number of passes through a metal sieve with rect-
angular pores with 11.6 µm width operated at (⚪) 50, (⬦) 100, and (Δ) 200 kPa pressure. The
sieves were operated from the front (empty mark) and back (filled mark); for more informa-
tion see Figure 10.7. (Data from Nazir, A. et al., J. Membr. Sci., 383, 1–2, 116–123.)
300
250
Droplet Weber number, Wed (−)
200
150
100
50
0
0 100 200 300 400 500
Droplet Reynolds number, Red (−)
FIGURe 10.9 Droplet Weber number versus droplet Reynolds number: (◻) 4 μm square
pores, ( ◼) 7.1, (○) 10.6, (⚫) 11.6, (Δ) 12.8, and (▴) 13.2 μm sieve; the last five sieves all
have rectangular pores and only the width of the pore is given here; the length is approximately
300 μm for all.
As expected, the droplet size decreased with increasing number of passes and also
the applied pressure as illustrated in Figure 10.8 for a sieve with rectangular pores.
What was further noted was that the droplet size is independent of the way the pre-
mix is pushed through the sieve, and this indicates that the structure of the sieve is
not that relevant for droplet breakup. The actual fluxes at which the emulsions were
produced are very high (can be well beyond 1000 m3/m2 h depending on the applied
pressure), and that is orders of magnitude higher as reported for cross-flow emul-
sification, for example, by Vladisavljević, Surh, and McClements (2006). Besides,
the flux stays constant during operation, indicating that the system is not sensitive
to fouling. On the downside, the span values of the emulsions are on the high side,
typically around 1.2–1.4 (Nazir, Schroën, and Boom, 2011, 2013a).
For squared pores, the droplet size reduction was much less in spite of their much
smaller size, and this is due to a difference in droplet breakup mechanism, as illus-
trated in a dimensionless plot (Figure 10.9) in which the droplet Weber number (ratio
of shear and interfacial tension) is plotted versus the droplet Reynolds number for
data for the first pass.
The following definitions were used by Nazir, Schroën, and Boom (2013a). The
hydraulic Reynolds number inside the pore, Rep, is
ρvdh
Re p = (10.4)
ηc
with ηc being the continuous phase viscosity and dh, the pore hydraulic diameter,
being 2lw/(l + w), with l the length of the pore, and w its width. The droplet Reynolds
number, Red, was defined by Nazir (2013) using the pore Reynolds number suggested
by Van Dinther et al. (2012):
2
d 
Re d = Re p  32,in  (10.5)
 2 dh 
where:
d32,in is the ingoing Sauter mean droplet diameter
The droplet Weber number, Wed, which is a ratio between the inertial forces (as a
result of local pressure fluctuations) and interfacial tension forces on a droplet, was
defined in Nazir, Schroën, and Boom (2011) as
d32,outρv 2
We d = (10.6)
2σ
in which d32,out is the Sauter mean diameter of the droplet that is produced and σ is
the droplet interfacial tension, for which the equilibrium value of 5.8 mN/m is used.
The sieves with rectangular pores have about five times higher droplet Weber
number, and all behave similarly but different from square pores. The inertial forces
seem to be more important for sieves having rectangular pores and are facilitated
at higher droplet Reynolds number. The flow through the rectangular pores was not
well developed, and partial turbulent conditions may have existed after the pores
therewith contributing to droplet breakup (Nazir, Schroën, and Boom, 2011). For
the square pores, at low values for Reynolds number, spontaneous droplet breakup
due to Laplace pressure differences may be more important, indicating that under
these conditions there is an effect of pore geometry on the mean droplet size.
Basically, Figure 10.9 contains all parameters needed to design emulsification pro-
cesses with metal sieves, but as mentioned the size distribution of the prepared emulsions
is not that sharp at span values between 1.2 and 1.4. To improve on this, Nazir (2013) and
Nazir, Schroën, and Boom (2013b) decided to deposit a layer of beads on the metal sieves
as was previously suggested by van der Zwan, Schroën, and Boom (2008), in order to
allow for more breakup mechanisms to occur as was deduced from microfluidic observa-
tions done by the same author (van der Zwan et al., 2006). The results of the investiga-
tions with packed beds are summarized in the next section.
10.2.5 Metal sieves in coMbination with Packed beds

The sieves with the rectangular pores depicted in Figure 10.7 were used as supports
for glass beads of different size (55, 65, 78, and 90 µm) that were used in packed
beds of different heights (1, 2.5, 5, 10, 15, and 20 mm). To be able to compare the
various experiments, a dimensionless diameter (droplet size/void size in the packed
bed) was plotted as a function of the Reynolds number in the packed bed (as previ-
ously defined but now adjusted for the porosity and tortuosity of the bed; Nazir,
2013). The result is shown in Figure 10.10, in which also the span values of the
respective emulsions are shown.
In Figure 10.10, there are two series of experiments, one in which the bed height
was varied and one in which different sized beads were used in the bed, both yielding
different hydrodynamic resistances and therewith different velocities and Reynolds
0.5 1.2
0.4
1.0
d32/dv (−)
0.3
δ (−)
0.8
0.2
Decreasing dv 0.6 Decreasing dv

0.1
Increasing H Increasing H
0.0 0.4
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70 80
(a) Re (−) (b) Re (−)
FIGURe 10.10 (a) Dimensionless droplet diameter, d32/dv, and (b) droplet span, δ, as a func-
tion of Reynolds number; Re: (Δ) particle size (related to void size dv) and (○) bed height varied, H.
numbers. Both data series start off separately at relatively low Reynolds number but
then merge at Reynolds number above 40. At low Reynolds numbers, the void size
has a greater influence on the droplet size than the velocity, indicating that the con-
striction needed for spontaneous droplet snap-off is here dominant in the size reduc-
tion; this is corroborated by the reduction of the span. With increasing Reynolds
number, both curves are the same, which indicates that the breakup mechanism
becomes similar (a decreasing bed height or an increasing void size), and this is due
to the dominance of inertial effects (i.e., droplet breakup due to local shear forces).
This inertial droplet breakup region is characterized by a decrease in droplet unifor-
mity (Nazir, Schroën, and Boom, 2013b). The size of the droplets is also influenced
by the viscosity ratio of both phases, and that is depicted in Figure 10.11a, in which
the ratio of droplet size over void size is shown.
The droplet size consistently became smaller with increasing number of passes
and was smallest at low viscosity ratio because the droplets can be effectively elon-
gated by the highly viscous continuous phase, after which the liquid threads will break
into small droplets. The flux that is shown in the right-hand side plot in Figure 10.11
increases till a viscosity ratio of 3 due to a decrease in continuous phase viscos-
ity, which also results in flow conditions that are less effective for droplet breakup,
because the exerted shear on the dispersed phase will decrease. At a viscosity ratio of
0.4 600
0.3
J (m3 m−2 hr−1)
400
d32/dv (−)
0.2
200
0.1
0 0
0.01 0.1 1 10 100 1000 0.01 0.1 1 10 100 1000
(a) ηd/ηc (−) (b) ηd/ηc (−)
FIGURe 10.11 Dimensionless droplet diameter, d32 /dv, (a) and flux, J, (b) obtained after
different passes as a function of viscosity ratio: (Δ) 1st pass, (□) 3rd pass, and (♦) 5th pass.
(Data from Nazir, A. et al., Chem. Eng. Sci., 116: 547–557, 2014.)
3, the flux reaches its maximum after which it decreases again, most probably because
of the emulsion droplets that now have to be broken up by a different mechanism, as
they cannot be easily elongated by the continuous phase. Interaction of the droplets
with the pore walls (glass beads) and constriction effects now become important, most
probably leading to higher effective viscosity inside the beds and lower fluxes. Still the
actual flux values that were found are high, and as was the case with the metal sieves,
the values stay constant, indicating that the packed beds are not prone to fouling.
At a low viscosity ratio, it is expected that droplet breakup will take place upon exit
of the bed, as was also found for metal sieves (Nazir, Schroën, and Boom, 2011, 2013a).
At higher viscosity ratio, the dispersed phase will still be able to intrude into the porous
bed, but breakup will mainly take place due to constriction inside the bed, leading to a
constant droplet size, as the droplet size now is determined by the average pore size in
the packed bed. This is also reflected in the slightly lower fluxes at high viscosity ratio,
indicating that the dispersed phase inside the packed bed impedes the flow. This con-
clusion is strengthened by the observation that the flux increases systematically, albeit
slightly, at higher number of passes corresponding to smaller droplets.
All droplet sizes were much smaller than the pore size. The droplet-to-pore size
ratio was as low as 0.1, which is the lowest ratio already reported for premix mem-
brane emulsification (Vladisavljević, Kobayashi, and Nakajima, 2012), whereas in
cross-flow membrane emulsification, the droplet size is typically 2–10 times the pore
size (Charcosset, Limayem, and Fessi, 2004). This also indicates that pore sizes need
to be much smaller for cross-flow emulsification to produce droplets of the same size,
which is also reflected in the typical fluxes in cross-flow membrane emulsification
that are orders of magnitude lower than the fluxes observed for the metal sieves and
the combination with packed beds. Besides, the span of the emulsions was consider-
ably reduced compared to the use of metal sieves only, and depending on the emul-
sion composition and process condition, this could be as low as 0.75.
For both the effects of viscosity, and the variation of bed height and bead size, scaling
relations were successfully derived (Nazir, 2013). The general shape of the equation is
γ ζ
d32 H  η 
= αEV−β    d  (10.7)
dv  D   ηc 
with Ev being the energy density; H the bed height; D the particle diameter; ηc and
ηd the viscosity of the continuous and dispersed phases, respectively; and α, β, γ,
and ζ fit parameters. These equations can be used to estimate the droplet size pro-
duced with all systems based on metal sieves and cover all the process conditions
and product properties (Nazir, 2013; Nazir, Schroën, and Boom, 2013b).
10.3 CoMPARIson oF eMULsIFICAtIon

WItH MICRostRUCtUReD sYsteMs
As mentioned in the introduction, cross-flow membrane emulsification is able to pro-
duce emulsions at low energy input, and that is also the case for other microstructured
emulsification systems such as microsieves, metal sieves (with or without packed beds),
tABLe 10.1
A Comparison of Conventional and Microstructured emulsification systems
with Premix emulsification
Premix emulsification
Conventional Microstructure
emulsification emulsification Membrane sieve Packed Bed
Small droplet size ++ + + + +
Droplet uniformity − +++ ++ + ++
Energy efficiency −− ++ ++ + +
Productivity ++ −− + ++ +
Fouling issues −− ++ ++ − −
Note: Number of + or − signs indicates how well a technique performs on a specific characteristic.
and microfluidic chips such as straight-through microchannels (Sugiura et al., 2001;

Kobayashi et al., 2005; van Dijke, Schroën, and Boom, 2008) and edge-based droplet
generation (EDGE) (van Dijke et al., 2010). All methods have their pros and cons, and
these are summarized in Table 10.1.
Conventional emulsification techniques are very good at reducing droplet size
at high throughputs, albeit also at high energy input. Emulsification with micro-
structures such as cross-flow membrane emulsification, straight-through micro-
channels and EDGE, is known for the monodispersity of the emulsions produced,
albeit at low throughputs. For cross-flow membrane emulsification, this monodis-
persity is found only for emulsions with relatively low dispersed phase fraction
(e.g., Vladisavljević and Schubert, 2002). For emulsions with higher dispersed
phase fractions, mostly multiple passes are needed, and this reduces droplet mono-
dispersity. For these emulsions, premix emulsification is an interesting alternative
that can be used for the controlled production of small and uniform sized droplets
at high production rates, albeit at a loss of monodispersity. Fouling may occur if a
“classic” membrane is used, and this is a serious drawback that can be mediated
by the use of the previously discussed metal sieves, possibly in combination with
a packed bed. Microfluidic systems may also be prone to fouling, and surface
properties need to be tended to as was described for microsieves (e.g., Rosso et al.,
2009). Typical droplet size distributions for various emulsification techniques have
been reported in literature; the dispersed phase fraction was ≤5% in all cases (see
Figure 10.12).
The Microfluidizer, a typical high-pressure homogenizer, results in a wide drop-
let size distribution. The EDGE system is an example of a microstructured emul-
sification system that produces very narrow droplet size distributions. Cross-flow
membrane emulsification then follows with a distribution that is somewhat less nar-
row. The packed bed system results in a fairly good distribution, and bearing the
low energy inputs and the high throughputs in mind, the system is an interesting
alternative for emulsions having ingredients causing (depth) fouling in conventional
membranes.
100
10
Volume (%)
0.1
0.01 0.1 1 10 100
FIGURe 10.12 Droplet size distributions obtained using different emulsification systems:
(□) Microfluidizer, 1100 bar, n = 2; (×) Microfluidizer, 50 bar, n = 1; (Δ) cross-flow mem-
brane emulsification, dp = 0.4 μm, 3.3 bar. (All three from Vladisavljević, G.T. et al., Colloids
Surf., A, 232, 199–207, 2004.); (●) packed bed system, dp = 23 μm, 2 bar, n = 5. (Data from
Nazir, A. et al., Chem. Eng. Sci., 92, 190–197, 2013b.); (⬦) EDGE system, 1.2 μm. (Data from
van Dijke, K.C. et al., Am. Inst. Chem. Eng. J., 56, 3, 833–836, 2010.); (+) Microchannel,
16 μm. (Data from Vladisavljević, G.T. et al., Colloids Surf. A, 232, 199–207, 2004.)
10.4 ConCLUDInG ReMARKs

The use of microstructures with uniform pores has led to increased understanding
of the overall emulsification processes. From the previous sections it is clear that it is
not just about having a device with uniform pores, but that the entire process needs
to be designed and the emulsion ingredients chosen in such a way that monodisperse
emulsion formation is possible. Although considerable progress is made, challenges
remain regarding the combination of high throughput and monodispersity.
ReFeRenCes
A.J. Abrahamse, R. van Lierop, R.G.M. van der Sman, A. van der Padt, R.M. Boom. 2002.
Analysis of droplet formation and interactions during cross-flow membrane emulsifica-
tion. Journal of Membrane Science 204: 125–137.
A.J. Abrahamse, A. van der Padt, R.M. Boom, W.B.C. de Heij. 2001. Process fundamentals
of membrane emulsification: Simulation with CFD. American Institute of Chemical
Engineers Journal 47 (6): 1285–1291.
A. Arafat, M. Giesbers, M. Rosso, E.J.R. Sudhölter, C.G.P.H. Schroën, R.G. White et al. 2007.
Covalent biofunctionalization of silicon nitride surfaces. Langmuir 23: 6233–6244.
A. Arafat, K. Schroën, L. de Smet, E. Sudhölter, H. Zuilhof. 2004. Tailor-made functionalization
of silicon nitride surfaces. Journal of the American Chemical Society 126: 8600–8601.
N. Aryantia, R.A. Williams, R. Houa, G.T. Vladisavljević. 2006. Performance of rotating

membrane emulsification for o/w production. Desalination 200: 572–574. Also pre-
sented at Euromembrane 2006, September 24–28, 2006, Giardini Naxos, Italy.
O. Behrend, H. Schubert. 2001. Influence of hydrostatic pressure and gas content on continu-
ous ultrasound emulsification. Ultrasonics Sonochemistry 8: 271–276.
C. Charcosset. 2009. Preparation of emulsions and particles by membrane emulsification for
the food processing industry. Journal of Food Engineering 92(3): 241–249.
C. Charcosset, I. Limayem, H. Fessi. 2004. The membrane emulsification process—A review.
Journal of Chemical Technology and Biotechnology 79: 209–218.
K.C. van Dijke, K. Schroën, R.M. Boom. 2008. Microchannel emulsification: From computa-
tional fluid dynamics to predictive analytical model. Langmuir 24: 10107–10115.
K.C. van Dijke, G. Veldhuis, C.G.P.H. Schroën, R.M. Boom. 2010. Simultaneous formation
of many droplets in a single microfluidic droplet formation unit. American Institute of
Chemical Engineers Journal 56 (3): 833–836.
A.M.C. Van Dinther, C.G.P.H. Schroën, F.J. Vergeldt, R.G.M. van der Sman, R.M. Boom. 2012.
Suspension flow in microfluidic devices—A review of experimental techniques focussing on
concentration and velocity gradients. Advances in Colloid and Interface Science 173: 23–34.
V. Eisner-Schadler. 2007. Emulsion processing with a rotating membrane (ROME). PhD
Dissertation, ETH Zürich, Switzerland, number 17153.
V. Eisner-Schadler, E.J. Windhab. 2006. Continuous membrane emulsification by using a mem-
brane system with controlled pore distance. Desalination 189: 130–135. Also presented
at the 10th Aachen Membrane Colloquium, March 16–17, 2005, Aachen, Germany.
T. Fuchigami, M. Toki, K. Nakanishi. 2000. Membrane emulsification using sol-gel derived
macroporous silica glass. Journal of Sol-Gel Science and Technology 19: 337–341.
A.J. Gijsbertsen-Abrahamse. 2003. Membrane emulsification: Process principles. PhD thesis,
Wageningen University, the Netherlands.
A.J. Gijsbertsen-Abrahamse, A. van der Padt, R.M. Boom. 2004. Status of cross-flow mem-
brane emulsification and outlook for industrial application. Journal of Membrane
Science 230: 149–159.
S. van der Graaf, T. Nisisako, C.G.P.H. Schroën, R.G.M. van der Sman, R.M. Boom. 2006.
Lattice Boltzmann simulations of droplet formation in a T-shaped microchannel.
Langmuir 22: 4144–4152.
S. van der Graaf, C.G.P.H. Schroën, R.M. Boom. 2005. Preparation of double emulsions by
membrane emulsification—A review. Journal of Membrane Science 251: 7–15.
S. van der Graaf, C.G.P.H. Schroën, R.G.M. van der Sman, R.M. Boom. 2004. Influence
of dynamic interfacial tension on droplet formation during membrane emulsification.
Journal of Colloid and Interface Science 277: 456–463.
S. van der Graaf, M.L.J. Steegmans, R.G.M. van der Sman, C.G.P.H. Schroën, R.M. Boom. 2005.
Droplet formation in a T-shaped microchannel junction: A model system for membrane emul-
sification. Colloids and Surfaces A: Physicochemical Engineering Aspects 266: 106–116.
S.M. Joscelyne, G. Trägårdh. 2000. Membrane emulsification—A literature review. Journal of
Membrane Science 169: 107–117.
H. Karbstein, H. Schubert. 1995. Developments in the continuous mechanical production of
oil-in-water macro-emulsions. Chemical Engineering and Processing 34: 205–211.
I. Kobayashi, X.F. Lou, S. Mukataka, M. Nakajima. 2005. Preparation of monodisperse water-
in-oil-in-water emulsions using microfluidization and straight-through microchannel
emulsification. Journal of the American Oil Chemists Society 82 (1): 65–71.
U. Lambrich, H. Schubert. 2005. Emulsification using microporous systems. Journal of
Membrane Science 257: 76–84.
E. Lepercq-Bost, M.-L. Giorgi, A. Isambert, C. Arnaud. 2008. Use of the capillary number for
the prediction of droplet size in membrane emulsification. Journal of Membrane Science
314: 76–89.
A.A. Maan, K. Schroën, R. Boom. 2011. Spontaneous droplet formation techniques for mono-
disperse emulsions preparation—Perspectives for food applications. Journal of Food
Engineering 107 (3–4): 334–346.
T. Nakashima, M. Shimizu. 1986. Porous glass from calcium alumino boro-silicate glass.
Ceramics Japan 21: 408.
T. Nakashima, M. Shimizu, M. Kukizaki. 1991. Membrane emulsification by microporous
glass. Key Engineering Materials 61–62: 513–516.
A. Nazir. 2013. Premix emulsification systems. PhD thesis, Wageningen University, the Netherlands.
A. Nazir, R.M. Boom, K. Schroën. 2014. Influence of the emulsion formulation in premix
emulsification using packed. Chemical Engineering Science 116: 547–557.
A. Nazir, K. Schroën, R. Boom. 2010. Premix emulsification: A review. Journal of Membrane
Science 362 (1–2): 1–11.
A. Nazir, K. Schroën, R. Boom. 2011. High-throughput premix membrane emulsification
using nickel sieves having straight-through pores. Journal of Membrane Science 383
(1–2): 116–123.
A. Nazir, K. Schroën, R. Boom. 2013a. The effect of pore geometry on premix membrane
emulsification using nickel sieves having uniform pores. Chemical Engineering Science
93: 173–180.
A. Nazir, K. Schroën, R. Boom. 2013b. Droplet break-up mechanism in premix emulsification
using packed beds. Chemical Engineering Science 92: 190–197.
S.J. Peng, R.A. Williams. 1998. Controlled production of emulsions using a crossflow mem-
brane. Part I: Droplet formation from a single pore. Chemical Engineering Research and
Design 76: 894–901.
M. Rayner, G. Trägårdh. 2002. Membrane emulsification modelling: How can we get from
characterisation to design? Desalination 145: 165–172.
M. Rayner, G. Trägårdh, C. Trägårdh, P. Dejmek, P. 2004. Using the surface evolver to
model droplet formation processes in membrane emulsification. Journal of Colloid and
Interface Science 279 (1): 175–185.
C.J.M. van Rijn (ed.) 2004. Nano and Micro Engineered Membrane Technology. Membrane
Science and Technology Series 10. Elsevier, Amsterdam, the Netherlands. ISBN 0444514899,
9780444514899, p. 384.
C.J.M. van Rijn, M.C. Elwenspoek. 1995. Micro filtration membrane sieve with silicon micro
machining for industrial and biomedical applications. Proceedings of the Institute of Electrical
and Electronics Engineers 29: 83–87.
C.J.M. van Rijn, M. van der Wekken, W. Nijdam, M. Elwenspoek. 1997. Deflection and
maximum load of microfiltration membrane sieves made with silicon micromachining.
Journal of Microelectromechanical Systems 6: 48–54.
M. Rosso, M. Giesbers, A. Arafat, K. Schroën, H. Zuilhof. 2009. Covalently attached organic
monolayers on SiC and SixN4 surfaces: Formation using UV light at room temperature.
Langmuir 25 (4): 2172–2180.
V. Schröder. 1999. Herstellen van Öl-in-Wasser Emulsionen mit Microporösen Membranen.
PhD thesis, Technische Hochschule Karlsruhe, Germany.
V. Schröder, O. Behrend, H. Schubert. 1998. Effect of dynamic interfacial tension on the
emulsification process using microporous, ceramic membranes. Journal of Colloid and
Interface Science 202: 334–340.
M.L.J. Steegmans, A. Warmerdam, C.G.P.H. Schroën, R.M. Boom. 2009. Dynamic interfacial
tension measurements with microfluidic Y-junctions. Langmuir 25 (17): 9751–9758.
M.T. Stillwell, R.G. Holdich, S.R. Kosvintsev, G. Gasparini, and I.W. Cumming. 2007. Stirred
cell membrane emulsification and factors influencing dispersion drop size and unifor-
mity. Industrial & Engineering Chemistry Research 46: 965–972.
S. Sugiura, M. Nakajima, S. Iwamoto, M. Seki. 2001. Interfacial tension driven monodis-
persed droplet formation from microfabricated channel array. Langmuir 17: 5562–5566.
G.T. Vladisavljević, I. Kobayashi, M. Nakajima. 2012. Production of uniform droplets using

membrane, microchannel, and microfluidic emulsification devices. Microfluidics and
Nanofluidics 13: 151–178.
G.T. Vladisavljević, I. Kobayashi, M. Nakajima, R.A. Williams, M. Shimizu, T. Nakashima.
2007. Shirasu porous glass membrane emulsification: Characterisation of membrane
structure by high-resolution X-ray microtomography and microscopic observation of
droplet formation in real time. Journal of Membrane Science 302: 243–253.
G.T. Vladisavljević, U. Lambrich, M. Nakajima, H. Schubert. 2004. Production of O/W emul-
sions using SPG membranes, ceramic α-aluminium oxide membranes, microfluidizer
and a silicon microchannel plate—A comparative study. Colloids and Surfaces A:
Physicochemical Engineering Aspects 232: 199–207.
G.T. Vladisavljević, H. Schubert. 2002. Preparation and analysis of oil-in-water emulsions
with a narrow droplet size distribution using Shirasu-porous-glass (SPG) membranes.
Desalination 144: 167–172.
G.T. Vladisavljević, J. Surh, J.D. McClements. 2006. Effect of emulsifier type on droplet
disruption in repeated Shirasu porous glass membrane homogenization. Langmuir 22:
4526–4533.
G.T. Vladisavljević, R.A. Williams. 2005. Recent developments in manufacturing emulsions
and particulate products using membranes. Advances in Colloid and Interface Science
113: 1–20.
L. Vogelaar, R.G.H. Lammertink, J.N. Barsema, W. Nijdam, L.A.M. Bolhuis-Versteeg, C.J.M.
van Rijn et al. 2005. Phase separation micromolding: A new generic approach for micro-
structuring various materials. Small 1: 645–655.
Q. Yuan, R. Houa, N. Aryantia, R.A. Williams, S. Biggs, S. Lawson, H. Silgram, M. Sarkar,
and R. Birch. 2008. Manufacture of controlled emulsions and particulates using mem-
brane emulsification. Desalination 224: 215–220.
E.A. van der Zwan, C.G.P.H. Schroën, R.M. Boom. 2008. Premix membrane emulsification by
using a packed layer of glass beads. American Institute of Chemical Engineers Journal
54 (8): 2190–2197.
E. van der Zwan, K. Schroën, K. van Dijke, R.M. Boom. 2006. Visualization of droplet
break-up in pre-mix membrane emulsification using microfluidic devices. Colloids and
Surfaces A: Physicochemical Engineering Aspects 277 (1–3): 223–229.
11 Formation and
Modification of
Dispersions Using
Shirasu Porous Glass
Membranes
Goran T. Vladisavljević
Contents
11.1 Introduction ................................................................................................. 256
11.1.1 Membrane Dispersion Processes................................................... 257
11.1.2 Membrane Treatment of Dispersions ............................................ 258
11.1.3 Comparison of Membranes with Other Methods to Generate
and Treat Dispersions ....................................................................260
11.2 SPG Membrane ........................................................................................... 261
11.2.1 Fabrication of SPG Membrane ...................................................... 261
11.2.2 Properties of SPG Membrane........................................................264
11.2.3 Surface Modification of SPG Membrane ......................................266
11.3 Emulsification Using SPG Membrane ........................................................ 267
11.3.1 Factors Affecting Droplet Size in DME........................................ 270
11.3.1.1 Influence of Transmembrane Pressure and Flux .......... 271
11.3.1.2 Influence of Membrane Pore Size and Shear Stress
on the Membrane Surface ............................................. 273
11.3.1.3 Influence of Surfactant ................................................. 275
11.3.2 Factors Affecting Droplet Size in PME ........................................ 275
11.3.3 Applications of Direct and Premix Membrane Emulsification
Using SPG Membrane ................................................................... 278
11.4 Gas Dispersion Using SPG Membrane ....................................................... 285
11.5 Conclusions ................................................................................................. 286
References .............................................................................................................. 286
255
ABSTRACT This chapter deals with the production, properties, and macrofluidic
applications of Shirasu porous glass (SPG) membrane. The first section provides an
overview of the membrane microfluidic processes used for production and modification
of liquid–liquid and gas–liquid micro and nanodispersions, such as direct and premix
membrane emulsification with and without phase inversion, membrane demulsifica-
tion, membrane micromixing/direct precipitation, and micro and nanobubbling. In the
last section of this chapter, SPG membranes are compared with conventional homog-
enizers and microfluidic drop generators in terms of production rate, droplet size uni-
formity, and applied shear stresses. The second section deals with the fabrication of
SPG membrane by spinodal decomposition in Na2O─CaO─Al2O3─B2O3─SiO2 type
glass and morphological, mechanical, and hydrodynamic properties of SPG mem-
brane. This chapter also covers modification of surface charge, contact angle, and
porosity of SPG membrane using different physical and chemical methods, such as
deposition of silica nanoparticles onto membrane surface, coating with silicon resin,
filling the pores with solvent-responsive polymer chains, and chemical modification
with silane coupling agents. The fourth section is focused on the effects of physical
properties of the dispersed and continuous phases, operating parameters, and mem-
brane properties on the droplet size in direct and premix SPG membrane emulsifi-
cation. In addition, the most common classes of micro and nanoparticles fabricated
using SPG membrane are reviewed, and their fabrication routes are discussed. It was
concluded that a broad variety of different chemical and physicochemical processes
can be combined with SPG membrane emulsification to convert droplets into uniform
particle. The last section briefly discusses the generation of micro and nanobubbles
using SPG membrane.
Keywords: membrane emulsification, Shirasu porous glass membrane, nanoparticles,
polymeric microspheres, microbubbles, Janus particles, core-shell particles
11.1 IntroduCtIon
Synthetic membranes are mainly used for separation purposes and to achieve a
chemical or biochemical conversion. Membrane separation processes are character-
ized by the fact that a feed stream is divided into two product streams of differ-
ent chemical compositions: retentate and permeate (Figure 11.1a) (Mulder, 1996).
A shear rate is applied at the retentate–membrane interface to limit concentration
polarization and accumulation of the rejected solids on the high pressure side of the
membrane. In the last two decades, microfluidic applications of membranes (forma-
tion of droplets and bubbles, micromixing of miscible liquids, droplet breakup, and
coalescence, etc.) are gaining in popularity as a result of rising global interest in
microfluidic technologies. Membrane microfluidic processes can be classified into
two groups: (1) formation of dispersions (gas–liquid, liquid–liquid, and solid–liquid)
(Figure 11.1b) and (2) treatment of dispersions (demulsification, homogenization,
and phase inversion). In a membrane dispersion process (Figure 11.1b), phase I is
injected through a microporous membrane into phase II for the purpose of (1) mix-
ing of two miscible fluids, usually two liquid phases; (2) forming droplets or bubbles
of phase I into phase II. Membrane treatment of dispersions (Figure 11.1c) involves
Formation and Modification of Dispersions 257
Product
Feed Retentate Phase II Product
P1 P2 P2
P2 P1 P1
(a) Permeate (b) Phase I (c) Feed
FIGure 11.1 A comparison between pressure-driven membrane separation and membrane

microfluidic processes, where P1 > P2. In a membrane separation process (a), feed stream
is split into two product streams of different chemical composition. (Data from Mulder, M.,
Basic Principles of Membrane Technology, Kluwer Academic Publishers, Dordrecht, the
Netherlands, 1996.) In a membrane dispersion process (b), two streams (miscible or immis-
cible) are combined together to form one product stream. Membrane treatment of dispersions
(c) involves passing a whole dispersion through the membrane, which results in the modifica-
tion of the particle size distribution in the original dispersion and/or phase inversion.
passing dispersion through the membrane, which results in the physicochemical

and mechanical interactions between the dispersed phase entities (bubbles/droplets/
particles) and the pore walls leading to the modification of the original particle size
distribution.
11.1.1 MeMbrane Dispersion processes

Membrane dispersion processes that involves injecting one fluid through the mem-
brane into another fluid are direct membrane emulsifications (DME) (Nakashima
et al., 2000), membrane micro and nanobubbling (Kukizaki and Goto, 2006, 2007a),
and membrane micromixing (Chen, Luo, Xu et al., 2004). A shear is applied at the
membrane surface to improve mixing efficiency or facilitate the detachment of bub-
bles or droplets from the membrane surface. In DME, one liquid (a dispersed phase)
is injected through a microporous membrane into another immiscible liquid (the
continuous phase) (Nakashima et al., 1991), leading to the formation of droplets at
the membrane–continuous phase interface (Figure 11.2a). Hydrophobic membranes
O/W Gas–liquid Nano-

emulsion dispersion dispersion
Water or
water-
Oil Gas soluble
organic
solvent
(a) (b) (c)
FIGure 11.2 Membrane dispersion processes with hydrophilic membrane: (a) production
of O/W emulsion by DME (Data from Nakashima, T. et al., Adv. Drug Deliv. Rev., 45, 47−56
2000.); (b) production of microbubbles (Data from Kukizaki, M. and Goto, M., Colloids Surf. A,
293, 87−94, 2007.) and nanobubbles (Data from Kukizaki, M. and Goto, M., J. Membr. Sci.,
281, 386−396, 2006.); and (c) production of nanoparticles by membrane micromixing/direct
precipitation method. (Data from Chen, G.G. et al., Powder Technol., 139, 180−185, 2004.)
are needed to produce water-in-oil (W/O) emulsions (Cheng et al., 2008; Jing et al.,
2006), and hydrophilic membranes are required to prepare oil-in-water (O/W)
emulsions (Figure 11.2a). In membrane microbubbling, a pressurized gas is forced
through a hydrophilic membrane into aqueous continuous phase, leading to the for-
mation of microbubbles (1 μm < dbubble < 1 mm) or nanobubbles (dbubble < 1 μm),
depending on the pore size of the membrane (Figure 11.2b). Micromixing is the
interpenetration of miscible solutions at the molecular level, and it is a crucial step
in any homogeneous reaction (Okhonin et al., 2011). In membrane micromixing,
an organic solution containing water-miscible organic solvent or an aqueous solu-
tion penetrates through a hydrophilic membrane into another aqueous phase for the
purpose of mixing two solvents rapidly with each other. Membrane micromixing can
be combined with direct precipitation to produce inorganic (Chen, Luo, Xu et al.,
2004) and organic (Laouini et al., 2011) nanoparticles. Precipitation of inorganic
compounds requires dispersion of an aqueous solution of water-soluble salt A into
an aqueous solution of water-soluble salt B, and nanoparticles are formed as a result
of chemical reaction between the two salts: A + B → C + D, where one of the
products is sparingly soluble in water (Table 11.1). Precipitation of organic com-
pounds requires dispersion of a water-miscible organic solvent containing particle-
forming organic compounds into an aqueous phase (antisolvent), and precipitation
occurs as a result of the lower solubility of the organic solutes in the aqueous phase
(Figure 11.2c). Particle-forming organic compounds in pharmaceutical nanodisper-
sions are active principle ingredients (APIs); excipients and typical organic phase
compositions are listed in Table 11.2.
11.1.2 MeMbrane TreaTMenT of Dispersions

Membrane processes used to modify particle size distribution of dispersions can
be classified into four groups: (1) simple premix membrane emulsification (PME)
(Suzuki et al., 1996); (2) PME with phase inversion (Suzuki et al., 1996); (3) mem-
brane demulsification (Kukizaki and Goto, 2008); and (4) homogenization of sus-
pensions by extrusion through membrane (Olson et al., 1979). In PME (Figure 11.3a),
tABLe 11.1
Formation of Inorganic nanoparticles by Membrane Micromixing/direct
Precipitation Method
Membrane
salt A salt B and Pore size references
BaSO4 nanoparticles 0.1−0.3M BaCl2 0.1M NaSO4 5 μm stainless steel, Chen, Luo, Xu
(d = 20−200 nm) 0.2−0.9 μm Ni et al. (2004)
Anatase-TiO2 0.03−0.15M 0.1−0.3M 0.2 μm Ni Chen, Luo, Yang
nanoparticles Ti(SO4)2 NH4HCO3 et al. (2004)
(d = 9−20 nm)
ZnO nanoparticles 0.2−1.2M ZnSO4 2.25M 5 μm stainless steel Wang et al.
(d = 9.4−14 nm) NH4HCO3 (2010)
tABLe 11.2
Formation of organic nanoparticles by Membrane Micromixing/direct
Precipitation Method
Membrane
excipients solvent and API and Pore size references
BDP-loaded 20−60 mg ml–1 Lipoïd® Ethanol + 0.4 0.4−10.2 μm Jaafar-Maalej
liposomes E80 + 4−12 mg mg/ml BDP SPG et al. (2011)
(d = 60−200 nm) ml–1 Chl
SPL-loaded 20−80 mg/ml Lipoïd® Ethanol + 3 mg 40 nm PP Laouini et al.
liposomes E80 + 4−16 mg ml–1 SPL hollow fiber (2011)
(d = 110−190 nm) ml−1 Chl
Vitamin E-loaded 5 mg/ml PCL Acetone + 4 mg 0.2−10.2 μm Khayata
PCL nanoparticles ml–1 vitamin E SPG et al. (2012)
(d = 250−350 nm)
Caffeine and 105 mM Tw + 105 mM Ethanol + 10 mg 0.9 μm SPG Pham et al.
SPL-loaded Chl + 23.3 mM DCP ml–1 caffeine or (2012)
niosomes 3 mg ml–1 SPL
(d = 111−115 nm)
API, Active principle ingredient; BDP, beclomethasone dipropionate; Chl, cholesterol; DCP, dicetyl
phosphate; Lipoïd® E80, egg yolk lecithin from Lipoïd GmbH; PCL, polycaprolactone; PP, polypropyl-
ene; SPL, spironolactone; Tw, Tween 60.
O/W W/O
emulsion emulsion Oil
layer
Hydrophilic Hydrophobic
membrane membrane Hydrophilic
membrane
O/W
W/O
premix
emulsion
(a) (b) (c)
FIGure 11.3 Treatment of emulsions using membranes: (a) production of O/W emulsion
by PME (Data from Suzuki, K. et al., Food Sci. Technol. Int. Tokyo, 2, 43−47, 1996.);
(b) production of W/O emulsion by PME with phase inversion (Data from Suzuki, K. et al.,
Food Sci. Technol. Int. Tokyo, 5, 234−238, 1999.); and (c) demulsification of W/O emulsion.
(Data from Kukizaki, M. and Goto, M., J. Membr. Sci., 322, 196−203, 2008.)
a pre-emulsion is forced through a microporous membrane (Suzuki et al., 1996) or

a packed bed of uniform particles (van der Zwan et al., 2008; Yasuda et al., 2010).
As in DME, hydrophobic and hydrophilic membranes are needed to produce W/O
and O/W emulsions, respectively. If the transmembrane pressure is lower than the
capillary pressure in a pore, the membrane will reject the droplets, while allow-
ing the continuous phase liquid to pass through, which will lead to the separation
of the emulsion into droplet-free continuous phase and concentrated emulsion

(Koltuniewicz et al., 1995; Park et al., 2001).
When the dispersed phase of the feed emulsion wets the membrane wall, the
rate of droplet coalescence in the membrane pores is faster than the rate of droplet
breakup, which leads to inversion of phases in the emulsion (Figure 11.3b) or separa-
tion of the feed emulsion into two distinct phases (Figure 11.3c). In PME with phase
inversion, an O/W or W/O/W emulsion undergoes inversion into a W/O emulsion as
a result of permeation through a hydrophobic membrane (Kawashima et al., 1991;
Suzuki et al., 1999). Similarly, a W/O emulsion can be inverted into O/W emul-
sion after permeation through a hydrophilic membrane. A successful phase inver-
sion requires that feed emulsion contains a blend of surfactants with a low and high
hydrophilic–lipophilic balance (HLB) number (Suzuki et al., 1999), or, otherwise,
the emulsion breaking is more likely to occur than the phase inversion.
11.1.3 coMparison of MeMbranes wiTh oTher MeThoDs

To GeneraTe anD TreaT Dispersions
Generation of droplets/bubbles in microfluidic devices such as T-junctions (Thorsen

et al., 2001) and flow focusing devices (Anna et al., 2003) usually involves injection
of one fluid through a single microchannel into a stream of another immiscible fluid
(Vladisavljević et al., 2012). The droplets/bubbles generated in microfluidic devices
are highly uniform in size, with a typical coefficient of variation in dripping regime
of 3% or less, and the drop generation frequency can exceed 10,000 Hz (Yobas et al.,
2006). However, the volume flow rate of the dispersed phase in microfluidic devices is
very low, usually 0.01−10 ml h−1, because there is typically only one droplet-generation
unit. Membranes overcome this low throughput limitation by providing countless
pores that serve as massively parallel drop generation units. Considering that mem-
brane modules can easily be integrated into a system with large total membrane area,
while the integration of microfluidic devices is often challenging due to significant
pressure drop in microfluidic channels and difficulties of controlling the flow rates of
individual streams in complicated channel networks, it is clear that membranes are
more suitable for large throughput applications. An advantage of microfluidic chan-
nels over membranes lies in their ability to produce droplets with a complex morphol-
ogy and to manipulate individual droplets with high precision after production.
Compared to high shear rotor–stator devices, high-pressure valve homogenizers,
and ultrasonic and static mixers, membrane dispersion devices operate under mild
shear stress conditions, allowing high yields of inner droplets in multiple emulsion
production (Dragosavac et al., 2012; Surh et al., 2007; Vladisavljević and Williams,
2008). Conventional emulsification techniques are not suitable when dealing with
shear-sensitive ingredients, because they apply more energy than needed to disrupt
droplets (Karbstein and Schubert, 1995). In DME, a shear rate on the membrane
surface is in the range of (1−50) × 103 s–1, but droplets can be produced even in the
absence of shear (Kukizaki, 2009a; Kukizaki and Goto, 2009; Kosvintsev et al.,
2008). A shear rate in rotor-stator devices such as high shear in-line mixers and
colloid mills is (1–2) × 105 s–1, and it is up to 107 s−1 in Microfluidizers®. In PME,
a pressure drop across the membrane is typically 1−10 bar, while in high-pressure
valve homogenizers it ranges from 50 to over 2000 bar. In addition, energy input
in conventional dispersion devices is not spatially uniform; for example, in rotor-
stator devices, shear forces are high in close proximity to a rotor and low in “dead
zones,” leading to the production of polydispersed emulsions. On the other hand, in
the majority of membrane dispersion processes, shear is uniformly distributed over
the membrane surface.
Another advantage of membrane emulsification compared to conventional emul-
sification devices is that membrane systems allow integration of emulsification step
and emulsion post-processing to achieve simultaneous drop generation and sepa-
ration, chemical/biochemical conversion, or physicochemical transformation. The
examples include integration of DME or PME with liquid–liquid extraction (Chen,
Luo, Sun et al., 2004, Xu et al., 2005), biphasic enzymatic transformation (Li and
Sakaki, 2008; Mazzei et al., 2010), pervaporation (Chang and Hatton, 2012), and
complex coacervation (Piacentini et al., submitted).
11.2 sPG MeMBrAne

Membranes used to produce and treat dispersions should have the following proper-
ties: (1) uniform pores with a broad range of available mean pore sizes to suit different
applications; (2) low hydrodynamic resistance; (3) high mechanical strength and
thermal and chemical resistance; (4) membrane material should be suitable for sur-
face modification (modification of contact angle, surface charge, permeability, etc.);
(5) membrane fabrication process should allow precise control over the pore size
and pore geometry. Shirasu porous glass (SPG) meets the majority of the abovemen-
tioned criteria, and it is by far the most widely used microporous membrane in mem-
brane dispersion processes. Advantages of SPG membrane over micro-engineered
membranes are in typically higher porosity, more versatile surface chemistry that
can be applied, and broader range of pore sizes available.
11.2.1 fabricaTion of spG MeMbrane

SPG membrane is fabricated from Na2O─CaO─Al2O3─B2O3─SiO2 or Na2O─
CaO─MgO─Al2O3─B2O3─SiO2 type mother glass through phase separation by
spinodal decomposition (Kukizaki and Nakashima, 2004; Nakashima and Kuroki,
1981; Nakashima and Shimizu, 1986). The mother glass is prepared by mixing and
melting raw materials (Shirasu, limestone, and boric acid) at about 1350°C. Typical
mixing ratios of raw materials for SPG membrane are given in Table 11.3. Soda ash
(Na2CO3) and sometimes MgO and ZrO2 are added to molten glass to adjust the rate
and temperature of phase separation and alkaline durability of the glass. Shirasu
is a volcanic ash deposit from southern Kyushu, which contains 72−77 wt% SiO2,
10−15 wt% Al2O3, and small amounts of other inorganic oxides (Table 11.4). Molten
mother glass is shaped into tubes or flat discs by blowing or casting and then heat
treated at 650°C−750°C for the period ranging from several hours to several tens
of hours. The thermal treatment causes a homogeneous glass melt to separate into
an acid-insoluble (Al2O3–SiO2 rich) phase and acid-soluble (CaO–B2O3 rich) phase
(Figure 11.4). The phase-separated glass is then immersed into a hydrochloric acid
tABLe 11.3
typical Mixing ratios of raw Materials in the Production
of sPG from na2o–Cao–Al2o3 –B2o3 –sio2 Mother Glass
wt%
Shirasu 51
Limestone 23
Boric acid 22
Soda ash 4
Source: Nakashima, T. 38th International SPG Forum on Membrane and

Particle Science and Technology in Food and Medical Care, Sadowara,
Japan, 2002.
Note: MgO (≈5 wt%) can also be added.
tABLe 11.4
Composition of Primary Glass,a sPG,a and Porous Vycor
Glass and Pyrex Glass
Primary Glass sPG Vycor® Glass Pyrex® Glass
for sPG (wt%) (wt%) (wt%) (wt%)
SiO2 49 69 94−99.5 81
Al2O2 10 13 0−0.5 2
CaO 17 2 – –
B2O3 16 7 0.2−6.0 13
Na2O 5 5 < 0.1 4
K2O 2 4 – –
Fe2O3 1 0.4 – –
Sources: Nakashima, T. et al., J. Ceram. Soc. Jpn. Int. Ed., 100, 1389−1393,
1992; Nakashima, T., 38th International SPG Forum on Membrane
and Particle Science and Technology in Food and Medical Care,
Sadowara, Japan, 2002.
a Based on proportions of raw materials given in Table 11.3.
solution to dissolve CaO–B2O3-rich phase, which results in the formation of porous

skeleton, whose composition is shown in Table 11.4. The porosity of SPG membrane
is determined by the volume fraction of the acid-soluble phase in the phase-separated
mother glass and ranges between 50% and 60% (Vladisavljević et al., 2005). If the
fraction of acid-soluble phase is too low or too high, separation may take place by the
nucleation and growth mechanism. The nucleation and growth mechanism occurs in
the metastable region of the phase diagram, between the spinodal and binodal lines
(Figure 11.5), and leads to the formation of discrete spherical particles of one phase
embedded in a continuous matrix of the other. This morphology is undesirable in the
fabrication of SPG membrane and must be avoided.
Shirasu Boric acid Limestone
Additives (Na2CO3, MgO, and ZrO2)

Primary glass
(Na2O–CaO– Glass fusion at 1350°C
Al2O3–B2O3–SiO2)
Acid soluble phase
(CaO–B2O3 rich)
Forming
Cooling to 760°C–750°C
Acid leaching HCl

Acid insoluble phase
(SiO2-Al2O3 rich)
SPG membrane
FIGure 11.4 A flow diagram of different steps involved in the fabrication of Shirasu
porous glass (SPG) membrane.
1
T1
UCST Binodal Interconnected
-type pores
Temperature
Spinodal
2
T2 Droplet
-type pores
Xs X1 Xi
Molar fraction of acid insoluble phase
FIGure 11.5 Spinodal decomposition of glass induced by cooling mother glass from an initial
temperature T1 to temperature T2 lying in the spinodal region (within the spinodal line). To pre-
vent phase separation via nucleation, a transition from the stable to the spinodal region of the
phase diagram must proceed quickly or through the upper critical solution temperature (UCST).
Figure 11.5 depicts a spinodal decomposition induced by cooling a homogeneous

glass melt from a temperature T1 at which all components are miscible in all
proportions to a temperature T2, which lies within the spinodal (unstable) region.
A homogeneous glass with composition of x1 is separated into two immiscible phases
with compositions of xs and xi. The ratio of acid-soluble to acid-insoluble phase can
be found by the lever rule and is equal to (xi − x1)/(x1 − xs). The mean pore diameter
dp of SPG membrane can be controlled by adjusting the time, t, and temperature, T2,
of the heat treatment process (Kukizaki, 2010):
 V  1/ 2  − Ea 
d p = 4 K 1/ 2  p  t exp  (2 RT )  (11.1)
 mm   2 
where:
K is a constant depending on the composition of mother glass
Ea is the activation energy for diffusion during phase separation [400−600 kJ
mol–1 according to Nakashima (2002) and Kukizaki (2010)]
R is the universal gas constant
Vp/mm is the total pore volume per unit mass of dry membrane
Therefore, the mean pore diameter of SPG membrane is proportional to the square
root of the heating time at any constant temperature, whereas a logarithm of the
mean pore diameter is inversely proportional to 1/T2 at constant heating time.
11.2.2 properTies of spG MeMbrane

SPG membrane is available from SPG Technology Co., Ltd. (Sadowara, Japan) over
a wide spectrum of mean pore sizes ranging from 0.040 to 20 μm (Table 11.5). The
membrane has a uniform internal microstructure, as confirmed by X-ray microto-
mography (XMT) (Vladisavljević et al., 2007), characterized by interconnected and
tortuous cylindrical pores with a tortuosity factor of ξ ≈ 1.3. On scanning electron
microscope (SEM) and XMT images, it was found that the pores have a noncylindri-
cal shape (Figure 11.6), because they extend in all directions and include pore junc-
tions. The number of pores per unit cross-sectional area of SPG membrane is given
by (Vladisavljević et al., 2005):
N 0.56
= 2 (11.2)
Am dp
where:
N/Am and dp are in m–2 and m, respectively
The hydraulic resistance of isotropic SPG membrane is given by (Vladisavljević

et al., 2005):
32ξ2δm
Rm,sym = (11.3)
d p2ε
tABLe 11.5
Properties of Commercial Isotropic (symmetric) sPG Membrane
shape tubes or Flat discs
Thickness, δm 0.4−1 mm
Compressive strength 200−280 Mpa
Pore diameter, dp 0.04−20 μm
Porosity, ε 50%−60%
True density 2000−2500 kg m–3
Zeta potential at pH = 3−10 and CNaCl = 1−100 mol m–3 −15−(−45) mV
Pore tortuosity, ξ 1.25−1.4
Number of pores per unit cross-sectional area, N/Am 109−1014 m–2
Specific pore volume, Vp/mm 0.5−0.6 dm3 kg–1
Hydraulic resistance, Rm,sym 108−1012 m–1
Sources: Kukizaki, M., Sep. Sci. Technol., 69, 87−96, 2009; Nakashima, T., 38th International
SPG Forum on Membrane and Particle Science and Technology in Food and Medical
Care, Sadowara, Japan, 2002; Nakashima, T. et al., J. Ceram. Soc. Jpn. Int. Ed., 100,
1389−1393, 1992; Vladisavljević, G.T. et al., J. Membr. Sci., 250, 69−77, 2005.
SEM image XMT image
200 μm
(a) (b)
FIGure 11.6 (a) Scanning electron micrograph of the surface of SPG membrane polished
with diamond paste and used for visualization of membrane emulsification by metallographic
microscope; (b) X-ray microtomography image of SPG membrane. (Data from Vladisavljević,
G.T. et al., J. Membr. Sci., 302, 243−253, 2007.)
where:
δm is the membrane thickness
ε is the membrane porosity
The hydraulic resistance of isotropic SPG membrane is relatively high (Table 11.5),
due to its substantial thickness of 400−1000 μm, but can be reduced if the membrane
is fabricated with anisotropic structure (Kukizaki and Goto, 2007b). Assuming that
the pore tortuosity and porosity, ξ and ε, are independent of the pore size, the hydrau-
lic resistance of anisotropic SPG membrane is given by (Kukizaki and Goto, 2007b):
32ξ2δskin 32ξ2δsup
Rm,asym = + 2 (11.4)
2
d p,skin ε d p,supε
where:
δskin and δsup are the thicknesses of the skin and support layer, respectively
dp,skin and dp,sup their mean pore diameters
According to Kukizaki and Goto (2007b), the thickness of the skin layer is 6% of
the overall membrane thickness and the ratio of the pore diameters in the skin and
support layer is around 7, so it can be written as:
 ξ2   δ δ   0.06 0.94 
Rm,asym = 32    2skin + 2sup  = Rm,sym  + 2  ≈ 0.08 Rm,sym
 ε   d p,skin d p,sup   1 7 
Therefore, the hydraulic resistance of asymmetric SPG membrane is just 8% of the

Rm value for symmetric membrane.
SPG is more stable in water and alkaline solutions than porous Vycor® glass,
because it contains less SiO2 and more Al2O3 (Table 11.4). However, the durability of
both membranes at high pH is poor due to the attack of hydroxide ions on siloxane
(Si─O─Si) bonds:
≡ Si − O − Si ≡ + OH − → ≡ Si − O − + ≡ Si − OH
Alkaline durability of SPG can be improved by incorporating about 3 mol% ZrO2

into the glass skeleton, which results in the formation of stable Zr─O─Si bonds
in the silicate network (Kukizaki, 2010). A compressive strength of SPG of
200−280 MPa is much higher than that of porous alumina or zirconia of the same
porosity (Nakashima et al., 1992), because SPG is made up of a continuous glass
skeleton with very few defects, while porous alumina or zirconia is composed of
skeletal grains joined together discontinuously via grain boundaries.
11.2.3 surface MoDificaTion of spG MeMbrane

The surface of SPG membrane can be rendered hydrophobic by chemical modifica-
tion with organosilane compounds such as chlorosilanes (Kukizaki and Wada, 2008)
or coating with silicone resin (Vladisavljević et al., 2005). Monochlorosilanes such
as trimethylchlorosilane (TMS) and octadecyldimethylchlorosilane (ODS) are the
most suitable for hydrophobization because they contain only one chlorine atom,
which means that no polymerization between silane molecules can occur when
they react with a silanol group on the pore surface (Figure 11.7a) (Kai et al., 2006).
The longer the carbon chain length in the organosilane compound, more hydropho-
bic the membrane surface becomes (Kukizaki and Wada, 2008). The membrane
hydrophobicity can be enhanced by depositing silica nanoparticles onto the surface
of SPG membrane prior to treatment with TMS (Meng et al., 2013). The surface
of SPG membrane can be made with thermoresponsive hydrophilic–hydrophobic
properties by depositing silica nanoparticles containing poly(N-isopropylacryl-
amide) (PNIPAM) brushes grafted on their surface (Meng et al., 2010). The porosity
−OH −OSi(CH3)2R
Cl−Si(CH3)2R
SiO2 −OH SiO2 −OSi(CH3)2R
TMS: R = CH3
ODS: R = C18H37
−OH −OSi(CH3)2R
(a)
RO
−OH −O
RO Si NH2
RO +
SiO2 −OH + H2O SiO2 −O Si NH3
APTMS: R = CH3
−
−OH APTES: R = C2H5 −O + OH
(b)
FIGure 11.7 Chemical modification of SPG surface by treatment with organosilane com-
pounds: (a) hydrophobic treatment with monochlorosilanes (TMS—trimethylchlorosilane,
ODS—octadecyldimethylchlorosilane) (Data from Kai, T. et al., J. Polym. Sci., Part A-1:
Polym. Chem., 44, 846−856, 2006.); (b) introduction of amino groups by amino trialkoxysi-
lanes to render the surface positively charged [APTMS—(3-aminopropyl)-trimethoxysilane,
APTES—(3-aminopropyl)-triethoxysilane].
and hydraulic resistance of SPG membrane can be modified over a wide range by
incorporating dextran macromolecules within the pores (Kawakita et al., 2009; Seto
et al., 2011). Dextran can be synthesized by in situ enzymatic reaction between dex-
transucrase immobilized within the pores and sucrose from an aqueous solution that
is passed through the membrane. A reversible change in the hydraulic resistance of
the modified SPG membrane is a consequence of reversible extension and shrinkage
of solvent-responsive dextran chains inside the pores.
The surface of untreated SPG surface has a negative zeta potential between
–15 and –45 mV within a pH range of 2−8 due to dissociation of silanol groups
(≡Si-OH Δ ≡SiO− + H+) (Kukizaki, 2009b). A positive charge on the membrane sur-
face can be induced by treating the membrane with amino trialkoxysilanes, such as
(3-aminopropyl)-trimethoxysilane (APTMS) and (3-aminopropyl)-triethoxysilane
(APTES) (Figure 11.7b). Amino trialkoxysilanes undergo hydrolysis in aqueous
solution resulting in the formation of silanol groups, which can be condensed with
a silanol group on the SPG surface to form stable siloxane bond (Si–O–Si) required
for surface modification.
11.3 eMuLsIFICAtIon usInG sPG MeMBrAne

SPG membrane was widely used both in DME (Vladisavljević and Schubert, 2002;
Vladisavljević, Lambrich et al., 2004) and PME (Vladisavljević, Shimizu et al.,
2004, 2006; Vladisavljević, Surh et al., 2006). The advantages of PME over DME
are in smaller droplet sizes (Figure 11.8) and higher transmembrane fluxes that can
be achieved for any given pore size. On the other hand, more severe membrane foul-
ing and broader particle size distribution can be expected, compared to DME.
Premix SPG Emulsion
30 μm
(a) (b)
FIGure 11.8 (a) A micrograph of droplets formed on the surface of SPG membrane in
DME (Data from Vladisavljević, G.T. et al., J. Membr. Sci., 302, 243−253, 2007.); (b) micro-
graphs of droplets before PME and after passing five times through 8-μm SPG membrane.
(Data from Vladisavljević and McClements, 2010.)
Various SPG membrane devices have been used in DME: (1) cross-flow module with
tubular SPG membrane with an effective length of up to 500 mm, (2) a short SPG mem-
brane tube with an effective length of 7−15 mm in a stirred vessel (internal or external
pressure micro kit), and (3) rotating SPG membrane tube in a stagnant continuous phase.
In the cross-flow DME system, a continuous phase liquid circulates from a storage tank
through the bore of the membrane tube and back to the tank (Figure 11.9). A dispersed
phase-forming liquid stored in a pressure vessel is fed to the outside of the membrane
tube and force to penetrate through the membrane under the pressure difference which
is 1.1–5 times higher than the capillary pressure (Vladisavljević and Schubert, 2003a).
The apparatus is operated continuously until a desired dispersed phase concentration is
achieved in the emulsion. A transmembrane flux in cross-flow DME should be kept
below 1−30 l m−2 h−1 to obtain uniform droplets with a relative span factor of droplet size
distribution of 0.25−0.45. To increase transmembrane flux by 2 orders of magnitude,
the continuous phase can be introduced into SPG membrane tube radially, as shown in
Figure 11.10. A tangential introduction of the continuous phase generates spiral stream-
lines in the axial direction (swirl flow) that exert a strong centrifugal force onto the inner
surface of the membrane helping to sweep away droplets from the membrane surface
(Shimoda et al., 2011). At the swirl-flow velocity of 0.85−5.4 m s−1 and the transmem-
brane flux of 0.3−3 m3 m−2 h−1, a relative span factor of droplet size distribution of
0.45−0.64 was achieved with an oil/water phase volume ratio in single-pass operation
of up to 0.4 (Shimoda et al., 2011). Insertion of static turbulence promoters is an alter-
native method of increasing shear at the membrane surface in cross-flow DME, while
maintaining a low shear in the recirculation loop (Koris et al., 2011).
Cross-flow systems are easy to scale up and offer a constant shear stress along
the membrane surface. However, at least several hundred millilitres of the continu-
ous phase is required in the system to provide recirculation. The SPG test kit shown
in Figure 11.11a requires a much smaller amount of continuous phase (<50 ml) and
can be operated with a very low hold-up volume of both phases, which is useful
for expensive samples such as medical preparations (Higashi and Setoguchi, 2000).
Vent
1 Membrane
module
Continuous
phase
Compressed Dispersed
gas phase
FIGure 11.9 An apparatus for cross-flow DME using tubular SPG membrane. During
initial start-up, a valve 1 is open to remove any trapped air from the module. (Data from
Nakashima, T. et al., Monodisperse single and double emulsions and method of producing
same, US Patent 5,326,484, 1994.)
The continuous phase is kept under agitation by a magnetic stir bar, while the
dispersed phase is injected through the membrane tube from outside to inside. The
membrane tube serves as a draft tube, which results in more effective circulation of
the continuous phase than in an internal pressure SPG kit.
In addition to DME with static SPG membrane, where shear stress is controlled
by fluid flow over the membrane surface, dynamic SPG membrane systems have
been investigated, where shear is controlled by rotating the membrane within a static
continuous phase (Pawlik and Norton, 2012, 2013). Rotating membrane systems can
be operated batchwise or continuously. In a continuous flow operation, surface shear
is decoupled from the cross-flow velocity, which means that sufficient shear on the
membrane surface can be achieved no matter how small the flow rate of the continu-
ous phase may be. Therefore, emulsions with a high dispersed phase concentration
can be produced without emulsion recycling that can help prevent damage to shear
sensitive components and secondary breakup of the drops formed by the membrane.
Streamline of the
continuous phase
Tubular
SPG membrane
Continuous
phase
(a)
Continuous Dispersed
phase phase
Emulsion
(b)
FIGure 11.10 Introduction of continuous phase in cross-flow module from the tangential
direction to improve a dispersed phase flux through the membrane (a) cross-sectional view of the
module in the tangential direction, that is, perpendicular to the direction of cross-flow; (b) cross-
sectional view of the module in the axial direction, that is, parallel to the direction of cross-flow.
(Data from Shimoda et al., 2011.)
An SPG membrane rig used for PME is shown in Figure 11.11b. A pressurized
premix from a pressure vessel is passed through the membrane tube from outside
to inside under the driving force of pressure difference ranging from several bars
(for a 10-μm membrane) to more than 10 bar (for 1-μm membrane) and up to 50
bar for the membrane with submicron pore sizes. The product emulsion is collected
inside the membrane tube and discharged from the bottom of the tube. In order to
reduce the droplet size additionally and improve droplet size uniformity, product
emulsion is passed repeatedly through the membrane (Vladisavljević, Shimizu et al.,
2004, 2006; Vladisavljević, Surh et al., 2006). Membrane homogenization using
repeated cycles was first developed by Olson et al. (1979) and used for homogeniza-
tion of lipid vesicles using track-etch polycarbonate filters.
11.3.1 facTors affecTinG DropleT size in DMe

The size distribution of droplets produced in DME depends on a variety of factors,
such as the pore size and wetting properties of the membrane, transmembrane flux,
shear stress generated on the membrane surface, physical properties of the dispersed
Dispersed
phase Vent
Continuous
phase
(a)
Premix
Emulsion
(b)
FIGure 11.11 External pressure type micro kits available by SPG Technology Co., Ltd.
(Sadowara, Japan) for (a) DME and (b) PME. The kits are supplied with SPG membrane tube
with an effective length of 10−15 mm.
and continuous phases, nature of the surfactant used and the surfactant concentra-
tion, and emulsion formulation (Joscelyne and Trägårdh, 2000).
11.3.1.1 Influence of transmembrane Pressure and Flux

The minimum transmembrane pressure for driving the oil phase through the pores is
known as the capillary pressure, Pcap, and is given by the Young–Laplace equation:
4γ wo cos θ
Pcap = (11.5)
dp
Oil
Water phase phase
Oil Water
phase phase
θ θ
γmw γmo γmo γmw
γwo γwo
γmw = γmo + γwo cosθ γmo = γmw + γwo cosθ
Hydrophilic membrane (θ < 90°) Hydrophobic membrane (θ > 90°)

(a) (b)
FIGure 11.12 Typical contact angles through the water phase and phase pressures encoun-
tered in membrane emulsification: (a) production of O/W emulsion (θ < 90°, Po > Pw); (b) pro-
duction of W/O emulsion (θ > 90°, Po < Pw). The contact angle θ is the angle measured through
the water phase, where a liquid–liquid interface meets a membrane surface (γmw = interfacial
tension between the membrane and water phase, γmo = interfacial tension between the mem-
brane and oil phase, γwo = interfacial tension between the water and oil phase).
where:
γwo is the equilibrium interfacial tension between the water and oil phase
θ is the contact angle; that is, the angle formed by a water phase at the three-
phase boundary where the water phase, oil phase, and membrane intersect
(Figure 11.12)
A hydrophilic membrane (θ < 90°) is used in the production of O/W emulsion; thus
Pcap > 0 and Po > Pw. A hydrophobic membrane (θ > 90°) is used in the production
of W/O emulsion, and thus Pcap < 0 and Po < Pw; that is, the water phase pressure
should be higher than the oil phase pressure by Pcap to drive the water phase through
the membrane. The droplet-generation regime is determined by capillary number
given by Ca = Udηd /γwo, where Ud is the velocity of the dispersed phase in a pore
and μd is the viscosity of the dispersed phase. For low capillary numbers in the pores
(Ca < Cacr), droplets are formed in the dripping regime. In this regime, the interfa-
cial tension force dominates the viscous force (Sugiura et al., 2002), and the droplet
size is virtually independent on the transmembrane flux (Figure 11.13). For high
capillary numbers (Ca > Cacr), droplets grow to large sizes (dd > 10dp) before being
detached from the membrane surface, which is termed as continuous outflow regime
(Kobayashi et al., 2003). In this regime, the viscous force dominates the interfacial
tension force and the droplet size sharply increases with increasing the dispersed
phase velocity. The critical flux, Jcr, that is, the transmembrane flux at which the
transition from dripping to continuous outflow regime occurs, is independent of
the pore size and increases with decreasing the viscosity of the dispersed phase.
Emulsions produced in the continuous outflow regime are highly polydisperse due
Jcr
Dripping regime Continuous outflow regime

Droplet size, dd
Transmembrane flux, J
FIGure 11.13 Mean droplet size in DME as a function of transmembrane flux, J. Dripping
regime is characterized by the formation of small droplets at high frequency and occurs at
J < Jcr. Continuous outflow regime is characterized by the formation of large droplets at low
frequency and occurs at J > Jcr.
to the random nature of droplet formation process. In addition, flow transition from
dripping to continuous outflow does not occur simultaneously for all pores, leading
to the large variations in the droplet size for the pores operating in the dripping and
continuous outflow regime.
11.3.1.2 Influence of Membrane Pore size and shear

stress on the Membrane surface
In a dripping regime, a linear correlation between the mean droplet size and the
mean pore size of SPG membrane exists: dd = Kdp (Figure 11.14), where K typically
ranges between 2.8 and 3.5 (Kukizaki and Goto, 2007c, 2009; Nakashima et al.,
1991; Vladisavljević, Shimizu et al., 2006). The gradient of dd versus dp increases
with decreasing shear stress on the membrane surface, but even in the absence of any
shear, K is 3.3 for O/W emulsions stabilized with 1% Tween 80 surfactant (Kukizaki
and Goto, 2009). The mean droplet size is determined by a balance between the
shear force exerted on the liquid–liquid interface by the continuous phase, Fd, and
the capillary force, Fca (Kosvintsev et al., 2005):
Fca = πd p γ (11.6)
Pore size, dp
Droplet size, dd
Shear stress on membrane surface, σw
FIGure 11.14 Mean droplet size in DME as a function of mean pore size of SPG membrane
and shear stress on the membrane surface.
2
d  (11.7)
Fd = 9πσw dd  d  − rp2
 2 
where:
rp is the pore radius
σw is the shear stress on the membrane surface
The equation for the droplet diameter can be obtained by solving Equations 11.6 and
11.7 for dd:
18σ2wrp2 + 2 81σw4 rp4 + 4rp2σ2w γ 2

dd = (11.8)
3σw
Therefore, the mean drop diameter decreases with increasing shear stress on the
membrane surface until it reaches a constant value at sufficiently high shear stresses
(Figure 11.14). In cross-flow DME, σw is a function of the mean velocity of the con-
tinuous phase inside the membrane tube Uc (Vladisavljević and Schubert, 2003b):
 ρ U2 
σw = λ  c c  (11.9)
 8 
where:
ρc is the density of the continuous phase
λ is the Moody friction factor
For laminar flow inside the membrane tube (Re < 2300): λ = 64/Re and σw = 8ηcUc/dmi,
where dmi is the inner diameter of the membrane tube and ηc is the viscosity of continu-
ous phase. For the rotating SPG membrane, σw can be estimated from (Vladisavljević
and Williams, 2006):
2ηc rmo
2
ω
σw = (11.10)
rb − rmo
2 2
where:
ω is the angular velocity of the membrane
rmo is the outer radius of the membrane tube
r b is the inner radius of the cylinder in which the membrane tube is rotating
11.3.1.3 Influence of surfactant

The role of surfactant in membrane emulsification is to rapidly adsorb to the newly
formed oil–water interface to facilitate droplet detachment and stabilize the formed
droplet against coalescence by reducing the interfacial tension. The effect of kinet-
ics of adsorption of surfactant at oil–aqueous interface on the droplet size has been
investigated by several groups (Rayner et al., 2005; Schröder et al., 1998; van der
Graaf et al., 2004). As a rule, the faster the surfactant molecules adsorb to the newly
formed interface, the smaller the droplet size of the emulsion becomes. Surfactant
molecules must not adsorb to the membrane surface, since otherwise the dispersed
phase will spread over the membrane surface. This means that the functional groups
of surfactant molecules must not carry a positive charge to avoid electrostatic deposi-
tion onto the negatively charged surface of SPG membrane (Nakashima et al., 1993).
The use of cationic surfactants, for example, alkyltrimethylammonium salts such as
cetyltrimethyl-ammonium bromide (CTAB), leads to polydispersed O/W emulsions
with dd/dp > 20 (Nakashima et al., 1993). The use of zwitterionic surfactants must
also be avoided, even when they carry a net negative charge. For example, lecithin
at pH 3 fouls SPG membrane due to electrostatic interactions between positively
charged groups [ −N(CH3 )3+ and −NH 3+ ] on phospholipid molecules and negatively
charged silanol groups on SPG surface, although at pH 3 the net charge on lecithin
molecules is negative (Surh et al., 2008). To produce cationic droplets using SPG
membrane, the membrane must be treated with amino trialkoxysilanes to induce a
positive charge on the surface (Figure 11.7b), or the charge of anionic droplets can
be altered after membrane emulsification by surfactant displacement (Vladisavljević
and McClements, 2010).
11.3.2 facTors affecTinG DropleT size in pMe

The mean droplet size in PME depends on several parameters such as the mean pore
size of SPG membrane, transmembrane pressure, number of passes through the mem-
brane, viscosity of the continuous and dispersed phases, and interfacial tension (Nazir
et al., 2010). The mean droplet size is a nonlinear function of the mean pore size of
SPG membrane (Figure 11.15): dd = K(dp)n, where n < 1. The dd/dp ratio decreases
with increasing the mean pore size and ranges from 1.5 to 1 at dp = 5−20 μm and the
Pore size, dp
DME
Droplet size, dd
PME
Transmembrane flux, J
FIGure 11.15 Mean droplet size in PME as a function of mean pore size of SPG membrane
(at σw,p = const) and transmembrane flux (at dp = const). For comparison, a relationship between
mean droplet size and mean pore size in DME is shown by the dashed line.
shear stress on the pore walls of 200 Pa (Vladisavljević, Shimizu et al., 2006). The
critical pressure in PME is given by (Park et al., 2001):
γ[2 + 2a6 / 2a6 − 1 × arccos(1 / a3 ) − 4 a2 ]

Pcap = (11.11)
a + a2 − 1
where:
a = d1 /dp, and d1 is the mean droplet size in premix
If d1 /dp >> 1, the capillary pressure is given by Equation 11.5. In PME, the transmem-
brane pressure resulting in the most uniform droplets is typically 10−50 times larger
than Pcap (Vladisavljević, Shimizu et al., 2004). The mean droplet size decreases
with increasing the mean shear stress on the pore walls, given by:
8ηe J ξ
σw,p = (11.12)
εdp
where:
ηe is the viscosity of emulsion inside the pores
According to Equation 11.12, the mean droplet size decreases with increasing trans-
membrane pressure, as shown in Figures 11.15 and 11.17b. The pressure energy is
used for flow through the membrane pores and droplet disruption:
 1 1 
∆ptm = ηe (Rm + Rf )J + Cφ  −  γ
d1 d p 
 (11.13)
∆p
flow
∆pdisr
where:
C is a constant
ϕ is the volume fraction of the dispersed phase
Rm is the membrane resistance
Rf is the fouling resistance
The second term in Equation 11.13 is based on the assumption that the energy needed
for droplet disruption is proportional to the resultant increase in surface area. The
fouling resistance occurs as a result of accumulation of the dispersed phase on the
membrane surface (external fouling) and inside the pores (internal fouling). External
fouling dominates at high dd/dp ratios in the feed emulsion and low transmembrane
pressures, whereas internal fouling dominates at high transmembrane pressures
and small droplet sizes relative to the pore size. In repeated PME (Vladisavljević,
Shimizu et al., 2004):
1 1 
∆ ptm = ηe (Rm + Rf, i )Ji + Cφ  − γ
 di di −1  (11.14)
∆p
flow ∆pdisr
where:
Ji and Rf,i are J and Rf during ith pass through the membrane
di and di−1 are the mean droplet diameters after ith and (i−1)th pass, respectively
The effect of varying droplet size on the viscosity of emulsion was disregarded
in Equation 11.14. As the number of passes through the membrane increases at
Δptm = const, the mean droplet size tends to a constant minimum value, that is, di → di−1
(Figure 11.17b), which means that Δpdisr → 0 and Δpflow → Δptm. Therefore, the term
accounting for droplet disruption (Δpdisr) becomes progressively less important than
the flow term (Δpflow), and pressure energy of the feed mixture is increasingly used for
providing emulsion flow through the membrane (Figure 11.16). As a consequence of
redistribution of pressure terms in Equation 11.14, the transmembrane flux at constant
operating pressure increases after each pass through the membrane until a maximum
flux is established. The maximum transmembrane flux in PME is limited by the mem-
brane resistance, emulsion viscosity, and transmembrane pressure (Figure 11.17a).
The effect of continuous phase viscosity, dispersed phase concentration, trans-
membrane pressure on the mean droplet size, and transmembrane flux in repeated
PME is shown in Figure 11.17. The largest increase in flux between the two passes was
observed between the first and second pass, because the most significant reduction in
the mean droplet size was observed in the first pass. Under the same conditions, the
limiting flux was substantially lower at the higher dispersed phase content, which was
a consequence of both the higher viscosity of emulsion, ηe, and the higher Δpdisr term
150
140 Δpflow
130
Pressure difference, Δpflow or Δpdisr/kPa
Δpflow, 150 kPa

120
φi = 10 vol% Δpdisr, 150 kPa
φo = 10 vol%
110 Δpflow, 20 kPa
Δpdisr, 20 kPa
40
Δpdisr
30
Δpflow
20
10 Δpdisr
0
1 2 3 4 5
Number of emulsification cycles, n
FIGure 11.16 The pressure difference used to overcome the hydraulic resistances in the
system and interfacial tension force as a function of the number of passes through the mem-
brane at two different transmembrane pressures. Production of W/O/W emulsion using PME
at the viscosity of the continuous phase of 126 mPa⋅s, the concentration of W/O drops in
W/O/W emulsion of 10 vol%, the concentration of inner water phase in the W/O emulsion of
10 vol%, and the mean pore size of the membrane of 10.7 μm. (Data from Vladisavljević, G.T.
et al., Colloids Surf. A, 232, 199−207, 2004.)
in Equation 11.14. Although the transmembrane fluxes were significantly higher at

the lower viscosity of the continuous phase, the lowest droplet sizes were obtained at
the higher viscosity of the continuous phase (128 mPa s), because of the higher shear
stress acting on the pore walls; at Δptm = 150 kPa and ϕo = 20 vol%, the shear stress
acting on the pore walls in the fifth pass was σw,p = 80 Pa at ηc = 1 mPa s, whereas σw,p
was 1880 Pa at ηc = 126 mPa s, in spite of the smaller transmembrane flux.
11.3.3 applicaTions of DirecT anD preMix MeMbrane

eMulsificaTion usinG spG MeMbrane
SPG membrane was initially used for the preparation of simple O/W and W/O
emulsions with a narrow particle size distribution and adjustable mean particle size
(Nakashima et al., 1991). Since the early 1990s, applications of SPG membrane
emulsification technique have been extended to the production of multiple emul-
sions, such as solid-in-oil-in-water (S/O/W) (Kukizaki, 2009c), oil-in-water-in-oil
(O/W/O) (Cho et al., 2005; Wei et al., 2013), and water-in-oil-in-water (W/O/W)
240
200
160
120
Transmembrane flux, J/m3 m−2 h−1
80
20
18
16
ηc = 1 mPas, φo = 1 vol%, Δptm = 150 kPa
14 ηc = 1 mPas, φo = 20 vol%, Δptm = 150 kPa
12 ηc = 1 mPas, φo = 20 vol%, Δptm = 100 kPa

10
8
4
1 2 3 4 5
(a) Number of passes through membrane, n
3
2
Span
0.5
0.4
0.3
11 dp = 10.7 μm
Mean size of outer drops d50/μm
10
1 2 3 4 5
(b) Number of passes through membrane, n
FIGure 11.17 The effect of the number of passes through the membrane on: (a) transmembrane
flux and (b) median diameter and relative span factor of W/O drops. Production of W/O/W
emulsion using PME at different transmembrane pressures (100 or 150 kPa), viscosities of the
continuous phase (1 or 126 mPa s), and concentrations of W/O drops in W/O/W emulsion (1 or
20 vol%). The mean pore size of the SPG membrane was 10.7 μm and the concentration of inner
water phase in the W/O emulsion was 30 vol%. (Data from Vladisavljević, G.T. et al., J. Membr.
Sci., 284, 373−383, 2006.)
(Surh et al., 2007); nano and microemulsions (Choi et al., 2012; Koga et al., 2010;
Laouini et al., 2012; Oh et al., 2011, 2013; Pradhan et al., 2013); emulsions with
droplets laminated with multilayered biopolymer films (Gudipati et al., 2010; Nazir
et al., 2012; Vladisavljević and McClements, 2010); microbubbles (Kukizaki and
Goto, 2007a); nanobubbles (Kukizaki and Goto, 2006); micro and nanoparticles
(Vladisavljević and Williams, 2005, 2010); and vesicles (liposomes and niosomes)
(Hwang et al., 2011; Pham et al., 2012).
Some examples of particles fabricated by DME or PME using SPG membrane are
given in Table 11.6 and Figure 11.18. Emulsion droplets were transformed into solid
particles by implementing a variety of chemical reactions or physicochemical processes
within the droplets, such as cross-linking of hydrogel-forming polymers (Wei et al.,
2013), polymerization of monomer mixtures (Omi et al., 1995), solidification from a
melt (Kukizaki and Goto, 2007c), polymer precipitation induced by solvent evaporation
or extraction (Liu, Ma, Meng et al., 2005), redox reaction (Kakazu et al., 2010), com-
plex coacervation (Kage et al., 1997), and thermal coagulation (El-Mahdy et al., 1998).
tABLe 11.6
examples of Microparticles Fabricated using dMe and PMe with sPG
Membrane
secondary reaction/Process
Product type example after dMe or PMe references
Ceramic particles Silica nano or Polymerization of silicic acids Kandori et al. (1992)
microparticles by interfacial or internal
reaction
Liquid crystal Thermochromic Melt crystallization in O/W Segura et al. (2013)
particles liquid crystal emulsion
particles
Carbon particles Carbon cryogel Sol–gel polycondensation Yamamoto et al.
followed by freeze-drying and (2010)
carbonization
Metal particles Solder metal Solidification of liquid metal in Torigoe et al. (2011)
microparticles M/W or M/O emulsion
Silver nanoparticles Reduction of silver ions in W/O Kakazu et al. (2010)
microemulsions
Solid lipid W/S microcarrier Melt crystallization in W/O/W Kukizaki and Goto
particles emulsion (2007c)
S/S microcarrier Melt crystallization in S/O/W Kukizaki (2009c)
emulsion
Coherent particles Melt crystallization in O/W D’oria et al. (2009);
emulsion Li et al. (2011)
Gel micro and Ca alginate Cross-linking of sodium Liu et al. (2003);
nanoparticles alginate with Ca2+ in W/O You et al. (2001);
emulsion Akamatsu et al.
(2011)
(Continued)
tABLe 11.6 (Continued )

Membrane
Chitosan Cross-linking of chitosan with Wang et al. (2005);
glutaraldehyde in W/O Wei et al. (2010);
emulsion Yue et al. (2011);
Akamatsu et al.
(2012)
Cross-linking of chitosan with Wei et al. (2013)
glutaraldehyde in O/W/O
emulsion
HTCC/GP Thermal gelation in W/O Wu et al. (2008)
emulsion
Alginate/chitosan Coalescence of Na alginate Zhang et al. (2011)
droplets with Ca2+ droplets and
particle coating with chitosan
Agarose Helix-coil transition induced by Zhou et al. (2007,
cooling 2008, 2009)
Protein Albumin Heat or chemical denaturation El-Mahdy et al.
microspheres of albumin in W/O emulsion (1998); Muramatsu
and Kondo (1995);
Muramatsu and
Nakauchi (1998)
Composite Polymer particles Solvent evaporation from oil Supsakulchai et al.
organic– with embedded phase in S/O/W emulsion (2002a, 2002b);
inorganic TiO2/Fe3O4 Omi et al. (2001);
particles nanoparticles or Wang et al. (2013);
quantum dots Yang et al. (2010);
Zhou et al. (2012)
Polymeric particles Solvent evaporation followed Ito et al. (2010)
coated with silica by electrostatic layer-by-layer
nanoparticles deposition
Coherent PSt, P(St-co-DVB), One-stage suspension Yuyama, Hashimoto
polymeric micro P(St-co-MMA), polymerization in O/W et al. (2000); Omi
or nanospheres PUU-VP, etc. emulsion et al. (1994); Nuisin
et al. (2000); Ma,
An et al. (2003);
PSt-PAAm One-stage suspension Ma et al. (2004)
composite polymerization in W/O/W
emulsion
P(AAm-co-AA) and One-stage suspension Nagashima et al.
PNaAMPS polymerization in W/O (1998); Hu et al.
hydrogel emulsion (2011)
(Continued)

Membrane
PMMA Two-stage suspension Omi et al. (1995,
microspheres and polymerization in O/W 1997)
large P(St-co-DVB) emulsion
spheres
PUU, PSt-PMMA, Solvent evaporation from oil Yuyama, Yamamoto
phase droplets in O/W et al. (2000); Ma
emulsion et al. (1999a, 1999b,
1999c)
Synthetic Coherent PLA and Solvent evaporation from oil Ito et. (2011); Yue et al.
biodegradable PLGA spheres phase droplets in O/W (2012); Kanakubo
polymer emulsion et al. (2010)
particles PLA or PLGA Solvent evaporation from oil Liu, Ma, Meng et al.
capsules for phase in W/O/W emulsion (2005); Liu, Ma,
hydrophilic actives, Wan (2005); Doan
DFB-loaded PLA et al. (2011); Hou
capsules et al. (2009)
mPEG-PLA Solvent extraction from oil Wei et al. (2008,
capsules for phase in W/O/W emulsion 2011)
hydrophilic actives
Core/shell and P(St-co- One-stage suspension Ma et al. (2001,
hollow particles DMAEMA), polymerization and internal 2002); Ma, Chen
P(St-co-DVB), phase separation in O/W et al. (2003); Lee
PDVB emulsion et al. (2010); Hao
et al. (2009)
Polymer-supported One-stage suspension Liu et al. (2010a,
palladium catalyst polymerization, internal phase 2010b)
separation, and ligand exchange
P(St-co-DVB-co- Two-stage suspension Wang et al. (2012)
MAA) polymerization and internal
phase separation in O/W
emulsion
ENB-P(M-co-U- In situ polymerization Liu et al. (2011)
co-F) core-shell
capsules
Chitosan Cross-linking of chitosan onto Akamatsu et al.
alginate particles and core (2010)
dissolution
Molecularly imprinted Molecular imprinting using Kou et al. (2012)
P(MMA-co-EDMA) CAP as a template molecule
particles
(Continued)

Membrane
PGPR-PE2CA Interfacial polymerization Lee et al. (2009)
core-shell particles followed by solvent
evaporation
Hollow porous silica One-stage suspension Kong et al. (2010,
nanocapsules polymerization, followed by 2012, 2013)
loaded with Fe3O4 sol-gel process and calcination
nanoparticles
Thermo- Porous PA shells Interfacial polymerization Chu et al. (2002,
responsive with P(NIPAM) 2003)
capsules gates in the pores
P(NIPAM-co-AA) Suspension polymerization in Si et al. (2011); Wang
capsules W/O emulsion et al. (2013)
Janus particles PS/PPC Solvent pervaporation and Chang and Hatton
internal phase separation (2012)
PMMA/P(S-BIEM)- Solvent evaporation, followed Tanaka et al. (2010);
g-PDMAEMA or by internal phase separation Ahmad (2008)
PS/P(MMA- and atom transfer radical
CMS)-b- polymerization
PDMAEMA
PS/PMMA Solvent evaporation followed Yamashita et al.
by internal phase separation (2012)
Complex Gelatin/acacia Complex coacervation in O/W Kage et al. (1997)
coacervate microcapsules emulsion
microcapsules
Nonspherical Hemispherical Cleavage of Janus particles Yamashita et al.
particles polymer particles (2012)
3D colloidal Clusters containing Solvent pervaporation and Chang and Hatton
assemblies silica-encapsulated coating of clusters with silica (2012)
magnetite
nanoparticles
AA, acrylic acid; CAP, chloramphenicol; CMS, chloromethylstyrene; DFB, decafluorobutane; DMAEMA,
dimethylaminoethyl methacrylate; DVB, divinylbenzene; EDMA, ethylene dimethacrylate; ENB
5-ethylidene-2-norbornene; HTCC, N-[(2-hydroxy-3-trimethylammonium) propyl] chitosan chloride; GP,
glycerophosphate; MAA, methacrylic acid; MMA, methyl methacrylate; mPEG, poly(monomethoxypoly
ethylene glycol); NIPAM, N-isopropylacrylamide; PAAm, PAAm: polyacrylamide; PE2CA, poly(ethyl
2-cyanoacrylate); PLA, polylactic acid or polylactide; PLGA, poly(lactic-co-glycolic acid); P(M-co-U-co-F),
Poly(melamine-co-urea-co-formaldehyde); PNaAMPS, poly[sodium 2-(acrylamido)-2-methylpropane-
sulfonate]; PPC, poly(propylene carbonate); P(S-BIEM), poly[styrene-2-(2-bromoisobutyryloxy)ethyl
methacrylate]; PUU, polyurethaneurea; St, styrene; TPP, tripolyphosphate; VP, vinyl polymer.
40°C
SiO2
25°C
PDMAEMA
PMMA
(a) (b) (c) (d)
Porous
Silica nanoparticles SiO2 shell
− −
−
− −
− + + ++ −
− PLGA −
+ + ++
− −
− −
(e) − − (f) (g) (h)
Hydrophilic
PS-HEMA Fe3O4 drug
Hydrophilic
drug nanoparticle
solution
Hydrophobic
Solid
surfactant
lipid
(i) (j) (k) (l)
FIGure 11.18 Examples of particles fabricated using SPG membrane emulsification:

(a) doxorubicin (DOX)-loaded liposomes prepared by a film-hydration method combined with
repeated SPG membrane homogenization and remote loading of DOX (Data from Hwang,
T. et al., Colloids Surf. A, 392, 250−255, 2011.); (b) porous thermoresponsive capsules with
poly(N-isopropylacrylamide) (PMIPAM) gates prepared by DME, interfacial polymeriza-
tion, and plasma-graft pore-filling polymerization (Data from Chu, L.Y. et al., Langmuir, 18,
1856−1864, 2002.); (c) “mushroom-like” Janus particles prepared by DME, internal phase
separation, and surface-initiated atom transfer radical polymerization (ATRP) (Data from
Tanaka, T. et al., Langmuir, 26, 7843–7847, 2010.); (d) silica-encapsulated magnetite nanopar-
ticle clusters prepared by DME, solvent pervaporation, and sol-gel coating (Data from Chang,
E.P. and Hatton, T.A., Langmuir, 28, 9748−9758, 2012.); (e) PLGA particles coated with silica
nanoparticles prepared by layer-by-layer electrostatic deposition of poly(allylamine hydro-
chloride) (PAH) and silica nanoparticles onto PLGA particles produced by DME (Data from
Ito, F. et al., Colloids Surf. A, 361, 109−117, 2010.); (f) hemispherical particles produced by
cleavage of Janus particles fabricated by PME (Data from Yamashita, N. et al., Langmuir, 28,
12886−12892, 2012.); (g) porous silica shells loaded with magnetic nanoparticles and antican-
cer drug prepared by DME, polymerization of styrene droplets, silica sol-gel coating of PS
particles, removing PS core by thermal treatment, and drug loading (Data from Kong, S.D.
et al., Nano Lett., 10, 5088–5092, 2010.); (h) Janus PMMA/PS particles produced by DME and
evaporation of toluene from homogeneous PMMA/PS/toluene droplets (Data from Yamashita,
N. et al., Langmuir, 28, 12886−12892, 2012.); (i) chitosan shells prepared by coating chitosan
onto alginate particles produced by DME, followed by cross-linking the shell and dissolution of
the alginate core (Data from Akamatsu, K. et al., Langmuir, 26, 14854−14860, 2010.); (j) mag-
netic polymer microspheres prepared from W/O/W emulsion by PME followed (Continued)
FIGure 11.18 (Continued) by chemical coprecipitation of Fe3O4 within the inner

water phase and solvent evaporation (Data from Yang, J. et al., Ind. Eng. Chem. Res., 49,
6047–6053, 2010.); (k) droplets of hydrophilic drug solution embedded in solid lipid matrix
prepared from W/O/W emulsion by temperature-controlled DME and melt crystallization
(Data from Kukizaki, M. and Goto, M., Colloids Surf. A, 293, 87−94, 2007.); (l) surfactant-
coated hydrophilic drug nanoparticles embedded in solid lipid matrix prepared from S/O/W
emulsions by temperature-controlled PME and melt crystallization. (Data from Kukizaki,
M., J. Membr. Sci., 327, 234−243, 2009.)
Cross-linking of gel-forming polymers within the droplets can be carried out

using physical or chemical cross-linking methods (Wang et al., 2005). Physical
cross-linking methods are helix-coil transition induced by cooling below the
phase transition temperature (Zhou et al., 2007), thermal gelation induced by
heating to about 37°C (Wu et al., 2008), and ionotropic gelation induced by the
addition of multivalent ions (Liu et al., 2003). Melt solidification involves per-
forming membrane emulsification above the melting point of the dispersed phase
followed by emulsion cooling. This approach was used for fabrication of solid
lipid particles for drug delivery applications (Kukizaki, 2009c), low-melting-
point metal particles for soldering microcomponents in microelectronics (Torigoe
et al., 2011), and thermochromic liquid crystal particles for heat transfer research
(Segura et al., 2013).
Polymeric particles were produced by SPG membrane emulsification and subse-
quent suspension polymerization (Omi et al., 1994) or solvent evaporation (Ito et al.,
2011). Suspension polymerization can be carried out in O/W (Ma, An et al., 2003),
W/O (Hu et al., 2011), or W/O/W emulsion (Ma et al., 2004) and can be combined
with droplet swelling technique (which is known as two-stage suspension polymeriza-
tion) to produce microspheres from hydrophilic monomers (Omi et al., 1997). Hollow
particles were produced by combining SPG membrane emulsification with interfa-
cial polymerization (Chu et al., 2003), internal phase separation (Liu et al., 2010a),
molecular imprinting (Kou et al., 2012), and coating a shell around polymer particles
by a sol-gel process (Kong et al., 2013) or interfacial cross-linking (Akamatsu et al.,
2010) followed by core disintegration by chemical dissolution or calcination.
11.4 GAs dIsPersIon usInG sPG MeMBrAne

Microbubbles or nanobubbles can be produced by injecting gas phase through a
hydrophilic SPG membrane into an aqueous surfactant solution (direct injection
method) or by loading porous particles fabricated by SPG membrane emulsifica-
tion with a suitable gas (Hou et al., 2009). Monodispersed microbubbles with a
relative span factor of about 0.5 were generated when the contact angle at mem-
brane–water–air interface was in the range of 0° < θ < 45° and the bubble-to-pore
size ratio was 7.9 (Kukizaki and Wada, 2008). Nanobubbles with a mean diameter
of 360–720 nm and relative span factor of 0.45–0.48 were produced by inject-
ing air through SPG membranes with a mean pore diameter of 43–85 nm into
0.05−0.5 wt% sodium dodecyl sulfate (SDS) solution (Kukizaki and Goto, 2006).
The mean size of nanobubbles was 8.6 times larger than the mean pore size and
unaffected by the flow velocity of air in the pores within a range of 0.5–3.7 m s−1
(Kukizaki and Goto, 2006). Microbubbles generated by SPG membranes can find
applications in the production of aerated food products (Zúñiga and Aguilera,
2008), ultrasound contrast agents for ultrasonic examinations (Hou et al., 2009),
and aerobic wastewater treatment (Liu, Tanaka, Zhang et al., 2012, 2013), which
can be combined with UV irradiation (Tasaki et al., 2009) or activated sludge pro-
cess (Liu, Tanaka, Ma et al., 2012).
11.5 ConCLusIons
SPG membranes are increasingly being used in microfluidic applications aiming at
generating uniform micro and nanodroplets, nanobubbles, and nanoparticles. They
have also been used for modification of emulsions (phase inversion, demulsification,
and homogenization) as well as in micromixing/direct nanoprecipitation processes
for production of inorganic and organic nanoparticles. SPG membranes can over-
come low throughput limitations of conventional microfluidic junctions and flow
focusing devices by providing countless number of pores that serve as massively
parallel T-junctions. Direct and premix membrane emulsification (DME and PME)
are two main modes of operation of SPG membrane emulsification devices. In DME,
the mean droplet size is proportional to the mean pore size and the proportionality
constant is typically around 3, whereas in PME, the ratio of the mean droplet size
to the mean pore size is less than 1.5 and can be below unity. To form uniformly
sized particles, DME or PME can be combined with a variety of physicochemical
or chemical processes that can be applied individually or in combination, such as
polymerization, cross-linking, solvent evaporation, electrostatic deposition, internal
phase separation, coagulation, calcination, sol-gel chemistry, crystallization, etc.
reFerenCes
Ahmad, H., Saito, N., Kagawa, Y. and Okubo, M. (2008) “Preparation of micrometer-sized,
monodisperse ‘Janus’ composite polymer particles having temperature-sensitive poly-
mer brushes at half of the surface by seeded atom transfer radical polymerization,”
Langmuir, 24: 688−691.
Akamatsu, K., Chen, W., Suzuki, Y., Ito, T., Nakao, A., Sugawara, T., Kikuchi, R. and Nakao, S.
(2010) “Preparation of monodisperse chitosan microcapsules with hollow structures
using the SPG membrane emulsification technique,” Langmuir, 26: 14854−14860.
Akamatsu, K., Ikeuchi, Y., Nakao, A. and Nakao, S. (2012) “Size-controlled and monodis-
perse enzyme-encapsulated chitosan microspheres developed by the SPG membrane
emulsification technique,” J. Colloid Interface Sci., 363: 707−710.
Akamatsu, K., Maruyama, K., Chen, W., Nakao, A. and Nakao, S. (2011) “Drastic difference
in porous structure of calcium alginate microspheres prepared with fresh or hydrolyzed
sodium alginate,” J. Colloid Interface Sci., 371: 46−51.
Anna, S.L., Bontoux, N. and Stone, H.A. (2003) “Formation of dispersions using ‘flow focus-
ing’ in microchannels,” Appl. Phys. Lett., 82: 364−366.
Chang, E.P. and Hatton, T.A. (2012) “Membrane emulsification and solvent pervaporation
processes for the continuous synthesis of functional magnetic and Janus nanobeads,”
Langmuir, 28: 9748−9758.
Chen, G.G., Luo, G.S., Sun, Y., Xu, J.H. and Wang, J.D. (2004) “A ceramic microfiltration
tube membrane dispersion extractor,” AIChE J., 50: 382−387.
Chen, G.G., Luo, G.S., Xu, J.H. and Wang, J.D. (2004) “Membrane dispersion precipitation
method to prepare nanopartials,” Powder Technol., 139: 180−185.
Chen, G., Luo, G., Yang, X., Sun, Y. and Wang, J. (2004) “Anatase-TiO2 nano-particle prepa-
ration with a micro-mixing technique and its photocatalytic performance,” Mater. Sci.
Eng. A, 380: 320−325.
Cheng, C.J., Chu, L.Y., Xie, R. and Wang, X.W. (2008) “Preparation of highly monodisperse
W/O emulsions with hydrophobically modified SPG membranes,” Chem. Eng. Technol.,
31: 377−383.
Cho, Y.H., Lee, J.J., Park, I.B., Huh, C.S., Baek, Y.J. and Park, J. (2005) “Protective effect of
microencapsulation consisting of multiple emulsification and heat gelation processes on
immunoglobulin in yolk,” J. Food Sci., 70: E148−E151.
Choi, Y.K., Poudel, B.K., Marasini, N., Yang, K.Y., Kim, J.W., Kim, J.O., Choi, H.G. and Yong,
C.S. (2012) “Enhanced solubility and oral bioavailability of itraconazole by combining
membrane emulsification and spray drying technique,” Int. J. Pharm., 434: 264−271.
Chu, L.Y., Park, S.H., Yamaguchi, T. and Nakao S. (2002) “Preparation of micron-sized mono-
dispersed thermoresponsive core-shell microcapsules,” Langmuir, 18: 1856−1864.
Chu, L.Y., Rui, X., Zhu, J.H., Chen, W.M., Yamaguchi, T. and Nakao, S. (2003) “Study of
SPG membrane emulsification processes for the preparation of monodisperse core-shell
microcapsules,” J. Colloid Interface Sci., 265: 187−196.
Doan, T.V.P., Couet, W. and Olivier, J.C. (2011) “Formulation and in vitro characterization
of inhalable rifampicin-loaded PLGA microspheres for sustained lung delivery,” Int.
J. Pharm., 414: 112–117.
D’oria, C., Charcosset, C., Barresi, A.A. and Fessi, H. (2009) “Preparation of solid lipid par-
ticles by membrane emulsification—Influence of process parameters,” Colloids Surf. A,
338: 114−118.
Dragosavac, M.M., Holdich, R.G., Vladisavljević, G.T. and Sovilj, M.N. (2012) “Stirred cell
membrane emulsification for multiple emulsions containing unrefined pumpkin seed oil
with uniform droplet size,” J. Membr. Sci., 392−393: 122−129.
El-Mahdy, M., Ibrahim, E.S., Safwat, S., el-Sayed, A., Ohshima, H., Makino, K., Muramatsu,
N. and Kondo, T. (1998) “Effects of preparation conditions on the monodispersity of
albumin microspheres,” J. Microencapsulation, 15: 661−673.
van der Graaf, S., Schroën, C.G.P.H., Van der Sman, R.G.M. and Boom, R.M. (2004) “Influence
of dynamic interfacial tension on droplet formation during membrane emulsification,”
J. Colloid Interface Sci., 277: 456−463.
Gudipati, V., Sandra, S., McClements, D.J. and Decker, E.A. (2010) “Oxidative stability and
in vitro digestibility of fish oil-in-water emulsions containing multilayered membranes,”
J. Agric. Food Chem., 58: 8093–8099.
Hao, D.X., Gong, F.L., Hu, G.H., Lei, J.D., Ma, G.H. and Su, Z.G. (2009) “The relation-
ship between heterogeneous structures and phase separation in synthesis of uniform
PolyDVB microspheres,” Polymer, 50: 3188−3195.
Higashi, S. and Setoguchi, T. (2000) “Hepatic arterial injection chemotherapy for hepatocel-
lular carcinoma with epirubicin aqueous solution as numerous vesicles in iodinated
poppy-seed oil microdroplets: Clinical application of water-in-oil-in-water emulsion
prepared using a membrane emulsification technique,” Adv. Drug Deliv. Rev., 45:
57−64.
Hou, Z., Lin, C. and Zhang, Q. (2009) “Preparation and characterization of PLA ultrasound
contrast agents by combining an ultrasound method and a Shirasu Porous Glass (SPG)
membrane emulsification technique,” 3rd International Conference on Bioinformatics
and Biomedical Engineering, June 11–13, 2009. Beijing, People’s Republic of China:
ICBBE.
Hu, J., Hiwatashi, K., Kurokawa, T., Liang, S.M., Wu, Z.L. and Gong, J.P. (2011) “Microgel-
reinforced hydrogel films with high mechanical strength and their visible mesoscale
fracture structure,” Macromolecules, 44: 7775–7781.
Hwang, T., Park, T.J., Koh, W.G., Cheong, I.W., Choi, S.W. and Kim, J.H. (2011) “Fabrication
of nano-scale liposomes containing doxorubicin using Shirasu porous glass membrane,”
Colloids Surf. A, 392: 250−255.
Ito, F., Kanakubo, Y. and Murakami, Y. (2011) “Rapid preparation of monodisperse biodegrad-
able polymer nanospheres using a membrane emulsification technique under low gas
pressure,” J. Polym. Res., 18: 2077–2085.
Ito, F., Uchida, Y. and Murakami, Y. (2010) “Facile technique for preparing organic–inorganic
composite particles: Monodisperse poly(lactide-co-glycolide) (PLGA) particles having
silica nanoparticles on the surface,” Colloids Surf. A, 361: 109−117.
Jaafar-Maalej, C., Charcosset, C. and Fessi, H. (2011) “A new method for liposome prepara-
tion using a membrane contactor,” J. Liposome Res., 21: 213−220.
Jing, C., Chin, C.Y. and Xie, R. (2006) “Preparation of highly monodisperse W/O emulsions
with hydrophobically modified SPG membranes,” J. Colloid Interface Sci., 300: 375−382.
Joscelyne, S.M. and Trägårdh, G. (2000) “Membrane emulsification—A literature review,”
J. Membr. Sci., 169: 107−117.
Kage, H., Kawahara, H., Ogura, H. and Matsuno, Y. (1997) “Microencapsulation of mono-
dispersed droplets by complex coacervation method and membrane thickness of gener-
ated capsules,” Kagaku Kogaku Ronbun., 23: 652−658.
Kai, T., Suma, Y., Ono, S., Yamaguchi, T. and Nakao, S. (2006) “Effect of the pore surface modifi-
cation of an inorganic substrate on the plasma-grafting behavior of pore-filling-type organic/
inorganic composite membranes,” J. Polym. Sci., Part A-1: Polym. Chem., 44: 846−856.
Kakazu, E., Murakami, T., Akamatsu, K., Sugawara, T., Kikuchi, R. and Nakao, S. (2010)
“Preparation of silver nanoparticles using the SPG membrane emulsification technique,”
J. Membr. Sci., 354: 1–5.
Kanakubo, Y., Ito, F. and Murakami, Y. (2010) “Novel one-pot facile technique for preparing
nanoparticles modified with hydrophilic polymers on the surface via block polymer-
assisted emulsification/evaporation process,” Colloids Surf. B, 78: 85−91.
Kandori, K., Kishi, K. and Ishikawa T. (1992) “Preparation of uniform silica hydrogel par-
ticles by SPG filter emulsification method,” Colloids Surf., 62: 259−262.
Karbstein, H. and Schubert H. (1995) “Developments in the continuous mechanical produc-
tion of oil-water macro-emulsions,” Chem. Eng. Process., 34: 205−211.
Kawakita, H., Hamamoto, K., Seto, H., Ohto, K., Harada, H. and Inoue, K. (2009) “Porosity
estimation of a membrane filled with dextran produced by immobilized dextransucrase,”
AIChE J., 55: 275−278.
Kawashima, Y., Hino, T., Takeuchi, H., Niwa, T. and Horibe, K. (1991) “Shear-induced phase
inversion and size control of water/oil/water emulsion droplets with porous membrane,”
J. Colloid Interface Sci., 145: 512−523.
Khayata, N., Abdelwahed, W., Chehna, M.F., Charcosset, C. and Fessi, H. (2012) “Stability
study and lyophilization of vitamin E-loaded nanocapsules prepared by membrane con-
tactor,” Int. J. Pharm., 439: 254−259.
Kobayashi, I., Nakajima, M. and Mukataka, S. (2003) “Preparation characteristics of oil-in-
water emulsions using differently charged surfactants in straight-through microchannel
emulsification,” Colloid. Surf. A, 229: 33−41.
Koga, K., Takarada, N. and Takada, K. (2010) “Nano-sized water-in-oil-in-water emulsion
enhances intestinal absorption of calcein, a high solubility and low permeability com-
pound,” Eur. J. Pharm. Biopharm., 74: 223–232.
Koltuniewicz, A.B., Field, R.W. and Arnot, T.C. (1995) “Cross-flow and dead-end microfiltra-
tion of oily-water emulsion. Part I: Experimental study and analysis of flux decline,”
J. Membr. Sci., 102: 193−207.
Kong, S.D., Choi, C., Khamwannah, J. and Jin, S. (2013) “Magnetically vectored delivery of cancer
drug using remotely on–off switchable nanocapsules,” IEEE Trans. Magn., 49: 349−352.
Kong, S.D., Zhang, W., Lee, J.H., Brammer, K., Lal, R., Karin, M. and Jin, S. (2010)
“Magnetically vectored nanocapsules for tumor penetration and remotely switchable
on-demand drug release,” Nano Lett., 10: 5088–5092.
Kong, S.D., Zhang, W., Lee, J.H., Choi, C., Khamwannah, J., Karin, M. and Jin, S. (2012)
“Externally triggered on-demand drug release and deep tumor penetration,” J. Vac. Sci.
Technol. B: Microelectron. Nanometer Struct., 30: 02C102−02C102-7.
Koris, A., Piacentini, E., Vatai, G., Bekassy-Molnar, E., Drioli, E. and Giorno, L. (2011)
“Investigation on the effects of a mechanical shear-stress modification method during
cross-flow membrane emulsification,” J. Membr. Sci., 371: 28−36.
Kosvintsev, S.R., Gasparini, G. and Holdich, R.G. (2008) “Membrane emulsification: Droplet
size and uniformity in the absence of surface shear,” J. Membr. Sci., 313: 182−189.
Kosvintsev, S.R., Gasparini, G., Holdich, R.G., Cumming, I.W. and Stillwell, M.T. (2005)
“Liquid-liquid membrane dispersion in a stirred cell with and without controlled shear,”
Ind. Eng. Chem. Res., 44: 9323−9330.
Kou, X., Li, Q., Lei, J., Geng, L., Deng, H., Zhang, G., Ma, G., Su, Z. and Jiang, Q. (2012)
“Preparation of molecularly imprinted nanospheres by premix membrane emulsification
technique,” J. Membr. Sci., 417–418: 87–95.
Kukizaki, M. (2009a) “Shirasu porous glass (SPG) membrane emulsification in the absence
of shear flow at the membrane surface: Influence of surfactant type and concentration,
viscosities of dispersed and continuous phases, and transmembrane pressure,” J. Membr.
Sci., 327: 234−243.
Kukizaki, M. (2009b) “Relation between salt rejection and electrokinetic properties on Shirasu
porous glass (SPG) membranes with nano-order uniform pores,” Sep. Sci. Technol., 69:
87−96.
Kukizaki, M. (2009c) “Preparation of solid lipid microcapsules via solid-in-oil-in-water dis-
persions by premix membrane emulsification,” Chem. Eng. J., 151: 387−396.
Kukizaki, M. (2010) “Large-scale production of alkali-resistant Shirasu porous glass (SPG)
membranes: Influence of ZrO2 addition on crystallization and phase separation in
Na2O─CaO─Al2O3─B2O3─SiO2 glasses; and alkali durability and pore morphology of
the membranes,” J. Membr. Sci., 360: 426−435.
Kukizaki, M. and Goto, M. (2006) “Size control of nanobubbles generated from Shirasu-
porous-glass (SPG) membranes,” J. Membr. Sci., 281: 386−396.
Kukizaki, M. and Goto, M. (2007a) “Spontaneous formation behavior of uniform-sized micro-
bubbles from Shirasu porous glass (SPG) membranes in the absence of water-phase
flow,” Colloids Surf. A, 296: 174−181.
Kukizaki, M. and Goto, M. (2007b) “Preparation and characterization of a new asymmetric
type of Shirasu porous glass (SPG) membrane used for membrane emulsification,”
J. Membr. Sci., 299: 190−199.
Kukizaki, M. and Goto, M. (2007c) “Preparation and evaluation of uniformly sized solid lipid
microcapsules using membrane emulsification,” Colloids Surf. A, 293: 87−94.
Kukizaki, M. and Goto, M. (2008) “Demulsification of water-in-oil emulsions by permeation
through Shirasu-porous-glass (SPG) membranes,” J. Membr. Sci., 322: 196−203.
Kukizaki, M. and Goto, M. (2009) “A comparative study of SPG membrane emulsification in
the presence and absence of continuous-phase flow,” J. Chem. Eng. Jpn., 42: 520−530.
Kukizaki, M. and Nakashima, T. (2004) “Acid leaching process in the preparation of porous
glass membranes from phase-separated glass in the Na2O–CaO–MgO–Al2O3–B2O3–
SiO2 system,” Membrane, 29: 301−308.
Kukizaki, M. and Wada, T. (2008) “Effect of the membrane wettability on the size and size
distribution of microbubbles formed from Shirasu-porous-glass (SPG) membranes,”
Colloids Surf. A, 317: 146−154.
Laouini, A., Fessi, H. and Charcosset, C. (2012) “Membrane emulsification: A promising alterna-
tive for vitamin E encapsulation within nano-emulsion,” J. Membr. Sci., 423–424: 85−96.
Laouini, A., Jaafar-Maalej, C., Sfar, S., Charcosset, C. and Fessi, H. (2011) “Liposome prepa-
ration using a hollow fiber membrane contactor—Application to spironolactone encap-
sulation,” Int. J. Pharm., 415: 53−61.
Lee, S.H., Baek, H.H., Kim, J.H. and Choi, S.W. (2009) “Core-shell poly(d,l-lactide-co-
glycolide)/poly(ethyl 2-cyanoacrylate) microparticles with doxorubicin to reduce initial
burst release,” Macromol. Res., 17: 1010−1014.
Lee, J., Hwang, D.R. and Shim, S.E. (2010) “Controlling morphology of polymer micro-
spheres by Shirasu porous glass (SPG) membrane emulsification and subsequent polym-
erization: From solid to hollow,” Macromol. Res., 18: 1142−1147.
Li, Y., Fessi, H. and Charcosset, C. (2011) “Preparation of indomethacin-loaded lipid particles
by membrane emulsification,” Adv. Sci. Lett., 4: 591−595.
Li, N. and Sakaki, K. (2008) “Performance of an emulsion enzyme membrane reactor com-
bined with premix membrane emulsification for lipase-catalyzed resolution of enantio-
mers,” J. Membr. Sci., 314: 183−192.
Liu, X.D., Bao, D.C., Xue, W.M., Xiong, Y., Yu, W.T., Yu, X.J., Ma, X.J. and Yuan, Q. (2003)
“Preparation of uniform calcium alginate gel beads by membrane emulsification cou-
pled with internal gelation,” J. Appl. Polym. Sci., 87: 848−852.
Liu, Y., Feng, X.J., Bao, D.C., Li, K.X. and Bao, M. (2010a) “Preparation of microcapsule-
supported palladium catalyst using SPG (Shirasu Porous Glass) emulsification tech-
nique,” Chin. Chem. Lett., 21: 979–982.
Liu, Y., Feng, X.J., Bao, D.C., Li, K. and Bao, M. (2010b) “A new method for the preparation
of microcapsule-supported palladium catalyst for Suzuki coupling reaction,” J. Mol.
Catal. A: Chem., 323: 16−22.
Liu, X., Lee, J.K. and Kessler, M.R. (2011) “Microencapsulation of self-healing agents with
melamine-urea-formaldehyde by the Shirasu porous glass (SPG) emulsification tech-
nique,” Macromol. Res., 19: 1056−1061.
Liu, R., Ma, G.H., Meng, F.T. and Su, Z.G. (2005) “Preparation of uniform-sized PLA micro-
capsules by combining Shirasu Porous Glass membrane emulsification technique and
multiple emulsion-solvent evaporation method,” J. Controlled Release, 103: 31−43.
Liu, R., Ma, G.H., Wan, Y.H. and Su, Z.G. (2005) “Influence of process parameters on the
size distribution of PLA microcapsules prepared by combining membrane emulsifica-
tion technique and double emulsion-solvent evaporation method,” Colloids Surf. B, 45:
144−153.
Liu, C., Tanaka, H., Ma, J., Zhang, L., Zhang, J., Huang, X. and Matsuzawa, Y. (2012) “Effect
of microbubble and its generation process on mixed liquor properties of activated sludge
using Shirasu porous glass (SPG) membrane system,” Water Res., 46: 6051−6058.
Liu, C., Tanaka, H., Zhang, L., Zhang, J., Huang, X., Ma, J. and Matsuzawa, Y. (2012) “Fouling
and structural changes of Shirasu porous glass (SPG) membrane used in aerobic waste-
water treatment process for microbubble aeration,” J. Membr. Sci., 421−422: 225−231.
Liu, C., Tanaka, H., Zhang, J., Zhang, L., Yang, J., Huang, X. and Kubota, N. (2013) “Successful
application of Shirasu porous glass (SPG) membrane system for microbubble aeration
in a biofilm reactor treating synthetic wastewater,” Sep. Purif. Technol., 103: 53–59.
Ma, G.H., An, C.J., Yuyama, H., Su, Z.G. and Omi, S. (2003) “Synthesis and characterization
of polyurethaneurea-vinyl polymer (PUU-VP) uniform hybrid microspheres by SPG
emulsification technique and subsequent suspension polymerization,” J. Appl. Polym.
Sci., 89: 163−178.
Ma, G.H., Chen, A.Y., Su, Z.G. and Omi, S. (2003) “Preparation of uniform hollow poly-
styrene particles with large voids by a glass-membrane emulsification technique and a
subsequent suspension polymerization,” J. Appl. Polym. Sci., 87: 244−251.
Ma, G.H., Nagai, M. and Omi, S. (1999a) “Study on preparation and morphology of uniform
artificial polystyrene–poly(methyl methacrylate) composite microspheres by employ-
ing the SPG (Shirasu Porous Glass) membrane emulsification technique,” J. Colloid
Interface Sci., 214: 264−282.
Ma, G.H., Nagai, M. and Omi, S. (1999b) “Effect of lauryl alcohol on morphology of uniform
polystyrene-poly(methyl methacrylate) composite microspheres prepared by Porous
glass membrane emulsification technique,” J. Colloid Interface Sci., 219: 110−128.
Ma, G.H., Nagai, M. and Omi, S. (1999c) “Preparation of uniform poly(lactide) microspheres
by employing the Shirasu porous glass (SPG) emulsification technique,” Colloids Surf.
A, 153: 383−394.
Ma, G.H., Nagai, M. and Omi, S. (2001) “Study on preparation of monodispersed poly(styrene-
co-N-dimethylaminoethyl methacrylate) composite microspheres by SPG (Shirasu
porous glass) emulsification technique,” J. Appl. Polym. Sci., 79: 2408−2424.
Ma, G.H., Omi, S., Dimonie, V.L., Sudol, E.D. and El-Aasser, M.S. (2002) “Study of the
preparation and mechanism of formation of hollow monodisperse polystyrene micro-
spheres by SPG (Shirasu Porous Glass) emulsification technique,” J. Appl. Polym. Sci.,
85: 1530−1543.
Ma, G.H., Sone, H. and Omi, S. (2004) “Preparation of uniform-sized polystyrene-polyacrylamide
composite microspheres from a W/O/W emulsion by membrane emulsification technique
and subsequent suspension polymerization,” Macromolecules, 37: 2954−2964.
Mazzei, R., Drioli, E. and Giorno, L. (2010) “Biocatalytic membrane reactor and membrane
emulsification concepts combined in a single unit to assist production and separation of
water unstable reaction products,” J. Membr. Sci., 352: 166−172.
Meng, T., Xie, R., Chen, Y.C., Cheng, C.J., Li, P.F., Ju, X.J. and Chu, L.Y. (2010) “A thermo-responsive
affinity membrane with nano-structured pores and grafted poly(N-isopropylacrylamide) sur-
face layer for hydrophobic adsorption,” J. Membr. Sci., 349: 258−267.
Meng, T., Xie, R., Ju, X.J., Cheng, C.J., Wang, S., Li, P.F., Liang, B. and Chun, L.Y. (2013)
“Nano-structure construction of porous membranes by depositing nanoparticles for
enhanced surface wettability,” J. Membr. Sci., 427: 63–72.
Mulder, M. (1996) Basic Principles of Membrane Technology, Dordrecht, the Netherlands:
Kluwer Academic Publishers.
Muramatsu, N. and Kondo, T. (1995) “An approach to prepare microparticles of uniform size,”
J. Microencapsulation, 2: 129−136.
Muramatsu, N. and Nakauchi, K. (1998) “A novel method to prepare monodisperse micropar-
ticles,” J. Microencapsulation, 15: 715−723.
Nakashima, T. (2002) “Porous glass material and its recent applications,” 38th International
SPG Forum on Membrane and Particle Science and Technology in Food and Medical
Care, November 21–22, Sadowara, Japan.
Nagashima, S., Ando, S., Tsukamoto, T., Ohshima, H. and Makino, K. (1998) “Preparation of
monodisperse poly(acrylamide-co-acrylic acid) hydrogel microspheres by a membrane
emulsification technique and their size-dependent surface properties,” Colloids Surf. B,
11: 47−56.
Nakashima, T. and Kuroki, Y. (1981) “Effect of composition and heat treatment on the phase
separation of NaO─B2O3─SiO2─Al2O3─CaO glass prepared from volcanic ashes,”
Nippon Kagaku Kaishi, 8: 1231−1238.
Nakashima T. and Shimizu M. (1986) “Porous glass from calcium alumino boro-silicate
glass,” Ceramics Japan, 21: 408−412.
Nakashima T., Shimizu M. and Kukizaki, M. (1991) “Membrane emulsification by micropo-
rous glass,” Key Eng. Mater., 61−62: 513−516.
Nakashima, T., Shimizu, M. and Kukizaki M. (1992) “Mechanical strength and thermal resis-
tance of porous glass,” J. Ceram. Soc. Jpn. Int. Ed., 100: 1389−1393.
Nakashima, T., Shimizu, M. and Kukizaki M. (1993) “Effect of surfactant on production of

monodispersed O/W emulsion in membrane emulsification,” Kagaku Kogaku Ronbun.,
19: 991−997.
Nakashima, T., Shimizu, M. and Kukizaki M. (1994) “Monodisperse single and double emul-
sions and method of producing same,” US Patent 5,326,484, July 5.
Nakashima, T., Shimizu, M. and Kukizaki, M. (2000) “Particle control of emulsion by mem-
brane emulsification and its applications,” Adv. Drug Deliv. Rev., 45: 47−56.
Nazir, A., Schroën, K. and Boom, R. (2010) “Premix emulsification: A review,” J. Membr.
Sci., 362: 1−11.
Nazir, H., Wang, L., Lian, G., Zhu, S., Zhang, Y., Liu, Y. and Ma, G. (2012) “Multilayered
silicone oil droplets of narrow size distribution: Preparation and improved deposition on
hair,” Colloids Surf. B, 100: 42−49.
Nuisin, R., Ma, G.H., Omi, S. and Kiatkamjornwong, S.J. (2000) “Dependence of morpho-
logical changes of polymer particles on hydrophobic/hydrophilic additives,” J. Appl.
Polym. Sci., 77: 1013−1028.
Oh, D.H., Balakrishnan, P., Oh, Y.K., Kim, D.D., Yong, C.S. and Choi, H.G. (2011) “Effect
of process parameters on nanoemulsion droplet size and distribution in SPG membrane
emulsification,” Int. J. Pharm., 404: 191–197.
Oh, D.H., Din, F.U., Kim, D.W., Kim, J.O., Yong, C.S. and Choi, H.G. (2013) “Flurbiprofen-
loaded nanoparticles prepared with polyvinylpyrrolidone using Shirasu porous glass
membranes and a spray-drying technique: Nano-sized formation and improved bioavail-
ability,” J. Microencapsulation, 30: 674–680, DOI:10.3109/02652048.2013.774447.
Okhonin, V., Petrov, A.P., Krylova, S.M. and Krylov, S.N. (2011) “Quantitative characterization of
micromixing based on uniformity and overlap,” Angew. Chem. Int. Ed., 50: 11999–12002.
Olson, F., Hunt, C.A. and Szoka, F.C. (1979) “Preparation of liposomes of defined size distribu-
tion by extrusion through polycarbonate membranes,” Biochim. Biophys. Acta, 557: 9−23.
Omi, S., Kanetaka, A., Shimamori, Y., Supsakulchai, A., Nagai, M. and Ma, G.H. (2001)
“Magnetite (Fe3O4) microcapsules prepared using a glass membrane and solvent
removal,” J. Microencapsulation, 19: 749−765.
Omi, S., Katami, K., Taguchi, T., Kaneko, K. and Iso, M. (1995) “Synthesis of uniform
PMMA microspheres employing modified SPG (Shirasu Porous Glass) emulsification
technique,” J. Appl. Polym. Sci., 57: 1013−1024.
Omi, S., Katami, K., Yamamoto, A. and Iso, M. (1994) “Synthesis of polymeric microspheres
employing SPG emulsification technique,” J. Appl. Polym. Sci., 51: 1−11.
Omi, S., Taguchi, T, Nagai, M. and Ma, G.H. (1997) “Synthesis of 100 μm uniform porous
spheres by SPG emulsification with subsequent swelling of the droplets,” J. Appl. Polym.
Sci., 63: 931−942.
Park, S.H., Yamaguchi, T. and Nakao, S. (2001) “Cut-off of dilute O/W emulsions through a
microfiltration membrane,” J. Membr. Sci., 190: 167−178.
Pawlik, A.K. and Norton, I.T. (2012) “Encapsulation stability of duplex emulsions prepared
with SPG cross-flow membrane, SPG rotating membrane and rotor-stator techniques—
A comparison,” J. Membr. Sci., 415−416: 459−468.
Pawlik, A.K. and Norton, I.T. (2013) “SPG rotating membrane technique for production of
food grade emulsions,” J. Food Eng., 114: 530−537.
Pham, T.T., Jaafar-Maalej, C., Charcosset, C. and Fessi, H. (2012) “Liposome and niosome
preparation using a membrane contactor for scale-up,” Colloids Surf. B, 94: 15−21.
Piacentini, E., Giorno, L., Dragosavac, M.M., Vladisavljević, G.T. and Holdich, R.G. (2013)
“Microencapsulation of oil droplets using cold water fish gelatine/gum arabic complex
coacervation by membrane emulsification,” Food Res. Int., 53: 362–372.
Pradhan, R., Lee, D.W., Choi, H.G., Yong, C.S. and Kim, J.O. (2013) “Fabrication of a uniformly
sized fenofibrate microemulsion by membrane emulsification,” J. Microencapsulation,
30: 42−48.
Rayner, M., Trägårdh, G. and Trägårdh, C. (2005) “The impact of mass transfer and interfacial
expansion rate on droplet size in membrane emulsification processes,” Colloids Surf. A,
266: 1−17.
Schröder, V., Behrend, O. and Schubert H. (1998) “Effect of dynamic interfacial tension on
the emulsification process using microporous, ceramic membrane,” J. Colloid Interface
Sci., 202: 334−340.
Segura, R., Cierpka, C., Rossi, M., Joseph, S., Bunjes, H. and Kähler, C. (2013) “Non-encapsulated
thermo-liquid crystals for digital particle tracking thermography/velocimetry in microfluid-
ics,” Microfluid. Nanofluid., 14: 445–456.
Seto, H., Ohto, K. and Kawakita, H. (2011) “Reversible extension and shrinkage of solvent-
responsive dextran chains produced by enzymatic reaction,” J. Membr. Sci., 370: 76−81.
Shimoda, M., Miyamae, H., Nishiyama, K., Yuasa, T., Noma, S. and Igura, N. (2011) “Swirl-
flow membrane emulsification for high throughput of dispersed phase flux through
Shirasu porous glass (SPG) membrane,” J. Chem. Eng. Jpn., 44: 1−6.
Si, T., Wang, Y., Wei, W., Lv, P., Ma, G. and Su, Z. (2011) “Effect of acrylic acid weight per-
centage on the pore size in poly(N-Isopropyl acrylamide-co-acrylic acid) microspheres,”
React. Funct. Polym., 71: 728–735.
Sugiura, S., Nakajima, M., Kumazawa, N., Iwamoto, S. and Seki, M. (2002) “Characterization
of spontaneous transformation-based droplet formation during microchannel emulsifi-
cation,” J. Phys. Chem. B, 106: 9405−9409.
Supsakulchai, A., Ma, G.H., Nagai, M. and Omi, S. (2002a) “Microencapsulation of fine tita-
nium dioxide powders from (S/O)/W emulsion with subsequent solvent evaporation,”
ACS Symp. Ser., 801: 260−275.
Supsakulchai, A., Ma, G.H., Nagai, M. and Omi, S. (2002b) “Uniform titanium dioxide (TiO2)
microcapsules prepared by glass membrane emulsification with subsequent solvent
evaporation,” J. Microencapsulation, 19: 425−449.
Surh, J., Jeong, Y.G. and Vladisavljević, G.T. (2008) “On the preparation of lecithin-stabilized
oil-in-water emulsions by multi-stage premix membrane emulsification,” J. Food Eng.,
89: 164−170.
Surh, J., Vladisavljević, G.T., Mun, S. and McClements, D.J. (2007) “Preparation and char-
acterization of water/oil and water/oil/water emulsions containing biopolymer-gelled
water droplets,” J. Agric. Food Chem., 55: 175−184.
Suzuki, K., Fujiki, I. and Hagura, Y. (1999) “Preparation of high concentration of O/W and
W/O emulsions by the membrane phase inversion emulsification using PTFE mem-
branes,” Food Sci. Technol. Int. Tokyo, 5: 234−238.
Suzuki, K., Shuto, I. and Hagura, Y. (1996) “Characteristics of the membrane emulsification
method combined with preliminary emulsification for preparing corn oil-in-water emul-
sions,” Food Sci. Technol. Int. Tokyo, 2: 43−47.
Tanaka, T., Okayama, M., Kitayama, Y., Kagawa, Y. and Okubo, M. (2010) “Preparation of
‘mushroom-like’ janus particles by site-selective surface-initiated atom transfer radical
polymerization in aqueous dispersed systems,” Langmuir, 26: 7843–7847.
Tasaki, T., Wada, T., Fujimoto, K., Kai, S., Ohe, K., Oshima, T., Baba, Y. and Kukizaki, M.
(2009), “Degradation of methyl orange using short-wavelength UV irradiation with
oxygen microbubbles,” J. Hazard. Mater., 162: 1103−1110.
Thorsen, T., Roberts, R.W., Arnold, F.H. and Quake, S.R. (2001) “Dynamic pattern formation
in a vesicle-generating microfluidic device,” Phys. Rev. Lett., 86: 4163−4166.
Torigoe, K., Shimizu, M., Yamamoto, K., Mizozoe, M., Takahashi, H., Suzuki, T. and Murase,
M. (2011) “Method and apparatus for manufacturing low melting point metal fine par-
ticles,” US patent 7,976,608, July 12.
Vladisavljević, G.T., Kobayashi, I. and Nakajima, M. (2012) “Production of uniform droplets
using membrane, microchannel and microfluidic emulsification devices,” Microfluid.
Nanofluid., 13: 151−178.
Vladisavljević, G.T., Kobayashi, I., Nakajima, M., Williams, R.A., Shimizu, M. and Nakashima,
T. (2007) “Shirasu Porous Glass membrane: Characterisation of microstructure by high
resolution x-ray microtomography and visualisation of droplet formation in real time,”
J. Membr. Sci., 302: 243−253.
Vladisavljević, G.T., Lambrich, U., Nakajima, M. and Schubert, H. (2004) “Production of
O/W emulsions using SPG membranes, ceramic α-Al2O3 membranes, microfluidizer
and a microchannel plate: A comparative study,” Colloids Surf. A, 232: 199−207.
Vladisavljević, G.T. and McClements, D.J. (2010) “Modification of interfacial characteristics of
monodisperse droplets produced using membrane emulsification by surfactant displace-
ment and/or polyelectrolyte electrostatic deposition,” Colloids Surf. A, 364: 123−131.
Vladisavljević, G.T. and Schubert, H. (2002) “Preparation and analysis of oil-in-water emul-
sions with a narrow droplet size distribution using Shirasu-porous-glass (SPG) mem-
branes,” Desalination, 144: 167−172.
Vladisavljević, G.T. and Schubert, H. (2003a) “Preparation of emulsions with a narrow par-
ticle size distribution using microporous α-alumina membranes,” J. Dispersion Sci.
Technol., 24: 811−819.
Vladisavljević, G.T. and Schubert, H. (2003b) “Influence of process parameters on droplet
size distribution in SPG membrane emulsification and stability of prepared emulsion
droplets,” J. Membr. Sci., 225: 15−23.
Vladisavljević, G.T., Shimizu, M. and Nakashima, T. (2004) “Preparation of monodisperse
multiple emulsions at high production rates by multi-stage premix membrane emulsifi-
cation,” J. Membr. Sci., 244: 97−106.
Vladisavljević, G.T., Shimizu, M. and Nakashima, T. (2005) “Permeability of hydrophilic and
hydrophobic Shirasu-porous-glass (SPG) membranes to pure liquids and its microstruc-
ture,” J. Membr. Sci., 250, 69−77.
Vladisavljević, G.T., Shimizu, M. and Nakashima, T. (2006) “Production of multiple emul-
sions for drug delivery systems by repeated SPG membrane homogenization: Influence
of mean pore size, interfacial tension and continuous phase viscosity,” J. Membr. Sci.,
284: 373−383.
Vladisavljević, G.T., Surh, J. and McClements, D.J. (2006) “Effect of emulsifier type on drop-
let disruption in repeated Shirasu porous glass membrane homogenization,” Langmuir,
22: 4526−4533.
Vladisavljević, G.T. and Williams, R.A. (2005) “Recent developments in manufacturing emul-
sions and particulate products using membranes,” Adv. Colloid Interface Sci., 113, 1−20.
Vladisavljević, G.T. and Williams, R.A. (2006) “Manufacture of large uniform droplets using
rotating membrane emulsification,” J. Colloid Interface Sci., 299: 396−402.
Vladisavljević, G.T. and Williams, R.A. (2008) “Recent developments in manufacturing
particulate products from double-emulsion templates using membrane and microflu-
idic devices,” in Aserin, A. (ed.), Multiple Emulsions: Technology and Applications,
Hoboken, NJ: John Wiley & Sons, Inc., pp. 121−164.
Vladisavljević, G.T. and Williams, R.A. (2010) “Recent developments in manufacturing
nanoparticles from emulsion droplets,” in Starov, V. (ed.), Nanoscience: Colloidal and
Interfacial Aspects, Boca Raton, FL: CRC Press, pp. 437–491.
Wang, G., Dou, H. and Sun, K. (2012) “Facile synthesis of hollow polymeric microparticles
possessing various morphologies via seeded polymerization,” Colloid Polym. Sci., 290:
1867−1877.
Wang, G., Leng, Y., Dou, H., Wang, L., Li, W., Wang, X., Sun, K. et al. (2013) “Highly effi-
cient preparation of multiscaled quantum dot barcodes for multiplexed hepatitis B detec-
tion,” ASC Nano, 7: 471−481.
Wang, L.Y., Ma, G.H. and Su, Z.G. (2005) “Preparation of uniform sized chitosan micro-
spheres by membrane emulsification technique and application as a carrier of protein
drug,” J. Controlled Release, 106: 62−75.
Wang, Y., Qin, J., Yi, W., Li, C. and Ma, G. (2013) “Preparation strategies of thermo-sensitive
P(NIPAM-co-AA) microspheres with narrow size distribution,” Powder Technol., 236:
107−113.
Wang, Y., Zhang, C., Bi, S. and Luo, G. (2010) “Preparation of ZnO nanoparticles using the
direct precipitation method in a membrane dispersion micro-structured reactor,” Powder
Technol., 202: 130−136.
Wei, W., Lv, P.P., Chen, X.M., Yue, Z.G., Fu, Q., Liu, S.Y., Yue, H. and Ma, G.H. (2013)
“Codelivery of mTERT siRNA and paclitaxel by chitosan-based nanoparticles promoted
synergistic tumor suppression,” Biomaterials, 34: 3912–3923.
Wei, W., Ma, G.H., Wang, L.Y., Wu, J. and Su, Z.G. (2010) “Hollow quaternized chitosan
microspheres increase the therapeutic effect of orally administered insulin,” Acta
Biomater., 6: 205–209.
Wei, Y., Wang, Y., Wang, L., Hao, D. and Ma, G.H. (2011) “Fabrication strategy for amphiphi-
lic microcapsules with narrow size distribution by premix membrane emulsification,”
Colloids Surf. B, 87: 399−408.
Wei, Q., Wei, W., Tian, R., Wang, L.Y., Su, Z.G. and Ma, G.H. (2008) “Preparation of uni-
form-sized PELA microspheres with high encapsulation efficiency of antigen by premix
membrane emulsification,” J. Colloid Interface Sci., 323: 267−273.
Wu, J., Wei, W., Wang, L.Y., Su, Z.G. and Ma, G.H. (2008) “Preparation of uniform-sized
pH-sensitive quaternized chitosan microsphere by combining membrane emulsification
technique and thermal-gelation method,” Colloids Surf. B, 63: 164−175.
Xu, J.H., Luo, G.S., Chen, G.G. and Tan, B. (2005) “Mass transfer performance and two-phase
flow characteristic in membrane dispersion mini-extractor,” J. Membr. Sci., 249: 75−81.
Yamamoto, T., Ohmori, T. and Kim, Y.H. (2010) “Synthesis of monodisperse carbon cryo-
gel microspheres using membrane emulsification of a phenol–formaldehyde solution,”
Carbon, 48: 912−928.
Yamashita, N., Konishi, N., Tanaka, T. and Okubo, M. (2012) “Preparation of hemispherical
polymer particles by cleavage of a Janus poly(methyl methacrylate)/polystyrene com-
posite particle,” Langmuir, 28: 12886−12892.
Yang, J., Hao, D.X., Bi, C.X., Su, Z.G., Wang, L.Y. and Ma, G.H. (2010) “Rapid synthesis of
uniform magnetic microspheres by combing premix membrane emulsification and in
situ formation techniques,” Ind. Eng. Chem. Res., 49: 6047–6053.
Yasuda, M., Goda, T., Ogino, H., Glomm, W.R. and Takayanagi, H. (2010) “Preparation of
uniform monomer droplets using packed column and continuous polymerization in tube
reactor,” J. Colloid Interface Sci., 349: 392−410.
Yobas, L., Martens, S., Ong, W.L. and Ranganathan, N. (2006) “High-performance flow-focusing
geometry for spontaneous generation of monodispersed droplets,” Lab Chip, 6: 1073−1079.
You, J.O., Park, S.B., Park, H.Y., Haam, S., Chung, C.H. and Kim, W.S. (2001) “Preparation
of regular sized Ca-alginate microspheres using membrane emulsification method,”
J. Microencapsulation, 18: 521−532.
Yue, H., Wei, W., Fan, B., Yue, Z., Wang, L., Ma, G. and Su, Z. (2012) “The orchestration of
cellular and humoral responses is facilitated by divergent intracellular antigen traffick-
ing in nanoparticle-based therapeutic vaccine,” Pharmacol. Res., 65: 189−197.
Yue, Z.G., Wei, W., Lv, P.P., Yue, H., Wang, L.Y., Su, Z.G. and Ma, G.H. (2011) “Surface
charge affects cellular uptake and intracellular trafficking of chitosan-based nanopar-
ticles,” Biomacromolecules, 12: 2440–2446.
Yuyama, H., Hashimoto, T., Ma, G.H., Nagai, M. and Omi, S. (2000) “Mechanism of suspen-
sion polymerization of uniform monomer droplets prepared by glass membrane (Shirasu
Porous Glass) emulsification technique,” J. Appl. Polym. Sci., 78: 1025−1043.
Yuyama, H., Yamamoto, K., Shirafuji, K., Nagai, M. and Ma, G.H. (2000) “Preparation of
polyurethaneurea (PUU) uniform spheres by SPG membrane emulsification technique,”
J. Appl. Polym. Sci., 77: 2237−2245.
Zhang, Y., Wei, W., Lv, P., Wang, L. and Ma, G. (2011) “Preparation and evaluation of alginate–
chitosan microspheres for oral delivery of insulin,” Eur. J. Pharm. Biopharm., 77: 11−19.
Zhou, Q.Z., Ma, G.H. and Su, Z.G. (2009) “Effect of membrane parameters on the size and
uniformity in preparing agarose beads by premix membrane emulsification,” J. Membr.
Sci., 326: 694–700.
Zhou, Q.Z., Wang, L.Y., Ma, G.H. and Su, Z.G. (2007) “Preparation of uniform-sized agarose
beads by microporous membrane emulsification technique,” J. Colloid Interface Sci.,
311: 118−127.
Zhou, Q.Z., Wang, L.Y., Ma, G.H. and Su, Z.G. (2008) “Multi-stage premix membrane emul-
sification for preparation of agarose microbeads with uniform size,” J. Membr. Sci., 322:
98–104.
Zhou, Q., Zhang, M.C., Shuang, C.D., Li, Z.Q. and Li, A.M. (2012) “Preparation of a novel
magnetic powder resin for the rapid removal of tetracycline in the aquatic environment,”
Chin. Chem. Lett., 23: 745–748.
Zúñiga, R.N. and Aguilera, J.M. (2008) “Aerated food gels: Fabrication and potential applica-
tions,” Trends Food Sci. Technol., 19: 176−187.
van der Zwan, E.A., Schroën, C.G.P.H. and Boom, R.M. (2008) “Premix membrane emulsifi-
cation by using a packed layer of glass beads,” AIChE J., 54: 2190−2197.
Food and Culinary Science
RAYN E R • DE JME K
Emulsions are found in a wide variety of food products, pharmaceuticals, paints,
and cosmetics, thus emulsification is a truly multidisciplinary phenomenon.
Therefore, understanding of the process must evolve from the combination of (at
Engineering Aspects
least) three different scientific specializations. Engineering Aspects of Food
Emulsification and Homogenization describes the state-of-the-art technology
E N G I N E E R I N G A S P E C T S O F F O O D E M U L S I F I C AT I O N A N D H O M O G E N I Z AT I O N
and brings together aspects from physical chemistry, fluid mechanics, and
chemical engineering. The book explores the unit operations used in emulsifica-
of
tion and homogenization processes, using fundamental theory from different fields
to discuss design and function of different emulsification techniques.
This book summarizes the present understanding of the involved physical–chemi-

cal processes as well as specific information about the limits and possibilities for
the different types of emulsifying equipment. It covers colloidal chemistry and Food Emulsification
engineering aspects of emulsification and discusses high-energy and low-energy
and
Homogenization
emulsification methods. The chapters highlight low-energy emulsification process-
es such as membrane emulsification that are now industrially feasible. Dramatical-
ly more energy-efficient processes are being developed, and this book clarifies
their present limitations, such us scale-up and achievable droplet sizes.
Features
• Describes state-of the-art technology of emulsification and homogenization

processes
• Brings together aspects from physical chemistry, fluid mechanics, and
chemical engineering
• Presents formulation aspects of emulsions with respect to stability
and function
• Brings together fundamental theory from different fields to discuss
design and function of different emulsification techniques
• Compares high-energy, low thermodynamic efficiency methods
with alternative low-energy, higher-efficiency processes
The present literature on emulsification is, to a large degree, influenced by the

division between physical chemistry, fluid dynamics, and chemical engineering.
Written by experts drawn from academia and industry, this book brings those
areas together to provide a comprehensive resource that gives a deeper under-
standing of emulsification and homogenization in food product development.
EDITED BY
K16909
Marilyn R ayner
ISBN: 978-1-4665-8043-5
90000 P et r D ejm ek
9 781466 580435

Engineering Aspects of Food Emulsification and Homogenization

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

Engineering Aspects of Food Emulsification and Homogenization

Uploaded by

Copyright:

Available Formats

Food and Culinary Science

This book summarizes the present understanding of the involved physical–chemi-

• Describes state-of the-art technology of emulsification and homogenization

The present literature on emulsification is, to a large degree, influenced by the

Engineering Aspects of Food Emulsification and Homogenization,

Boca Raton London New York

CRC Press is an imprint of the

No claim to original U.S. Government works

International Standard Book Number-13: 978-1-4665-8044-2 (eBook - PDF)

For our families, teachers, and mentors.

Section i emulsion Fundamentals

Chapter 1 Scales and Forces in Emulsification .....................................................3

Chapter 2 Emulsion Formation and Instability ................................................... 33

Chapter 3 Formulation of Emulsions .................................................................. 51

Chapter 4 Particle-Stabilized Emulsions .......................................................... 101

Section ii High-energy Processes

Chapter 5 Droplet Breakup in High-Pressure Homogenizers .......................... 125

Chapter 6 High-Pressure Homogenizer Design ................................................ 149

Chapter 7 High-Pressure Homogenization with Microstructured Systems ...... 169

Chapter 8 Rotor–Stator Devices ....................................................................... 195

Section iii Low-energy Processes

Chapter 9 Microchannel Emulsification: Aspects of Droplet Generation,

Chapter 10 Emulsification with Microsieves and Other Well-Defined

Chapter 11 Formation and Modification of Dispersions Using Shirasu

EMULSIONS AND EMULSIFICATION

For micrometer-sized emulsion drops, this would amount to an order of magnitude of

THE EMULSIFICATION PROCESS

A comprehensive discussion on emulsification methods and their application must

THE NEED FOR NEW LITERATURE ON EMULSIFICATION

Petr Dejmek, a native Czech, graduated as a process

Björn Bergenståhl Lars Nilsson

Andreas Håkansson Marilyn Rayner

Fredrik Innings Karin Schroën

SECTION II—HIGH-ENERGY PROCESSES

A novel, more efficient homogenizer design suitable for partial homogenization is

SECTION III—LOW-ENERGY PROCESSES

1.2 droplet sIze dIstrIbutIons

Premixer Emulsification machine

4. Mechanical energy Intense mechanical energy input

Premixing— Emulsification in the droplet disruption zone— Stabilization of new interface—

1. Continuous phase 4. Energy input for

Slow stabilization or deficient

2. Dispersed phase 4. Energy input to press dispersed

studies on formulation, production, and evaluation of emulsion focus on droplet size

Thus, any type of average dab is given by

Furthermore, in many texts, the coefficient of variation (CV) of a given mean as a

1.3 overvIew oF emulsIFIcatIon machInes

FIGure 1.2 Schematic illustration of various designs of emulsification machines and

Inlet Dispersed phase is pressed through the

FIGure 1.2 (Continued ) Schematic illustration of various designs of emulsification

(e.g., Ultraturrax) 103–108 several m3

1.4 droplet dIsruptIon Fundamentals

successively higher concentrations of surfactant to lower the interfacial tension. In

1.4.1 InterfacIal forces

1.4.2 DIsruptIve forces

Cross section of oil in a pore where

Droplet necking inside pore

(c) Droplet breakup in spontaneous transformation based droplet formation

1.4.3 flow types

inertial forces ULρ

1.5 Flow reGImes

Droplet deformation timescale (τDEF ) ηd /(ηc G) ηd / ε ηc ηd / 3 ε 2 d 2 ρc

Surfactant adsorption timescale (τADS ) (6 π Γ)/(d mc G) [(6 π Γ)/(d mc )] ( ηc /ε ) (Γ/mc )  3 ρc /(d ε) 

1.5.1 lv flow regIme