Desen Produto

MECHANICAL ENGINEERING THEORY AND APPLICATIONS
MECHANICAL DESIGN,
MATERIALS AND
MANUFACTURING
No part of this digital document may be reproduced, stored in a retrieval system or transmitted in any form or
by any means. The publisher has taken reasonable care in the preparation of this digital document, but makes no
expressed or implied warranty of any kind and assumes no responsibility for any errors or omissions. No
liability is assumed for incidental or consequential damages in connection with or arising out of information
contained herein. This digital document is sold with the clear understanding that the publisher is not engaged in
rendering legal, medical or any other professional services.
MECHANICAL ENGINEERING THEORY
AND APPLICATIONS
Additional books and e-books in this series can be found on Nova’s

website under the Series tab.
MECHANICAL ENGINEERING THEORY AND APPLICATIONS
MECHANICAL DESIGN,
MATERIALS AND
MANUFACTURING
SANDIP A. KALE
EDITOR
Copyright © 2019 by Nova Science Publishers, Inc.
All rights reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted
in any form or by any means: electronic, electrostatic, magnetic, tape, mechanical photocopying,
recording or otherwise without the written permission of the Publisher.
We have partnered with Copyright Clearance Center to make it easy for you to obtain permissions to
reuse content from this publication. Simply navigate to this publication’s page on Nova’s website and
locate the “Get Permission” button below the title description. This button is linked directly to the
title’s permission page on copyright.com. Alternatively, you can visit copyright.com and search by
title, ISBN, or ISSN.
For further questions about using the service on copyright.com, please contact:
Copyright Clearance Center
Phone: +1-(978) 750-8400 Fax: +1-(978) 750-4470 E-mail: info@copyright.com.
NOTICE TO THE READER

The Publisher has taken reasonable care in the preparation of this book, but makes no expressed or
implied warranty of any kind and assumes no responsibility for any errors or omissions. No liability is
assumed for incidental or consequential damages in connection with or arising out of information
contained in this book. The Publisher shall not be liable for any special, consequential, or exemplary
damages resulting, in whole or in part, from the readers’ use of, or reliance upon, this material. Any
parts of this book based on government reports are so indicated and copyright is claimed for those parts
to the extent applicable to compilations of such works.
Independent verification should be sought for any data, advice or recommendations contained in this
book. In addition, no responsibility is assumed by the publisher for any injury and/or damage to
persons or property arising from any methods, products, instructions, ideas or otherwise contained in
this publication.
This publication is designed to provide accurate and authoritative information with regard to the subject
matter covered herein. It is sold with the clear understanding that the Publisher is not engaged in
rendering legal or any other professional services. If legal or any other expert assistance is required, the
services of a competent person should be sought. FROM A DECLARATION OF PARTICIPANTS
JOINTLY ADOPTED BY A COMMITTEE OF THE AMERICAN BAR ASSOCIATION AND A
COMMITTEE OF PUBLISHERS.
Additional color graphics may be available in the e-book version of this book.
Library of Congress Cataloging-in-Publication Data

Names: Kale, Sandip A., editor.
Title: Mechanical design, materials and manufacturing / editor: Sandip A. Kale
(Mechanical Engineering Department, Trinity College of Engineering
and Research, Pune, India; Technology Research and Innovation Centre, Pune, India).
Description: Hauppauge, New York: Nova Science Publishers, Inc., [2019] |
Series: Mechanical engineering theory and applications | Includes bibliographical references and index.
Identifiers: LCCN 2018060478 (print) | LCCN 2019000032 (ebook) | ISBN
9781536147926 (ebook) | ISBN 9781536147919 (hardcover) | ISBN 9781536147926 (Ebook)
Subjects: LCSH: Machinery. | Materials.Classification: LCC TJ7 (ebook) | LCC TJ7 .M378 2019 (print) | DDC
621.8--dc23 LC record available at https://lccn.loc.gov/2018060478
Published by Nova Science Publishers, Inc. † New York

CONTENTS
Preface vii
Chapter 1 Finite Element Analysis of Three DoF Strain
Gauge Force Transducer 1
Ankur Jaiswal, H. P. Jawale and Atul Sane
Chapter 2 The Basic Principle of Calculation and Analysis of
the Defective Structure of Solids 17
V. I. Talanin, I. E. Talanin, V. V. Zhdanova,
D. I. Yakymchuk and A. V. Rybalko
Chapter 3 Advanced Methods Used in Molecular Dynamics
Simulation of Macromolecules 57
Hiqmet Kamberaj
Chapter 4 Study on Error in Sigmoidal Function Generation
of 4R Mechanism 135
Ankur Jaiswal and H. P. Jawale
Chapter 5 Identifying Transfer Vertex from the Adjacency
Matrix for Epicyclic Gear Trains 153
Mallu Chengal Reddy, Rudraraju Manish
and Yendluri Daseswara Rao
vi Contents
Chapter 6 3-D Simulation Studies on Strengthening of

Beam-Column Joint 165
Khan Mohammad Firoz, Garg Aman
and H. D. Chalak
Chapter 7 Nanomaterials, Ceramic Bulk and Bioceramics:
Synthesis, Properties and Applications 175
H. H. M. Darweesh
Chapter 8 Simulation of Low Cost Automation and
Life Cycle Cost Analysis for a
Special Purpose Machine 263
Pattanayak Satyajit
and Hauchhum Lalhmingsanga
Chapter 9 Estimation of Axial Force in
Incremental Sheet Metal Forming 291
Bohra Murtaza and Y. V. D. Rao
About the Editor 303
Index 305
PREFACE
Design and manufacturing processes are the important aspects of all

industrial sectors. Though, the developments in the field of electronics and
digital industries are significant, the importance of the basic mechanical
industry remains always on the top side. The mechanical design, materials
and manufacturing are noteworthy areas of research at any time. The
purpose of this book is to present some advanced research studies on
mechanical design, materials and manufacturing.
First five chapters discuss some important design methods, tools and
techniques for various mechanical engineering applications. The first
chapter presents an analysis of a novel force transducer which has a special
shape that allows strategic placement of the strain gauges and senses axial
forces by ignoring the moments. The strain developed in all the three
directions of load application is estimated by finite element analysis.
The combination of physical research with the tools and means of
modern information technologies leads to the creation of a new method for
verifying the validity of theoretical constructions. In the field of physics
and material science of silicon, this approach makes it possible to use it for
the development of the foundations of defect structure management. The
second chapter explains the basic principle of calculation and analysis of
the defective structure of solids. The theoretical models of the physical
viii Sandip A. Kale
process of solid state physics, material science, programming and the

creation of material and devices based on it are combined in this work.
The third most interesting chapter presents advanced methods used in
molecular dynamics simulation of macromolecules. These methods include
generalized ensemble methods, metadynamics method, umbrella sampling
methods, transition path sampling methods, accelerated molecular
dynamics method, conformational flooding method. This work also
proposes new possible improvement of these methods, which could result
in further enhancement of conformation sampling.
Chapter Four explores an extended method of mathematical modelling
of Freudenstein-Chebyshev approximation theory for sigmoidal function
applied to four and five precision points. The sigmoidal function has been
analyzed based on the structural error between the generated function and
desired function. The generated sigmoidal function is analyzed and the
structural parameters have been obtained.
Chapter Five presents an algorithm to find the transfer vertex of a
given epicyclic gear train (EGT). The algorithm used searches for gear pair
and finding the transfer vertex through corresponding turning pairs. It also
identifies cases where more than one transfer vertex, are possible. This
algorithm reduces the number of EGTs to be checked for rotational
isomorphism after displacement isomorphism.
The use of composite fibers is one of the interesting research areas in
all industrial fields. In the sixth chapter, an analytical study (using
ABAQUS/CAE) on the strengthening of beam-column joint under seismic
conditions using carbon fibre reinforced polymer (CFRP) sheets has been
carried out to draw some important results.
Chapter Seven discusses the hottest and most interesting topic covers
recent information about preparation, properties and applications of
nanomaterials, ceramics and bioceramics. It almost incorporates into all
applied science branches as physics, chemistry, materials sciences,
biology, agriculture, medicine, tissue engineering, bones, caffolds, dentists,
cement and concrete and all types of building materials particularly bricks,
ceramic and nanoceramics or in general advanced ceramics industries,
biomaterials and many others.
Preface ix
The manufacturing industry is always focusing to reduce the

production cost without compromising the quality. The simulation of low
cost automation and life cycle cost analysis for special purpose machine is
a major area of research in industry. Chapter Eight presents a methodology
to design, develop and simulate a twin spindle turning special purpose
machine based on the data collected from hydraulic, pneumatic, electro
pneumatic which will serve as low cost automation. The design is based on
low cost automation principles rather than pneumatic and programmable
logic controller principles. The last chapter is about the estimation of axial
force in incremental sheet metal forming.
I believe the use of this book for the researchers and students engaged
in the field of mechanical engineering and material science. I am thankful
to authors for contributing in this book and reviewers for valuable reviews
and suggestions.
Sandip A. Kale
Editor
In: Mechanical Design, Materials … ISBN: 978-1-53614-791-9
Editor: Sandip A. Kale © 2019 Nova Science Publishers, Inc.
Chapter 1
FINITE ELEMENT ANALYSIS OF THREE DOF

STRAIN GAUGE FORCE TRANSDUCER
Ankur Jaiswal, H. P. Jawale* and Atul Sane

Department of Mechanical Engineering,
Visvesvaraya National Institute of Technology, Nagpur, India
ABSTRACT
Strain Gauge Force transducers are the instruments deployed to

measure the magnitude of force applied on the body. This chapter
presents an analysis of a novel force transducer which has a special shape
that allows strategic placement of the strain gauges. The arrangement
facilitates to sense only axial forces and ignore the moments, simplifying
the measurements and calculations. The geometry of the transducer is
conceptualized with three beams on a frame like structure, such that each
of the frame independently acts as a single degree of freedom cantilever
beam. Due to the property of isotropic material and single body structure,
it is having a proportional force conversion. The strain developed in an
individual frame is estimated by Finite Element Analysis of a CAD
model of the transducer. The behaviour of the transducer for uni-axial and
tri-axial loadings is presented.

Corresponding Authors Emails: jaiswalankur13@gmail.com; hpjawale@rediffmail.com.
2 Ankur Jaiswal, H. P. Jawale and Atul Sane
Keywords: force transducers, finite element analysis, strain gauges
1. INTRODUCTION
Strain gauge based force and torque sensors are widely used in
measurement technology due to their stability at room temperature and
linear static characteristics. In such applications strain gauges are mounted
on the primary sensing elements of a desired cross-section. The applied
force or torque gets converted into displacement making the strain gauge to
change the resistance which is proportional to the applied force or torque.
There are different configurations of primary sensing elements that are
devised and calibrated to give desired sensitivity. However, inherently
these configurations also have moment components in supplement to the
force component. This makes the conversion of resistance into the readable
electrical signals complex. Additionally, due to the cross sensitivity,
analysis becomes tedious.
The performance of the strain gauge is a function of all other elements
like the position of the gauge, gauge factor, input voltage applied, bridge
connection and other circuitry to estimate the strain in terms of voltage.
Thus the location of dipicted strain gauges directly affects the
performance. Measurement of externally applied forces can be better
estimated with the accurate positioning of the strain gauge on the
transducer at the place of maximum strain.
The six-axis force sensor with four T-shaped bars and wrist force
sensor is seen to be analyzed using the FEM and analytical method. The
optimized design is used for high measurement sensitivities application
(Chao and Chen, 1997, Liu and Tzo, 2002). This chapter elaborates a new
frame/truss type of sensor body design. The unified criteria are based on
static and dynamic stiffness of the sensor and sensitivity for strain gauge.
The results have been compared with the previous design (Bayo and
Stubbe, 1989) on the basis of rigidity, force sensitivity and flexibility
parameters.
Finite Element Analysis of Three DoF Strain Gauge … 3
The static and kinematic equations are developed through a

compliance matrix form of the model. The developed model is analyzing to
the two- and three-dimensional sensor applications for regular polygons
and polyhedrons (Svinin and Uchiyama, 1995). The six-axis force sensor is
found to be developed with the combination of two same three-axis force
sensors is idealized, describing the theoretical analysis, providing the
characteristic matrix connecting the load and sensor output vectors
(Kaneko, 1993) Stewart platform is seen to be used for force sensing,
where derived closed-form solution of forward kinematics is applied for
analysis of force and torque 6-axis transducers by obtaining force and
torque acting on the upper plate from six leg forces of the parallel platform
(Kang, 2001). The theoretical and Finite Element Method (FEM) analysis
for six-component force moment sensor (Kang, 2001)is presented. The
comparative analysis of rated strains in FEM analysis, analytical equations
and experiments (Kim, et al., 1999)is dealt in. Further, compact force–
torque sensor is designed and developed for Stewart platform structure for
light weight capabilities of sensing elements applications (Hirose and
Yoneda, 1990). This chapter presents the parallel loads sharing principle of
six-dimensional heavy force/torque. Analyzing the results for multi-
dimensional time-varying heavy load conditions of dynamic measurement
problems and comparing the FEM analysis (Liu, et al., 2011), Further,
design and kinematic analysis of compliant parallel mechanism (CPM) for
six-component Force/Torque (F/T) sensor is seen to be worked out (Liang,
et al., 2013), The Design of Experiments (DOE) technique is used to
analyzed and compare the Finite Element Analysis (FEA) results. The
design and optimization of a novel six-axis force/torque sensor based on
strain gauges for space robot is elaborated in (Sun, et al., 2015). The
geometry parameters of elastic body are optimized by response surface
methodology (RSM) and validating the results by FEM. The literature
presented work on force and the torque sensors which were to be mounted
on the robotic manipulator, being carried with atm. The complexity is
involved due to which the accuracy is limited A novel approach of
mounting sensors in stationary platform is developed along with
calibration and testing of three axis force Sensor is presented in

(Deshpande, et al., 2016).
Present work in this chapter is extension of the three axis force sensor
analysis sited above. The difficulty found with this work is tapping of
unequal strains due to improper locations for depicting strain gauges. Thus,
it was essential to estimate strains exactly for depiction of gauges. The
estimations of stresses and strains for exiting compact transducer is the
motive for work in this chapter. Also it has led to modification of
transducer design followed by comparative analysis as presented herewith.
2. STRAIN GAUGE BASED FORCE SENSING
Force Transducers are devices which provide an observer the

numerical value corresponding to the applied force or torque on the body,
which ideally should be equal to the true value. The output of a transducer
is dependent upon the predetermined relationship between the measured
value and the input parameter being measured. Force Transducers usually
use an elastic load bearing element or their combination. Due to
application of force there is a deflection in the member and that deflection
is measured by the strain gauges which convert it into a measurable output.
The output of a strain gauge is in the form of an electric signal with current
varying according to the magnitude of applied force (Stefanescu, 2011).
In recent days, there are many methods which are used to measure
torque or force applied to any object, among those methods Strain gauges
are the most widely used. The Strain gauge is capable of distinguishing
tensile and compressive strains by a positive or negative sign. Thus it’s
possible to detect expansion as well as contraction of the objects on which
strain gauges are attached. The strain gauges are strategically mounted on
the force sensing element which experiences a change in resistance which
is induced by the deformation of the sensor. The Sensitivity of the strain
gauge is decided by a fundamental factor known as the gauge factor. It is
the ratio of the relative change in electrical resistance R with respect to the
mechanical strain ε.
L
R
A (1)
R / R R / R
GF  
L / L  (2)
where,
ρ: Resistivity/Conductivity constant
L: Length of the sensor
A: Cross sectional area of sensor
R: Electrical resistance of the sensor
ε: Mechanical Strain
The strain gauge cell is based on an elastic element to which a number

of electrical resistance strain gauges are bonded. The strain produce due to
application of force is dependent on the geometric shape as well as the
modulus of elasticity of that element. Each strain gauge responds to the
local strain at its location, and the measurement of force is determined
from a combination of these individual measurements of strain (Yurish,
2014) (Yurish, 2014).
In actual life scenarios forces are usually not unidirectional. Mostly
applied forces are multi-directional and dynamic in nature, thus making it
difficult for a single axis sensor to detect applied load. This situation is
very well addressed by multi-axis Force transducers which convert input
force/torque signals to voltage signals; serve to perceive external
force/torque information of different dimensions. They are highly
applicable in scenarios including Robotic control and manipulation,
aerospace and industries (Chen, et al., 2015).
3. TRANSDUCER DESIGN
This chapter discusses a novel three DOF force transducer which has a
special shape that allows strategic placement of the strain gauges in Figure
1. Because of such arrangements they only sense axial forces and ignore
the moments. This makes measurements and calculations very simple. The
simplest primary sensing elements of any force sensing transducer is
generally similar to a cantilever configuration. Combining three cantilevers
for each of the axis of Cartesian coordinate system in a single
configuration can function as a three DOF transducer for force sensing.
This design felicitates to study the application and results of uniaxial,
biaxial and tri-axial force loading with the means of pulleys and dead
weights directly on the transducer.
Figure 1. A CAD model describing the Force Transducer (green), Pulleys and dead
weight for loading (brown).
(a) (b)
Figure 2. CAD model depicting the complete assembly of FT1 and FT2.
There is also a new proposed design for the same transducer having
some alterations to the frame and the members so that the applied force on
the same region is more easily distributed in Figure 2. This helps in better
stress dispersion.
3.1 Converting of Cartesian Coordinates
The spherical coordinates of a point (r, θ, φ) can be obtained from its

Cartesian coordinates in Figure 3 (x, y, z) by the equations given below.
2
r x y z
2 2
(3)
 
 y   y
  a cos   a cos 
 2 2  r
x y z 
2
 (4)
 x
  a tan 
z (5)
Figure 3. Spherical Co-ordinate System.

Conversely, the Cartesian coordinates may be retrieved from the

spherical coordinates (radius r, inclination θ, azimuth φ), where r ∈ [0, ∞],
θ ∈ [0, π], φ ∈ [0, 2π], by:
x  r sin  sin  (6)
y  r cos (7)
z  r sin  cos  (8)
4. FINITE ELEMENT ANALYSIS
4.1 Material Properties
The material considered for design and analysis is Structural steel with
following in Table 1.
Table 1. Material Properties of Structural Steel
Young’s Modulus 200 GPa

Poisson's Ratio 0.3
Density 7850 kg/ m3
4.2 Meshing
A ten node quadratic tetrahedron element C3D10 is used for meshing

both the types of transducers. The Force Transducer (FT1) type model
contains 6686 elements and FT2 type model contains 10100 elements.
Figure 4 (a) and 4 (b) shows the meshed model for FT1 and FT2 type of
transducers respectively.
4.3 Loading Conditions
The base of the transducer is made fixed support in order to replicate

actual real life scenario of the transducer firmly placed on a solid base.
This improvises the results obtained by the FEM analysis on ABAQUS.
Loading is applied on the topmost surface of the transducers as shown in
Figure 5. The applied loads implies to a resultant force which is in range of
0.2 to 2 N. The loads are applied in the simulation packages to find out the
components of this force in Cartesian coordinate system. These
components (Fx, Fy and Fz) were calculated using equation (6-8).
(a) FT1 (b) FT2
Figure 4. Final Meshing of the geometry for the analysis of FT1 and FT2.
(a) FT1 (b) FT2
Figure 5. Loading conditions on FT1 and FT2.

5. RESULTS
The stresses and strains in the members of the transducer were

calculated by applying ten different sets of loading conditions ranging from
0.2 N to 2 N in x, y and z directions. The calculated strains for FT1 and
FT2 are shown in Table 2 and Table 3 respectively. The loads were applied
on the hemispherical region on the top of the transducer and the base link
of the transducer was fixed to replicate test conditions as shown in
Figure 5.
Figure 6 (a) and 6 (b) shows the maximum principal strain contour on
the different links of transducer FT1 and FT2 respectively, at a load of 0.2
N on all the three directions. Figure 6 (c) and 6 (d) shows the maximum
principal strain contour on the different links of transducer FT1 and FT2
respectively, at a load of 1 N on all the three directions Figure 6 (e) and 6
(f) shows the maximum principal strain contour on the different links of
transducer FT1 and FT2 respectively, at a load of 2 N on all the three
directions.
Table 2. Stress values from ABAQUS for Maximum Principal Elastic

Strain in FT1
load (N) Maximum Principal Strain (*e^-5) for FT1

x y z xyz
0.2 1.513 0.459 0.594 0.829
0.4 3.026 0.918 1.189 1.645
0.6 4.551 1.381 1.788 2.474
0.8 6.068 1.844 2.38 3.29
1 7.587 2.32 2.983 4.113
1.2 9.106 2.767 3.584 4.954
1.4 10.625 3.229 4.171 5.755
1.6 12.144 3.681 4.77 6.569
1.8 13.663 4.143 5.366 7.41
2 15.182 4.61 5.959 8.218
(a) (b)
(c) (d)
(e) (f)
Figure 6. (a-f) Principal Strain distribution for TF1 and TF2.

Table 3. Stress values from ABAQUS for Maximum Principal Elastic

Strain in FT2
load (N) Maximum Principal Strain (*e^-5) for FT2

x y z xyz
0.2 1.551 0.59 0.619 1.194
0.4 3.101 1.181 1.238 2.371
0.6 4.664 1.777 1.862 3.565
0.8 6.215 2.368 2.481 4.745
1 7.766 2.958 3.1 5.93
1.2 9.316 3.549 3.719 7.121
1.4 10.87 4.14 4.338 8.299
1.6 12.43 4.735 4.962 9.485
1.8 13.98 5.326 5.581 10.6
2 15.53 5.917 6.2 11.84
6. OBSERVATIONS
1. When load is applied in any one direction (uni-axial loading),

maximum strain is obtained in the member corresponding to the
respective direction. Also about 5-10% of the reaming strain is
observed to be distributed to other two directional members.
2. Under tri-axial loading conditions maximum strain occurs in the
bottom-most member, moderate in the middle member and least in
the uppermost member.
3. The transducer exhibits a linear relationship for loading and
induced strain in both FT1 and FT2 type of transducers as shown
in Figure 7. This is the most favourable condition for this
transducer as a primary sensing element for force measurement.
Due to this property the hysteresis errors would be eliminated.
However, there is a minute difference in the calculated strain
between the FT1 and FT2 transducers.
4. The difference between induced strain in FT1 and FT2 also

increases linearly as shown in Figure 8, and it is maximum in case
of tri-axial loading scenario due to cross sensitivity of the
members.
5. Under the same loading conditions, it is observed that the middle
member has the least magnitude of the strain induced.
16
Force in X- direction (FT1)
Maximum Principal Strain (*e5)
14 Force in X- direction (FT2)

-
Force in Y- direction (FT1)

12 Force in Y- direction (FT2)
Force in Z- direction (FT1)
10 Force in Z- direction (FT2)
Force in XYZ- direction (FT1)
8 Force in XYZ- direction (FT2)
0
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Load (N)
Figure 7. Maximum Principal Strain V/s Applied Load.
4
X direction
3.5
Y direction
3 Z direction
Strain Difference
XYZ diection
2.5
1.5
0.5
0
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Load (N)
Figure 8. Strain difference curve between TF1 and TF2.

CONCLUSION
Estimation of strains in a compact force transducer for application of

platform type sensor is presented herewith. The observations have led to
modifications in the transducer profile so that the strains are concentrated
at a specific location. The comparative strain estimations are also presented
which has facilitated in designing and manufacturing of the transducer.
The strain profile is used for finalizing the location of gauge pasting. It
helped in successful theoretical estimators of strains in terms of output
voltage of the connecting bridge circuit and further calibrations.
The future scope lies in the demonstration of the effectiveness of this
platform type force sensor automated assembly task.
REFERENCES
Bayo, Eduardo, and JR Stubbe. (1989), Six‐Axis Force Sensor Evaluation

and a New Type of Optimal Frame Truss Design for Robotic
Applications, Journal of Field Robotics 6 (2), 191-208.
Chao, Lu-Ping, and Kuen-Tzong Chen. (1997), Shape Optimal Design and
Force Sensitivity Evaluation of Six-Axis Force Sensors, Sensors and
Actuators A: Physical 63 (2), 105-12.
Chen, Danfeng, Aiguo Song, and Ang Li. (2015), Design and Calibration
of a Six-Axis Force/Torque Sensor with Large Measurement Range
Used for the Space Manipulator, Procedia Engineering 99 (1), 1164-
70.
Deshpande, MS, HP Jawale, and HT Thorat. “Development, Calibration
and Testing of Three Axis Force Sensor.” Paper presented at the
Mechanical and Aerospace Engineering (ICMAE), 2016 7th
International Conference on, 2016.
Hirose, Shigeo, and Kan Yoneda. “Development of Optical Six-Axial
Force Sensor and Its Signal Calibration Considering Nonlinear
Interference.” Paper presented at the Robotics and Automation, 1990.

Proceedings., 1990 IEEE International Conference on, 1990.
Kaneko, Makoto. “A New Design of Six-Axis Force Sensors.” Paper
presented at the Robotics and Automation, 1993. Proceedings., 1993
IEEE International Conference on, 1993.
Kang, Chul-Goo. (2001), Closed-Form Force Sensing of a 6-Axis Force
Transducer Based on the Stewart Platform, Sensors and Actuators A:
Physical 90 (1-2), 31-37.
Kim, Gab-Soon, Dae-Im Kang, and Se-Hun Rhee. (1999), Design and
Fabrication of a Six-Component Force/Moment Sensor, Sensors and
Actuators A: Physical 77 (3), 209-20.
Liang, Qiaokang, Dan Zhang, Yaonan Wang, and Yunjian Ge. (2013),
Design and Analysis of a Novel Six-Component F/T Sensor Based on
Cpm for Passive Compliant Assembly, Measurement Science Review
13 (5), 253-64.
Liu, Sheng A, and Hung L Tzo. (2002), A Novel Six-Component Force
Sensor of Good Measurement Isotropy and Sensitivities, Sensors and
Actuators A: Physical 100 (2-3), 223-30.
Liu, Wei, Ying-jun Li, Zhen-yuan Jia, Jun Zhang, and Min Qian. (2011),
Research on Parallel Load Sharing Principle of Piezoelectric Six-
Dimensional Heavy Force/Torque Sensor, Mechanical Systems and
Signal Processing 25 (1), 331-43.
Stefanescu, Dan Mihai. (2011), Handbook of Force Transducers:
Principles and Components, Springer Science & Business Media.
Sun, Yongjun, Yiwei Liu, Tian Zou, Minghe Jin, and Hong Liu. (2015),
Design and Optimization of a Novel Six-Axis Force/Torque Sensor for
Space Robot, Measurement 65 (135-48.
Svinin, Mikhail M., and Masaru Uchiyama. (1995), Optimal Geometric
Structures of Force/Torque Sensors, The International journal of
robotics research 14 (6), 560-73.
Yurish, Sergey. (2014), Modern Sensors, Transducers and Sensor
Networks, Lulu. com.
Chapter 2
THE BASIC PRINCIPLE OF CALCULATION

AND ANALYSIS OF THE DEFECTIVE
STRUCTURE OF SOLIDS
V. I. Talanin, I. E. Talanin, V. V. Zhdanova,

D. I. Yakymchuk and A. V. Rybalko
Department of Computer Science and Software Engineering,
Institute of Economics and Information Technologies,
Zaporozhye, Ukraine
ABSTRACT
To describe the defective structure of semiconductor silicon, a triad
is created: physical plus mathematical models - computational algorithm -
program. This solution can be used both in studying the properties of
crystals and in industrial production. Such a solution is the basic principle
in the study of structural imperfections in any solid. A structural diagram
of the diffusion model for the formation of structural imperfections in a
crystal and a computational algorithm is presented.

Corresponding Author Email: v.i.talanin@mail.ru.
18 V. I. Talanin, I. E. Talanin, V. V. Zhdanova et al.
Keywords: grown-in microdefects, defective structure, model,

computational algorithm, defect engineering
1. INTRODUCTION
Physics, and in particular solid-state physics, is the basis for

application in various fields of human activity. Modern solid-state physics
can not be imagined without a wide application of mathematical modeling.
The mathematical model, with one or another degree of approximation to
the real physical process, is a computational algorithm for the use of
computer technology. This method of connecting simulation with
computer technology allows you to quickly and thoroughly explore
physical objects in any practical and imaginary situations. For its
implementation, it is necessary to create a triad adequate to this physical
process: physical plus mathematical model - computational algorithm –
program. A prerequisite is the development at first of a physical model of
the physical process under study, which should be adequate to the
experimental results of the research. Then such a triad can serve as a key
for solving various applied engineering problems in this or that sphere of
human activity.
The problems of solid state physics are closely intertwined with the
problems of materials science and solid state chemistry. Material science is
a science that studies the structure and properties of materials, which
establishes a relationship between their composition, structure and
behavior of materials, depending on the environmental impact. The
chemistry of the solid state overlaps the issues considered by both these
sections of knowledge, but it particularly touches on the issues of
synthesizing new materials. The structure and composition of a solid
determine its properties. In turn, the properties of a solid determine the
possibilities of engineering applications of various materials.
The Basic Principle of Calculation and Analysis ... 19
Semiconductor silicon, due to the requirements imposed on it, is one of

the most highly pure materials of our time. At the same time, it is also the
most studied material both experimentally and theoretically. Since the
properties of semiconductor silicon depend critically on its structural
perfection, the main task is the development on the basis of experimental
data of a theoretical description of the formation and development of
structural imperfections during crystal growth and the production of
devices. The solution of this problem was sought with the help of the
classical theory of a solid (Born & Huang, 1954) and the classical theory of
the nucleation and subsequent growth of particles of the second phase
(Cristian, 1965).
From these theories, a physical mechanism was developed for defects
formation, called grown-in microdefects, which are formed during crystal
growth (Voronkov, 1982). Numerous variations of mathematical models,
which were built on this physical mechanism, have received a general
name of the model of the dynamics of point defects (PDs) (Kulkarni,
2005). This model describe the formation of dislocation loops and
microvoids. It is assumed that the formation of impurity imperfections is
associated with subsequent thermal treatments of the crystals (Kulkarni,
et al., 2004).
In contrast to these statements, we developed a model for high-
temperature impurity precipitation, which was based on the methods and
approaches of classical theory the nucleation and subsequent growth of
second-phase particles (Talanin & Talanin, 2016a). In contrast to the
models of the dynamics of PDs in this model, the entire variety of the
defect structure of silicon is described on the basis of the formation of
impurity precipitates during crystal cooling. To confirm the results of the
high-temperature impurity precipitation model, the solid-state model A. A.
Vlasov was used. This model is an alternative to the classical theory of
solids (Talanin & Talanin, 2016b).
The general physical model of formation and development of
semiconductor silicon defect structure has been called the diffusion model.
It includes mathematical models of a high-temperature model for the
precipitates formation, growth and coalescence, models for formation of
secondary microdefects. The interrelation of these mathematical models

with parameters of crystal growth makes it possible to develop the
algorithm for defect structure calculating. This opens up wide opportunities
for solving the most important problem of the technology of ultrapure
semiconductor materials - the control of the crystals defective structure.
2. HIGH TEMPERATURE IMPURITY PRECIPITATION
Despite numerous experimental proofs of the existence of impurity

precipitates in grown dislocation-free silicon single crystals (Talanin &
Talanin, 2006), it was assumed in the theoretical description of the defect
formation process that the precipitation process begins only when the
already grown crystals are heat treated. This assumption was made on the
basis of the theory of nucleation and growth of particles of the second
phase in solids, in which the distribution function of nuclei of various sizes
∆𝐺(𝑅)
𝑓0 (𝑅)~𝑒𝑥𝑝 {− }
𝑘𝑇
where 𝑘 is a Boltzmann constant; ∆𝐺 is a minimum energy that must be

expended to create an nucleus of a given size 𝑅; 𝑇 is the temperature. Then
the formation of a critical nucleus at high temperature becomes impossible
and the formation and growth of the nucleus occurs only with an increase
in temperature. This led to the impossibility of a theoretical description of
precipitation upon cooling of the crystal during its growth (Cristian, 1965).
In our reasoning, we proceeded from the firmly established fact that
the formation and development of the growth microdefect structure of
silicon is due to the thermal conditions (Talanin & Talanin, 2016a).
Numerous researches have made it possible to establish that micropores
and also loops are formed in divers crystal fields and depending on
𝑉𝑔 ⁄𝐺𝑎 = 𝜀 (where 𝑉𝑔 is a crystal growth rate, 𝐺𝑎 is an axial temperature
gradient at the crystal center, and 𝜀 is some constant) (Talanin & Talanin,
2016a). There was an idea to introduce in the theoretical calculations the

growth parameter and determine the defective structure of silicon. Thus, it
was necessary to solve two problems: to determine the temperature
distribution along the length of the ingot during its cooling and to solve the
problem of the rapid recombination process of intrinsic point defects
(IPDs) near the front of crystallization. The first problem was solved very
simply. It is established that during the cooling of the crystal, depending on
the thermal growth parameters, the temperature distribution varies
according to the law
1⁄𝑇 = 1⁄𝑇𝑚 + 𝐺(𝑟)𝑧⁄𝑇𝑚2
where 𝑧 is the distance from the crystallization front, 𝑇𝑚 is the melting

point (Talanin & Talanin, 2016a). In this formula, the radial
inhomogeneity of the temperature field is taken into account in accordance
with formula 𝐺(𝑟) = 𝐺𝑎 + (𝐺𝑒 − 𝐺𝑎 )(𝑟⁄𝑅𝑆 )2 (where 𝑅𝑆 is the radius of
the crystal; 𝐺𝑒 is an axial temperature gradient at the crystal edge; 𝑟 is a
current coordinate in range from 0 to 𝑅𝑆 (Talanin & Talanin, 2016a).
Hence, it is very simple to obtain a temperature dependence on time, taking
into account the growth parameters of the crystal
𝑇𝑚2
𝑇(𝑡) =
𝑇𝑚 + 𝑉𝑔 𝐺𝑎 𝑡
The solution of the second problem required to abandon the

assumption of rapid recombination of IPDs near crystallization front. This
gave, depending on the value of the growth parameter, the homogeneous
formation of microvoids or interstitial dislocation loops and to construct
the simplest mathematical models (Voronkov, 1982; Kulkarni, 2005).
However, this assumption was not correct, since it completely excluded
any admixture from defect formation processes (Kulkarni, et al., 2004).
But even in the purest unalloyed dislocation-free silicon single crystals
obtained by the method of crucible-free zone melting, the concentration of
residual impurities of oxygen and carbon is one to three orders of

magnitude higher than the concentration of IPDs near the crystallization
front. Moreover, the distribution of intuitively understandable and
experimentally established low-temperature recombination acts to
processes occurring at high temperatures was erroneous. In (Talanin &
Talanin, 2007a), we showed that the process of recombination of IPDs in
dislocation-free single crystals of silicon near the crystallization front is
practically impossible due to the presence of a recombination barrier. This
means that the interaction “IPD-impurity” has the advantage over
interaction “IPD – IPD.” This result and taking into account the influence
of the thermal growth conditions of silicon single crystals in the form of
the 𝑇(𝑡) dependence make it possible to theoretically describe the
conditions for nucleation, growth, and coalescence of precipitates from the
crystallization temperature to room temperature.
The theoretical description was based on an experimentally established
physical model: (i) formation near the front of crystallization the impurity
complexes; (ii) precipitates formation, growth and coalescence during
cooling from crystallization to room temperatures; (iii) formation at 𝑇 ≤
1423 К micropores or loops depending on the constant 𝑉𝑔 ⁄𝐺𝑎 . Along with
the developed mathematical models for all structural defects, the physical
mechanism is a diffusion model for the grown-in microdefects formation
(Talanin & Talanin 2016a).
The central place in the diffusion model is occupied by the high-
temperature impurity precipitation model (Talanin & Talanin 2010b;
Talanin et al., 2007b). The precipitation process occurs throughout the
entire cooling range of the crystal, while microvoids and dislocation loops
are formed in a narrow temperature range at 𝑇 ≤ 𝑇𝑚 − 300𝐾.
Consideration of the elastic interaction of atoms of oxygen, carbon,
vacancies and intrinsic interstitial silicon atoms in undoped silicon single
crystals led to the understanding that near the crystallization front there are
nucleation processes (Talanin et al., 2007b).
In diffusion model a combined modeling method based on the solution

of differential equations for small clusters and the Fokker-Planck equation
for large-size clusters in the cooling temperature range 1683-1373 K was
used (Talanin & Talanin, 2016a). The model of dissociative diffusion
(migration of impurities) was used to calculate the formation of
precipitates in the high-temperature region T ~ 1683–1403 K (Talanin
et al., 2007b). When the dimensions of the nuclei are small, this
approximation is used. The edge of the reaction formation front of the
complex (that is, the formation of the oxygen-vacancy complexes and the
carbon-interstitial silicon complexes) is located at a distance of ~ 3·10-4
mm from the crystallization front (Talanin et al., 2007b). Recombination of
IPDs is absent and then their excessive concentration arises. This interval
is a diffusion layer on which this excessive concentration is applied. The
calculation of the defect structure at the stage of precipitates formation and
growth in the model described in (Talanin et al., 2007b) agrees well with
the experimental results available in the literature (Talanin & Talanin,
2006). This is true in regard to the temperatures of formation of growning
microdefects; the experiments on quenching of the crystals; and the
concentrations of precipitates, which were determined from the results of
electron microscopy investigations (~1013–1014 cm–3) (Talanin & Talanin,
2006; Talanin & Talanin, 2010b).
Consider a system consisting of oxygen and carbon atoms, vacancies,
and intrinsic interstitial silicon atoms. It will allow us to describe the
kinetics of nucleation and growth of new phase particles in a
supersaturated solid impurity solution. A comparative analysis of the joint
evolution of oxygen and carbon deposition and the optimization of the
computational algorithm for the numerical solution of equations requires
the carrying out of a dimensional analysis of kinetic equations and
conservation laws using characteristic time constants and critical sizes of
defects (Talanin & Talanin 2010b). Such an analysis, in particular, made it
possible to obtain a system of equations where the mass balance of PDs is
described by diffusion equations for silicon interstitials, vacancies and
impurities:
Co  2Co CoSiO2

 Do 
t z 2 t
Cc  2Cc CcSiC (1)
 Dc 
t z 2 t
Ci  2Ci CiSiO2 CiSiC
 Di 2  
t z t t
Cv  Cv Cv
2 SiO2
CvSiC
 Dv  
t z 2 t t
where Cv , Ci , Cс , Cо is the concentration of vacancies, intrinsic interstitials

of silicon, carbon and oxygen in the crystal, respectively. The equations
can be used in dimensionless form
~
f SiO2 I SiO2

 O ~ (2)
~
f SiC t I SiC
 0 ~
 tc c
t
where   is the dimensionless time. The time constants in (2) are
to
given by tO  (nOcr ,0 )2 / g SiO
0
2
; tc  (nccr ,0 )2 / g SiC
0
, where the critical growth
rates of the precipitates are defined as
0
g SiO2
 NO0O exp( Gact
SiO2 0
/ kT ); g SiC  NO0c exp(Gact
SiC
/ kT ) .
The precipitates normalized sizes are determined in (2):
~O  nO / nOcr ,0 ;~c  nc / nccr ,0 , where nOcr , nccr 
are normalizing precipitates critical sizes. The quantities
NO0  4 (rOcr ,0 )2  SiO2 COeq ; Nc0  4 (rccr ,0 )2  SiCCceq

are the numbers of particles. In the system of equations (2), the

normalization to the initial concentrations of the nucleation centers was
made:
~ f SiO ~ f
f SiO2  0 2 ; f SiC  SiC0
(3)
f SiO2 f SiC
The particles fluxes on right hand sides of system of equations (2) are
described as
~ BSiO ~ B
ASiO2  ( g~SiO2  d SiO2 )nOcr ,0  ~ 2 ; ASiC  ( g~SiC  d SiC )nOcr ,0  ~SiC ; (4)
O O
in which the following notation is used for the normalized kinetic

coefficients:
~ BSiO ~ B
ASiO2  ( g~SiO2  d SiO2 )nOcr ,0  ~ 2 ; ASiC  ( g~SiC  d SiC )nOcr ,0  ~SiC ; (5)
O O
~ ~
g~SiO2  d SiO2 g~SiC  d SiC
BSiO2  ; BSiC  (6)
2 2
The normalized rates of growth and dissolution of the precipitates in

(5) and (6) look like
g SiO g ~ d SiO ~ d
g~SiO2  0 2 ; g~SiC  SiC
0
; d SiO2  0 2 ; d SiC  SiC
0
(7)
g SiO2 g SiC g SiO2 g SiC
Critical size of precipitates
2uV p
rOcr  (8)
kT ln( S0 Si i Sv v )  6uV p
2uV p
rCcr  (9)
kT ln( Sс Si i Sv v )  6uV p
where So  Co Coeg , Sc 
Cc
, Si 
Ci
, Sv 
Cv
Cceg Cieg Cveg
Are the supersaturations of the oxygen, carbon, silicon interstitials, and

vacancies;  is a density of a surface energy of interface between the
precipitate and the matrix;  is a silicon shear modulus;  and  are the
relative linear and volume misfit strains of the precipitate and the matrix,
respectively;  i and  v are the fractions of the intrinsic interstitial silicon
atoms and vacancies per impurity atom attached to the precipitate,
respectively; V p is the molecular volume of the precipitate; and
1  3
u  (1   i x   v x)1  ( ) (Talanin & Talanin 2010b).
1 
The critical dimensions of the microdefects and the characteristic time
constants associated with them indicate the scale of the changes in the
distribution function over the size of grown-in microdefects with time.
These parameters are important in the analysis of the evolution of growth
microdefects.
The processes of nucleation pass near the crystallization front (Talanin
& Talanin 2010b). The dissociative diffusion model analyzes phenomena
in the diffusion region of the crystallization front, and the Fokker-Planck
equation determines the distribution functions of the critical size of oxygen
and carbon precipitates. Both these models complement each other
(Talanin et al., 2007b).
A growing undoped silicon crystal can be represented as a
multicomponent multiphase system in which the role of components is
played by oxygen and silicon atoms, intrinsic interstitial silicon atoms, and
vacancies, whereas oxygen and carbon precipitates and agglomerates of
IPDs (microvoids and interstitial dislocation loops) should be treated as a
new phase. The kinetics of the precipitation process includes three stages:
formation the nuclei of new phase, and clusters growth, and stage of
coalescence (Talanin & Talanin 2004). This process is a first-order phase
transition.
The growth of clusters is accompanied by a significant decrease of the
supersaturation solid solution. Their number in the second stage does not
change. This process is described by a system of equations (Talanin &
Talanin 2011):
dN 0
 k0 NN 0  g1 N1 ,
dt
  N i ki N  gi   gi 1 N i 1  ki 1 NN i 1 ,
dN i (10)
dt
dN
  N  ki N i   g i N i ,
dt i 0 i 1
where N i is a volume average concentration of nucleation centers that

attach i particles, N is a monomer concentration, ki N is the rate of
attachment of a monomer for a nucleation center, and g i is the rate of
detachment of monomer for nucleation center. At the initial time, in system
are monomers and centers of nucleation. The precipitates growth is limited
by monomer diffusion. The kinetic coefficients are given by ki  4Ri D ,
where Ri is a radius of free particle attachment and D is a free particle
diffusion coefficient (Talanin & Talanin 2011).
The system (10) obeys the law of conservation of nucleation centers
N c   Ni (t ) and the law of conservation of the total number of
i 1
particles, including both monomers and particles involved in precipitates,

i.e., N 0  N t    iN i , where N 0 is the monomer concentration at
i 1
the initial instant of time. Therefore, the average number of particles at the
nucleation centers can be represented in the form
 iN i
N 0  N t 
i i 0
 (11)
N i 0
i Nc
For i »1, we can use the Fokker–Planck equation. In accordance with

the balance g i   k i  1N ECE i  1 CE i   k i N E , the Fokker–
Planck equation takes the form,
C i, t )  2
 N t   N E  k i C i, t   N t   N E  2 k i C i, t  (12)
t i 2i
where N E is a equilibrium concentration of monomers (Talanin & Talanin

2011). The mathematical expectation i(t):
 k0 N  N E it   m ,
di 
(13)
dt
where k0  4Ri D , m is a initial precipitate size, and  is a parameter

of cluster geometry.
The change in monomer concentration during the decomposition of the
solid solution (Talanin & Talanin 2011):
dN t 
 k0 Nc1 N t   N E   N 0  mNc N t 

(14)
dt
The kinetics of decrease in the monomer concentration as a result of

the decomposition of the solid solution at can be written in the form
m  0 can be written in the form
N t   N E 
 

1
 exp  Nc 1   N 0  N E  k0t 1 

(15)
N 0  N E  
Average sediment radius at the growth stage:
3bit 
Rt   3 (16)
4
Depending on the temperature growth conditions, the duration of the

growth phase varies from 𝑇 ≈ 𝑇𝑚 − 20𝐾 for small-sized single crystals of
silicon to 𝑇 ≤ 𝑇𝑚 − 300𝐾 for large crystals (Talanin & Talanin 2011).
In coalescence occurs a dissolution of small-sized particles and the
growth of coarse-grained particles. At this stage, new particles are not
formed. The condition providing for changeover to the coalescence stage is
Rt 
the ratio u t    1 , where Rcr t  is a precipitate critical radius.
Rcr t 
Precipitate grows at Rt   Rcr t  and dissolves at Rt   Rcr t  . If the
supersaturation of the solute tends to zero, then the solution is possible
(Talanin & Talanin 2011).
The system of equations describing the nonisothermal coalescence has
the form
f R, t  
  f R, t dR / dt   0, f0 R   f R,0 (16a)
t R

QTc   cc t     f R, t R3dR (17)
0

L
Tc t   Tc 0    f R, t   f R R dR
3
(18)
c p t
0
0
dR 2D 2c0 Tc 

 R / Rk  1 при D  R (19)
dt kTc R 2
dR 2D 3c0 Tc 

 R / Rk  1 при D  R (20)
dt kTc R
where expression (16) is the equation of continuity in the space of sizes for
the size distribution function of precipitates, expression (17) is the mass
balance equation, expression (18) is the equation accounting for the
amount of released heat, f 0 R  is the initial size distribution function of
precipitates, Tc 0 is the initial temperature of the solid solution,
  4 / 3 is the volume per atom in the precipitates of the new phase,
QTc  is the total amount of the material in the precipitates and the
solution, cc t  is the solute concentration in the solid solution, L is the
heat of the phase transition per atom of the precipitated phase, c p is the
heat capacity at constant pressure per unit mass of the solid solution, and
 е is the density of the solid solution.
Equations (16) - (20) form a complete system of equations describing
the process of nonisothermal decomposition of the solid solution in the
coalescence stage. The solution of equations (16) - (20) is carried out by
the method developed in (Talanin & Talanin 2011). The simultaneous
nucleation and growth of new phase particles (oxygen and carbon
precipitates) during the cooling of silicon crystals after growth leads to a
strong interaction between the evolution processes of these two subsystems
of grown-in microdefects Absorption of increasing vacancy oxygen
precipitates leads to the emission of silicon atoms into the interstitial. The
intrinsic interstitial silicon atoms, in turn, interact with growing carbon
precipitates, which in the process of their growth supply vacancies for
growing oxygen precipitates. Such interaction leads to an accelerated
transition of the subsystems of oxygen and carbon precipitates to the
coalescence stage, in comparison with the independent evolution of these
two subsystems (Talanin & Talanin, 2011).
The kinetic model of growth and coalescence of oxygen and carbon

precipitates together with the kinetic models of their formation (Talanin &
Talanin, 2010b; Talanin et al., 2007b) represent a unified model of the
precipitation process in dislocation-free single crystals of silicon. The
mathematical apparatus of this model makes it possible, in the future, to
take into account and analyze the interaction of IPDs not only with
background impurities of oxygen and carbon, but also with other
impurities (for example, transition metals, nitrogen, hydrogen, etc.)
“impurity-impurity” (Talanin & Talanin 2011). The obtained results on
high-temperature impurity precipitation required confirmation with the
help of another representation, different from the classical ones. To this
end, we used the solid state model of Vlasov.
3. PRECIPITATION OF THE IMPURITY IN ACCORDANCE

WITH THE MODEL OF THE SOLID STATE OF VLASOV
In 1945, the Soviet scientist Vlasov suggested that the physical

properties of gases, liquids and solids can be described with the help of the
proposed earlier to describe the peculiar properties of the plasma of the
kinetic equation (Vlasov, 1945). For that time such an idea seemed to be
non-trivial and was subjected to severe criticism of leading Soviet
theoretical physicists (Ginzburg et al., 1946).
On the basis of his approach, Vlasov put the introduction of a single
distribution function that depends on all the coordinates and their
derivatives up to any order (non-local statistical mechanics) (Vlasov,
1978). In the classical theory of the crystal, the main conclusions about the
dynamics of the system are made on the basis of the mechanical theory of
atomic vibrations in the vicinity of equilibrium positions (Born & Huang,
1954). The equations of the oscillations make it possible to obtain only the
time functions, while for the description of the wave processes in the
original equations, explicit derivatives with respect to the spatial
coordinates must be explicitly contained. According to Vlasov, the theory
of crystal growth cannot be created on the basis of a method that uses

classical or quantum Hamiltonians with model interaction potentials
between atoms (Vlasov, 1978). During phase transformations, the electron
configurations of atoms and the functional form of the interaction
potentials change. Field equations give a temporal evolution of the
interactions of particle systems, the change in shape of which with time is a
fundamental property of the kinetics of crystal growth. Consequently, the
basis for the kinetics of phase transitions should be equations with a self-
consistent field (Vlasov, 1978).
To describe the stationary properties of a crystal, the concept of the
particle distribution density 𝜌(𝑟) = ∫ 𝑓(𝑟, 𝑣)𝑑𝑣 is used. Vlasov showed
that the non-linear model of the crystal is based on nonlinear equations that
make it possible to calculate the molecular potential and density of the
particle location under thermal equilibrium (Vlasov, 1950):
∞
𝐾1,2 (𝑟)
𝑈(𝑟) = 𝜆𝑘𝑇 ∫ 𝐾1,2 (𝑟)𝑒𝑥𝑝 (− ) 𝑑𝑟
𝑘𝑇
−∞
𝐾1,2 (𝑟)
𝜌(𝑟) = 𝜆𝑘𝑇𝑒𝑥𝑝 (− 𝑘𝑇
) (21)
where 𝑘 is the Boltzmann constant; 𝐾1,2 is the potential of pair interaction;

𝜆 is some characteristic number. The initial equations represent the
𝜕
equations for two particles under steady-state conditions (𝜕𝑡 = 0) (Vlasov,
1950). In order to distinguish the types of interactions, one usually
considers the systems of Vlasov equations (the Vlasov-Poisson, Vlasov-
Maxwell, Vlasov-Einstein, and Vlasov-Yang-Mills equations).
In applications of the mathematical theory of non-linear equations to
physical problems, the main mathematical problem arises, which consists
in finding the characteristic numbers. By characteristic numbers, we mean
here the values of a parameter 𝜆 for which equations of the type (21) have
solutions that are different from the trivial ones. By trivial solutions are
meant solutions of equations of the type (21), corresponding to the case of
uniform density (Vlasov, 1950). The characteristic number 𝜆 is determined
(Vlasov, 1950) from the main criterion for the existence of the crystalline
state, and the crystallization condition can be written as follows:
4𝜋𝑁 ∞ ∗
∫ 𝐾1,2 (𝜌)𝜌2 𝑑𝜌
𝑘𝑇𝑚 0
=1 (22)
∗
where 𝑁 is the number of particles; 𝐾1,2 = −𝐾1,2 (Vlasov, 1978).
For the crystal case, Vlasov showed that the state space is determined
by the temperature changes (Vlasov, 1950). The density of complexes
distribution we defined as a function of crystal cooling temperature
𝑈1𝑚𝑖𝑛,2𝑚𝑖𝑛
𝜌(𝑇) = 𝜆𝑘𝑇𝑒𝑥𝑝 (− 𝑘𝑇
) (23)
where 𝑈1𝑚𝑖𝑛 and 𝑈2𝑚𝑖𝑛 are the minima of interatomic potentials in silicon-
oxygen and silicon-carbon complexes, respectively.
Vlasov showed that the spatial periodic distribution is one of the
particular states of particle motion (Vlasov, 1950). This solution was
carried out by Vlasov for an ideal (defect-free) crystal.
However, real crystals always contain structural imperfections. For
defective (real) crystals, we considered the possibility of pair formation of
an IPD-impurity with subsequent impurity precipitation (Talanin &
Talanin, 2016b). The potential of interatomic interaction is the basis for
describing the interaction between the atoms of matter. Analytic
consideration of the potentials of the interaction of two atoms is extremely
difficult, since after quantum-mechanical calculations they are represented
as functions with a large number of parameters. In this connection, model
potentials with a small number of parameters are used (Mazhukin et al.,
2014; Mattoni et al., 2007). To estimate the formation parameters of
silicon-carbon and silicon-oxygen complexes, we presented interaction in
form of Mee-Lennard-Jones potential.
To determine the characteristic numbers of silicon-oxygen (𝜆1 ) and
silicon-carbon (𝜆2 ) complexes, we use (22) for the number of particles in
𝐾1,2
the complex 𝑁 = 2 and 𝐾(𝜌) = 𝑘𝑇 (1 − 𝑒 − 𝑘𝑇 ). The calculation gave the
following values 𝜆1 = 4.482·108 eV-1 and 𝜆2 = 1.099·109 eV-1. In this

computational experiment, it was assumed that the concentration of
nucleation centers for carbon and oxygen complexes is ~ 1012 cm-3.
We determined the evolution of the density of the complex distribution
as a function of the crystal cooling temperature. The calculations of
complexation with the Vlasov solid state model show that during the
growth of dislocation-free silicon single crystals, silicon impurity
complexes form near the crystallization front, which coalesce during the
cooling of the crystal, leading to the formation of carbon precipitates (SiC)
and oxygen (SiO2) (Talanin & Talanin, 2016b).
Two theories of nucleation of particles of the second phase (the
classical theory of nucleation and the Vlasov model for solids) lead to
similar results. This allows us to say that:
 Vlasov's solid body model can be used to describe the formation of

a defective structure of real crystals;
 Vlasov's approach to the description of substances may have a
universal character.
The model of high-temperature precipitation, built on classical

concepts, suggests that nucleation is possible both during cooling of the
crystal during growth and during its heating (Talanin & Talanin, 2013).
The Vlasov solid state model unequivocally indicates that the formation of
complexes does not occur during low-temperature crystal treatments
(up to ~11000 С) (Talanin et al., 2017a). The appearance of donor and
acceptor centers in silicon during thermal treatments is directly related to
the original defect structure. Precipitates of impurities that were created
during crystal growth are in it during the coalescence stage. Therefore, the
heating of the crystals promotes the dissolution of some and the coarsening
of other precipitates. For the formation of electrical centers, it is precisely
soluble precipitates that are responsible.
The formation of silicon-impurity complexes is the determining stage
in the formation of the defect structure. Depending on the thermal growth
conditions, interstitial dislocation loops or microvoids form during further

cooling of the crystal. High-temperature impurity precipitation is
associated with the subsequent transformation of growth microdefects in
the production process of silicon devices and is the foundation for the
creation and development of a defective structure of semiconductor silicon.
4. FORMATION OF SECONDARY
STRUCTURAL IMPERFECTIONS
Under certain thermal conditions for the growth of crystals, depending

on the growth parameter V , microvoids or interstitial dislocation loops
G
are formed in the temperature range of cooling 1403 ... 1223 K. The author
of (Voronkov, 1982) made the assumption of rapid recombination of IPDs
near the crystallization front. Remaining after the recombination, the IPDs,
depending on the temperature growth conditions, form either microvoids or
dislocation loops. Based on this were developed various interpretations of
model the dynamics of PDs (Kulkarni, 2005; Kulkarni et al., 2004; Sinno,
1999; Brown et al., 2001). All of these variations papers never allowed
impurity precipitation during crystal cooling (Kulkarni et al., 2004).
The model of the dynamics of IPDs, in the general case, includes three
approximations: rigorous, simplified, and discrete-continuum approaches
(Kulkarni, 2005). The most rigorous model requires the solution of integro-
differential equations for the concentration fields of PDs, and the
distribution of growth microdefects in this model is a function of the
coordinates, the time, and the time of evolution of the size distribution of
microdefects (Kulkarni et al., 2004; Sinno, 1999). In the simplified
(lumped) model, the average radius of the defects is approximated by the
square root of their average surface area (Kulkarni, 2005). This
approximation is taken into account in the additional variable, which is
proportional to the total surface area of the defects. The lumped model is
effective in calculating the two-dimensional distribution of growth
microdefects (Kulkarni, 2005). Both models use the classical nucleation
theory and suggest the calculation of the formation of stable nuclei and the
kinetics of diffusion-limited growth of defects. The discrete-continuum
approximation (discrete-continuous model) involves a comprehensive
approach: the solution of discrete equations for the smallest defects and the
solution of the Fokker–Planck equations for large-sized defects (Brown
et al., 2001).
We calculated the formation of microvoids and interstitial dislocation
loops in accordance with the strict approximation of the model of PD
dynamics, provided that there is no high-temperature recombination of
IPDs. We confirm the homogeneous character of microvoid formation
process (Talanin & Talanin, 2010a). Microvoids are formed on the already
existing precipitates. The data from the computational experiment for
determining the concentration of vacancy microvoids correlate well with
the experimentally observed results (104-105 cm-3) (Kulkarni et al., 2004).
For dislocation loops, where the experiments give a value of ~ 106-107 cm-3
(Brown et al., 2001), the discrepancy reaches three orders of magnitude.
This may be due to the fact that, in contrast to microvoids, which are
formed only by the action of the coagulation mechanism, the formation of
dislocation loops occurs due to the action of both the coagulation
mechanism and the mechanism of prismatic extrusion (deformation
mechanism). The results of calculations show that the main contribution to
the formation of dislocation loops is made by the mechanism of prismatic
extrusion, when the formation of interstitial dislocation loops is associated
with the removal of stresses around the growing precipitate.
We know that the kinetics of high-temperature precipitation
encompasses three stages: the inception of a new phase, the growth of
sediment and the stage of fusion. The elastic interaction between IPDs and
impurity atoms causes the formation of precipitates. The transfer of excess
(scarce) matter from the sediment or vice versa is caused by elastic
deformations and associated mechanical stresses. During the growth of the
precipitate, its coherence with the matrix is lost. The formation and
displacement of dislocation loops is due to the structural relaxation of
precipitates.
Consider the simplest precipitate in the form of a sphere. The

formation of a circular interstitial dislocation loop of nonconformity occurs
due to the growth of precipitate. The total energy of deformation of the
system decreases. The matrix material in the volume of the crystal is
displaced by precipitate. The dislocation loop around the precipitate is
formed by intrinsic interstitial atoms. On the precipitate itself, a contour of
misfit dislocation is formed. The critical size of the precipitates
corresponds to the critical size of the dislocation loops (Talanin & Talanin,
2012a).
In silicon precipitate produces a stress field caused by mismatch
between the lattice parameters of precipitate a1  and surrounding matrix
a2  . Then, the precipitate intrinsic deformation can be defined as
a1  a2
 (24)
a1
The intrinsic deformation of precipitate in the volume of matrix

determined as
  xx  xy  xz 
 
     xy  yy  yz   pr  , (25)
  zz 
 zx  zy
where the diagonal terms constitute a dilatation mismatch (between the

precipitates and matrix lattices); the other terms are shear components;
  pr  is the Kronecker delta. The elastic fields of precipitate (stress  ij
and deformation  ij ) and the field of total displacement are calculated
accounting for intrinsic deformation (25) and regions of precipitate
 
localization   pr . The elastic fields of precipitate are calculated
according to the known scheme using elastic moduli, elastic Green
function or its Fourier transform.
Let us consider the easiest model of a spherical precipitate with

uniaxial intrinsic deformation, i.e.,  ii   ,  ij  0i  j; i, j,  x, y, z  .
The elastic strain energy of a spheroidal defect rises according to the cube
 
law as the precipitate radius R pr increases (Kolesnikova et al., 2007):
32  
E pr   J   2  R3pr , (26)
45  1   
where J is a shear modulus;  is a Poisson's ratio. Starting from a Rcrit

radius, the elastic strain mechanism begins to function. This mechanism
give a prismatic interstitial dislocation loop (Kolesnikova et al., 2007). The
energy criterion for this mechanism is E initial  E final condition, where
Einitial, E final constitute elastic energy of the system with precipitate before
and after relaxation (Kolesnikova et al., 2007).
Assume that a dislocation loop of mismatch has equatorial location on
the spheroidal precipitate RD  Rpr , and intrinsic energy of a prismatic
loop is equal (Kolesnikova et al., 2007)
J  b 2  RD  2  RD 
Eloop    ln  2 , (27)
2  1    
 f 

where f is the radius of the core loop; b is the magnitude of the Burgers
vector. A critical value of the precipitate radius corresponds to a value at
which the loop is formed on the precipitate (Kolesnikova et al., 2007)
3b  1.08Rcrit 
Rcrit   ln , (28)
8 1     b 
where  is a constant contribution of the dislocation core.

The dislocation loops with R  Rcrit will increase. The dislocation

loops with R  Rcrit will dissolve. The supersaturation of silicon by
intrinsic interstitial atoms and the dissolution of small dislocation loops
lead to the growth of large dislocation loops. In this case, the crystal
growth ratio is V  crit (where 0.12 mm2/K·min  crit 
G
0.3 mm2/K·min (Kulkarni, 2005)). When oversaturation of vacancies
( V  crit ) occurs, the interstitial dislocation loops start to dissolve.
G
Increase in the radius of interstitial dislocation loop can be defined by the
formula depending on the crystal cooling time (Burton & Speight, 1985):
R(t )  Rcrit
2
 j  D(t )  t , (29)
where D(t ) is a diffusion coefficient of silicon interstitials; t is a time of

cooling; j is factor of proportionality. The crystal cooling time is defined
Tm2
from: T (t )  (Kulkarni et al., 2004; Talanin & Talanin, 3013),
Tm  U  t
where U  Vg  G is the cooling rate of the crystal.
The concentration of dislocation loops should be considered as a
function of the concentration of precipitates. At the stage of precipitates
growth and coalescence the dislocation loops formation is determined by
the deformation mechanism. Then, the loop concentration depends on the
crystal cooling time (Brown et al., 2001):
M (t )
N (t )  , (30)
1  D(t )  t 2  Rcrit
2
where M (t ) is the concentration of precipitates.

The kinetic model of the formation and growth of interstitial
dislocation loops formed in dislocation-free single crystals of silicon as a
result of the structural relaxation of precipitates of background impurities

of oxygen and carbon shows that high-temperature precipitation of oxygen
and carbon is a fundamental process in the formation of a defective silicon
structure during crystal cooling after growing. The removal of the elastic
deformation caused by the growing precipitate occurs through the
formation and movement of dislocation loops.
Within the framework of the proposed model, the obtained results were
compared with experimental data. A result confirms the reliability of the
proposed model and allows it to be used to correctly take into account the
influence of the parameters of precipitates on the formation of interstitial
dislocation loops.
5. CONTROLLING THE DEFECTIVE

STRUCTURE OF CRYSTALS
The results obtained radically change the physical picture of our ideas
of defect formation in crystals. The entire process is responsible for high-
temperature precipitation, which occurs from the time the crystal is grown
until its end. Microvoids and dislocation loops are derived from the
emerging precipitate structure, their formation is determined by the
specific growth conditions of semiconductor silicon. The formed defect
structure, in turn, is the basis for the transformation processes that occur in
the crystal as a result of technological treatments when creating various
devices. The computational experiments carried out made it possible to
describe the process of defect structure formation.
For a theoretical description of the defect structure of semiconductor
silicon, we used several mathematical models with different levels of
complexity of the computational experiment. In the framework of the
diffusion model, they can be represented in the form of Figure 1
Each of them to some extent describes the details of the general picture
of physical phenomena occurring in the original system (the formation of
structural imperfections during the growth of semiconductor silicon).
The general requirement is one - the correspondence between the

results of computational experiments and the results obtained
experimentally.
Figure 1. Scheme diffusion model of defect formation.
The application of methods of mathematical modeling makes it

possible to improve the technological process and accelerate the
development of new technologies for growing dislocation-free single
crystals of silicon with a specified minimized content of growth
microdefects. In the engineering of growth microdefects, the problem of
controlling the composition and state of the ensemble of PDs in crystals is
put forward, requiring deep penetration into the problem of their
interaction. The absence of a single algorithm for solving defect formation
problems in solids leads researchers to the necessity of their
phenomenological analysis for each type of crystals separately. At the
same time, many phenomena occurring in different crystals are of a general
nature, for example, the interaction of PDs among themselves; their
interaction with dislocations; formation and transformation of dislocations,
etc. Semiconductor silicon, being an extremely pure and perfect in
structure material, can itself be considered as an initial model for

constructing theoretical models of defect formation in other semiconductor
materials and metals. The obtained mathematical models and the proposed
methods for their solution in silicon make it possible to formulate and
solve many problems in the kinetics of diffusion processes in solids. The
primacy of the processes of high-temperature precipitation is a
fundamental feature that determines the general kinetics of defect
formation in highly perfect crystals, both semiconductors and metals. The
specificity of growing and distinguishing the crystal structure of various
dislocation-free single crystals must necessarily be taken into account in
the physical model and mathematical constructions of the diffusion model.
The mathematical apparatus of the diffusion model with the help of
modern information technologies is used as the basis for a software
package. After developing mathematical models of each of the stages of
this process, an algorithm was developed to calculate the formation of a
defective crystal structure during growth (Talanin et al., 2017b) and on the
way of creating the instrument (Talanin et al., 2012b).
The algorithm is based on the dominant role of thermal growth
conditions on defect formation in a crystal. From the scheme of the
algorithm (Figure 2), a special role of high-temperature precipitation is
seen in the formation of the defective crystal structure. Changing the
thermal growth conditions, one can create a given defective crystal
structure. And, conversely, the defect structure can determine the
temperature conditions for crystal growth. Therefore, the software complex
constructed on the basis of this algorithm will have the functions of both a
research tool and devices for controlling the process of growing silicon
crystals. To create such software products, a research (calculation)
technique is needed that would link each point of the crystal to its real
structure. The method of virtual investigation of a defective structure
should be verified by experimental researches.
Figure 2. Algorithm for calculating of formation of a defective crystal structure during

growth.
The method that determines and calculates the growth defect structure
is based on a parameter that connects the main thermal conditions of
crystal growth Vg V
 Ccrit . The condition g  Ccrit in real crystals
Ga Ga
agrees to the V-shaped distribution of precipitates. In Figure 3 shows a
standard picture of the distribution of growth microdefects in dislocation-
free single crystals of silicon grown by the Czochralski method with
variable growth rate. The crystal diameter was 50 mm, the growth rate
varied from 𝑉𝑔𝑚𝑖𝑛 = 0.5 mm/min to 𝑉𝑔𝑚𝑎𝑥 = 3.0 mm/min.
Three areas of defect formation can be identified: region 1 (above the
V-shaped precipitates distribution), region 2 (V-shaped precipitates
distribution region) and region 3 (below the V-shaped precipitates
distribution). In small-sized silicon crystals in region 1, growth
microdefects (precipitates) of interstitial and vacancy types are formed at
comparable concentrations; in region 2 within the V-shaped distribution
the same defects as in region 1 are formed, and outside the V-shaped
distribution interstitial dislocation loops and predominantly interstitial
precipitates are formed. In crystals with a diameter of more than 80 mm in
regions 1, 2 (inside the V-shaped distribution) microvoids begin to form. In

region 3, dislocation loops and precipitates of predominantly interstitial
type are formed.
Figure 3. The standard picture of the distribution of grown-in microdefects in

dislocation-free single crystals of silicon grown by the Czochralski method with a
variable growth rate.
With increasing crystal diameter, the thermal growth conditions of the

crystal change and the picture of defect formation in the longitudinal
section of the crystal undergoes changes that occur in the rather rapid
disappearance of regions 1 and 3. For example, in modern large-sized
single crystals of silicon (diameter 150…300 mm) grown by the
Czochralski method, the picture of defect formation corresponds only to
region 2 in Figure 3 (Ammon et al., 1999).
When Vg  Ccrit in small crystals in region 1 the concentration of

Ga
vacancies is comparable with the concentration of intrinsic interstitial
silicon atoms (high growth rates of the crystal). In small crystals in region
3 ( Vg  Ccrit ), the concentration of vacancies is small compared with the
Ga
concentration of intrinsic interstitials of silicon (low crystal growth rates).
For large-sizes crystals ( Vg  Ccrit ), supersaturation in vacancies is due to
Ga
the presence of the V-shaped precipitate distribution, which leads to an
impoverishment of impurity atoms within the V-shaped distribution of
precipitates and the creation of conditions for the homogeneous formation
of microvoids. The growth parameter V g controls a system of interacting
Ga
PDs during crystal cooling.
The first step of the technique is to specify a certain diameter of the
crystal. Since Vg  Ccrit it is determined theoretically and experimentally
Ga
in a certain range of values (0.06 mm2/K·min ≤ C crit ≤ 0.3 mm2/K·min),
then for calculation we select a certain value C crit .
In the second step of the procedure, the value of the axial temperature
gradient at the center of the crystal (𝐺𝑎 ) is chosen, as well as the values of
the minimum (𝑉𝑔𝑚𝑖𝑛 ) and maximum (𝑉𝑔𝑚𝑎𝑥 ) crystal growth rate. For each
crystal diameter, these values can be determined from an analysis of the
experimental and theoretical data.
The third step is the choice of the axial temperature gradient at the
𝐺
edge of the crystal (𝐺𝑒 ) in the range of values 𝑒⁄𝐺 = 1.0 ... 2.5. The
𝑎
parabolic radial distribution of the axial temperature gradient 𝐺(𝑟) = 𝐺𝑎 +
2
(𝐺𝑒 − 𝐺𝑎 ) ∙ (𝑟⁄𝑅 ) , where 𝑅𝑠 is the radius of the crystal; 𝑟 is the current
𝑠
coordinate in the range from 0 to 𝑅𝑠 .
In Figure 3, some value 𝑉𝑔 = 𝑐𝑜𝑛𝑠𝑡 is given as an example. In this
case, we fall into region 2. Based on the analysis of the defect structure of
this region within the V-shaped distribution of precipitates, the formation
of precipitates is calculated, and the formation of precipitates and

dislocation loops is done outside the V-shaped precipitates distribution.
The technique is fairly simple and can easily be implemented as a
software product that can become a convenient virtual experimental device
when investigating various properties of dislocation-free single crystals of
silicon. The second advantage lies in the fact that the analysis and
calculation of the defect formation process is determined only by the
thermal growth conditions.
The technique of analysis and calculation of the formation of grown-in
microdefects in dislocation-free single crystals of silicon is easy to
implement on a personal computer in technological and research practice.
Theoretical research of the real structure of crystals depending on their
thermal growth conditions with the help of an original virtual technique for
analyzing and calculating the formation of grown-in microdefects is a new
experimental technique. In addition, this technique allows you to replace
the experimental researches of the structure with adequate theoretical
researches, so the software product developed on the basis of the proposed
method will be a new virtual experimental device.
The development of a software package for the analysis and
calculation of the formation of grown-in microdefects in dislocation-free
silicon single crystals became possible after the creation of a precipitation
model during the cooling of the crystal after growth, which, in conjunction
with the kinetic models of the formation and growth of interstitial
dislocation loops and microvoids, makes it possible to theoretically
describe the processes of formation and transformation of grown-in
microdefects in dislocation-free silicon single crystals of any diameter,
obtained by methods floating zone method and the Czochralski method.
Since the adequacy of the developed triad “mathematical model -
algorithm - program” of the original physical model is certified by carrying
out computational experiments and their comparison with the results of
physical researches, thereat replacing the experimental studies of the
structure with adequate theoretical researches using the software package
volume, the complex is a new virtual experimental device. Since the
method of application of the software is original, it represents a new

technique for the experiment.
At present, the computational algorithm is implemented by us on the
basis of programming languages C++, C # and JavaScript (Talanin et al.,
2018). These software products allow you to get information in tabular,
graphical and animation forms. Software systems require conventional
computer technology, are simple and convenient to use.
6. THE POSSIBILITY OF APPLYING THE MODEL OF HIGH-

TEMPERATURE PRECIPITATION FOR OTHER SOLIDS
When creating a model of high-temperature precipitation in

dislocation-free silicon single crystals, an elastic interaction was made
between impurity atoms and intrinsic point defects. It is possible to take
into account the elastic interaction “impurity-impurity”, as well as the
Coulomb interaction. However, even without these approximations, the
high-temperature precipitation model is in good agreement with the
experimental data (Talanin and Talanin, 2016a; Talanin and Talanin,
2006).
The phenomenon of high-temperature precipitation accompanies the
growth of crystals and other substances. The similarity of the structure of
dislocation-free single crystals of silicon and germanium makes it possible,
as a test of this assumption, to choose germanium crystals.
The creation of a high-temperature precipitation model was impossible
without calculating the recombination parameters. The authors of (Cowern
et al., 2013) found two separate forms of their existence in the study of
point defects in germanium. The first (low-temperature) form is a widely
known point defect. The second form of the defect has a structure similar
to that of the amorphous pocket. This second form was called the morph
(Cowern et al., 2013). This high-temperature shape has a configurational
entropy value of ~30𝑘. It exists both for an interstitial defect and for a
vacancy. We will estimate the recombination parameters for high and low
temperature regions in germanium
High temperature area. The temperature dependence of the
configuration entropy for the defect model considered above can be
represented in the following form (Hodge, 1997):
𝑆𝑐 (𝑇) = 𝑆∞ (1 − 𝑇𝑘 ⁄𝑇) (31)
where 𝑆∞ is the limiting value of the configurational entropy 𝑆𝑐 (at 𝑇 →

𝑇𝑚 ); 𝑇𝑚 is the melting temperature and 𝑇𝑘 is the characteristic temperature.
It is assumed that the characteristic temperature 𝑇𝑘 is the minimum
temperature at which structural imperfections arise in dislocation-free
germanium single crystals. This temperature can be estimated to be 𝑇𝑘 =
573𝐾 as the minimum temperature of formation of thermal donors in
germanium (Clauws, 2007). If we take 𝑆∞ = −30𝑘 (Cowern et al., 2013),
then we obtain 𝑆𝑐 (𝑇) = −30𝑘(1 − 573⁄𝑇).
Since the contribution of the enthalpy term ∆𝐻 is negligible, the free
energy of the recombination barrier ∆𝐺 = −𝑇 ∙ ∆𝑆. The temperature
dependence of the height of the recombination barrier is controlled by the
entropy of formation of point defects
∆𝐺(𝑇) = −𝑇 ∙ [−𝑆𝑐 (𝑇)] = 𝑇 ∙ 𝑆𝑐 (𝑇) (32)
Approximate estimation at a temperature 𝑇 = 𝑇𝑚 leads to the free

energy of the recombination barrier ∆𝐺(1211𝐾) ≈ 1.65 𝑒𝑉.
The experimental results on self-diffusion in germanium show that the
main contribution is made by vacancies, and the contribution of intrinsic
interstitial atoms is insignificant (Hüger et al., 2008; Śpiewak et al., 2007).
The diffusion coefficient obeys the Arrhenius dependence over a wide
range of temperatures 𝐷(𝑇) = 13.6𝑒𝑥𝑝(−3.094𝑒𝑉/𝑘𝑇) (Werner et al.,
1985). Approximate estimation at a temperature 𝑇 = 𝑇𝑚 leads to
𝐷(1211𝐾) ≈ 1.81·10-12 cm2·s-1.
The recombination time at high temperatures (𝜏1 ) can be evaluated

from the expression
𝜏1 = Ω⁄4𝜋 ∙ 𝐷(𝑇) ∙ 𝑟0 · 𝑒𝑥𝑝(− ∆𝐺(𝑇)⁄𝑘𝑇) (33)
where Ω is the volume of the crystal lattice and 𝑟0 = 3 ∙ 10−8cm is the

recombination radius. Approximate estimation at a temperature 𝑇 = 𝑇𝑚
leads to 𝜏1 ≈ 243 s.
The recombination factor 𝑘𝐼𝑉 (𝑇) is described by the theory of
diffusion-limited reactions together with the kinetic activation barrier
(Vanhellemont et al., 2008). At high temperatures, the recombination
factor can be written in the following form
𝑘𝐼𝑉 (𝑇) = 4𝜋 ∙ 𝑟0 ∙ 𝐷(𝑇) ∙ 𝑒𝑥𝑝(−∆𝐺(𝑇)⁄𝑘𝑇)⁄Ω ∙ 𝑐𝑠 (34)
where 𝑐𝑠 = 5·1022 cm–3 is the atomic density. The estimation at a

temperature 𝑇 = 𝑇𝑚 leads to the recombination factor 𝑘𝐼𝑉 (1211𝐾) ≈
8.22·10-26 cm·s-1. Lemke and Sudkamp (Lemke and Sudkamp, 1999)
introduced the following criterion for “rapid recombination”:
𝑘𝐼𝑉 (1211𝐾) ∙ 𝐶𝑉𝑚 ≥ 20 𝑠 −1 where 𝐶𝑉𝑚 = 1.3·1015 cm–3 is the vacancy
concentration at 𝑇 = 𝑇𝑚 (Clauws, 2007). The calculation shows that the
“fast recombination” criterion under the conditions of Voronkov's model is
not satisfied.
Low temperature area. The estimation of the recombination time at
low temperature (𝜏2 ) is carried out by the formula 𝜏2 (𝑇) = 𝜏∞ ∙
𝑒𝑥𝑝(−∆𝐺(𝑇)⁄𝑇𝑆𝑐 (𝑇)). The quantity 𝜏∞ = 661s is determined under the
condition 𝑇 = 𝑇𝑚 . Then
𝜏2 (𝑇) = 661 ∙ 𝑒𝑥𝑝(−∆𝐺(𝑇)⁄𝑇𝑆𝑐 (𝑇)) (35)
The estimate at 𝑇 = 583 K gives the value 𝜏2 → 0 (when estimating

without taking into account the vibrational entropy). Consequently, under
conditions of low-temperature studies, recombination of intrinsic point
defects proceeds at a sufficiently high rate. The theoretical calculations

carried out by us confirm the model of the entropy barrier in germanium,
the essence of which is that the decrease in the barrier is due to a decrease
in the configuration entropy with decreasing temperature (Gosele et al.,
1980; Gosele et al., 1983).
In essence, this model is as follows: the dependence of the barrier
height on the temperature is determined by the configuration of intrinsic
point defects at high temperatures. At high temperatures, the intrinsic
interstitials of silicon and vacancies are stretched into several atomic
volumes (11 atoms occupy 10 cells). around the point defect there is a
disordered region extending isotropically up to the atoms of the second
coordination sphere (Gosele et al., 1980; Gosele et al., 1983).
Recombination can occur only in the case of simultaneous compression of
both defects in the vicinity of one atomic volume. Since extended defect
configurations have a greater number of microstates than a point defect,
this compression reduces entropy. As the temperature is lowered, the
barrier decreases significantly, disappearing at low temperatures, and
defects easily recombine. This is due to a change in the configuration of
intrinsic point defects (Gosele et al., 1983).
Thermodynamic calculations show that the process of aggregation of
point defects in germanium predominates over the process of
recombination between intrinsic point defects. The contribution of the
recombination process at high temperatures to the aggregation process is
negligible. Consequently, vacancies and intrinsic interstitial atoms coexist
in thermal equilibrium. Therefore, both types of intrinsic point defects
participate in the aggregation process simultaneously. The decay of the
supersaturated solid solution of point defects is caused by cooling by two
mechanisms: vacancy and interstitial, leading to the formation of impurity-
vacancy and impurity-interstitial agglomerates. This means that defect
formation processes in silicon and germanium occur in an identical manner
and can be described using the model of high-temperature precipitation.
CONCLUSION
Theoretical description of problems of modern physics should be

accompanied by the development, with the help of information
technologies, of software products that allow obtaining and analyzing the
main results of research. The combination of physical research with the
tools and means of modern information technologies leads to the creation
of a new method for verifying the validity of theoretical constructions. In
the field of physics and material science of silicon, this approach makes it
possible to use it for the development of the foundations of defect structure
management. Actually, the theoretical models of the physical process of
solid state physics, material science, programming and the creation of
material and devices based on it are combined in one research.
This solution to the problem of defect formation in semiconductor
silicon can be considered as a standard for other crystalline materials. This
solution is based on model of high-temperature impurity precipitation.
Defective structure is transformed under the influence of technological
treatments and determines the parameters and properties of the
instruments.
For other solids, the proposed model can meet certain difficulties. For
example, for dislocation-free single-crystal germanium, many parameters
are missing, which are determined from experimental studies. The same
situation is characteristic of other solids. The solution to this problem
requires further research to determine the microscopic and macroscopic
parameters of solids.
The set of defects determines the properties of real crystals and
structures, and the additional introduction of non-equilibrium point defects
during technological actions can significantly change both the
thermodynamic state of crystals and structures, and their most important
physical properties. The management of atomic processes at the interfaces,
defect-impurity reactions in materials and structures becomes one of the
main directions in modern materials science and the basis of the new
direction in materials science of semiconductors - defect engineering.
Technology engineering defects are a necessary and inseparable part of the

technology for the production of semiconductor silicon and devices based
on it.
The use of software products makes it possible to proceed to real
defect engineering. These software products can be used:
- to simulate the engineering of defects in crystals of solids during

their growth;
- to simulate the engineering of defects in crystals of solids during
the manufacture of devices;
- for use as a tool for studying the processes of defect formation in
solids.
In turn, defect engineering allows you to control the properties and quality
of the materials used.
REFERENCES
Ammon von W., Dornberger, E. and Hansson P. O. (1999). “Bulk

properties of very large diameter silicon single crystal.” J. Cryst.
Growth. 198-199: 390-398.
Born M. and Huang K. (1954). “Dynamical theory of crystal lattices.”
Oxford Clarendon Press.
Brown R. A., Zhihong W. and Mori T. (2001). “Engineering analysis of
microdefect formation during silicon crystal growth.” J. Cryst.
Growth. 225: 97-109.
Burton B. and Speight M. V. (1986). “The coarsening and annihilation
kinetics of dislocation loops.” Phil. Mag. A. 53: 385-402.
Clauws P. (2007). “Oxygen in Germanium.” In: Germanium-Based
Technologies from Materials to Devices, edited Cor Claeys and Eddy
Simoen, 97-130. Amsterdam: Elsevier.
Cowern N.E.B., Simdyankin S., Ahn C., Bennett N.S., Goss J.P.,
Hartmann J.M., Pakfar A., Hamm S., Valentin J., Napolitani E.,
Salvador D., Bruno E., Mirabella S. (2013). “Extended point defects in
crystalline materials: Ge and Si.” Phys. Rev. Letters. 110: 155501.
Cristian J. W. (1965). “The theory of transformations in metals and
alloys.” London: Pergamon Press.
Ginzburg V. L., Landau L. D., Leontovich M. A. and Fok V. A. (1946).
“About inconsistency of works by A. A. Vlasov on general theory of
plasma and physics of solid body.” J. Exp. Theor. Phys. 16: 246-252.
Gosele U., Frank W., and Seeger A., (1980). “About the secret of self-
interstitial atoms in silicon.” J. Appl. Phys. 23: 361-367.
Gosele U., Frank W., and Seeger A., (1983). “An entropy barrier against
vacancy-interstitial recombination in silicon.” Solid State Commun.,
45: 31-33.
Hodge M.I. “Adam-Gibbs formulation of enthalpy relaxation near the glass
transition”. (1997). J. Res. Natl. Inst. Stand. Technol. 102: 195-202.
Hüger E., Tietze U., Lott D., Bracht H., Bougeard D., Haller E.E., and
Schmidt H. (2008). “Self-diffusion in germanium isotope multilayers
at low temperatures.” Applied Physics Letters 93: 162104.
Kolesnikova A. L., Romanov A. E., and Chaldyshev V. V. (2007).
“Elastic-energy relaxation in heterostructures with strained
nanoinclusions.” Physics of the Solid State. 49: 667-674.
Kulkarni M. S., Voronkov V. V. and Falster R. (2004). “Quantification of
defect dynamics in unsteady-state and steady-state Czochralski growth
of monocrystalline silicon.” J. Electrochem. Soc. 151: G663-G669.
Kulkarni M. S. (2005). “A selective review of the quantification of defect
dynamics in growing Czochralski silicon crystals.” Ind. Eng. Chem.
Res. 44: 6246-6263.
Lemke H., and Sudkamp W., (1999). “Analytical approximations for the
distributions of intrinsic point defects in grown silicon crystal.” Phys.
Stat. Sol. (a). 176: 843-865.
Mattoni A., Ippolito M. and Colombo L. (2007). “Atomistic modeling of
brittleness in covalent materials.” Phys. Rev. B, 76: 224103-224111.
Mazhukin V. I., Shapranov A. V. and Rudenko A. V. (2014).

“Comparative analysis of interatomic interaction potentials for
crystalline silicon.” Mathematica Montisnigri XXX: 56-75.
Sinno T. (1999). “Modeling microdefect formation in Czochralski silicon.”
J. Electrochem. Soc. 146: 2300-2312.
Śpiewak P., Kurzydłowski K.J., Sueoka K., Romandic I., and
Vanhellemont J., (2008). “First Principles Calculations of the
Formation Energy of the Neutral Vacancy in Germanium.” Solid State
Phenomena 131-133: 241-246.
Talanin V. I. and Talanin I. E., (2004). “Mechanism of formation and
physical classification of the grown-in microdefects in semiconductor
silicon.” Defect & Diffusion Forum 230-232: 177-198.
Talanin V. I. and Talanin I. E., (2006). “Formation of grown-in
microdefects in dislocation-free silicon monocrystals.” In: New
Research on Semiconductors, edited by Thomas B. Elliot, 31-68. New
York: Nova Sci. Publ., Inc.
Talanin V. I. and Talanin I. E., (2007a). “On the recombination of intrinsic
point defects in dislocation-free silicon single crystals.” Physics of the
Solid State 49: 467-471.
Talanin V. I., Talanin I. E., and Voronin A. A. (2007b). “About formation
of grown-in microdefects in dislocation-free silicon single crystals.”
Can. J. Phys. 85: 1459-1471.
Talanin V. I. and Talanin I. E., (2010a). “Kinetics of formation of vacancy
microvoids and interstitial dislocation loops in dislocation-free silicon
single crystals.” Physics of the Solid State. 52: 1880-1886.
Talanin V. I. and Talanin I. E., (2010b). “Kinetics of high-temperature
precipitation in dislocation-free silicon single crystals.” Physics of the
Solid State 52: 2063-2069.
Talanin V. I. and Talanin I. E., (2011). “Kinetic model of growth and
coalescence of oxygen and carbon precipitates during cooling of as-
grown silicon crystals.” Physics of the Solid State 53: 119-126.
Talanin V. I. and Talanin I. E., (2012a). “A kinetic model of the formation
and growth of interstitial dislocation loops.” J. Cryst. Growth. 346: 45-
49.
Talanin V. I., Talanin I. E., and Ustimenko N. Ph. (2012b). “Structural

scheme of information system for defect engineering of dislocation-
free silicon single crystals.” Science and Technology 2: 130-134.
Talanin V. I. and Talanin I. E., (2013). “Diffusion model of the formation
of growth microdefects as applied to the description of defect
formation in heat-treated silicon single crystals.” Physics of the Solid
State 55: 282-287.
Talanin V. I. and Talanin I. E., (2016a). “Diffusion model of the formation
of growth microdefects: a new approach to defect formation in crystals
(review).” Physics of the Solid State 58: 427-437.
Talanin V. I. and Talanin I. E., (2016b). “Complex formation in
semiconductor silicon within the framework of the Vlasov model of a
solid state.” Physics of the Solid State 58: 2050-2054.
Talanin V. I., Talanin I. E., and Lashko V. I. (2017a). “Description of the
real monocrystalline structure on the basis of the Vlasov model for
solids.” In: New research on silicon – structure, properties, technology
edited by V. I. Talanin, 5-9. Rijeka: INTECH Publ.
Talanin V. I., Talanin I. E., and Lashko V. I. (2017b). “Algorithm for
calculating the initial defect structure of semiconductor silicon”
Engineering Physics 2(4): 60-71.
Talanin V. I., Talanin I. E., and Lashko V. I. (2018). “The new software
for research and the modelling of grown-in microdefects in
dislocation-free silicon single crystals.” Engineering and Applied
Sciences 3: 1-5.
Vanhellemont J., Spiewak P., Sueoka K., Romandic I., and Simoen E.,
(2008). “Intrinsic point defect properties and engineering in silicon and
germanium Czochralski crystal growth.” In: The 5th International
Symposium on Advanced Science and Technology of Silicon Materials
(JSPS Si Symposium), Nov. 10-14, 2008, Kona, Hawaii, USA.
Vlasov A. A. (1945). “On the theory of the solid state.” Journal of Physics
9: 130-139.
Vlasov A. A. (1961). “Many-Particle Theory and Its Application to
Plasma”. New York, Gordon and Breach.
Vlasov A. A. (1978). “Nonlocal Statistical Mechanics”. Moscow, Nauka.
Voronkov V. V. (1982). “Mechanism of swirl defects formation in silicon”

J. Cryst. Growth. 59: 625-642.
Werner M., Mehrer H., and Hochheimer H.D., (1985). “Effect of
hydrostatic pressure, temperature, and doping on self-diffusion in
germanium.” Phys. Rev. B 32: 3930.
In: Mechanical Design Materials ISBN: 978-1-53614-791-9
Editor: Sandip A. Kale
c 2019 Nova Science Publishers, Inc.
Chapter 3
A DVANCED M ETHODS U SED IN M OLECULAR

DYNAMICS S IMULATION OF
M ACROMOLECULES
Hiqmet Kamberaj∗
Department of Computer Engineering
Faculty of Engineering
International Balkan University
Skopje, Republic of Macedonia
Abstract
Molecular dynamics (MD) simulations at atomic level have widely
been used in studying macromolecular systems, such as protein, DNA
and their complexes, mainly because the classical statistical mechanic’s
laws can explain different phenomena occurring at specified experimental
conditions. In this study, we will present the most advanced methods used
in the MD simulation of macromolecular systems. Furthermore, a discus-
sion of applications of these methods and perspective on developing new
approaches will be introduced. This study aims to review the methods
that are developed to enhance the conformation sampling of molecular
simulations, in particular, for observing rare events in complex molecular
systems. In the summary, we also present a discussion and perspective
∗ Corresponding Author Email: h.kamberaj@gmail.com
58 Hiqmet Kamberaj
on the methods described in this chapter and propose the new possible
improvement of these approaches, which could result in further enhance-
ment of conformation sampling.
Keywords: molecular dynamics simulation, enhanced sampling, rare events,

conformation transitions.
1. Introduction
MD approach at atomic resolution is often used to study complex biomolecu-
lar systems (M. Karplus and J.A. McCammon, 2002), mainly because the clas-
sical statistical mechanic’s laws can explain different phenomena occurring
at specified experimental conditions (van Gunsteren et al., 2006). In partic-
ular, MD is used to study the internal fluctuations (A. Amadei et al., 1993;
M. Karplus and J.A. McCammon, 2002), protein folding dynamics (Rogal and
Bolhuis, 2008), transition path sampling (P. G. Bolhuis et al., 2002), protein-
DNA, protein-protein and protein-ligand complexes, and free energy calcula-
tions (S. L. Seyler and O. Beckstein, 2014). However, standard MD simula-
tion has limited time and size scale, which makes it difficult to study typi-
cal phenomena of macromolecular systems, such as slow conformation mo-
tions (Clarage et al., 1995; Palmer, 1982). Therefore, it has been argued else-
where (G. Ciccotti and E. Vanden-Eijnden, 2015; M. K. Transtrum et al., 2015)
that these limitations may be avoided by employing new statistical and compu-
tational approaches to be studied efficiently.
There have been different efforts in developing new approaches for en-
hancement of conformation sampling of simulations using MD technique, as
discussed elsewhere (van Gunsteren et al., 2006). These MD approaches are
used in many applications for lowering conformation transition barriers by in-
creasing the rate of rare events occurrence by introducing a bias that can be
rigorously removed a posterior, or even without the bias term. Different ap-
proached have been used to introduce bias during MD simulation, such as by
changing the shape of potential energy surface until a (quasi) flat landscape is
obtained (A. Piela et al., 1989; Hamelberg et al., 2004b; Hünenberger and van
Gunsteren, 1997; Laio and Parrinello, 2002a), using soft-core potential interac-
tions (Hünenberger and van Gunsteren, 1997), conformational flooding (Grub-
Advanced Methods Used in Molecular Dynamics ... 59
müller, 1995), (geometrical) constraints (Wells et al., 2005), or using Tsallis

dynamics (Andricioaei and Straub, 1997). Other approaches include parallel
tempering, such as replica-exchange (Berg and Neuhaus, 1992; Earl and Deem,
2005; Frantz et al., 1990; Marinari and Parisi, 1992; Wang and Swendsen,
1986), multi-canonical algorithms (Y. Okamoto, 2004), and swarm-like dynam-
ics (H. Kamberaj, 2015, 2018; Huber and van Gunsteren, 1998; K. K. Burusco
et al., 2015). For gaining an increase in both time a size scale of the systems,
the coarse-grained models have also shown a great interest, for instance, by de-
creasing the number of interacting particles (I. Bahar and R. L. Jernigan, 1997;
Irbäck et al., 2000; McCammon et al., 1980; Oldziej et al., 2004; Smith and
Hall, 2001a,b; Tozzini, 2005; Tozzini and McCammon, 2005; Tozzini et al.,
2006; Y. Ueda et al., 1978), or reducing the dimensional space to only essential
degrees of freedom (Kamberaj, 2011; Lange and Grubmüller, 2006; Stepanova,
2007). Recently (Dror et al., 2011; Friedrichs et al., 2009), using computer engi-
neering, longer MD simulation runs have been reported scaling from hundreds
of microseconds to milliseconds timescale. Besides, development of multiple
time step integration numerical schemes, such as reference system propaga-
tor algorithm (RESPA), have provided other approaches for extending the time
scales of MD simulations (M.E. Tuckerman et al., 1992; Minary et al., 2004;
Tuckerman and Martyna, 2000).
In this chapter, we will describe in details some of these methods which
are most often used to improve the sampling of configuration space in the MD
simulations. We aim to critically review these methods and provide a discussion
and perspective of the approaches introduced here. Also, we will further discuss
possible improvements of some these methods, which could yield an increase
of sampling efficiency of MD simulation in studying more complex phenomena
of macromolecular systems.
2. Multiple Time Step Integrator

MD simulations of complex molecular systems, such as biomolecules char-
acterized by multiple time scales, show some disadvantage due to the small
time steps used to ensure the stability of numerical integration of the fast mo-
tions. Hence, too many time steps are needed for observation of slow confor-
mation transitions, which practically requires a large number of force computa-
60 Hiqmet Kamberaj
tions. For these reasons, the Reference System Propagator Algorithm (RESPA)
method is introduced to reduce computational efforts for simulations of such
system (Tuckerman and Berne, 1991a,b; Tuckerman et al., 1990, 1991). The
time-reversible forms of the RESPA methods have also been developed, named
r-RESPA, which have shown to be very stable concerning the order and stability
of numerical integrators (M.E. Tuckerman et al., 1992). The r-RESPA, which
will be discussed below in more details, uses Trotter factorization of the classi-
cal Liouville propagation operator (Creutz and Goksch, 1989; H.D. Raedt and
B.D. Raedt, 1983; Takahashi and Imada, 1984).
Following the discussion in literature (M.E. Tuckerman et al., 1992) (see
also Ref. (M.P. Allen and D.J. Tildesley, 1989)), for a system with f degrees of
freedom the Liouville operator, L, is defined as
f
∂ ∂
iL = {· · · , H} = ∑ ẋ j + ṗ j (1)
j=1 ∂x j ∂p j
where Cartesian coordinates are used with (x j , p j ) ≡ Γ the position and con-
jugate momenta of the system, ṗ j gives the force along the jth direction, and
{· · ·} represents the Poisson bracket of the system. L is a linear Hermitian oper-
ator of square integrable function on the phase space of Γ. The time propagation
operator as a function of L is defined by
U(t) = exp (iLt)
which is a unitary: U(−t) = U −1 (t). The position and conjugate momenta state
point of the system at a given time t is defined as Γ(t) = U(t)Γ(0), which allows
determining one time step propagation as the following:
Γ(∆t) = exp (iL∆t)Γ(0)
where ∆t = t/P is the size of a time step. Here, t is the total evolution time and
P are the number of integration points.
By splitting the Liouville operator into n different terms, like the following:
n
iL = ∑ iLk
k=1
and use the Trotter factorization scheme (H.F. Trotter, 1959), then the propaga-
tor becomes
(" #
n−1
U(t) = ∑ Uk (∆t/2) Un (∆t) (2)
k=1
" #)P
n−1
× ∑ Un−k(∆t/2) + O(t 3 /P2 )
k=1
where Uk (h) = exp (iLk h). Denoting

" #
n−1
G(∆t) = ∑ Uk(∆t/2) Un (∆t)
k=1
" #
n−1
× ∑ Un−k(∆t/2)
k=1
As shown in Ref. (M.E. Tuckerman et al., 1992), G(∆t)G(−∆t) = 1, therefore,

G(∆t) generates time-reversible dynamics.
The multiple time step integrator is based on splitting the system into the fast
and slow degrees of freedom. Equivalently, decomposing the forces entering
into the equations of motion into long-range forces, Fl (r) and short-range forces
Fs (r) (M.E. Tuckerman et al., 1992):
F(r) = Fs (r) + Fl (r)
The short-range forces in the system are related to the slow degrees of freedom,
and thus, they determine the multiple time step of the integrator δt. On the other
hand, the long-range forces are related to the fast degrees of freedom, and thus,
they determine the most extended time step of the integrator ∆t. The relationship
is established as
∆t
δt = (3)
NMT S
where NMT S is the number of multiple steps. Here, the short-range forces are
calculated every time step δt, and long-range forces are calculated after every
NMT S time steps (i.e., every time step ∆t). Hence, the degrees of freedom are ad-
vanced using ∆t as a time step. In the r-RESPA implementation, this procedure
62 Hiqmet Kamberaj
decreases the number of calls for forces calculations, which reduces, in turn, the
overall computational time.
The basic idea of r-RESPA implementation, as discussed else-
where (M.E. Tuckerman et al., 1992; Minary et al., 2004; Tuckerman and Mar-
tyna, 2000), is on determining a reference system force Fs (r) for short range
interactions. Then, Eq. (1) can be written in the following form:
f
∂ ∂ ∂
iL = ∑ ẋ j + Fs (x j ) + Fl (x j ) (4)
j=1 ∂x j ∂p j ∂p j
f
∂
= iLs + ∑ Fl (x j )
j=1 ∂p j
and the propagator operator is factorized as

f
∆t ∂
G(∆t) = ∏ exp Fl (x j ) (5)
j=1 2 ∂p j
× exp (iLs ∆t)

f
∆t ∂
× ∏ exp Fl (x j )
j=1 2 ∂p j
where the operator exp (iLs ∆t) propagates the state vector using the short range
forces with a shorter time step δt (see Eq. 3). Here, this operator is factorized
using the Trotter formula (M.E. Tuckerman et al., 1992):
"
f
δt ∂
exp (iLs ∆t) = ∏ exp Fs (x j ) (6)
j=1 2 ∂p j
f
∂
× ∏ exp δtFs (x j )ẋ j
j=1 ∂x j
f #NMT S
δt ∂
× ∏ exp Fs (x j )
j=1 2 ∂p j
Here, NMT S is usually chosen a priory to guarantee the stability of numerical

integrator (M.E. Tuckerman et al., 1992). Usually, when the operator G(∆t) is
applied to an initial state (r(0), p(0)), it gives a solution for both position and
velocity similar to Verlet numerical integrator (M.E. Tuckerman et al., 1992).
Following the discussions in Refs. (M.E. Tuckerman et al., 1992;
M. Tuckerman and M. Parrinello, 1994; W.B. Street et al., 1978), for a Lennard-
Jones type of fluid, exists only the translational relaxation time characteristic.
In that case, the integration time step can easily be chosen. On the other hand,
for biomolecules, indeed there exists more than one time-scale. For example, in
addition to the translational and rotational relaxation times, there exists the time
characterizing intra-molecular motion, such as bond stretching, angle bending,
and dihedral angle motion. Furthermore, the inter-molecular motion, including
van der Waals and electrostatic interactions, is of the typical timescale of one or
more orders in magnitude larger than intra-molecular motion. In such cases, the
system is characterized by stiff nonlinear differential equations, which require
the use of a small enough time step to observe fast motion, if treated using one
time-scale.
Other systems that are characterized by more than one timescales are those
consisting of high-frequency oscillators interacting with a bath of slow mo-
tion (Tuckerman et al., 1990), and the systems consisting of large mass particles
(slow degrees of freedom) interacting with lighter ones (fast particles) (Tucker-
man et al., 1991).
The method is also used to treat systems coupled to a Nosé heat bath (Nosé,
1984; S. Nosé, 1984a; W.G. Hoover, 1985) used to keep temperature and/or
pressure fixed during MD simulations. Here, heat bath includes extra fast
degrees of freedom into the system, treated using multiple time stepping al-
gorithms (M.E. Tuckerman et al., 1992), typically, two or more time steps.
The method has been used by many molecular simulation software codes in
performing simulations of complex systems, for example, in CHARMM pro-
gram (B. R. Brooks et al., 2009).
However, the approach is limited by the so-called resonance phenomena,
which restricts the use of time steps higher than ∆t < 8 fs by r-RESPA in
MD simulations of biomolecular systems (Bishop et al., 1997; Ma et al., 2003;
Schlick et al., 1998). It must be noted that not just time-reversible integra-
tors, but also multiple time step symplectic integrators (Skeel et al., 1997)
show numerical instability limiting the use of large time steps (Wolfram, 2002).
According to Ref. (Schlick et al., 1998), the resonance phenomena is the re-
64 Hiqmet Kamberaj
sult of using the perturbation techniques to derive the numerical integrators.

To overcome these problems, numeric methods have been introduced to in-
crease the time steps in molecular dynamics simulations. For example, non-
symplectic Langevin Molly (LM) integration method (Izaguirre et al., 2001) and
the so-called LN integrator, which combines the force separation approach with
Langevin dynamics (Barth and Schlick, 1998a,b). These methods allow using
more substantial time steps in MD simulations using stochastic approaches to
increase the numerical stability of integration. A stable version of r-RESPA
integrator has also been introduced, named the Targeted Mollified Impulse
method (Ma and Izaguirre, 2003), which includes the Langevin dynamics to
improve the accuracy of multiple time stepping integrator.
In Ref. (Minary et al., 2004), authors discuss a reversible, resonance-free
integrator which allows for using time steps of the order up to 100 fs or even
larger depending on the time length correlations studied. This integrator uses
non-Hamiltonian dynamics, which are shown to sample a canonical distribution
of physical configuration space (Minary et al., 2004)
(q1 , q2 , · · · , q3N ) ≡ ((x1 , y1 , z1 ), · · · , (xN , yN , zN ))
Here, we have written the equations of motion governing dynamics by modify-

ing those given in Ref. (Minary et al., 2004) as the following:
pi
q̇i = , (7)
mi
(s)
ṗi = Fi − λi pi − P1 pi
 
(i) (i) 2
L Q1 (ξ1, j ) (i) M
(i)
η̇(i) = − ∑  ξ2, j − ∑ ξk, j 
j=1 kB T k=2
(i) (i) (i) (i) (i) (i)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi ξ1, j
j = 1, · · · , L
(i)
(i) Gk, j (i) (i)
ξ̇k, j = (i)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
(i)
(i) GM, j
ξ̇M, j = (i)
, j = 1, · · · , L
QM, j
(s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(s) Γi,k (s) (s)
Ṗi,k = − Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(s) Γi,M
Ṗi,M = ,
Wi,M
for i = 1, 2, · · · f ( f = 3N), where

1 pi Fi
λi = (8)
2K(p, ξ) mi
!
L
(i) 1 L−1 (i) (i) (i) 2
λb =
2K(p, ξ)
−
L ∑ Q1 ξ2, j (ξ1, j ) (9)
j=1
L
p2i L − 1 (i) (i)
2K(p, ξ) = + ∑ Q1 (ξ1, j )2 (10)
mi L j=1
which ensures that maximum total kinetic energy accumulated in each degree
of freedom is LkB T . In Eq. 7, M is the Nosé-Hoover chain length of ther-
(i)
mostats, ξk, j (k = 1, · · · , M and j = 1, · · · , L) are the thermostat velocities asso-
ciated with Lagrangian multiplier along the i degrees of freedom and η(i) is the
corresponding thermostat coordinate, which is used to control the accumulated
kinetic energy fluctuations. Fi is Newton’s force on the i degrees of freedom.
(i)
The thermostat forces Gk, j are defined as
(i) (i) (i)
Gk, j = Qk−1(ξk−1, j )2 − kB T, (k = 2, 3, · · · , M) (11)
for ( j = 1, · · · , L), were L is an adjustable parameter. Here, kB is the Boltz-

(i)
mann constant and Qk determine fictitious thermostat masses optimized in
Refs. (G. J. Martyna et al., 1992; S. Nosé, 1984b):
(i)
Qk = kB T τ, k = 1, 2, · · · , M (12)
66 Hiqmet Kamberaj
(i)
where τ is a time scale associated with the thermostat. λi and λb are the La-
grangian multipliers which are determined such that equations of motion have
to satisfy the following constraint:
2K(p, ξ) = LkB T
(s)
In Eqs. 7, si,k and Pi,k are, respectively, the thermostat coordinates and their
associated velocities (k = 1, · · · , M) at temperature T for ith degrees of freedom
of the real system. Thermostat forces Γi,k are defined as
p2i
Γi,1 = − kB T (13)
mi
2
(s)
Γi,k = Wi,k−1 Pi,k−1 − kB T, k = 2, · · · , M
Wi,k are thermostat masses determined by Eq. 12, as in Refs. (G. J. Martyna
et al., 1992; S. Nosé, 1984b).
Eqs. 7 can be numerically solved using the Liouville operator formalism and
Trotter factorization schemes as suggested elsewhere (Minary et al., 2004) (and
the references therein). The classical Liouville operator can be expressed as:
f
∂ ∂
iL = ∑ v̇d + q̇d (14)
d=1 ∂v d ∂q d
M L
(d) ∂ ∂
+ ∑ ∑ ξ̇k, j (d)
+ η̇(d)
k=1 j=1 ∂ξk, j ∂η(d)

M
(s) ∂ ∂ 
+ ∑ Ṗd,k (s)
+ s˙d,k
∂sd,k
k=1 ∂Pd,k
which can then be decomposed for every degree of freedom d as

" #
f f Nd
(d) (d) (d)
iL = ∑ iL(d) = ∑ iL1 + ∑ iL2,n + iLNHC
d=1 d=1 n=1
where
M
(d) ∂ ∂ ∂
iL1 = vd + η̇(d) (d) + ∑ ṡd,k (15)
∂qd ∂η k=1 ∂sd,k
(n)
!
(d) Fd (n) (n) (s) (n) ∂
iL2,n = (n)
− λd vd − Pd,1 vd (n)
md ∂vd
(n)
Γd,1 ∂ L
(n) (d) ∂
+ − ∑ λd ξ1, j n = 1, 2, · · · , Nd
Wd,1 ∂P(s) j=1 (d)
∂ξ1, j
d,1
M L G(d) M−1 L
(d) k, j ∂ (d) (d) ∂
iLNHC = ∑∑ (d) (d)
− ∑ ∑ ξk, j ξk+1, j (d)
k=2 j=1 Qk, j ∂ξk, j k=1 j=1 ∂ξk, j
L
(d) (d) ∂
− ∑ λb ξ1, j (d)
j=1 ∂ξ1, j
M M−1
Γd,k ∂ (s) (s) ∂
+∑ (s)
− ∑ Pd,k Pd,k+1 (s)
k=2 Wd,k ∂P k=1
d,k ∂P k
where Nd is the number of parts that the force on every degree of freedom can
be split, that is
Nd
(n)
Fd = ∑ Fd
n=1
for each degree of freedom d (d = 1, 2, · · · , f ). Here, vd is the velocity of

the d-the degree of freedom, vd ≡ q̇d = pd /md . Note that it is assumed that
the force’s strength is decreasing with n. Introducing the multiple time step
parameters (Minary et al., 2004):
Nd
∆t
δt = , NMT S = ∏ sn (16)
NMT S n=1
n−1
sNd = 1, wn = ∏ sk , w1 = 1
k=1
Using the Trotter factorization scheme for classical Liouville operator, as sug-
gested in Ref. (Minary et al., 2004), then the approximation of true evolution
can be written as:
(t) h is1 −2
(t)
iL̃N δt
Γ(∆t) ≈ e d · · · eiL̃2 δt eiL̃1 δt (17)
68 Hiqmet Kamberaj
s −2
iL̃2 δt 2 iL̃Nd δt
×e · · ·e Γ(0)
where
(d) δt
eiL̃k δt = eiLNHC 2 (18)
(d) (d) (d)
iL2,1 δt2 iL1 δt ∑kn=1 iL2,n wn δt2
×e e e
(d)
iLNHC δt2
×e
for each degree of freedom d (d = 1, 2, · · · , f ). Thus, the weak or long-range

forces correspond to large values of n, and hence are calculated less often, but
they are weighted with larger wn to equalize they time step with that of short-
range forces, where
wn+1
sn =
wn
gives the ratio of strengths between (n + 1) and n forces. It can be seen that the
number of n force evaluations is NMT S /wn . The error in one time step is O(∆t 3 ),
and for the entire trajectory of length t, it is O(t∆t 2) (Minary et al., 2004). An-
alytical solutions can be obtained for each of the exponential factorized parts of
the classical Liouville operator using the following relations:

∂
exp a f (x) = f (x + a), (19)
∂x

∂
exp ax f (x) = f (ea x)
∂x
where a is a constant. Furthermore, Nosé-Hoover part iLNHC of the classical op-

erator can also be decomposed using Trotter factorization schemes as suggested
in Ref. (G. J. Martyna et al., 1996).
The method, using an unmodified version of Eq. 7 as in Ref. (Minary et al.,
2004), has been implemented in the PINY-MD software (M. E. Tuckerman
et al., 2000) and it is applied for different test systems, including a protein
studied in vacuo using CHARMM22 force field (Jr et al., 1998). The results
published in Ref. (Minary et al., 2004) have shown that large time steps of
∆ = 100 f s provided perfect agreement with other methods using much smaller
time steps. Efforts should be made to also include the solvent as a part of the
system and check the efficiency of the method in the simulation of large macro-
molecular systems in the solvent. Future work should also focus on the compar-
ison of the efficiency of sampling conformation equilibrium space of such com-
plex systems using other methods discussed below or combining this method
with other enhanced sampling techniques.
3. Generalized Ensemble Methods

It has been suggested (Hansmann and Okamoto, 1993) that generalized-
ensemble can be used for a better sampling of configurations characterized by
lower energies in computer simulations. This class of methods includes ap-
proaches, such as multicanonical sampling (Berg and Neuhaus, 1991, 1992),
the broad histogram method (de Oliveira, 1998; de Oliveira et al., 1996), Wang-
Landau algorithm (Wang and Landau, 2001), Tsallis weights methods (Tsal-
lis, 1988), and parallel tempering or replica exchange method (Geyer, 1992;
K. Hukushima and K. Nemoto, 1996; Penna, 1995). These methods are often
used to study the dynamics of biomolecular systems (Hansmann and Okamoto,
1999).
All of the above mentioned generalized-ensemble approaches have the same
starting point, that is, the replacement of canonical Boltzmann-like weights at
temperature T
exp (−β∆E)
with non-Boltzmann weights, which allows the system escaping from the local
minimum states. Here, ∆E represents the energy barrier height and β is the
inverse temperature of the simulation, β = 1/kB T .
In the canonical ensemble (characterized by fixed N, V , and T ), each state
point, (r, p), in the phase space is associated with a Boltzmann weight, which is
defined in terms of the Hamiltonian function H(r, p):
WB(r, p, β) = exp (−βH(r, p)) (20)
Since momentum p and coordinates r are independent, we can integrate ac-

cording to the momentum space Eq. 20, and re-write the Boltzmann factor as a
70 Hiqmet Kamberaj
function of the instantaneous value, E, of the potential energy function U(r):
WB(E, β) = exp (−βE) (21)
The probability distribution function of a canonical ensemble is proportional

to the product of WB (E, β) and the density of states Ω(E):
P(E, β) ∝ Ω(E)WB(E, β)
Here, Ω(E) is a monotonically increasing function of the energy E. Since

WB (E, β) is a monotonically decreasing function of E, then P(E, β) has a Gaus-
sian shape distribution with a maximum around average energy E for a fixed
inverse temperature β. In a typical MD simulation, due to sampling problems,
accurate calculation of Ω(E) is not possible, especially, at low temperatures and
complex systems, which can be trapped at some local minimum energy state.
Here, we will discuss how these weights are chosen for those methods which
are most often used in molecular dynamics simulations.
3.1. Multicanonical Sampling Method

The main aim of the multicanonical ensemble (the so-called MUCA) is to mul-
tiply the states with a non-Boltzmann multicanonical factor, Wmu (E), which
yields a uniform probability energy distribution, Pmu (E) (Berg and Neuhaus,
1991, 1992):
Pmu (E) ∝ Ω(E)Wmu(E) ≡ constant (22)
Since probability is uniform (i.e., flat), the multicanonical ensemble achieves

free random walks is in the potential energy space. In this way the system
is able to escape faster any local energy minimum state, hence enhancing the
configuration phase space sampling in an MD simulation. From Eq. 22, we can
calculate the non-Boltzmann weight as
1
Wmu(E) ≡ exp (−βEmu (E, β0 )) ∝ (23)
Ω(E)
where Emu (E, β0 ) is the multicanonical potential energy function given by
1
Emu (E, β0 ) = kB T0 ln Ω(E) = S(E) (24)
kB β0
where S(E) = kB ln Ω(E) is the entropy function of the microcanonical ensemble

and β0 is the multicanonical inverse temperature.
The density of states is practically unknown a priory, therefore, the non-
Boltzmann’s weights Wmu (E) are determined, in general, using short MD simu-
lation runs (Berg and Neuhaus, 1991, 1992), and this is one of the limitations of
standard multicanonical ensemble approach, which can be overcome by com-
bining MUCA with other methods as discussed in the following sections.
The implementation of the MUCA in MD simulation is conveniently intro-
duced by modifying equations of motion with new forces, F̃i , acting on particles
as (Bartels and Karplus, 1998; Hansmann et al., 1996; Nakajima et al., 1997):
∂Emu (E, β0 )
F̃i = − (25)
∂qi
∂Emu (E, β0 )
= Fi , i = 1, 2, · · · , f (26)
∂E
where Fi is the Newton force acting on the i degree of freedom. Eqs. 7 describing
the dynamics of a system in the multicanonical ensemble are re-written as
pi
q̇i = (27)
mi
β(E) (s)
β0
 
(i) (i) 2
L Q 1 (ξ1, j ) (i)
M
(i)
η̇(i) = − ∑  ξ2, j − ∑ ξk, j 
j=1 kB T k=2
(i) (i) (i) (i) (i) (i)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi ξ1, j
j = 1, · · · , L
(i)
(i) Gk, j (i) (i)
ξ̇k, j = (i)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
(i)
(i) GM, j
ξ̇M, j = (i)
QM, j
j = 1, · · · , L
72 Hiqmet Kamberaj
(s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(s) Γi,k (s) (s)
Ṗi,k = − Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(s) Γi,M
Ṗi,M = ,
Wi,M
where β defines the simulation inverse temperature, such as

1 ∂S(E)
β(E0 ) =
kB ∂E E0
The multicanonical weighting factor is usually determined by short trial MD

simulation runs at high temperature T0 using a canonical ensemble (Berg and
Celik, 1992; Okamoto and Hansmann, 1995), as described by Eqs. 7. From
these trial runs, we can then determine
(
(1)
Emu (E, β0 ) = E
(1)
Wmu (E, β0 ) = WB(E, β0 ) = exp(−β0 E)
A maximum value of energy Emax is determined as an average of potential

energy function at temperature T0 :
Emax = hEiT0
Then, for E ≤ Emax , a flat energy distribution is achieved, and for E > Emax, we
obtain the canonical ensemble distribution at T0 . At every MD time step, t, the
probability distribution weighting factor is given by:

W (t) (E, β0 ) = exp −β0 E (t) (E, β0 )
(t)
Then, a histogram N (t) (E) is accumulated for distribution Pmu (E) of potential
(t)
energy. Denoting by Emin the minimum energy value obtained until the t time
step. For the (t + 1) time step, the multicanonical potential energy is obtained
as


 E, E ≥ Emax

 1 (t)
 (t) (t)

 E mu (E, β 0 ) + ln N (E) −c ,


 β0
(t+1) (t)
Emu (E, β0 ) = Emin ≤ E < Emax (28)
 (t)

 β(t+1) (Emin ) (t)



 E − Emin


 β0
 (t+1) (t) (t)
+Emu (Emin , β0 ), E < Emin
where c(t) are used to ensure the continuity of energy function at E = Emax ,
determined as
1 (t)
c(t) = ln N (Emax)
β0
The MD simulation continues until a reasonably flat potential energy func-
tion is obtained, which is determined by comparing the values of energy for
all E < Emax and requiring to be of the same order of magnitude. After this
(t)
convergence is reached, Emin should be equal to the global minimum potential
energy function value. Note that during MD simulation, a polynomial or some-
times a cubic spline function (Y. Sugita and Y. Okamoto, 2000) is used to fit the
histograms each MD simulation time step (Nakajima et al., 1997).
Long MD simulation in a multicanonical ensemble is performed, after the
optimal weighting factor is obtained. Then, the ensemble average of any phys-
ical quantity, A , is determined using the Weighted Histogram Analysis Method
(WHAM) (Gallicchio et al., 2005), which is described in details, for the general
case, in Section 3.10..
3.2. The Wang-Landau Multicanonical Method

In the Wang-Landau method (WLM), a random walk in the energy space with
probability proportional to the density of states Ω(E):
1
P(E) ∝ (29)
Ω(E)
generates a flat energy distribution (Wang and Landau, 2001). To achieve this
the estimated density of states is modified systematically until a flat distribution
74 Hiqmet Kamberaj
is produced in the energy space. In this procedure, simultaneously, the density

of states converges to the true value, by controlling a so-called modification
factor at each iteration step. At the end of the simulation, this modification
factor becomes very close to one, representing a random walk with the true
density of states (Wang and Landau, 2001).
Initially, the density of states, Ω(E), is unknown, therefore, it is set to one:
Ω(E) = 1, ∀E
Then, a sampling of energy space is performed with a probability given by

Eq. 29. In general, if E1 and E2 are two energy states, then the transition proba-
bility from state E1 to E2 is

Ω(E1 )
P(E1 → E2 ) = min ,1 (30)
Ω(E2 )
Every time an energy state E is visited, we multiply existing density of states
by the factor γ > 1 (Wang and Landau, 2001)
Ω(E) → Ω(E)γ (31)
or in algorithmic scale
ln(Ω(E)) → ln (Ω(E)) + ln (γ) (32)
If the move is rejected, then E remains unchanged, and we modify the current
Ω(E) with the same factor γ. In the first publication (Wang and Landau, 2001),
the suggested initial value of factor γ is
γ = γ0 = e1 = 2.71828
which allows faster convergence of Ω(E) to the true density of states even for a
very large system. On the other hand, as discussed in Ref. (Wang and Landau,
2001), if γ0 is too small, then the convergence is extremely slow. However,
values of γ0 being too large will produce high statistical errors.
During the simulation a histogram H(E) is accumulated, representing
counts for every visited energy bin, E, by the system. After histogram becomes
flat in the sampled energy range, we say Ω(E) has converged to the true value
with an accuracy proportional to the factor ln(γ). Then, the modification factor
is decreased according to (Wang and Landau, 2001)
√
γ1 = γ0
At this moment, the histogram is reset, and the random walk sampling restarts.
Now, the density of states is multiplied by a smaller value of factor γ1 at each
step. The algorithm continues in this way, and each time that the histogram
becomes flat, the modification factor is decreased as (Wang and Landau, 2001)
√
γi+1 = γi
The algorithm stops when

γfinal = exp 10−8 ≈ 1.00000001
It can be seen that the accuracy is controlled by estimating the density of states
and the length of a simulation by a factor γ. In addition to γfinal , the accuracy of
estimating Ω(E) depends on the complexity and size of the system, criterion of
the flat histogram, and algorithm’s implementation (Wang and Landau, 2001).
In MD simulations WLM is implemented by modifying equations of motion
for the multicanonical method, as in Refs. (Hansmann et al., 1996; Nakajima
et al., 1997). Here, we purpose to modify equations of motion, given by Eqs. 27,
as the following
pi
q̇i = (33)
mi
β (s)
β0
 
(i) (i) 2
L Q1 (ξ1, j ) (i) M
(i)
η̇(i) = − ∑  ξ2, j − ∑ ξk, j 
j=1 kB T k=2
(i) (i) (i) (i) (i) (i)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi ξ1, j
j = 1, · · · , L (34)
(i)
(i) Gk, j (i) (i)
ξ̇k, j = (i)
− ξk+1, j ξk, j
Qk
76 Hiqmet Kamberaj
j = 1, · · · , L; k = 2, · · · , M − 1
(i)
(i) GM, j
ξ̇M, j = (i)
QM, j
j = 1, · · · , L
(s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(s) Γi,k (s) (s)
Ṗi,k = − Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(s) Γi,M
Ṗi,M = ,
Wi,M
where
d lnΩ(E)
β0 = (35)
dE
with β and β0 being, respectively, the inverse simulation and multicanonical
temperatures. Note that the Wang-Landau application in MD simulation con-
sists in calculation of density of states Ω(E) from Eq. 31 or Eq. 32, then using
Eq. 35 to run MD simulation (T. Nagasima et al., 2007). In order to calculate
accurately Ω(E) and hence β0 , a histogram bin of energy distribution is esti-
mated, and the bin width will determine the accuracy of calculation of Ω(E)
and β0 since it defines the ruggedness of energy distribution (T. Nagasima et al.,
2007). To smooth the ruggedness, the energy distribution is approximated by
a Gaussian distribution, and then, WHAM can be used to estimate Ω(E) and
multicanonical inverse temperature.
The method has found application to Ising spin lattice systems (Landau
et al., 2004; Wang and Landau, 2001). It has been used to study the confor-
mation transitions of proteins using confined lattice models (Pattanasiri et al.,
2012), protein folding (Wüst and Landau, 2012), and optimizing temperature
distribution in replica exchange method (H. Kamberaj and A. van der Vaart,
2009).
3.3. Tsallis Statistics Molecular Dynamics Method

In the Tsallis statistics molecular dynamics (TSMD) approach (Tsallis, 1988),
the principle of maximum generalized entropy is employed to obtain the gener-
alized statistical mechanic’s formalism. The probability weights can be deter-

mined as (Tsallis, 1988)
q
−
WT (E, β) = [1 + (q − 1)β(E − E0 )] q − 1
q is an adjustable parameter taking real values and E0 is the system’s ground en-
ergy. Note that WT (E, β) > 0. Besides, for q → 1, the Boltzmann’s weight can
be obtained, and for q > 1, probability distribution has longer tails. The long
tails of the Tsallis distribution have inspired construction of generalized distri-
butions which will enhance the excursion towards regions with higher energy by
decreasing the magnitude of the force close to these regions. This increases the
rate of barrier crossing and hence allows the system escaping the local minimum
energy states (A. Karolak and A. van der Vaart, 2012; Andricioaei and Straub,
1997; H. Kamberaj and A. van der Vaart, 2007; J. Kim and J. E. Straub, 2009;
Sugita and Okamoto, 1999).
The aim of Tsallis statistical ensemble is to weight each state by a weighting
factor, WT (E, β) (Sugita and Okamoto, 1999):
PT (E, β) ∝ Ω(E)WT(E, β) (36)
The implementation of Tsallis statistics in MD simulations is obtained by defin-

ing the Tsallis weights as the following (Sugita and Okamoto, 1999)
WT (E, β) = exp (−βUeff )
where Ueff is an effective potential defined as

q
Ueff(E, β) = ln (1 + β(q − 1)(E − E0 )) (37)
β(q − 1)
In the new generalized ensemble, MD simulations use the new potential

function Ueff, which replaces the old one E. The new forces that drive Newton’s
equations of motion are written as (Sugita and Okamoto, 1999)
∂Ueff (E, β)
F̃i = −
∂qi
∂Ueff (E, β)
= Fi
∂E
78 Hiqmet Kamberaj
1
= Fi
1 + β(q − 1)(E − E0 )
Fi is the Newton force on particle i (i = 1, 2, · · · , N). Then, the equations of mo-
tion describing a generalized canonical ensemble according to Tsallis statistics
can be given as the following:
pi
q̇i = (38)
mi
1 (s)
1 + β(q − 1)(E − E0 )
 
(i) (i) 2
L Q 1 (ξ1, j ) (i)
M
(i)
η̇(i) = − ∑  ξ2, j − ∑ ξk, j 
j=1 kB T k=2
(i) (i) (i) (i) (i) (i)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi ξ1, j
j = 1, · · · , L
(i)
(i) Gk, j (i) (i)
ξ̇k, j = (i)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
(i)
(i) GM, j
ξ̇M, j = (i)
QM, j
j = 1, · · · , L
(s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(s) Γi,k (s) (s)
Ṗi,k = − Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(s) Γi,M
Ṗi,M = ,
Wi,M
TSMD has successfully been employed to different molecular systems, such
as simulation of atomic clusters (Andricioaei and Straub, 1996, 1997),
protein folding (I. Fukuda and H. Nakamura, 2002; S. Jang et al., 2008;
U. H. E. Hansmann and Y. Okamoto, 1997; Y. Pak and S. Wang, 1999), and
molecular docking (Y. Pak and S. Wang, 2000). The approach has also been
implemented with replica exchange method by replacing Boltzmann’s weights

with Tsallis weighting factors for each replica (J. Kim and J. E. Straub, 2009;
S. Jang et al., 2003; T. W. Whitfield et al., 2002).
3.4. Swarm Particle-Like Molecular Dynamics Method

As we mentioned above, Eqs. 7 can be used to describe the Nosé-Hoover dy-
namics (Hoover, 1985; S. Nosé, 1984b) of a system of N atoms coupled to a
chain of thermostats (G. J. Martyna et al., 1992). Recently (H. Kamberaj, 2015)
a new approach was introduced based on the swarm particle social intelligence,
which is tested to improve the conformational sampling (H. Kamberaj, 2015,
2018). In this approach, in addition to the Newtonian forces, a random force
is exerted on each particle (H. Kamberaj, 2015). This is similar to Langevin
dynamics (Schlick, 2010). In particular, the MD equations of motion given by
Eqs. 7 can be modified following Ref. (H. Kamberaj, 2015) as:
pi
q̇i = (39)
mi
pLbest
q̇Lbest
i = i δ U(q) < U qLbest ,
mi
Gbest
p
q̇Gbest
i = i δ U(q) < U qGbest ,
mi
(s)
ṗi = Fi − λi pi − P1 pi (40)
m
+ ∑ Pi j γ1 u1 (cLbest
j − c j ) + γ2 u2 (cGbest
j − cj)
j=1
ṗLbest
i = −γ1 u1 (qLbest
i − qi )
Gbest
ṗi = −γ2 u2 (qGbest
i − qi )
 
(i) (i) 2
L Q 1 (ξ1, j ) (i)
M
(i)
η̇(i) = − ∑  ξ2, j − ∑ ξk, j 
j=1 kB T k=2
(i) (i) (i) (i) (i) (i)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi ξ1, j
j = 1, · · · , L
80 Hiqmet Kamberaj
(i)
(i) Gk, j (i) (i)
ξ̇k, j = (i)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
(i)
(i) GM, j
ξ̇M, j = (i)
QM, j
j = 1, · · · , L
(s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(s) Γi,k (s) (s)
Ṗi,k = − Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(s) Γi,M
Ṗi,M = ,
Wi,M
where the vector c = (c1 , c2 , · · · , cm )T characterizes the essential degrees of free-
dom in the system. Projection operator, P, transforms the real coordinates q to
the so-called collective coordinates c according to:
f
c j = ∑ Pi j qi
i=1
In Eq. 39, {cLbest

j }mj=1 and {cGbest
j }mj=1 are defined as (H. Kamberaj, 2018):
f
cLbest
j = ∑ Pi j qLbest
i
i=1
f
cGbest
j = ∑ Pi j qGbest
i
i=1
which are updated every time step. Here, qLbest is configuration vector with the
lowest value of the potential energy of the system and qGbest is configuration
vector of the final state of the system.
In Eq. 39 ui (i = 1, 2) denotes a uniformly distributed random number in
(0, 1), and γ1 and γ2 are adjustable parameters.
In Eq. 39, the δ function is given as:

Lbest
1, if U(q) < U qLbest
δ U(q) < U q =
0, otherwise
and

Gbest
1, if U(q) < U qGbest
δ U(q) < U q =
0, otherwise
The augmented dynamical system, which is given by Eq. (39) , sample an equi-
librium canonical distribution with conserved total energy given by:
f
p2i
Eext = ∑ (41)
i=1 2mi
f
2 !
pLbest
i
+ ∑ δ U(q) < U qLbest
i=1 2mi
f
2 !
pGbest
i
+ ∑ δ U(q) < U qGbest
i=1 2mi
| {z }
Etot,kin
h f 2 i
1 Lbest
2 Gbest
+U(q) +
2 ∑ 11 j
u γ q − q j + u γ
2 2 q j − q j
j=1
| {z }
Ubias
!
f M Qi,k ξ2i,k
+∑ ∑ + kB T si,k
i=1 k=1 2
| {z }
Ethermo
where f is the total number of degrees of freedom of the system ( f = 3N).

These equations represent an extended phase space of the augmented dynamical
system with real variables:
Gbest Gbest
(qi , pi ) , qLbest
i , pLbest
i , qi , pi , i = 1, 2, · · · , f
and thermostats variables:
(si,k , ξi,k ), i = 1, 2, · · · , f ; k = 1, 2, · · · , M
In Eq. 41, Etot,kin is the total kinetic energy of augmented system, Ubias is the
total potential energy including bias term, and Ethermo is the thermostat energy.
82 Hiqmet Kamberaj
WHAM is used to recover the equilibrium canonical distribution of the real sys-
tem (H. Kamberaj, 2015, 2018). The augmented dynamical system is shown
to sample metastable (H. Kamberaj, 2018) and rare transition events (Hummer
and Szabo, 2010), and to enhance the conformation sampling (Andricioaei and
Straub, 1997; J. Kim and J. E. Straub, 2009). In Eq. (39), the first bias term
steers the system towards the state with the lowest energy, which has been vis-
ited at any instant time t and hence enhancing the local basin sampling. Besides,
the second bias term indicates the “information" about configuration with the
lowest energy ever visited, and hence enhancing the barrier crossing rate.
3.5. Replica Exchange Method

Another class of methods that use the generalized distributions for sampling the
conformation phase space is also the so-called temperature Replica Exchange
Method (REM) (Earl and Deem, 2005; Falcioni and Deem, 1999; Neal, 1996;
Sugita and Okamoto, 1999; Wang and Swendsen, 1986). REM is often used
to solve the problems of quasi-ergodicity in simulations of (bio)molecular sys-
tems. In REM, replicas representing the system are simulated independently
at different temperatures (Wang and Swendsen, 1986). In particular, consider
a system of N atoms each with a mass mi , position vector ri = (xi , yi , zi ), and
conjugated momentum pi = (pxi , pyi , pzi ). In standard REM the generalized
ensemble corresponds to L independent replications of the original system cou-
pled to L thermostats at different temperatures. Using Nosé-Hoover dynam-
ics (Hoover, 1985; S. Nosé, 1984b), each replica is in equilibrium with a chain
of thermostats (G. J. Martyna et al., 1992) and the equations of motion are given
here as the following for each replica α:
pi,α
q̇i,α = (42)
mi
(α,s)
ṗi,α = Fi,α − λi,α pi,α − P1 pi,α
 
(i,α) (i,α) 2
L Q 1 (ξ1, j ) (i,α)
M
(i,α)
η̇(i,α) = − ∑  ξ2, j − ∑ ξk, j 
j=1 k T
B α k=2
(i,α) (i,α) (i,α) (i,α) (i,α) (i,α)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi,α ξ1, j
j = 1, · · · , L
(i,α)
(i,α) Gk, j (i,α) (i,α)
ξ̇k, j = (i,α)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
(i,α)
(i,α) GM, j
ξ̇M, j = (i,α)
QM, j
j = 1, · · · , L
(α) (α,s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(α)
(α,s) Γi,k (α,s) (α,s)
Ṗi,k = (α)
− Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(α)
(α,s) Γi,M
Ṗi,M = (α)
Wi,M
Two neighboring thermostats (e.g., i and j) swap at regular interval of times

their configurations (replicas) with probability, Pacc , which preserves the de-
tailed balance (Sugita and Okamoto, 1999; Wang and Swendsen, 1986):

Pacc = min 1, exp(−(β j − βi )(Ei − E j )) (43)
where Ei and E j correspond to the total energies of replicas i and j, respec-

tively. In REM, high-temperature replicas are able to cross more often energy
barrier. On the other hand, low-temperature replicas sample more often poten-
tial energy basins. It has been suggested (H. Fukunishi et al., 2002) that the
number of replicas scales as the square root of system’s degrees of freedom.
Note that increasing the number of replicas requires longer simulation runtime,
which is necessary to optimize the rate of round trips between the two extreme
temperatures.
Omitting the solvent degrees of freedom through the use of implicit or
hybrid explicit/implicit solvent models (A.E. Garcia and J.N. Onuchic, 2003;
A. Okur et al., 2006; D. Bashford and D.A. Case, 2000; R. Zhou, 2003; R. Zhou
and B.J. Berne, 2002) have increased the efficiency of REM. Another approach
includes the use of separate heat baths for the solute and solvent (X. Cheng
et al., 2005). It has been argued (A.E. Garcia and J.N. Onuchic, 2003; P. Liu
84 Hiqmet Kamberaj
et al., 2005; R. Zhou, 2003; R. Zhou and B.J. Berne, 2002) that reductions in
system size may not accurately describe the structure and dynamics of the sys-
tem.
Other approaches, similar to REM, are also proposed. For instance, the
Hamiltonian Replica Exchange Method (HREM) (H. Fukunishi et al., 2002).
An HREM with biasing the backbone dihedral potentials yielded a reduction
in the number of replicas (K. Srinivasaraghavan and M. Zacharias, 2007). In
resolution HREM approach, which uses implicit solvent models only (E. Lyman
et al., 2006; P. Liu and G.A. Voth, 2007), in addition to different temperature
couplings, the replicas exchange a subset of configuration coordinates from a
coarse-grained model (E. Lyman et al., 2006).
Use of temperature scaling for the solvent-solvent and solvent-protein inter-
actions in REM has also shown to reduce the number of replicas (P. Liu et al.,
2005), which has further been improved by using the Tsallis biasing poten-
tial (H. Kamberaj and A. van der Vaart, 2007).
Efforts have also been made to optimize the distribution of tempera-
tures among the replicas as in Refs. (A. Kone and D.A. Kofke, 2005; Berg,
2004; C. Predescu et al., 2004; D.A. Kofke, 2002; Escobedo and Martinez-
Veracoechea, 2007; Gront and Kolinski, 2007; H.G. Katzgraber et al., 2006;
Li et al., 2007; Nadler and Hansmann, 2007; N. Rathore et al., 2005; Pre-
descu et al., 2005; Sabo et al., 2008; S. Trebst et al., 2004, 2006; Sugita and
Okamoto, 1999). Some of these methods (Escobedo and Martinez-Veracoechea,
2007; H.G. Katzgraber et al., 2006; Nadler and Hansmann, 2007; S. Trebst
et al., 2004, 2006) have particularly been important in studying the protein
folding/unfolding transitions, which represent a difficult case study in standard
REM because of the low rate of accepted swaps between replicas across the
transition temperature (Huang et al., 2007; S. Trebst et al., 2006).
To further increase the efficiency of REM, other approaches to REM have
also been proposed (H. Kamberaj and A. van der Vaart, 2009), which aims to
obtain a flat generalized probability distribution function in temperature space
using the Wang-Landau algorithm (F. Wang and D.P. Landau, 2001a,b). The
method addresses two problems of REM: it increases the probability of swap-
ping, and it decreases the bottleneck for exchange at the transition temperature.
Note that a WHAM is used for analyzing the data from all replicas (see
Section 3.10.).
3.6. Swarm Particle-Like Replica Exchange Method

More recently (H. Kamberaj, 2015), a combination of replica exchange
method with Swarm Particle-like Molecular Dynamics (SPMD) is introduced.
SPMD showed to improve conformation sampling when applied to Lennard-
Jones atomic cluster systems (H. Kamberaj, 2015) and protein folding prob-
lems (H. Kamberaj, 2018) when combined with replica exchange approach.
Here, the equations of motion given in Ref. (H. Kamberaj, 2018) are modified
as the following:
pi,α
q̇i,α = (44)
mi
pLbest
i,α
q̇Lbest
i,α = δ U(qα ) < U qLbest
α ,
mi
pGbest
i,α
q̇Gbest
i,α = δ U(qα ) < U qGbest
α ,
mi
(α,s)
ṗi,α = Fi,α − λi,α pi,α − P1 pi,α (45)
m
α,Lbest (α)
+ ∑ Pi j γ1 u1 (c j −cj )
j=1

(α)
+ γ2 u2 (cGbest
j −cj )
ṗLbest Lbest
i,α = −γ1 u1 (qi,α − qi,α )
ṗGbest
i,α = −γ2 u2 (qGbest
i,α − qi,α )
 
(i,α) (i,α) 2
L Q 1 (ξ1, j ) (i,α)
M
(i,α)
η̇(i,α) = − ∑  ξ2, j − ∑ ξk, j 
j=1 kB T α k=2
(i,α) (i,α) (i,α) (i,α) (i,α) (i,α)
j = 1, · · · , L
(i,α)
(i,α) Gk, j (i,α) (i,α)
ξ̇k, j = (i,α)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
86 Hiqmet Kamberaj
(i,α)
(i,α) GM, j
ξ̇M, j = (i,α)
QM, j
j = 1, · · · , L
(α) (α,s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(α)
Ṗi,k = (α)
− Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(α)
(α,s) Γi,M
Ṗi,M = (α)
,
Wi,M
where all the variables have the same meaning as in Eq. 39 for the replica α
and {cGbest
j }mj=1 is related to the global best coordinates qGbest corresponding to
configuration with the lowest energy among all replicas through the projection
operator P:
f
cGbest
j = ∑ Pi j qGbest
i
i=1
It has been shown elsewhere (H. Kamberaj, 2018) that the Eqs. 44 preserve
the detailed balance condition. Following Ref. (H. Kamberaj, 2018), assum-
ing that a Markovian chain of states is formed, the probability of obtaining a
trajectory in the configuration space of the replica k can be written as:
T −1
Pk (XkT ) = exp (−βk E(xk,0 )) ∏ π(xk,t → xk,t+1 ) (46)
t=0
with βk being the inverse temperature of the thermostat k. In Eq. 46 E(xk,t ) is the
total energy obtained for the configuration xk,t . Here, XkT represent T replicas
of the system:
XkT = {xk,0 → xk,1 → · · · → xk,T −1 }
The initial configurations of each replica are obtained from a canonically dis-
tributed with an initial unbiased energy of the system for replica k E(xk,0 ):
ρinit (xk,0 ) = exp (−βk E(xk,0 ))

In Eq. 46, π(xk,t → xk,t+1 ) is the propagation probability at each time step,
which depend on the details of deterministic or stochastic dynamics. In general,
the Markovian transition probability π(xk,t → xk,t+1 ) can have any distribution
that conserves the Boltzmann distribution. Here, π(xk,t → xk,t+1 ) represents the
action characterized by augmented system given in Eq. (44), which produces a
Boltzmann distribution in the extended phase space of variables. In the general
case of the Newtonian dynamics, we can write:
p(xt → xt+1 ) = δ(xt+1 − Φ∆t (xt ))
where δ is the delta function and Φ∆t (xt ) is the discrete flow map of one time
step ∆t propagation operator. In this case, a trajectory can be generated using
an initial state sampled from some canonical distribution and then propagating
in time using usual Hamiltonian dynamics. Note that for Hamiltonian dynamics
is easy to find a time-reversible discrete flow map. On the other hand, when
dynamics are governed by Eq. (44), the structure is not symplectic, but still, it
is time reversible.
It is important to note that WHAM can be used for analyzing the data
from all replicas in the case of REM:SPMD simulation as presented in
Refs. (H. Kamberaj, 2015, 2018) (see Section 3.10.).
3.7. Replica Exchange Multicanonical Method

To overcome the problems of standard REM (e.g., the large number of replicas
and high computational demands) and of MUCA (e.g., difficulties on deter-
mining the weighting factor), a new method has been proposed called replica
exchange multicanonical (REMUCA) (Y. Sugita and Y. Okamoto, 2000). In
REMUCA, the first step is to perform a short replica exchange method simula-
tion (e.g., of L replicas) to calculate the multicanonical weighting factor. Then,
a standard multicanonical simulation run is performed using this weighting fac-
tor. The multiple-histogram re-weighting technique (Ferrenberg and Swendsen,
1989; Kumar et al., 1992) can be used to calculate the energy density of states
as described in the next section (see 3.10.).
After we obtain the energy density of states, the multicanonical weighting
factor is obtained from Eq. 23 and Eq. 24. Note that the multicanonical energy
Emu (E, β0 ) obtained in this way is determined in the range between hEiT1 and
hEiTL , where T1 and TL denote the lowest and highest temperature, respectively,
88 Hiqmet Kamberaj
used in the replica exchange simulation. While outside this interval the potential
energy of multicanonical simulation is determined through extrapolation:

∂Emu (E, β0 )

 (E − E1 )



 ∂E E1

 +Emu (E1 , β0 ), E < E1

(0)
Emu (E) =
 mu (E, β0 ), E
E 1 ≤ E ≤ EL (47)

 ∂E mu (E, β 0 )

 (E − EL )


 ∂E EL

+Emu (EL , β0 ), E > EL
In multicanonical MD simulations, Eq. 25 (or Eq. 27) is used to govern New-
(0)
ton’s dynamics where Emu (E, β0 ) is replaced by Emu (E). Then, after the sim-
ulation run, the trajectories are analyzed using WHAM for a single run, as de-
scribed in the next section.
Eq. 47 can also be written as

T0 T0


 (E − E1 ) + T0 S(E1 ) = E + constant,

 T1 T1

 E < E1 = hEiT1

(0)
Emu (E) = T0 S(E), E1 ≤ E ≤ EL (48)

 T0 T0

 (E − EL ) + T0 S(EL ) = E + constant,

 TL TL


E > EL = hEiTL
and the dynamical equations of motion are defined here as:
pi,α
q̇i,α = (49)
mi

β1

 Fi,α , E < E1 = hEiT1

 β
 β0 (E)

(α,s) α
ṗi,α = −λi,α pi,α − P1 pi,α + Fi,α , E1 ≤ E ≤ EL

 β0

 β
 L Fi,α ,
 E > EL = hEiTL
β0
 
(i,α) (i,α)
(i,α)
L Q1 (ξ1, j )2 (i,α) M (i,α)
η̇ =−∑  ξ2, j − ∑ ξk, j 
j=1 kB Tα k=2
(i,α) (i,α) (i,α) (i,α) (i,α) (i,α)
j = 1, · · · , L
(i,α)
(i,α) Gk, j (i,α) (i,α)
ξ̇k, j = (i,α)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
(i,α)
(i,α) GM, j
ξ̇M, j = (i,α)
QM, j
j = 1, · · · , L
(α) (α,s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(α)
Ṗi,k = (α)
− Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(α)
(α,s) Γi,M
Ṗi,M = (α)
Wi,M
From Eq. 23, the Boltzmann’s factor depends on temperature T and energy
E, and hence, scaling potential energy (and so the force) by a constant κ is
similar to scaling the temperature by 1/κ (Hansmann et al., 1996; R. Yamamoto
(0)
and W. Kob, 2000). Therefore, Emu given by Eq. 47 (or Eq. 48) generates a
canonical ensemble distribution at T = T1 , multicanonical ensemble distribution
for E1 ≤ E ≤ EL , and canonical ensemble distribution simulation at T = TL for
E > EL .
3.8. Multicanonical Replica Exchange Method

The replica exchange method in multicanonical simulation, REMUCA, can
also be introduced as a multicanonical replica exchange method (MU-
CAREM) (Y. Sugita and Y. Okamoto, 2000). In MUCAREM, the final MD
simulation run is a replica exchange with fewer replicas, say L , in contrast to
REMUCA, where the final run is a standard multicanonical MD simulation.
Since the degree of energy probability distribution overlapping in a multicanon-
ical simulation is higher compared to canonical one, a fewer number of replicas
90 Hiqmet Kamberaj
are needed for the final simulation run to guarantee an optimal distribution of
replicas among thermostats.
In MUCAREM, similar to REMUCA, short replica exchange MD simula-
tions are performed with L replicas and L thermostats, covering a temperature
range from T1 to TL . During this short simulation run, we can estimate the en-
ergy density of states Ω(E) for all range of energy using WHAM techniques.
After we define the density of states Ω(E), we can chose L pairs of thermostats
(m) (m) (m) (m)
with temperatures (TL , TH ), for m = 1, 2, · · · , L , where TL < TH . In prac-
tice, the temperatures are arranged such that ensure sufficient overlapping be-
(1) (L )
tween neighboring pairs. Here, we have TL = T1 and TH = TL , and

 E (m) = hEi (m) ,
L TL
(50)
 EH(m) = hEi (m) , m = 1, 2, · · · , L
TH
Then, we chose L thermostats at temperatures T1 , T2 , · · · , TL and assign to

each the multicanonical potential as

∂Emu (E, Tm)
(E − E (m) )





 ∂E E
(m)
 L
 +Emu (EL(m), Tm ), E < EL(m)



(m) (m) (m)
Emu (E) = Emu (E, Tm), E L ≤ E ≤ EH
(51)


 ∂E mu (E, Tm )


 (E − E (m) )


 ∂E EH
(m)

 (m) (m)
+Emu (EH , Tm ), E > EH
where the multicanonical potential energy, Emu (E, T ), is obtained for the entire
(m)
interval of energy. Also, this choice of Emu (E) generates a canonical distribu-
(m) (m) (m) (m)
tion at T = TL for E < EL , a multicanonical distribution for EL ≤ E ≤ EH ,
(m) (m)
and a canonical distribution simulation run at T = TH for E > EH .
In final step of MUCAREM, the production run is defined as a replica ex-
change simulation with L different thermostats at temperatures T1 , T2 , · · · , TL
(1) (2) (L )
and multicanonical potential energies, Emu (E), Emu (E), · · · , Emu (E). The tran-
sition probability of swapping two replicas of neighboring temperatures is given
by

(i) ( j) 1, ∆≤0
w(xm | xm+1 ) = (52)
exp(−∆), ∆ > 0
where
h i
(m+1) (m+1)
∆ = βm+1 Emu (E(q(i))) − Emu (E(q( j))) (53)
h i
(m) (m)
− βm Emu (E(q(i))) − Emu (E(q( j)))
Here, the multicanonical potential energies, E (m) (E(q( j))) and E (m+1)(E(q(i))),
have to be calculated since E (m) (E) has different values for m (Y. Sugita et al.,
2000).
Using the multiple-histogram reweighting method, the canonical distribu-
tion can be obtained (Ferrenberg and Swendsen, 1989; Kumar et al., 1992), as
presented in Section 3.10..
3.9. Tsallis Replica Exchange Methods

Tsallis’s weight factor for a configuration q at inverse temperature β` has the
following general form (J. Kim and J. E. Straub, 2009):
q
h i `
(0) 1 − q
W`(q) = 1 − β` (1 − q` )(U(q) −U` ) ` (54)
(0)
where U` is a reference minimum value of the potential energy U(q) at Tsal-
lis entropy parameter q` of replica ` associated with thermostat held at inverse
temperature β` . Note that in the limit when q` → 1, the Boltzmann’s weight of
standard MD simulation can be obtained:

WB(q) ∼ exp −β` (U −U (0) )
There may exist two approaches of applying Tsallis-like replica exchange

method. In the first implementation (named q-REM) L replicas are initially
setup running at different values of q` for ` = 1, 2, · · · , L and equal temperature
T as proposed (S. Jang et al., 2003). While in a second implementation, the
Tsallis-like dynamics is incorporated with REM in the form of the generalized
ensemble (TSREM) (T. W. Whitfield et al., 2002). In TSREM implementation,
L replicas are associated with L thermostats at different inverse temperatures β`
and Tsallis entropy parameter q` (` = 1, 2, · · · , L). Usually, in both implemen-
tations the reference replica samples the phase space using MD simulation with
92 Hiqmet Kamberaj
original potential energy function, that is, it has q` = 1 and β` = β0 , where β0 is

the required inverse temperature. On the other hand, the other replicas sample
using biased potential energy function associated with effective potential energy
that is given by Eq. 37 (i.e., q` > 1 for all replicas ` > 1).
In Tsallis-like REM a swapping between two neighboring replicas 1 and 2
has an acceptance probability given by
Pacc(1 ↔ 2) = min [1, exp (−∆12 )] (55)
where
∆12 = β2 (Ueff,2 (q1 ; q2 , T2 ) −Ueff,2 (q2 ; q2 , T2 )) (56)

+ β1 (Ueff,1 (q2 ; q1 , T1 ) −Ueff,1 (q1 ; q1 , T1 ))
where Ueff,1 and Ueff,2 are the effective Tsallis potential energy of replica 1 and
2, respectively.
Close to the barrier regions, the magnitude of the force, because of low-
ering the barriers, is reduced for q larger than one (Andricioaei and Straub,
1997; Plastino and Anteneodo, 1997). Therefore, in these regions resistance on
the particles decreases and the barrier crossing rates increases. However, the
largest value of q has to be carefully determined after some preliminary test
runs depending on the system. It is empirically suggested (H. Kamberaj and
A. van der Vaart, 2007; U. H. E. Hansmann and Y. Okamoto, 1999) an upper
value as q = 1 + 1/ f , where f is the number of degrees of freedom.
A more general approach for optimization of q values has been suggested in
Ref. (van Giessen and Straub, 2005). Based on this approach, expression of ∆12
in Eq. 56 is written as
Zq1
1 1 1
∆12 = − dz (57)
kB T2(z) T1(z)
q2
where T is the effective Tsallis temperature given as

∂(βUeff) −1
kB T (q; q, T ) =
∂U U
The performance of either q-REM or TSREM will directly depend on the

rate of accepted replica attempted swaps, defined by the average Pacc of accep-
tance probability of each swapping attempt:
Z
Pacc(1 ↔ 2) = dq1 dq2 P1 (q1 )P2 (q2 ) min[exp (−∆12 )]
where Pi (qi ) is the probability that replica i has a configuration qi or energy

U(qi ), which can be written
Pi (qi ) ≡ Pi (U) = Ω(U)WT,i(U) (58)
where Ω(U) represents the energy density of states and WT,i (U) Tsallis weight-
ing factor of replica i. Eq. 58 can further be simplified as (van Giessen and
Straub, 2005):
Z
Pacc(1 ↔ 2) = dUdU 0 θ(∆12 )P1 (U)P2 (U 0 ) (59)
Z
+ dUdU 0 θ(−∆12)P1 (U 0)P2 (U)
where θ(∆) denotes the Heaviside step function:

1, x > 0
θ(x) =
0, otherwise
and ∆i (U,U 0) = −∆i (U 0 ,U), from which we can obtain (van Giessen and
Straub, 2005):
Z
Pacc(1 ↔ 2) = 2 dUdU 0 P1 (U)P2 (U 0)θ(∆12) (60)
The optimal Pacc(1 ↔ 2) is obtained by maximizing the overlapping integrals

between two neighboring Tsallis probability distribution functions. For that, we
can approximate the function of Pi (U) by local expansion around the stationary
point U0 as (van Giessen and Straub, 2005):
1
lnPi (U) = ln Pi (U0 ) − (U −U0 )2 + · · · (61)
2σq
where U0 is the energy where both Tsallis effective temperature and statistical
temperature, TS (U) are equal:
TS (U0 ) = T (U0 )
94 Hiqmet Kamberaj
where
−1
∂S
TS (U0 ) =
∂U U0
where S(U) is the microcanonical entropy:

S(U) = kB lnΩ(U)
In Eq. (61) σq denotes the width of Tsallis probability density function at Gaus-
sian approximation given by (van Giessen and Straub, 2005):
0
TS (U) Ti 0 (U) −1
σq (U0) = − (62)
TS2 (U) Ti 2 (U) U0
where
∂TS ∂Ti
TS0 (U) = , T 0 (U) = .
∂U i ∂U
If TS (U) is assumed to be linear function of U around U0 , then TS0 (U) is a con-
stant. Furthermore, the equivalence between the microcanonical and canonical
ensembles indicates that this constant is 1/CV (T0 ), thus
TS0 (U0 ) = 1/CV (T0 )

Then, we obtain (van Giessen and Straub, 2005)
σ0
σq (U0) = (63)
1−κ
where
κ = (qi − 1)CV (T0 ), σ0 = T02CV (T0 )
where T0 = TS (U0 ) and σ0 is the Gaussian width of the canonical probability

density function at temperature T0 .
It can be seen from Eq. 63 that in the limit of qi → 1, which is the limit
of Boltzmann distribution, σq → σ0 , corresponding to a Gaussian distribution.
Moreover, if 1 < qi < 1 + qc , where
1
qc = TS0 (U0 ) =
CV (T0 )
then σq > σ0 (Eq. 63), and hence Tsallis probability density function, Pi (U),
has a broader distribution than canonical function at T0 . Whereas, for qi <
1, Pi (U) becomes narrower compare to canonical probability density function.
However, in both cases, Tsallis distribution has its maximum at stable point U0
as canonical distribution at T0 . For qi = 1 + qc , the Tsallis effective temperature,
T (U), it is tangential to canonical temperature TS (U) function at U0, and Pi (U)
is locally flat around U0 , indicating only marginal stability. Thus, the choice qi =
1 + qc generates the most delocalized Tsallis distribution for standard Tsallis
MD simulation run. For qi > qc, the local minimum of Pi (U), namely U0 , is an
unstable crossing point (van Giessen and Straub, 2005).
The WHAM is used to estimate the averages of unbiased system quanti-
ties at required temperature T0 (H. Kamberaj and A. van der Vaart, 2007). The
configuration probability density for each replica k (k = 1, 2, · · · , K) at inverse
temperature β` (` = 1, 2, · · · , L) is written as
1
P(U; qk, β` ) = q (64)
Zk`k
h i qk
(0) 1−qk
× Ω(U) 1 + β` (qk − 1)(U −U )
where U is unbiased potential energy, Ω(U) is density of states, Zk` is configu-

ration partition function:
Z h iql /(1−qk)
Zk` = dUΩ(U) 1 + β` (qk − 1)(U −U (0) )
The canonical distribution at the required inverse temperature β0 is given by

P(U; β0) = f k`P(U; qk, β` ) exp βÙkbias − β0U (65)
where Ukbias is the bias potential energy function of replica k and

q
Zk`k
fk` =
Z0
or

−1
P(U; qk, β`) = f k` P(U; β0) exp − βÙkbias − β0U (66)
Further details of the method are presented in Section 3.10..

96 Hiqmet Kamberaj
3.10. The Weighted Histogram Analysis Method

The WHAM is often used to analyze the data from replica exchange molecular
dynamics simulation. This is considered an efficient technique of data pro-
cessing since it combines all the data from replicas. In WHAM, it assumed
that K copies of the same system (namely the replicas) are in equilibrium with
L thermostats at inverse temperature β` (` = 1, 2, · · · , L). In addition, to each
replica unbiased potential energy, U`(qk ) (` = 1, 2, · · · , L; k = 1, 2, · · · , K) a bi-
asing potential energy term is added ∆U`(qk ). Then, a histogram of M bins
is created for the unbiased potential energy combining all of the replicas, with
Um (m = 1, 2, · · · , M) being the energy at the center of the bin. Thus, for each
replica k and histogram unbiased potential energy bin m we count the number
of independent snapshots, namely Hkm. The probability of observing the system
at energy bin m and thermostat ` is defined as (Gallicchio et al., 2005):
P`m = Z`−1C`m Ωm e−β0Um (67)
where Ωm = Ω(Um) is the density of states at the energy bin m and the constant
C`m determines both the effect of temperature and biasing potential in probabil-
ity distribution as:
C`m = exp (− (β` − β0 )Um) × exp (−β` ∆U`) (68)
In Eq. 67, Ωm e−β0Um gives the unbiased probability of the bin m at the target
temperature and Z` is the partition function at β` . Note that ∑M
m=1 P`m = 1 must
be satisfied. Combining Eq. 67 and Eq. 68 we obtain:
bias −F )
P`m = Ωm e−β` (Um `
(69)
where Umbias gives the value of biased potential energy at the center of bin m and
F` is the Helmholtz free energy, which has to be estimated, given as
F` = −(1/β` ) lnZ`
Let nk` be the number of saved snapshots from replica k visiting thermostat `,
then the accumulated probability density for energy bin m can be determined as:
K L
nk` bias
Pm = Ωm ∑ ∑ Nk e−β (U ` m −F` ) (70)
k=1 `=1
where Nk is the total number of saved snapshots from the replica k. Pm can also
be approximated as (Chodera et al., 2007):
K
Hkm
Pm ≈ ∑ (71)
k=1 Nk
Using the last two equations, we obtain

K
∑ Hkm
k=1
Ωm = K L
(72)
bias −F )
∑ ∑ nkè−β` (Um `
k=1 `=1
M
1 bias
F` = − ln ∑ Ωm e−βÙm
β` m=1
To take into account any possible correlations between configurations saved

from simulations, the histogram bin statistical inefficiency for each energy bin m
from replica k, gkm, can be introduced (Chodera et al., 2007), which determines
the effective number of snapshots from replica k with unbiased potential energy
eff , and the effective number of snapshots from replica k in
falling in bin m, Hkm
equilibrium with thermostat `, neffk` :
eff Hkm nk`

Hkm = ; neff
k` =
gkm gkm
Then, the estimated value of the density of states Ω̂m is given as:
K
eff
∑ Hkm
k=1
Ω̂m = K L
(73)
bias −F )
∑ ∑ neff
k` e
−β` (Um `
k=1 `=1
M
1 bias
F` = − ln ∑ Ω̂m e−βÙm
β` m=1
From Eq. 73, Ω̂m depends on F` , and F` also depends on Ω̂m . Therefore,
F` and Ω̂m are usually determined iteratively from Eqs. 73, starting from some
98 Hiqmet Kamberaj
arbitrary choice F` = 0 (` = 1, 2, · · · , L) and continuing until a convergence is

reached. The statistical error σ2Ω̂ of Ω̂m is given by (Chodera et al., 2007):
m
Ω̂m
σ2Ω̂ = K L
(74)
m
bias −F )
∑ ∑ neff
k` e
−β` (Um `
k=1 `=1
The estimated average value of any physical quantity A of the system at the
target inverse temperature β0 is computed by summing the weighted values from
all configurations:
K Nk
∑ ∑ Wkn(β0 )Akn
k=1 n=1
Â(β0 ) = (75)
K Nk
∑ ∑ Wkn(β0 )
k=1 n=1
In Eq. (75) Wkn(β0 ) are the weights given by

M
Ω̂m −β0Um
Wkn(β0 ) = ∑ e
m=1 Hkm
The chain rule of error propagation is used to obtain the statistical error of
Â(β0 ) (Chodera et al., 2007):
2 2
2 hXi σX σY2 σ2XY
σÂ = + −2 (76)
hY i (hXi)2 (hY i)2 hXihY i
where
1 Nk
hXi = ∑ Wn(β0)An
Nk n=1
(77)
1 Nk
hY i = ∑ Wn(β0)
Nk n=1
(78)
Nk
gX
σ2X = ∑ (Wn(β0)An − hXi)2
Nk (Nk − 1) n=1
(79)
Nk
gY
σY2 = ∑ (Wn(β0) − hY i)2
Nk (Nk − 1) n=1
(80)
Nk
gXY
σ2XY = ∑ (Wn (β0 )An − hXi) (81)
Nk (Nk − 1) n=1
× (Wn (β0 ) − hY i)
Here, gX(Y,XY ) are the statistical inefficiencies determined from (auto)correlation

functions of replica exchange simulations.
If ∆U` = 0 (` = 1, 2, · · · , L), the standard WHAM of replica exchange simu-
lations is obtained, discussed already in the literature. (Chodera et al., 2007)
4. Metadynamics Method
Metadynamics method has been developed by Alessandra Laio & Michele Par-
rinello (Laio and Parrinello, 2002b). The method consists on finding a limited
number of essential collective coordinates, ci (i = 1, 2, · · · , m), upon which the
free energy depends on F(c). At any moment of time t, the free energy sur-
face is explored based on the dynamical equations of motion determined by the
forces acting on the system:
∂F
fit = −
∂cti
In metadynamics method, a bias potential function Ubias(c) is constructed that
is added to the Hamiltonian function of system. This bias potential is written
as a sum of Gaussian distributions, which are added at any time t of the tra-
jectory in subspace expanded by collective coordinate as generated during the
MD simulation. The main effect of the bias potential is to not allow the system
visiting configurations that are already explored. The mathematical form of the
bias potential is (Barducci et al., 2011; Laio and Parrinello, 2002b)
!
Z t m 0 2
(ci (t) − c i (t ))
Ubias(c,t) = dt 0ω exp − ∑
0 i=1 2σ2i
Here, ω gives the rate of energy change on time:

W
ω=
τ
100 Hiqmet Kamberaj
where W and τ are the Gaussian height and the time interval of deposition,
respectively (Barducci et al., 2011). Here, σi is the Gaussian distribution width
of collective coordinate i. W , τ and σ are adjustable parameters to optimize the
algorithm (Laio and Parrinello, 2002b).
The main benefit of using metadynamics method is being able to escape
from the local minimum free energy metastable states, and hence it increases
the rate of sampling rare events. Besides, the metadynamics method allows
sampling of new reaction pathways after the system escapes local minimum
states (Barducci et al., 2011). In the metadynamics method, there is no need for
a priory knowledge of the exact topology of the free energy landscape.
After a certain long time, metadynamics technique will eventually give a
bias potential Ubias (Laio et al., 2005):
Ubias(c,t → ∞) = −F(c) +C
where C is an integration constant, F(c) is the underlying free energy of system,

defined as
Z
1
F(c) = − ln drδ(c − c(r))e−βU(r)
β
where r is the vector of coordinates, β = 1/kB T and U(r) is the potential en-
ergy function. This formula has been tested for simplified models (Laio et al.,
2005) and for other complex systems (Gervasio et al., 2005; Laio and Parrinello,
2002b). A formal proof of this expression is shown in Ref. (Bussi et al., 2006).
The free energy surface can be obtained up to an uncertainty, which is in-
versely proportional to the inverse temperature β and intrinsic diffusion coeffi-
cient D of the system in collective coordinates subspace (Gervasio et al., 2005;
Laio et al., 2005):
1/2
ω
ε∝
Dβ
In practical applications of metadynamics method, ε is estimated by comparing
different independent simulation runs (Angioletti-Uberti et al., 2010; Barducci
et al., 2006; Provasi and Filizola, 2010) or using block averaging (Pfaendtner
et al., 2009).
The main advantage of using metadynamics method include parallelization,

which is an intrinsic property of metadynamics. For instance, one can run multi-
ple interacting copies of metadynamics simulations for reconstruction of a free
energy surface, where every simulation contributes to the time-dependent po-
tential (Raiteri et al., 2006). The implementation of method yields an algorithm
that scales very well linearly with the number of processors, independent on the
type of processor.
However, there are two disadvantages of the metadynamics simulations.
The first, in single metadynamics simulation the convergence of bias potential
Ubias is not reached to a constant value, but it oscillates about a constant value,
making the criteria for stopping the simulation too tricky in practice (Barducci
et al., 2011). The second disadvantage is related to the identification of collec-
tive coordinates for describing complex topological free energy is very difficult.
The well-tempered metadynamics method (Barducci et al., 2008) provides
a solution to the first problem of standard metadynamics. In this method, the
rate of bias accumulation decreases over the course of the simulation, which is
made possible using this expression for the bias potential:

0 ωN(c,t)
Ubias(c,t) = kB ∆T ln 1 +
kB ∆T
where N(c,t) is a histogram accumulated during the simulation for collective
variables c and ∆T a free parameter with dimensions of temperature. The deriva-
0
tive Ubias (c,t) with respect to time t is

0 ωδc,c(t) U(c,t)
U̇bias(c,t) = = ω exp − δc,c(t)
ωN(c,t) kB ∆T
1+
kB ∆T
The new approach can easily be related to standard metadynamics by taking
δc,c(t) to be a Gaussian function. This is practically implemented by defining
the height of a Gaussian W as

Ubias(c,t)
W = ωτ exp −
kB ∆T
There are two main characteristics of the well-tempered metadynamics
compare to standard metadynamics (Barducci et al., 2011). The first, rate of bias
102 Hiqmet Kamberaj
accumulation decreases with simulation time as 1/t and the deviations from the
equilibrium dynamics are small. Secondly, a convergence of the bias potential
is reached up to a constant value, C, though a complete compensation of the free
energy surface is not obtained:
∆T
Ubias(c,t → ∞) = − F(c) +C
T + ∆T
In long run simulation, the collective coordinates probability distribution be-
comes:

F(c)
P(c) ∝ exp −
kB (T + ∆T )
where for ∆T → 0, standard MD simulations are recovered, and for ∆T → ∞ we
have standard metadynamics. In contrast to standard metadynamics, in well-
tempered metadynamics, the bias potential converges to a finite value during
one run. For all other values of ∆T the extent of free energy surface exploration
is determined by adjusting ∆T (Barducci et al., 2011).
5. Umbrella Sampling Methods

Umbrella sampling method is developed in Refs. (G.M. Torrie and J. P. Valleau,
1977; Torrie and Valleau, 1974). This technique adds a bias term to the potential
energy function applied to the system, which ensures efficient sampling along
the path of a reaction coordinate. The bias term can be added in a single simula-
tion run or in multiple copies of simulations (often called windows) with over-
lapping distributions. The umbrella sampling method aims to connect regions
of phase space that are separated by high energy barriers, which is a reason for
naming it the umbrella sampling.
The bias potential as a function of the reaction coordinate, let say q, has the
following form:
Ubias(q) = −F(q)
where F(q) is not usually known a priory. The umbrella sampling aims to
calculate the F(q) by employing two main types of bias potentials, such as
harmonic biases consisting of a set of windows centered at different points along
q and adaptive bias modeled to match −F(q) in only one window enveloping
the whole range of q.
5.1. Harmonic Bias Potential

In this approach, the entire range of values of q is divided into a subset of small
size windows, Nw . Then, a bias potential function is employed in every window
allowing the system to fluctuate around a reference point qref i centered at the
window i of the form (Kästner, 2011):
ki 2
Ui,bias(q) = q − qref
i (82)
2
Free energy curves calculated in this way are combined using WHAM tech-
nique (Ferrenberg and Swendsen, 1989; Kumar et al., 1992). From Eq. (82), the
bias potential Ubias is characterized by ki , which can also be adjusted depending
on the window, number of windows, Nw , and qref i .
In general, qref
i are chosen uniformly distributed in all range of q. In prac-
tice (Frenkel and Smit, 2002), there is a compromise between the statistical
errors and computational time required. For instance, increasing the number
of windows results in a smaller statistical error, but longer computational time
is needed. However, the advantage is that MD simulations in each window
are completely independent, and hence they can run in parallel by producing
multiple MD simulation copies, which takes into account the advantage of the
parallelism of computer architecture available. This approach has already been
used in atomistic simulations of protein folding (Nymeyer et al., 2004).
Combination of the umbrella sampling with replica exchange have also been
suggested (Auer and Frenkel, 2004) to improve the conformational sampling.
Strength, ki , of the bias potential, has to be chosen before simulation runs, such
that the bias potential allows steering the system across potential energy barri-
ers. On the other hand, too large ki will cause very narrow probability distri-
butions. Often, if the probability distributions have too large gaps between the
windows, then additional windows could also be inserted. Overlapping between
distributions at each window has to be sufficiently large for WHAM and it could
also be advantageous in umbrella integration (Kästner and Thiel, 2006). Large
values of ki can also lead to instability of numerical integration of the equations
of motion unless small time steps are used. In addition, for too large values of
ki only the configurations with high energies will be sampled (Straatsma and
McCammon, 1991).
The statistical error can also be derived analytically as a function of ki (Käst-
ner and Thiel, 2006). Location of the next sampled window (qref i+1 ) can be chosen
104 Hiqmet Kamberaj
as (Woolf and Grossfield, 2002)
qref ref
i+1 = qi + wi
where wi is the window width. Alternatively, the experimental data can also be
used to determine optimized values of bias parameters (Mills and Andricioaei,
2008).
5.2. Adaptive Bias Umbrella Sampling

On the other hand, this method aims to screen the entire interval of interest for
the reaction coordinate q in a single simulation run (Bartels and Karplus, 1997,
1998; Hooft et al., 1992; Laio and Parrinello, 2002a; Mezei, 1987), by choosing
a bias potential of the following form:
Ubias(q) = −F(q)
Adding this bias term to the potential yields an exactly flat energy surface, and
hence the resulting probability distribution is uniform along q. Usually, the
simulations start with an initial guess for Ubias(q), because F(q) is not known a
priory. Then, iteratively, Ubias(q) is improved until a complete uniform distribu-
tion is obtained in q space.
6. Transition Path Sampling Methods

In typical chemical reactions in solutions, the difficulty of computer simula-
tions is in understanding the rare events occurring in complex systems when
moving from one basin of attraction to another on a multimodal potential en-
ergy landscape. In particular, determining the transition state of these processes
as a function of order parameters, which have also to be defined, is notoriously
difficult problem (D. J. Wales, 2015), which will allow sampling using the stan-
dard MD simulations starting from this initial state (Anderson, 1973; Chandler,
1978; Hänggi et al., 1990; Keck, 1962). The disadvantage of this approach is
that the transition state is not always known a priory. Furthermore, because of
the high dimensionality of the problem phase space, the energy landscape has a
complex topology with many transition states, and hence the reaction coordinate
may not be represented accurately by the order parameter (Dellago et al., 1998).
On the other hand, transition path sampling (Bolhuis et al., 2000; Dellago et al.,
1998), as an alternative method, does not require previous knowledge of tran-
sition states, but it relies on the calculation of isomorphic reversible work from
reactive flux correlation functions.
In this approach, L + 1 copies of the trajectories in phase space characterize
a path in space-time:
XL = {x0 → x1 → · · · → xL }
Here, xt (t = 0, 1, · · · , L) represent points in a 2D-dimensional phase space. A

relationship between the sequence time t and physical simulation run time exists
depending on the transition path (Dellago et al., 1998), which is represented by
2D(L + 1) coordinates.
If we assume visited states form a Markovian chain, then the probability of
simulating the trajectories is given by
L−1
P (XL) exp (−βE(x0)) ∏ p(xt → xt+1 ) (83)
t=0
where β is the inverse simulation temperature and E(xt ) is the total energy at
configuration xt . Here, the initial configuration (at t = 0) is generated from a
canonical distribution (Dellago et al., 1998):
ρinit (x0 ) = exp (−βE(x0 ))
In Eq. (83), p(xt → xt+1 ) gives the transition probability for each time step,
which is determined based on the natural dynamics. Usually, any Markovian
transition probability p(xt → xt+1 ) is such that it should obey to the Boltzmann
distribution and is normalized (Dellago et al., 1998). Typically, two approaches
are proposed as a choice for p(xt → xt+1 ): the Markovian action and Langevin
action (Dellago et al., 1998). If the natural dynamics are governed by the New-
ton’s equations of motion for the Hamiltonian systems, then
p(xt → xt+1 ) = δ(xt+1 − Φ∆t (xt ))
where δ is the delta function and Φ∆t (xt ) is the discrete flow map of one-time
step propagation.
106 Hiqmet Kamberaj
A transition path sampling can be performed by applying the constraints

hA (x0 ) and hB (xL ) at the endpoints in path probability as the following:
−1
PAB (XL ) ≡ ZAB (L) hA (x0 )P (XL)hB (xL ) (84)
Z
ZAB (L) = d L x hA (x0 )P (X; L)hB(xL )
where hA (x0 ) and hB (xL ) are indicator functions defined as

1 if x ∈ A, B
hA,B (x) =
0 if x ∈ / A, B
Here, hA (x0 ) constraints the trajectory path to start in the region A (i.e., reactant)
and hB (xL ) constraints the trajectory path to stop in the region B (i.e., product).
Typically, L steps are used to take the system from states A to the state B, with
action defined by Eq. (84).
This approach has been used to probe the dynamics of folding pathways for
the C-terminal β-hairpin of protein G-B1 using MD simulation at room temper-
ature of protein in explicit solvent (Bolhuis, 2003).
It can be suggested for time propagation of the system to be governed by the
swarm particle-like dynamics given in the following form (i = 1, 2, · · · , f ):
pi
ẋi = (85)
mi
(s)
+ γ1 u1 (xLbest
i − xi ) + γ2 u2 (xGbest
i − xi )
 
(i) (i) 2
L Q 1 (ξ1, j ) (i)
M
(i)
η̇(i) = − ∑  ξ2, j − ∑ ξk, j 
j=1 kB T k=2
(i) (i) (i) (i) (i) (i)
ξ̇1, j = −ξ1, j ξ2, j − λb ξ1, j − λi ξ1, j
j = 1, · · · , L
(i)
(i) Gk, j (i) (i)
ξ̇k, j = (i)
− ξk+1, j ξk, j
Qk
j = 1, · · · , L; k = 2, · · · , M − 1
(i)
(i) GM, j
ξ̇M, j = (i)
QM, j
j = 1, · · · , L
(s)
ṡi,k = Pi,k , k = 1, 2, · · · , M
(s) Γi,k (s) (s)
Ṗi,k = − Pi,k+1 Pi,k , k = 1, 2, · · · , M − 1
Wi,k
(s) Γi,M
Ṗi,M = ,
Wi,M
where xGbest represents the coordinates of final state B (product), i.e., xL . Then,
we can generate different trajectories starting in the region A (reactant) and bi-
ased towards the end in final region B. As advantage this method does not
postulate a priory knowledge of the transition state.
7. Accelerated Molecular Dynamics Method

Accelerated molecular dynamics approach is proposed as a robust method to
bias the potential energy function to efficiently enhance the barrier crossing
during the simulations (Hamelberg et al., 2004a), based on previously intro-
duced methods (Grubmüller, 1995; Rahman and Tully, 2002; Voter, 1997).
The method has been described in details elsewhere (Hamelberg et al., 2004a,
2007). In this approach, a reference boost potential energy U0 term is defined
with a value slightly lower in magnitude than the local potential energy min-
imum (Hamelberg et al., 2004a), then each step of simulations the potential
energy U(r) is modified by a continuous non-negative bias potential ∆U(r)
as (Hamelberg et al., 2004a, 2007)
Ubias(r) = U(r) + ∆U(r) (86)
where the bias term is given as

 2
 (U0 −U(r))
If U(r) < U0
∆U(r) =
 (U0 −U(r)) + α
0 Otherwise
108 Hiqmet Kamberaj
where α is used to adjust the depth of potential energy minimum and modu-
late local smoothness of the energy basins of Ubias. In this approach, the bias
term ∆U(r) raises the potential surfaces near the minimum states and leaves
unaffected surface points near the barriers.
Another form of the bias term ∆U(r) has also been proposed (de Oliveira
et al., 2008), such as
Ubias(r) = U(r) − ∆U(r) (87)
where the bias term is given as

 2
 (U(r) −U0 )
If U(r) ≥ U0
∆U(r) =
 (U(r) −U0 ) + α
0 Otherwise
In this new MD simulation approach, transitions are accelerated by lowering the
barriers instead. With increasing α the modified landscape becomes rougher,
and it moves closer to the original potential. It is interesting to note that taking
into account the relationship between average potential energy surface rough-
ness and the diffusivity, the method allows acquiring approximately the kinetics
of original potential energy landscape (Doshi and Hamelberg, 2011; Hamelberg
et al., 2005; Xin et al., 2010).
The method has successfully been used to study the sampling of slow dif-
fusive conformation transitions of torsion angles for biomolecules in timescales
longer than milliseconds (Hamelberg et al., 2007). The approach is also tested in
the ab initio molecular dynamics simulations (Bucher et al., 2011). The method
is efficiently used to increase accuracy and convergence of free energy com-
putations in condensed-phase systems when combined with thermodynamic in-
tegration simulations (de Oliveira et al., 2008). The approach is also used as
replica exchange by varying the degree of acceleration among the replicas for
gas-phase model systems (Fajer et al., 2008). The approach is used to study the
waters contribution to the energetic roughness from peptide dynamics (Johnson
et al., 2010). Besides, the method has also shown to retrieve the kinetic rate
constant when applied in simulations of the helix to beta strand transition of
alanine dipeptide in explicit solvent (de Oliveira et al., 2007). Recently (Pierce
et al., 2012), the approach has been used to access conformation changes in
time scales of milliseconds for bovine pancreatic trypsin inhibitor protein em-
phasizing one of the method’s advantage for not needing prior knowledge of
free energy landscape or reaction coordinate.
8. Conformational Flooding Method

Helmut Grubmüller (Grubmüller, 1995) introduced a new approach examining
conformation transitions in complex macromolecular systems at the atomistic
level. In this method, first, the so-called conformation space (Ansari et al., 1987)
is defined for the system characterized by the Hamiltonian H as a restricted
region of the configuration phase space, in which system spends a long time.
Typically, this determines the time needed for the system to sample enough
phase space for the correct determination of the statistical averages (Grubmüller,
1995).
In this confined space, the free energy landscape is determined in order to
k
find an effective Hamiltonian Heff , where k indicates one of the subspaces in
configuration space. Here, it is assumed that this configuration space is made
up by regions of low free energy F, which are separated from each other by
high energy barriers of order ∆F. The free energy landscape of the subspace
is expressed in terms of the so-called collective coordinates (A. Amadei et al.,
1993; Frauenfelder et al., 1989; Go and Scheraga, 1969; Grubmüller, 1994),
characterized by the vector
c = (c1 , c2 , · · · , cm )T
where m is the number of essential degrees of freedom of the configuration sub-

space. This describes a coarse-graining of the configuration space, leaving out
3N − m degrees of freedom, with N being the total number of atoms. According
to (Grubmüller, 1995), in the subspace describe by the collective coordinates,
the conformation space density ρ̃(c) is defined as
Z
ρ̃(c) = dx0 ρ(x0)δ c − c(x0 ) (88)
where x is a 3N-dimensional Cartesian vector of the N particle positions, and

ρ(x) is the configuration space density. Hence, the free energy landscape can be
110 Hiqmet Kamberaj
evaluated as (Grubmüller, 1995):

1
F(c = − ln ρ̃(c)
β
Calculation of the ρ̃(c) requires knowledge of ρ(x), which is an integra-
tion on a 3N configuration space difficult to be evaluated in practice, since the
system has to be ergodic (Grubmüller, 1995). However, in time scales covered
by MD simulations (typically of order a few hundred nanoseconds), systems
(e.g. biomolecular systems) are non-ergodic at all time scales (see for exam-
ple (Clarage et al., 1995) or (Grubmüller, 1995) and the references therein).
In (Grubmüller, 1995), the configuration density of the subspace k, ρk (x) is
approximated as

−1 1 T −1
ρk (x) = Z exp − (x − x̄) C (x − x̄) (89)
2
In Eq. (89), the partition function, Z, is given by

1
Z
T −1
Z = dx exp − (x − x̄) C (x − x̄)
2
Here C is the covariance matrix, which is a ℜ3N×3N symmetric and positive

matrix, defined as (Grubmüller, 1995) (and the references therein):
C = h(x − x̄)(x − x̄)T ik
where
x̄ = hxik
Here, h· · ·ik denotes ensemble average in the configuration subspace k. The

matrix C is calculated based on MD trajectories, which then is diagonalized:
C = ET Λ−1 E (90)
where E and Λ are the matrix of eigenvectors and the diagonal matrix of eigen-
values, respectively. Projection of the Cartesian coordinates fluctuations along
the space spanned by eigenvectors E is
q = ET (x − x̄)
Equation 89 can be simplified as

−1 1 T
ρk (x) = Z exp − q Λq (91)
2
Coarse-graining of the configuration subspace k allows the definition of m
essential collective coordinates (A. Amadei et al., 1993)
c = (q1 , q2 , · · · , qm)T
with the largest eigenvalues, which characterize the low-frequency fluctuation
modes, and the remaining 3N − m eigenvalues, which characterize the high-
frequency fluctuation modes, are assumed not to influence the conformation
transitions (Grubmüller, 1995). The justification of this choice is based on
the fact that the first m eigenmodes are anharmonic and with high amplitude,
whereas the other 3N − m eigenmodes are essentially harmonic and with small
amplitudes, and hence only m coordinates will dominate the collective motions
in biomolecular systems (see for example discussion in Ref. (Grubmüller, 1995)
and the references therein.)
Thus, the conformation subspace density is defined as

−1 1 T
ρ̃k (c) = Z̃c exp − c Λc c
2
where Λc is a reduced matrix of m diagonal elements and Z̃c is the correspond-
ing subspace partition function. Then, the effective Hamiltonian becomes,
k 1 T
Heff (c) = c Λc c (92)
2β
This coarse-grained model for the configuration subspace Heffk is fundamen-
tal in designing the so-called flooding potential Vfl (c) chosen as a multivariate
Gaussian in order to fulfill the criteria discussed in Ref. (Grubmüller, 1995):

1 T
Vfl(c) = Efl exp − c Λfl c (93)
2
where Efl characterizes the strength of Vfl (c) and Λfl determines the shape of
Vfl (c) in conformation space, which is chosen to be
Λfl = Λc /γ2
112 Hiqmet Kamberaj
where γ is a proportionality constant specified as a function of Vfl (Grubmüller,

1995):
γ = (βEfl )1/2
The method is applied to probe conformation transitions in argon clusters

and simplified protein model (Grubmüller, 1995). Other examples used to
demonstrate the application of flooding to accelerate conformation transitions
and chemical reactions are also examined (Lange et al., 2006).
9. Discussion and Perspectives

Both industry and academic research are often using the molecular dynamics
technique and its variants to a wide range of problems and systems, from inor-
ganic and organic fluids to macromolecular. Yet, there are several issues identi-
fied in applying molecular dynamics simulations to these systems as we probed
in this work, such as time and size scale, and rare events.
The recent advances in parallel supercomputing have made possible to brace
larger spatial scales, but increasing simulation timescales remains still a chal-
lenge. For instance, simulations spanning up to microseconds in the lifetime of a
macromolecule need to cover billions of numerical integration time steps, which
is a challenge from the computation point of view. This is because in simulations
of biomolecules, at each time step, only a relatively small amount of computa-
tion can be run on parallel among a large number of processors. Hence, indeed,
billions of simulation time steps can only be executed in a considerable amount
of time. In nowadays, for molecular dynamics method development scientists,
in particular, a significant challenge is to effectively employ the computers of
near future to perform simulations of systems with millions of atoms (Hardy
et al., 2011; Phillips et al., 2014; Stone et al., 2011, 2013, 2014).
Another approach is exploitation of hardware parallel supercomputers for
MD simulations with processors that can execute traditional MD codes orders
of magnitude faster, such as Anton supercomputer (Scarpazza et al., 2013).
In long MD simulation runs up to milliseconds timescale using fully atom-
istic physical models, force field accuracy will determine the overall accuracy
achieved. Very recent studies (Piana et al., 2014) (and the references therein)
have shown that prediction of native-structures and folding rated can be more
robust concerning error compare to the potential energy function. Moreover, the
numerical integrator used in MD simulations should guarantee energy conser-
vation and stable trajectories in long time scale simulations (Gray et al., 1994;
Kamberaj, 2005). Large computer storage is also needed to save, analyze, and
better computer graphics to handle a large amount of data produced.
Coarse-grained models of macromolecular systems have probed problems
of biologically relevant size and timescale during simulations when combined
with computer power (Tozzini and McCammon, 2005). However, the coarse-
grained models of proteins remain demanding, because of the challenges in
building useful energy potential functions representing all the physics of in-
teractions (Kamberaj, 2011; Lange and Grubmüller, 2006; Stepanova, 2007).
The most tested coarse-grained model is the bead-spring model of polymers.
Solvent effects are added using Brownian dynamics (Ermak and McCammon,
1978) or Stokesian dynamics (Phung et al., 1996).
In this study, we attempted to give a big picture of the methods used to
enhance sampling in molecular dynamics simulations and thus being able to
simulate rare events for complex molecular systems. Also, we presented the ad-
vantages and limitations of each method. Our final message from this study
is that probing relevant time and size scales of (bio)physical and chemical
phenomena in macromolecular systems may need new statistical models of
data processing and computational theoretical models to allow studying them
efficiently (G. Ciccotti and E. Vanden-Eijnden, 2015; M. K. Transtrum et al.,
2015).
Acknowledgments
The author thanks International Balkan University for the support.
Conflict of Interest
The author declares that there is no conflict of interest regarding the publication
of this chapter.
114 Hiqmet Kamberaj
References
A. Amadei, A. B. M. Linssen, and H. J. C. Berendsen. Essential dynamics of
proteins. Proteins: Struct., Funct., Genet., 17:412–425, 1993.
A.E. Garcia and J.N. Onuchic. Folding a protein in a computer: An atomic

description of the folding/unfolding of protein A. Proc. Natl. Acad. Sci. USA,
100:13898–13903, 2003.
A. Karolak and A. van der Vaart. Importance of local interactions for the sta-
bility of inhibitory helix 1 of Ets-1 in the apo state. Biophys. Chem., 165-166
(3):74–8, 2012.
A. Kone and D.A. Kofke. Selection of temperature intervals for parallel-

tempering simulations. J. Chem. Phys., 122:206101, 2005.
J. B. Anderson. Statistical theories of chemical reactions. Distributions in the

transition region. J. Chem. Phys., 58:4684, 1973.
I. Andricioaei and J. E. Straub. Generalized simulated annealing algo-

rithms using Tsallis statistics: Application to conformational optimization
of a tetrapeptide. Phys. Rev. E, 53:R3055–R3058, Apr 1996. doi: 10.
1103/PhysRevE.53.R3055. URL https://link.aps.org/doi/10.1103/
PhysRevE.53.R3055.
I. Andricioaei and J. E. Straub. On Monte Carlo and molecular dynamics meth-

ods inspired by Tsallis statistics: Methodology, optimization, and applica-
tion to atomic clusters. J. Chem. Phys., 107(21):9117–9124, 1997. doi:
10.1063/1.475203. URL https://doi.org/10.1063/1.475203.
S. Angioletti-Uberti, M. Ceriotti, P. D. Lee, and M. W. Finnis. Solid-liquid

interface free energy through metadynamics simulations. Phys. Rev. B, 81:
125416, 2010.
A. Ansari, J. Berendzen, D. Braunstein, B. R. Cowen, H. Frauenfelder, M. K.

Hong, I. E. T. Iben, J. B. Johnson, P. Ormos, T. B. Sauke, R. Scholl,
A. Schulte, P. J. Steinbach, J. Vittitow, and R. D. Young. Rebinding and
relaxation in the myoglobin pocket. Biophys. Chem., 26:337–355, 1987.
A. Okur, L. Wickstrom, M. Layten, R. Geney, K. Song, V. Hornak, and

C.J. Simmerling. Improved efficiency of replica exchange simulations
through use of hybrid explicit/implicit salvation model. J. Chem. Theory and
Comput., 2:420–433, 2006.
A. Piela, J. Kostrowicki, and H. A. Scheraga. On the multiple-minima problem

in the conformational analysis of molecules: Deformation of the potential
energy hypersurface by the diffusion equation method. J. Phys. Chem., 93:
3339–3346, 1989.
S. Auer and D. Frenkel. Numerical prediction of absolute crystallisation rates

in hard-sphere colloids. J. Chem. Phys., 120(6):3015–3029, 2004.
A. Barducci, R. Chelli, P. Procacci, V. Schettino, F. L. Gervasio, and M. Par-

rinello. Metadynamics simulation of prion protein: Beta-structure stability
and the early stages of misfolding. J. Am. Chem. Soc., 128:2705–2710, 2006.
A. Barducci, G. Bussi, and M. Parrinello. Well-tempered metadynamics: A

smoothly converging and tune-able free energy method. Phys. Rev. Lett.,
100:020603, 2008.
A. Barducci, M. Bonomi, and M. Parrinello. Metadynamics. Advanced Review,

1:826–843, 2011.
C. Bartels and M. Karplus. Multidimensional adaptive umbrella sampling: Ap-

plications to main chain and side chain peptide conformations. J. Comput.
Chem., 18:1450–1462, 1997.
C. Bartels and M. Karplus. Probability distributions for complex systems:

Adaptive umbrella sampling of the potential energy. J. Phys. Chem. B, 102:
865–880, 1998.
E. Barth and T. Schlick. Overcoming stability limitations in biomolecular dy-

namics. I. Combining force splitting via extrapolation with Langevin dynam-
ics in LN. J. Chem. Phys., 109(5):1617–1632, 1998a.
E. Barth and T. Schlick. Extrapolation versus impulse in multiple-time stepping

schemes. II. Linear analysis and applications to Newtonian and Langevin dy-
namics. J. Chem. Phys., 109(5):1633–1642, 1998b.
116 Hiqmet Kamberaj
B. A. Berg and T. Celik. New approach to spin-glass simulations. Phys. Rev.

Lett., 69:2292–2295, Oct 1992. doi: 10.1103/PhysRevLett.69.2292. URL
https://link.aps.org/doi/10.1103/PhysRevLett.69.2292.
B. A. Berg and T. Neuhaus. Multicanonical algorithms for first order phase

transitions. Phys. Lett. B, 267(2):249 – 253, 1991. ISSN 0370-2693.
doi: https://doi.org/10.1016/0370-2693(91)91256-U. URL http://www.
sciencedirect.com/science/article/pii/037026939191256U.
B. A. Berg and T. Neuhaus. Multicanonical ensemble: A new approach to

simulate first-order phase transitions. Phys. Rev. Lett., 68:9–12, Jan 1992. doi:
10.1103/PhysRevLett.68.9. URL https://link.aps.org/doi/10.1103/
PhysRevLett.68.9.
B.A. Berg. Markov chain Monte Carlo simulations and their statistical analysis.
World Scientific, Singapore, 2004.
T. C. Bishop, R. D. Skeel, and K. Schulten. Difficulties with multiple time

stepping and fast multipole algorithm in molecular dynamics. J. Comput.
Chem., 18(14):1785–1791, 1997.
P. G. Bolhuis. Transition-path sampling of β-hairpin folding. Proc. Natl. Acad.

Sci. USA, 100(21):12129–12134, 2003.
P. G. Bolhuis, C. Dellago, P. L. Geissler, and D. Chandler. Transition-path

sampling: Throwing ropes over mountains in the dark. J. Phys.: Condens.
Matter, 12:A147–A152, 2000.
B. R. Brooks, C. L. Brooks, A. D. MacKerell, L. Nilsson, R. J. Petrella,

B. Roux, Y. Won, G. Archontis, C. Bartels, S. Boresch, A. Caflisch,
L. Caves, Q. Cui, A. R. Dinner, M. Feig, S. Fischer, J. Gao, M. Hodoscek,
W. Im, K. Kuczera, T. Lazaridis, J. Ma, V. Ovchinnikov, E. Paci,
R. W. Pastor, C. B. Post, J. Z. Pu, M. Schaefer, B. Tidor, R. M. Venable,
H. L. Woodcock, X. Wu, W. Yang, D. M. York, and M. Karplus. CHARMM:
The biomolecular simulation program. J. Comput. Chem., 30(10):1545–
1614, 2009.
D. Bucher, L. C. T. Pierce, J. A. McCammon, and P. R. L. Markwick. On the

use of accelerated molecular dynamics to enhance configurational sampling
in ab initio simulations. J. Chem. Theory Comput., 7:890–897, 2011.
G. Bussi, A. Laio, and M. Parrinello. Equilibrium free energies from non equi-
librium metadynamics. Phys. Rev. Lett., 96:090601, 2006.
D. Chandler. Statistical mechanics of isomerization dynamics in liquids and

the transition state approximation. J. Chem. Phys., 68(6):2959–2970, 1978.
doi: 10.1063/1.436049. URL https://aip.scitation.org/doi/abs/10.
1063/1.436049.
J. D. Chodera, W. C. Swope, J. W. Pitera, C. Seok, and K. A. Dill. Use of the

weighted histogram analysis method for the analysis of simulated and parallel
tempering simulations. J. Chem. Theory Comput., 3:26–41, 2007.
J. B. Clarage, T. Romo, B. .K. Andrews, B. M. Pettitt, and G. N. Jr. Phillips.

A sampling problem in molecular dynamics simulations of macromolecules.
Proc. Natl. Acad. Sci. USA, 92:3288–3292, 1995.
C. Predescu, M. Predescu, and C. Ciobanu. The incomplete beta function law

for parallel tempering sampling of classical canonical systems. J. Chem.
Phys., 120(9):4119–4128, 2004.
M. Creutz and A. Goksch. Higher-order hybrid Monte Carlo algorithms. Phys.

Rev. Lett., 63:9–12, Jul 1989. doi: 10.1103/PhysRevLett.63.9. URL https:
//link.aps.org/doi/10.1103/PhysRevLett.63.9.
D.A. Kofke. On the acceptance probability of replica-exchange Monte Carlo

trials. J. Chem. Phys., 117(15):6911–6914, 2002. With erratum in Ref. ?.
D. Bashford and D.A. Case. Generalized Born model of macromolecular salva-

tion effects. Annu. Rev. Phys. Chem., 51:129–152, 2000.
C. A. F. de Oliveira, D. Hamelberg, and J. A. McCammon. Estimating kinetic

rates from accelerated molecular dynamics simulations: Alanine dipeptide in
explicit solvent as a case study. J. Chem. Phys., 127:175105, 2007.
118 Hiqmet Kamberaj
C. A. F. de Oliveira, D. Hamelberg, and J. A. McCammon. Coupling accelerated

molecular dynamics methods with thermodynamic integration simulations. J.
Chem. Theory Comput., 4:1516–1525, 2008.
P. M. C. de Oliveira. Broad Histogram Simulation: Microcanonical Ising Dy-

namics. Int. J. Mod. Phys. C, 9:497–503, 1998.
P. M. C. de Oliveira, T. J. P. Penna, and H. J. Herrmann. Broad histogram

method. Braz. J. Phys., 26:677–683, 1996.
C. Dellago, P. G. Bolhuis, F. S. Csajka, and D. Chandler. Transition path sam-

pling and the calculation of rate constants. J. Chem. Phys., 108(5):1964–
1977, 1998.
D. J. Wales. Perspective: Insight into reaction coordinates and dynamics from

the potential energy landscape. J. Chem. Phys., 142:130901–12, 2015.
U. Doshi and D. Hamelberg. Extracting realistic kinetics of rare activated pro-

cesses from accelerated molecular dynamics using Kramer’s theory. J. Chem.
Theory Comput., 7:575–581, 2011.
R. O. Dror, D. H. Arlow, P. Maragakis, T. J. Mildorf, A. C. Pan, H. Xu, and

D. W. Borhani. Activation mechanism of the β2-adrenergic receptor. Proc.
Natl. Acad. Sci. USA, 108:18684–18689, 2011.
D. J. Earl and M. W. Deem. Parallel tempering: Theory, applications, and new

perspectives. Phys. Chem. Chem. Phys., 7:3910, 2005.
E. Lyman, F. M. Ytreberg, and D. M. Zuckerman. Resolution exchange simu-

lation. Phys. Rev. Lett., 96:028105, Jan 2006. doi: 10.1103/PhysRevLett.96.
028105. URL https://link.aps.org/doi/10.1103/PhysRevLett.96.
028105.
D. L. Ermak and J. A. McCammon. Brownian dynamics with hydrodynamic

interactions. J. Chem. Phys., 64:1352, 1978.
F.A. Escobedo and F.J. Martinez-Veracoechea. Optimized expanded ensembles

for simulations involving molecular insertions and deletions. i. closed sys-
tems. J. Chem. Phys., 127:174103, 2007.
M. Fajer, D. Hamelberg, and J. A. McCammon. Replica-exchange accelerated

molecular dynamics (REXAMD) applied to thermodynamic integration. J.
Chem. Theory Comput., 4:1565–1569, 2008.
M. Falcioni and M. W. Deem. A biased Monte Carlo scheme for zeolite structure
solution. J. Chem. Phys., 110(3):1754, 1999.
A. M. Ferrenberg and R. H. Swendsen. Optimized Monte Carlo data analysis.

Phys. Rev. Lett., 63:1195–1198, 1989.
D. D. Frantz, D. L. Freeman, and J. D. Doll. Reducing quasi-ergodic behavior

in Monte Carlo simulations by J-walking: Applications to atomic clusters. J.
Chem. Phys., 93(4):2769–2784, 1990. doi: 10.1063/1.458863. URL https:
//doi.org/10.1063/1.458863.
H. Frauenfelder, P. J. Steinbach, and R. D. Young. Conformational relaxation

in proteins. Chem. Scr., 23A:145–150, 1989.
D. Frenkel and B. Smit. Understanding Molecular Simulations: From Algo-

rithms to Applications. Academic Press, San Diego, CA, second edition,
2002.
M. S. Friedrichs, P. Eastman, V. Vaidyanathan, M. Houston, S. Legrand, A. L.

Beberg, D. L. Ensign, C. M. Bruns, and V. S. Pande. Accelerating molecular
dynamics simulation on graphics processing units. J. Comput. Chem., 30:
864–72, 2009.
F. Wang and D.P. Landau. Efficient, multiple-range random walk algorithm to

calculate the density of states. Phys. Rev. Lett., 86(10):2050–2053, 2001a.
F. Wang and D.P. Landau. Determining the density of states for classical sta-
tistical models: A random walk algorithm to produce a flat histogram. Phys.
Rev. E, 64:056101–16, 2001b.
E. Gallicchio, M. Andrec, A. K. Felts, and R. M. Levy. Temperature weighted

histogram analysis method, replica exchange, and transition paths. J. Phys.
Chem. B, 109:6722–6731, 2005.
120 Hiqmet Kamberaj
G. Ciccotti and E. Vanden-Eijnden. The trees and the forest. Aims and objec-
tives of molecular dynamics simulations. Eur. Phys. J. Special Topics, 224:
2515–2518, 2015.
F. L. Gervasio, A. Laio, and M. Parrinello. Flexible docking in solution using

metadynamics. J. Am. Chem. Soc., 127:2600–2607, 2005.
G. J. Geyer. Practical Markov Chain Monte Carlo. Stat. Sci., 7(4):473–483,

1992.
G. J. Martyna, M. L. Klein, and M. E. Tuckerman. Nosé-Hoover chains: The

canonical ensemble via continuous dynamics. J. Chem. Phys., 97(4):2635–
2643, 1992. doi: 10.1063/1.463940. URL https://doi.org/10.1063/1.
463940.
G. J. Martyna, M. E. Tuckerman, D. J. Tobias, and M. L. Klein. Explicit re-

versible integrators for extended systems dynamics. Mol. Phys., 87(5):1117–
1157, 1996.
G.M. Torrie and J. P. Valleau. Nonphysical sampling distributions in Monte-

Carlo free-energy estimation: Umbrella sampling. J. Comput. Phys., 23(2):
187–199, 1977.
N. Go and H. A. Scheraga. Analysis of the contribution of internal vibrations

to the statistical weights of equilibrium conformations of macromolecules.
J. Chem. Phys., 51(11):4751–4767, 1969. doi: 10.1063/1.1671863. URL
https://doi.org/10.1063/1.1671863.
S. K. Gray, D. W. Noid, and B. G. Sumpter. Symplectic integrators for large

scale molecular dynamics simulations: A comparison of several explicit
methods. J. Chem. Phys., 101(5):4062–4072, 1994.
D. Gront and A. Kolinski. Efficient scheme for optimization of parallel temper-

ing Monte Carlo method. J. Phys.: Condens. Matter, 19:036225, 2007.
H. Grubmüller. PhD thesis, Technische Universität München, 1994.
H. Grubmüller. Predicting slow structural transitions in macromolecular sys-

tems: Conformational flooding. Phys. Rev. E, 52:2893–2906, 1995.
D. Hamelberg, J. Mongan, and J. A. McCammon. Accelerated molecular dy-

namics: A promising and efficient simulation method for biomolecules. J.
Chem. Phys., 120:11919, 2004a.
D. Hamelberg, J. Mongan, and J.A. McCammon. Accelerated molecular dy-
namics: A promising and efficient simulation method for biomolecules. J.
Chem. Phys., 120(24):11919–11929, 2004b.
D. Hamelberg, T. Shen, and J. A. McCammon. Relating kinetic rates and lo-
cal energetic roughness by accelerated molecular-dynamics simulations. J.
Chem. Phys., 122:241103, 2005.
D. Hamelberg, C. A. F. de Oliveira, and J. A. McCammon. Sampling of slow
diffusive conformational transitions with accelerated molecular dynamics. J.
Chem. Phys., 127(15):155102, 2007.
P. Hänggi, P. Talkner, and M. Borkovec. Reaction-rate theory: Fifty years after
Kramers. Rev. Mod. Phys., 62:251, 1990.
U. H. E. Hansmann and Y. Okamoto. In Annual reviews in computational
physics VI. (World Scientific, Singapore, 1999.
U. H. E. Hansmann and Y. Okamoto. Prediction of peptide conformation by
multicanonical algorithm: New approach to the multiple-minima problem. J.
Comput. Chem., 14:1333–1338, 1993.
U. H. E. Hansmann, Y. Okamoto, and F. Eisenmenger. Molecular dynamics,
Langevin and hydrid Monte Carlo simulations in a multicanonical ensemble.
Chem. Phys. Lett., 259:321–330, 1996.
D. .J. Hardy, J. .E. Stone, K. L. Vandivort, D. Gohara, C. Rodrigues, and
K. Schulten. Fast molecular electrostatics algorithms on GPUs. In In Wen
mei Hwu, editor, GPU Computing Gems, pages 43–58. Morgan Kaufmann
Publishers, 2011.
H.D. Raedt and B.D. Raedt. Applications of the generalized Trotter formula.
Phys. Rev. A, 28(6):3575, 1983.
H.F. Trotter. On the product of semi-groups of operators. Proc. Am. Math. Soc.,
10:545–551, 1959.
122 Hiqmet Kamberaj
H. Fukunishi, O. Watanabe, and S. Takada. On the Hamiltonian replica ex-

change method for efficient sampling of biomolecular systems: Aplacation
to protein structure prediction. J. Chem. Phys., 116:9058–9067, 2002.
H.G. Katzgraber, S. Trebst, D.A. Huse, and M. Troyer. Feedback-optimized

parallel tempering Monte Carlo. J. Stat. Mech., page P03018, 2006.
H. Kamberaj. Conformational sampling enhancement of replica exchange

molecular dynamics simulations using swarm particle intelligence. J. Chem.
Phys., 143:124105–8, 2015.
H. Kamberaj. Faster protein folding using enhanced conformational sampling

of molecular dynamics simulation. J. Mol. Graph. Model., 81:32–49, 2018.
H. Kamberaj and A. van der Vaart. Multiple scaling replica exchange for the
conformational sampling of biomolecules in explicit water. J. Chem. Phys.,
127:234102–234109, 2007.
H. Kamberaj and A. van der Vaart. An optimised replica exchange method for
molecular dynamics simulations. J. Chem. Phys., 130:074904, 2009.
R. W. .W. Hooft, B. P. van Eijck, and J. Kroon. An adaptive umbrella sam-

pling procedure in conformational analysis using molecular dynamics and its
application to glycol. J. Chem. Phys., 97:6690–6694, 1992.
W.G. Hoover. Canonical dynamics: Equilibrium phase-space distributions.

Phys. Rev. A, 31:1695–1697, 1985.
X. Huang, M. Hagen, B. Kim, R.A. Friesner, R. Zhou, and B.J. Berne. Replica
exchange with solute tempering: efficiency in large scale systems. J. Phys.
Chem. B, 111:5405–5410, 2007.
T. Huber and W. F. van Gunsteren. SWARM-MD: Searching conformational

space by cooperative molecular dynamics. J. Phys. Chem. A, 102:5937–5943,
1998.
G. Hummer and A. Szabo. Free energy profiles from single-molecule pulling

experiments. Proc. Natl. Acad. Sci. USA, 107(50):21441–21446, 2010.
P. H. Hünenberger and W. F. van Gunsteren. Computer Simulation of Biomolec-

ular Systems, Theoretical and Experimental Applications. Kluwer, Dor-
drecht, The Netherlands, 1997.
I. Bahar and R. L. Jernigan. Inter-residue potentials in globular proteins and the

dominance of highly specific hydrophilic interactions at close separations. J.
Mol. Biol., 266:195–214, 1997.
I. Fukuda and H. Nakamura. Tsallis dynamics using the Nosé-Hoover ap-

proach. Phys. Rev. E, 65:026105, 2002.
A. Irbäck, F. Sjunnesson, and S. Wallin. Three-helix-bundle protein in a Ra-

machandran model. Proc. Natl. Acad. Sci. U.S.A., 97:13614–13618, 2000.
J. A. Izaguirre, D. P. Catarello, J. M. Wozniak, and R. D. Skeel. Langevin sta-

bilization of molecular dynamics. J. Chem. Phys., 114(5):2090–2098, 2001.
J. Kim and J. E. Straub. Optimal replica exchange method combined with Tsal-
lis weight sampling. J. Chem. Phys., 130:144114–11, 2009.
Q. Johnson, U. Doshi, T. Chen, and D. Hamelberg. Water’s contribution to the

energetic roughness from peptide dynamics. J. Chem. Theory Comput., 6:
2591–2597, 2010.
A. D. MacKerell Jr, D. Bashford, M. Bellot, R. L. Dunbrack Jr., J. D. Evanseck,

M. J. Field, S. Fisher, J. Gao, H. Guo, S. Ha, D. Joseph-McCarthy, L. Kuchnir,
K. Kuczera, F. T. K. Lau, C. Mattos, S. Michnick, T. Ngo, D. T. Nguyen,
B. Prodhom, W. E. Reiher, B. Roux, M. Schlenkrich, J. C. Smith, R. Stote,
J. Straub, M. Watanabe, J. Wiorkiewicz-Kuczera, D. Yin, and M. Karplus.
All-atom empirical potential for molecular modelling and dynamics studies
of proteins. J. Phys. Chem. B, 102(18):3586–3616, 1998.
H. Kamberaj. A theoretical model for the collective motion of proteins by means

of principal component analysis. Cent. Eur. J. Phys., 9(1):96–109, 2011.
H. Kamberaj. PhD thesis, Manchester Metropolitan University, Manchester,

UK, May 2005.
J. Kästner. Umbrella sampling. Advanced Review, 00:1–11, 2011.

124 Hiqmet Kamberaj
J. Kästner and W. Thiel. Analysis of the statistical error in umbrella sampling

simulations by umbrella integration. J. Chem. Phys., 124:234106(1–7), 2006.
J. C. Keck. Statistical investigation of dissociation cross-sections for diatoms.

Discuss. Faraday Soc., 33:173–182, 1962.
K. Hukushima and K. Nemoto. Exchange Monte Carlo method and application

to spin glass simulations. J. Phys. Soc. Jpn., 65:1604–1608, 1996.
K. K. Burusco, N. J. Bruce, I. Alibay, and R. A. Bryce. Free energy calcu-

lations using a swarm-enhanced sampling molecular dynamics approach.
Chem. Phys. Chem., 16:3233–3241, 2015.
K. Srinivasaraghavan and M. Zacharias. Enhanced sampling of peptide and pro-

tein conformations using replica exchange simulations with a peptide back-
bone biasing potential. Proteins: Struct., Funct., Bioinf., 66:697–706, 2007.
S. Kumar, J. M. Rosenberg, D. Bouzida, R. H. Swendsen, and P. A. Koll-

man. The weighted histogram analysis method for free-energy calculations
on biomolecules. I. The method. J. Comput. Chem., 13(8):1011–1021, 1992.
A. Laio and M. Parrinello. Escaping free-energy minima. Proc. Natl. Acad. Sci.
USA, 99:12562–12566, 2002a.
A. Laio and M. Parrinello. Escaping free-energy minima. Proc. Natl. Acad. Sci.
USA, 99(20):12562–12566, 2002b.
A. Laio, A. F. L. Rodriguez, F. Gervasio, M. Ceccarelli, and M. Parrinello.

Assessing the accuracy of metadynamics. J. Phys. Chem. B, 109:6714–6721,
2005.
D. P. Landau, S.-H. Tsai, and M. Exler. A new approach to Monte Carlo simu-
lations in statistical physics: Wang-Landau sampling. Am. J. Phys., 72(10):
1294–1302, 2004.
O. F. Lange and H. Grubmüller. Can principal components yield a dimension

reduced description of protein dynamics on long time scales? J. Phys. Chem.
B, 110:22842–22852, 2006.
O. F. Lange, L. V. Schäfer, and H. Grubmüller. Flooding in GROMACS: ac-

celerated barrier crossing in molecular dynamics. J. Comput. Chem., 27(14):
1693–1702, 2006.
X. Li, C.P. O’Brien, G. Collier, N.A. Vellore, F. Wang, and R.A. Latour. An im-
proved replica-exchange sampling method: Temperature intervals with global
energy reassignment. J. Chem. Phys., 127:164116, 2007.
Q. Ma and J. A. Izaguirre. Targeted mollified impulse: A multi scale stochas-

tic integrator for long molecular dynamics simulations. Multiscale Model.
Simul., 2(1):1–21, 2003.
Q. Ma, J. A. Izaguirre, and R. D. Skeel. Verlet-I/r-RESPA/Impulse is limited by

nonlinear instabilities. SIAM J. Sci. Comput., 24:1951, 2003.
E. Marinari and G. Parisi. Simulated tempering: a new Monte Carlo scheme.

Europhys. Lett., 19:451, 1992.
J. A. McCammon, S. H. Northrup, M. Karplus, and R. M. Levy. Helix-coil tran-

sitions in a simple polypeptide model. Biopolymers, 19:2033–2045, 1980.
M.E. Tuckerman, B.J. Berne, and G.J. Martyna. Reversible multiple time scale
molecular dynamics. J. Chem. Phys., 97(3):1990, 1992.
M. E. Tuckerman, D. A. Yarne, S. O. Samuelson, A. L. Hughes, and

G. J. Martyna. Exploiting multiple levels of parallelism in molecular
dynamics based calculations via modern techniques and software paradigms
on distributed memory computers. Comput. Phys. Commun., 128:333, 2000.
M. Mezei. Adaptive umbrella sampling: Self-consistent determination of the

non-Boltzmann bias. J. Comput. Phys., 68:237–248, 1987.
M. Mills and I. Andricioaei. An experimentally guided umbrella sampling pro-

tocol for biomolecules. J. Chem. Phys., 129:114101(1–13), 2008.
P. Minary, M. E. Tuckerman, and G. J. Martyna. Long time molecular dynamics

for enhanced conformational sampling in biomolecular systems. Phys. Rev.
Lett., 93(150201-4), 2004.
126 Hiqmet Kamberaj
M. Karplus and J.A. McCammon. Molecular dynamics simulations of

biomolecules. Nat. Struct. Biol., 9(9):646–652, 2002. With corrigenda in
Nat. Struct. Biol. 9(10),788 (2002).
M. K. Transtrum, B. B. Machta, K. S. Brown, B. C. Daniels, C. R. Myers, and

J. P. Sethna. Perspective: Sloppiness and emergent theories in physics, biol-
ogy, and beyond. J. Chem. Phys., 143:010901–13, 2015.
M.P. Allen and D.J. Tildesley. Computer simulation of liquids. Oxford Univer-
sity Press, 1989.
M. Tuckerman and M. Parrinello. Integrating the Car-Parrinello equations. I.

Basic integration techniques. J. Chem. Phys., 101(2):1302, 1994.
W. Nadler and U.H.E. Hansmann. Generalized ensemble and tempering simu-

lations: A unified view. Phys. Rev. E, 75:026109, 2007.
N. Nakajima, H. Nakamura, and A. Kidera. Multicanonical ensemble generated

by molecular dynamics simulation for enhanced conformational sampling of
peptides. J. Phys. Chem. B, 101:817–824, 1997.
R. M. Neal. Sampling from multimodal distributions using tempered transitions.

Statistics and Computing, 6(4):353–366, 1996.
S. Nosé. A molecular dynamics method for simulations in the canonical ensem-

ble. Mol. Phys., 52:255, 1984.
N. Rathore, M. Chopra, and J.J. de Pablo. Optimal allocation of replicas in par-

allel tempering simulations. J. Chem. Phys., 122:024111, 2005.
H. Nymeyer, S. Gnanakaran, and A. E. Garcia. Atomistic simulations of protein

folding. using the replica exchange algorithm. Methods in Enzymology, 383:
119–149, 2004.
Y. Okamoto and U. H. E. Hansmann. Thermodynamics of helix-coil transi-

tions studied by multicanonical algorithms. J. Phys. Chem., 99:11276–11287,
1995.
S. Oldziej, A. Liwo, C. Czaplewski, J. Pillardy, and H. A. Scheraga. Optimiza-

tion of the UNRES force field by hierarchical design of the potential-energy
landscape. 2. Off-lattice tests of the method with single proteins. J. Phys.
Chem. B, 108:16934–16949, 2004.
R. Palmer. Broken ergodicity. Adv. Phys., 32:669–735, 1982.
B. Pattanasiri, Y. W. Li, D. P. Landau, T. Wüst, and W. Triampo. Conformation
transitions of a confined lattice protein: A Wang-Landau study. J. Phys.:
Conf. Series, 402:012048, 2012.
T. J. P. Penna. Traveling salesman problem and Tsallis statistics. Phys. Rev. E,
51:R1, 1995.
J. Pfaendtner, D. Brabduardi, M. Parrinello, T. D. Pollard, and G. A. Voth.
Nucleotide-dependent conformational states of actin. Proc. Natl. Acad. Sci.
USA, 106:12723–12728, 2009.
P. G. Bolhuis, D. Chandler, C. Dellago, and P. L. Geissler. Transition path sam-
pling: Throwing ropes over rough mountain passes, in the dark. Annu. Rev.
Phys. Chem., 53:291–318, 2002.
J. C. Phillips, J. .E. Stone, K. .L. Vandivort, T. .G. Armstrong, J. M. Wozniak,
M. Wilde, and K. Schulten. Petascale Tcl with NAMD, VMD, and Swift/T.
In SC’14 workshop on High Performance Technical Computing in Dynamic
Languages. IEEE Press, 2014.
T. N. Phung, J. F. Brady, and G. Bossis. Stokesian dynamics simulation of
brownian suspensions. J. Fluid Mech., 313:181–207, 1996.
S. Piana, J. L. Klepeis, and D. E. Shaw. Assessing the accuracy of physical
models used in protein-folding simulations: quantitative evidence from long
molecular dynamics simulations. Curr. Opin. Struct. Biol., 24:98–105, 2014.
L. C. Pierce, R. S. Ferrer, C. .A. F. de Oliveira, J. A. McCammon, and R. .C.
Walker. Routine access to millisecond time scale events with accelerated
molecular dynamics. J. Chem. Theory Comput., 8(9):2997–3002, 2012.
A. R. Plastino and C. Anteneodo. A dynamical thermostatting approach to
nonextensive canonical ensembles. Ann. Phys., 255(2):250–269, 1997.
128 Hiqmet Kamberaj
P. Liu and G.A. Voth. Smart resolution replica exchange: An efficient algorithm
for exploring complex energy landscapes. J. Chem. Phys., 126(4):045106–6,
2007.
P. Liu, B. Kim, R.A. Friesner, and B.J. Berne. Replica exchange with solute
tempering: A method for sampling biological systems in explicit water. Proc.
Natl. Acad. Sci. USA, 103(39):13749–13754, 2005.
C. Predescu, M. Predescu, and C.V. Ciobanu. On the efficiency of exchange in

parallel tempering Monte Carlo simulations. J. Phys. Chem. B, 109:4189–
4196, 2005.
D. Provasi and M. Filizola. Putative active states of a prototype G-protein-

coupled receptor from biased molecular dynamics. Biophys. J., 98:2347–
2355, 2010.
J. A. Rahman and J. C. Tully. Puddle-skimming: An efficient sampling of mul-

tidimensional configuration space. J. Chem. Phys., 116:8750, 2002.
P. Raiteri, A. Laio, F. L. Gervasio, C. Micheletti, and M. Parrinello. Efficient

reconstruction of complex free energy landscapes by multiple walkers meta-
dynamics. J. Phys. Chem. B., 110:3533–3539, 2006.
J. Rogal and P. G. Bolhuis. Multiple state transition path sampling. J. Chem.

Phys., 129:224107, 2008.
R. Yamamoto and W. Kob. Replica-exchange molecular dynamics simulation

for supercooled liquids. Phys. Rev. E, 61:5473–5476, 2000.
R. Zhou. Free energy landscape of protein folding in water: explicit vs. implicit
solvent. Proteins: Struct., Funct., Bioinf., 53(2):148–161, 2003.
R. Zhou and B.J. Berne. Can a continuum solvent model reproduce the free
energy landscape of a β-hairpin folding in water? Proc. Natl. Acad. Sci.
USA, 99:12777–12782, 2002.
D. Sabo, M. Meuwly, D.L. Freeman, and J.D. Doll. A constant entropy increase
model for the selection of parallel tempering ensembles. J. Chem. Phys., 128:
174109, 2008.
D. P. Scarpazza, D. J. Ierardi, A. K. Lerer, K. M. Mackenzie, A. C. Pan, J. A.

Bank, E. Chow, R. O. Dror, J. P. Grossman, D. Killebrew, M. A. Moraes,
C. Predescu, J. K. Salmon, and D. E. Shaw. Extending the generality of
molecular dynamics simulations on a special-purpose machine. In Proceed-
ings of the 27th IEEE International Parallel and Distributed processing Sym-
posium, pages 933–945, Boston, MA, USA, 2013. IEEE Computer Society.
T. Schlick. Molecular Modeling and Simulation. Springer, New York, NY, 2nd
edition, 2010.
T. Schlick, M. Mandziuk, R. D. Skeel, and K. Srinivas. Nonlinear resonance
artefacts in molecular dynamics simulations. J. Comput. Phys., 140(1):1–29,
1998.
S. Jang, S. Shin, and Y. Pak. Replica-exchange method using the generalized
effective potential. Phys. Rev. Lett., 91:058305, 2003.
S. Jang, E. Kim, and Y. Pak. All-atom level direct folding simulation of a ββα
miniprotein. J. Chem. Phys., 128:105102, 2008.
R. D. Skeel, G. Zhang, and T. Schlick. A family of symplectic integrators: sta-
bility, accuracy, and molecular dynamics applications. SIAM J. Sci. Comput.,
18(1):203–222, 1997.
S. L. Seyler and O. Beckstein. Sampling large conformational transitions:
adenylate kinase as a testing ground. Mol. Sim., 40(10-11):855–877, 2014.
A. V. Smith and C. K. Hall. Alpha-helix formation: discontinuous molecular
dynamics on an intermediate-resolution protein model. Proteins, 44(3):344–
60, 2001a.
A. V. Smith and C. K. Hall. Assembly of a tetrameric alpha-helical bundle:
Computer simulations on an intermediate-resolution protein model. Proteins,
44:376, 2001b.
S. Nosé. A unified formulation of the constant temperature molecular dynamics
methods. J. Chem. Phys., 81:511, 1984a.
S. Nosé. A molecular dynamics method for simulation in the canonical ensem-
ble. Mol. Phys., 52:255, 1984b.
130 Hiqmet Kamberaj
M. Stepanova. Dynamics of essential collective motions in proteins: Theory.

Phys. Rev. E, 76(5):051918, 2007.
J. .E. Stone, D. .J. Hardy, B. Isralewitz, and K. Schulten. GPU algorithms
for molecular modelling. In D. A. Bader and J. Kurzak, editors, Scientific
Computing with Multicore and Accelerators, pages 351–371. Chapman &
Hall, 2011.
J. E. Stone, K. L. Vandivort, and K. Schulten. GPU-accelerated analysis and
visualisation on petascale supercomputing platforms. In Proceedings of the
8th International Workshop on Ultrascale Visualization, pages 6:1–6:8, New
York, NY, USA, 2013. UltraVis’13.
J. E. Stone, R. McGreevy, B. Isralewitz, and K. Schulten. GPU-accelerated
analysis and visualisation of large structures solved by molecular dynam-
ics flexible-fitting. Faraday Discuss., 169:265–283, 2014. doi: 10.1039/
C4FD00005F. URL http://dx.doi.org/10.1039/C4FD00005F.
T. P. Straatsma and J. A. McCammon. Multiconfiguration thermodynamic inte-
gration. J. Chem. Phys., 95:1175–1188, 1991.
S. Trebst, D.A. Huse, and M. Troyer. Optimizing the ensemble for equilibrium
in broad-histogram Monte Carlo. Phys. Rev. E, 70:046701, 2004.
S. Trebst, M. Troyer, and U.H.E. Hansmann. Optimized parallel tempering sim-
ulations of proteins. J. Chem. Phys., 124:174903, 2006.
Y. Sugita and Y. Okamoto. Replica-exchange molecular dynamics method for
protein folding. Chem. Phys. Lett., 314:141–151, 1999.
M. Takahashi and M. Imada. Monte Carlo calculation of quantum systems. II.
higher order correction. J. Phys. Soc. Jpn., 53:3765–3769, 1984.
T. Nagasima, A. R. Kinjo, T. Mitsui, and K. Nishikawa. Wang-Landau molec-
ular dynamics technique to search for low-energy conformational space of
proteins. Phys. Rev. E, 75:066706, 2007.
G. M. Torrie and J. P. Valleau. Monte Carlo free energy estimates using non-
Boltzmann sampling: Application to the sub-critical Lennard-Jones fluid.
Chem. Phys. Lett.., 28:578–581, 1974.
V. Tozzini. Coarse-grained models for proteins. Curr. Opin. Struc. Biol., 15:
144, 2005.
V. Tozzini and J. McCammon. A coarse grained model for the dynamics of the
early stages of the binding mechanism of hiv-1 protease. Chem. Phys. Lett.,
413:123–128, 2005.
V. Tozzini, W. Rocchia, and J. A. McCammon. Mapping all-atom models onto

one-bead coarse-grained models: General properties and applications to a
minimal polypeptide model. J. Chem. Theory Comput., 2:667–673, 2006.
C. Tsallis. Possible generalization of Boltzmann-Gibbs statistics. J. Stat. Phys.,

52:479–487, 1988.
M. .E. Tuckerman and B. J. Berne. Molecular dynamics algorithm for multiple

time scales: Systems with long range forces. J. Chem. Phys., 94:6811, 1991a.
M. .E. Tuckerman and B. J. Berne. Molecular dynamics in systems with multi-

ple time scales: Systems with stiff and soft degrees of freedom and with short
and long range forces. J. Chem. Phys., 95:8362, 1991b.
M. .E. Tuckerman and G. J. Martyna. Understanding modern molecular dynam-

ics: Techniques and Applications. J. Phys. Chem. B, 104:159–178, 2000.
M. .E. Tuckerman, G. J. Martyna, and B. J. Berne. Molecular dynamics algo-

rithm for condensed systems with multiple time scales. J. Chem. Phys., 93:
1287, 1990.
M. .E. Tuckerman, B. J. Berne, and A. Rossi. Molecular dynamics algorithm

for multiple time scales: Systems with disparate masses. J. Chem. Phys., 94:
1465, 1991.
T. W. Whitfield, L. Bu, and J. E. Straub. Generalized parallel sampling. Phys-

ica A: Statistical Mechanics and its Applications, 305:157–171, 2002.
U. H. E. Hansmann and Y. Okamoto. Tackling the protein folding problem by

a generalized-ensemble approach with Tsallis statistics. Braz. J. Phys., 29:
187, 1999.
132 Hiqmet Kamberaj
U. H. E. Hansmann and Y. Okamoto. Generalized-ensemble Monte Carlo

method for systems with rough energy landscape. Phys. Rev. E, 56(2):2228–
2233, 1997.
A. E. van Giessen and J. E. Straub. Monte Carlo simulations of polyalanine

using a reduced model and statistics-based interaction potentials. J. Chem.
Phys., 122:024904, 2005.
W. van Gunsteren, D. Bakowies, R. Baron, I. Chandrasekhar, M. Christen,

X. Daura, P. Gee, D. P. Geerke, A. Glättli, P. H. Hünenberger, M. A. Kas-
tenholz, C. Oostenbrink, M. Schenk, D. Trzesniak, N. F. A. can der Vegt, and
H. B. Yu. Biomolecular modeling: Goals, problems, perspectives. Angew.
Chem. Int. Ed., 45(25):4064–4092, 2006.
A. F. Voter. Hyperdynamics: Accelerated molecular dynamics of infrequent

events. Phys. Rev. Lett., 78:3908, 1997.
F. Wang and D. P. Landau. Efficient, multiple-range random walk algorithm to

calculate the density of states. Phys. Rev. Lett., 86:2050, 2001.
J. S. Wang and R. H. Swendsen. Replica Monte Carlo simulation of spin glasses.

Phys. Rev. Lett., 57:2607–2609, 1986.
W.B. Street, D.J. Tildesley, and G. Saville. Multiple timestep methods in

molecular dynamics. Mol. Phys., 35:639, 1978.
S. Wells, S. Menor, B. Hespenheide, and M. Thorpe. Constrained geomet-

ric simulation of diffusive motion in proteins. Phys. Biol., 2(4):S127–S136,
2005.
W.G. Hoover. Canonical dynamics: Equilibrium phase-space distributions.

Phys. Rev. A, 31:1695, 1985.
S. Wolfram. A new kind of science. Wolfram Media, Inc., Champaign, IL, 2002.
T. B. Woolf and A. Grossfield. Interaction of tryptophan analogs with POPC

lipid bilayers investigated by molecular dynamics calculations. Langmuir,
18:198–210, 2002.
T. Wüst and D. P. Landau. Optimized Wang-Landau sampling of lattice poly-

mers: Ground state search and folding thermodynamics of HP model pro-
teins. J. Chem. Phys., 137(6):064903, 2012.
X. Cheng, G. Cui, V. Hornak, and C. Simmerling. Modified replica exchange

simulation methods for local structure refinement. J. Phys. Chem. B, 109:
8220–8230, 2005.
Y. Xin, U. Doshi, and D. Hamelberg. Examining the limits of time reweighing

and kramer’s rate theory to obtain correct kinetics from accelerated molecular
dynamics. J. Chem. Phys., 132:224101, 2010.
Y. Okamoto. Designing generalized statistical ensembles for numerical simula-

tions of biopolymers. J. Mol. Graphics Modell., 22:425–439, 2004.
Y. Pak and S. Wang. Folding of a 16-residue helical peptide using molecular

dynamics simulation with Tsallis effective potential. J. Chem. Phys., 111:
4359, 1999.
Y. Pak and S. Wang. Application of a molecular dynamics simulation method

with a generalized effective potential to the flexible molecular docking prob-
lems. J. Phys. Chem. B, 104:354–359, 2000.
Y. Sugita and Y. Okamoto. Replica-exchange multicanonical algorithm and

multicanonical replica-exchange method for simulating systems with rough
energy landscape. Chem. Phys. Lett., 329:261–270, 2000.
Y. Sugita, A. Kitao, and Y. Okamoto. Multidimensional replica-exchange

method for free-energy calculations. J. Chem. Phys., 113:6042–6051, 2000.
Y. Ueda, H. Taketomi, and N. Go. Studies on protein folding, unfolding, and

fluctuations by computer simulation. II. A. three-dimensional lattice model
of lysozyme. Biopolymers, 17:1531–1548, 1978.
Chapter 4
STUDY ON ERROR IN SIGMOIDAL FUNCTION

GENERATION OF 4R MECHANISM
Ankur Jaiswal and H. P. Jawale*

Visvesvaraya National Institute of Technology, Nagpur,
Maharashtra, India
ABSTRACT
This chapter presents an extended Freudenstein-Chebyshev
approximation method for sigmoidal function generation applied to four
and five precision points. The mathematical model for sigmoidal function
generation is presented and the structural error parameters are obtained.
The structural error between the generated function and desired function
is estimated. The least-square method is used for minimizing the
structural error. Comparison of the results of the least square method and
analytical method is carried out. The proposed methodology is
demonstrated for designing a four bar function generation planar
mechanism.

Corresponding Author Email: jaiswalankur13@gmail.com.
136 Ankur Jaiswal and H. P. Jawale
Keywords: kinematics synthesis, optimization, function generation,

precision points
1. INTRODUCTION
A mechanism is an assemblage of rigid bodies connected together for

the purpose of obtaining the desired motion. Dimensional synthesis is a
method of determining the geometric characteristics of the links. The four-
bar linkage has a long history in both the theoretical kinematics as well as
plenty applications for spcial motion and path generating tasks. Various
approaches of mechanisms synthesis are found in literature for path
generation, function generation and rigid body guidance. The function
generation is a approach to obtain desired characteristic at output link of
the menchaism. The approaches are many times extended to estimate
accuracy and optimize the performance.
Traditionally, the objective function of mechanism synthesis is
specified using the Freudenstein equation and the solved by the Chebyshev
approximation theory. Within the constrints in obtaining the solutions, the
desired accuracy is possible to obtain at limited precision points. In
literature, solution for three four and five precision points function
generation task has been dealt successfully using various approaches like
freudenstein-chebyshev method, Loop-closure equation, (George and
Arthur, 1984, Hartenberg and Denavit, 1964, Norton and Wang, 2004,
Rao, 1979, Saxena and Ananthasuresh, 2003, Sun, 1982, Todorov, 2002).
Also, the least square method, Galerkin’s method, non-linear programming
approach, continuation method and sequential quadratic programming
(SQP) method is used to optimize the structural error and compare the past
results of the mechanism (Akcali and Dittrich, 1989, Cossalter, et al., 1992,
Lin, 1998, Shariati and Norouzi, 2011, Wilde, 1982).
Study on Error in Sigmoidal Function Generation of 4R … 137
Analytical treatment on function generation in spherical four bar

mechanisms (Alizade and Kilit, 2005) and five accuracy points is dealt in
(Rose and Sandor, 1973, Sheu, et al., 2008) comparing the continuation
method (Huang, et al., 2009). Comparison of the error using heuristic
optimization techniques and gradient based method is presented (Mirmahdi
and Norouzi, 2013). A new approach for dimensional synthesis of planar
mechanisms with mixed exact - approximate points for path and function
generation task is found to be used (Cervantes-Sánchez, et al., 2009, Diab
and Smaili, 2008) for rigid links and for adjustable links is (Soong and
Chang, 2011, Zhou, 2009) structural error is analyzed. The interpolation
approximation, least squares approximation and chebyshev approximation
for solving the function generation five point synthesis problem (Alizade
and Gezgin, 2011), approximate precision-position synthesis
(Freudenstein, 2010), and classical chebyshev synthesis is extended to
obtain error minimization (Ceccarelli and Vinciguerra, 2000). The gauss
elimination and homotopy continuation method (Wu, 2005) applied to
slider crank mechanism, sum of square method (Simionescu and Beale,
2002) is also seen. Optimization of the path synthesis using adjustable
approaches (Peng and Sodhi, 2010).
The work on four and five precision point synthesis and therein
optimization of error is required to be carried out with newer techniques,
extending scope for the analysis to sigmoidal function generation. The
present script aims to extend the freudenstein-chebyshev method for four
and five point synthesis and optimization using least square method. The
objective function is part of exponential function. A sigmoidal function is
a mathematical functionhaving an “S” shape (sigmoid curve) and the limit
are 0 to1 (Costarelli and Spigler, 2014). The objective function has been
analyzed based on the structural error between the generated function and
desired function.
1
F ( x) 
1  e x
2. KINEMATIC SYNTHESIS
2.1. Four Precision Point Synthesis
The Chebyshev equation to determine the accuracy points of the

mechanism can be given as
x 0  x n 1 x n 1  x 0 cos(2 j  1)
x( j )   .
2 2 2n (1)
Freudenstein method (Hartenberg and Denavit, 1964) is an analytical

method for synthesizing four-bar linkage in Figure 1. For three-position
synthesis, except the three link length ratio, all other parameters were
chosen arbitrarily. In the case of four-position synthesis, four parameters
are required for satisfying four positions. One more parameter in addition
to the three parameters taken in three point synthesis is chosen (Jaiswal and
Jawale, 2017, Jaiswal and Jawale, 2017).
D1 cos   D2 cos  D3  cos(  )

(2)
Equation (2) i.e., Freudenstein equation is used to synthesize a four bar

mechanism for three accuracy points where:
L1 L
D1  D2  1
L4 , L2 ,
L12  L22  L23  L24

D3 
2 L2 L4
A
B
L3
L4 L2
ψ φ
L1
O4 O2
Figure 1. Four bar mechanism.
Equation (2) is modified for four accuracy points by substituting –
 j   i  ( ) j
where, ѱi = initial output angle; Substituting cos i  s1

Using values of D1, D2 and D3 on re-arranging
R1 sin( j  ( ) j )  R2 cos( j  ( ) j )  R3 cos  j  R4

 R5 sin( ) j  cos( ) j
(3)
where:
1
L (1  s12 ) 2
R1  2
L1s1 ,
L2
R2 
L1 ,
L2
R3 
L4 s1 , (4)
L L L L
2 2 2 2
R4  1 2 3 4
2 L1 L4 s1 ,
1
(1  s12 ) 2
R5 
s1
Equation (3) is the required design equation for synthesis of four bar
linkage up to four accuracy points.
In equation (3) and (4), there are five variables (R1, R2, R3, R4 and R5).
For four accuracy points, there are only four equations [29] and similarly
in five points case.
R1  R2 R5 (5)
R j  Aj  λB j
(6)
Here, Aj and Bj are selected so as to satisfy the following two sets of

four linear equations.
A1 sin( j  ( ) j )  A2 cos( j  ( ) j )  A3 cos  j  A4   cos( ) j

(7)
B1 sin( j  ( ) j )  B2 cos( j  ( ) j )  B3 cos  j  B4  sin( ) j

(8)
where, j = 1, 2, 3, 4 and Solving for Aj and Bj by using matrix method
 
1
( A  B )  ( A2  B1 ) 2  4 A1 B2 2
 1 2
2 B2 (9)
2.2. Five Precision Point Synthesis
The three link ratios, and two angles, as design variables in case of five
point synthesis. Let the crank and follower angles for five accuracy points
be φj and ψj (j = 1, 2, 3, 4, 5), then,
 j  i    j  j   i  ( ) j
and
Substituting the above in the Freudenstein equation (2),
R1 cos{ 1 j   1 j }  R2 sin{ 1 j   1 j }  R3 cos 1 j

 R4  1 j  R5  R6 sin  1 j  cos 1 j
(10)
where:
R1 
s1 [1  1  s . 1  s 
2
1
2
2
K 2 .s1 .s 2 ,
R2 

s1 1  s  1  s
2
1
2
2 
K 2 s1 s 2 ,
K 1 s1
R3 
K 2 s 2 (11)
R4 

K 1 . 1  s12 
K 2 .s 2 (11)
K3
R5 
K 2 s2 ,
R6 
1  s  2
1
s2
Equation (10) is the required design equation for synthesis of four bar
linkage up to five accuracy points. Using compatibility condition for
linearization in equation (10) and (11) and applying gauss elimination
method is used to solve the non-linear equations. Derive the linear equation
(12) and (13) is to determine the design parameters of the mechanism.
R1 R4  R3 R2

R1 R3  R2 R4
A1 cos(1q   1q )  A2 sin(1q   1q )  A3 cos( )1q  A4 sin( )1q
 A5  cos 1q  0
(12)
B1 cos(1q   1q )  B2 sin(1q   1q )  B3 cos( )1q  B4 sin( )1q

 B5  sin  1q  0
(13)
where, q = 1,2,3,4 and 5.
3. OPTIMIZATION OF THE STRUCTURAL ERROR
3.1. Modifying the Design Equation for Four and Five

Presicion Points
Here, least-square technique is used as the optimization approach for

mechanism synthesis. Now, on applying the least square technique to
minimize the errors [29], the overall error can be written as - rearranging
the equation (3)
D1 cos(q  ( )q )  D2 sin( )q  D3 cos q  D4  D5 sin(q  ( )q )

(14)
 cos   q  0
L2
D1 
L1
1
(1  s12 ) 2
D2 
s1
L2
D3 
L4 s1 (15)
L L L L
2 2 2 2
D4  1 2 3 4
2 L1 L4 s1
1  s 
1
2 2
D5  1
L1 s1
Rearranging the above equation and replacing R1-R6 by D1-D6

respectively.
The design equation (10) can be written as:
D1 cos{ 1q   1q }  D2 sin{ 1q   1q }  D3 cos 1q
 D4 sin  1q  D5  D6 sin  1q  cos 1q  0
(16)
where:
D1 
s1 [1  1  s . 1  s 
2
1
2
2
K 2 s1 s 2 ,
D2 

s1 1  s12  1  s 22 
K 2 s1 s 2 , (17)
D4 

K1 . 1  s 2
1 
K 2 .s 2
K3
D5 
K 2 s2 ,
D6 
1  s  2
1
s2
3.2. Optimization for Four and Five Presicion Points
From equations (15) and (17) –
D5 = D1D2and Assuming, D5  1 (for four points) (18)
D1 D4  D2 D3
D6 
D1 D3  D2 D4 (for five points) (19)

E    Aq    Bq 
2 2
 (20)
For minimizing overall error, Σ (ƐAq)2& (ƐBq)2 are minimized [30]. By

derivative method first derivative must be equal to zero, such that
  Aq 2   Bq 2
0 0
An , Bn (21)
 Aq  A1 sin( j  ( ) j )  A2 cos( j  ( ) j )  A3 cos  j  A4  cos( ) j

(22)
 Bq  B1 sin( j  ( ) j )  B2 cos( j  ( ) j )

(23)
 B3 cos  j  B4  sin( ) j
Aq  A1 cos(1q   1q )  A2 sin(1q   1q )  A3 cos( )1q

(24)
 A4 sin( )1q  A5  cos( )1q
Bq  B1 cos(1q   1q )  B2 sin(1q   1q )  B3 cos( )1q

(25)
 B4 sin( )1q  B5  sin( )1q
The equations (22) & (23) and (24) & (25) determine the optimal
design parameters for four and five accuracy points respectively [29, 30].
4. RESULTS AND DISCUSSION
The following Table 3.1 shows a comparative study of the four and
five point synthesis of sigmoidal function and their optimization using least
square technique. The results of synthesis for sigmoidal functions at four
and five accuracy points are summarized in Table 3.1. The structural errors
for sigmoidal functions are shown in Figure 2 and 3.
Figure 2. Plot of error for sigmoidal function generator with four precision points.
Table 3.1. Plot of error for sigmoidal function generator with four and
five accuracy points
Design Variables The results of four point The results of five point
synthesis by Sigmoidal synthesis by Sigmoidal
function [1/(1+e-x)] function [1/(1+e-x)]
Analytical Least Square Analytical Least Square
method Optimization method Optimization
Interval of ‘x’ -- 0x1 0x1 0x1 0x1
Range of φ , deg -- 90 90 90 90
Range of ψ , deg -- 60 60 60 60
Initial crank -- 15 9 73.1478 71.9893
angle, φi, deg
Link length L2/L1 3.0336 2.9389 -1.0389 -1.7078
ratio’s L3/L1 9.7791 6.3517 1.6085 1.6123
L4/L1 7.2282 8.8720 -1.7275 -1.0215
Follower angle, -- 51.2460 46.2122 61.6060 60.3716
ψj,deg
Maximum error -- 2.8614e-04 2.8033e-04 9.2931e-06 5.4204e-06
Maximum output -- 0.0391 0.0383 0.0013 7.4145e-04
error %
Figure 3. Plot of error for sigmoidal function generator with five precision points.
Present work depicts error for sigmoidal function by analytical method

0.0391% and 0.0013%. After optimization by least square method the
results are 0.0383% and 7.4145e-04%. The maximum output error with
four and five accuracy points synthesis 2.8614e-04 and 9.2931e-06 by an
analytical method.
After optimization the position of the accuracy points are different and
determine the new design parameter of the mechanism and maximum
output error with four and five accuracy points are 2.8033e-04 and
5.4204e-06. The error of mechanism is dependent on the position of the
accuracy point. The results obtained with the least square method for
synthesis of five point mechanism is optimal.
Figures 4 and 5 show the mapping of independent variable and input
angles of the syntheze mechanism. The precision points of the sigmoidal
function generating mechanism for four and five point case shown in
Figure 4 and 5. The deviation of the precision points using least square
method is 0.0001 to 0.01.
Evaluation of the four and five point synthesis after optimization
results is more accurate, it provides minimum error and mechanism
behaves constrained motion to the other mathematical functions.
100
90
Precision Points
80
Input Angle (  )
70
60
50
40
30
20
10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Independent Variable (x)
Figure 4. Independent variable Vs Input angle for four accuracy points.
90
80
Precision Points
70
Input Angle (  )
60
50
40
30
20
10
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Independent Variable (x)
Figure 5. Independent variable Vs Input angle for five accuracy points.
CONCLUSION
The work reported in this chapter describes a methodology to design a

four bar planar mechanism to generate sigmoidal function for four and five
point synthesis. The different technique for four and five point synthesis of
planar mechanisms has been discussed in the chapter. The proposed
method is simple and easy to implement in the four bar mechanism. The
function generation of four-bar linkage can be used in automobiles,

especially for control systems.
The results obtained by mathematical analysis have following silent
features:
1. The sigmoidal function generation by proposed method is found to

generate minimum error and mechanism behaves with predictable
motion than other function generation i.e., trignometric and
hyperbolic functions (Hartenberg and Denavit, 1964, Jaiswal and
Jawale, 2017).
2. The least square method is developed to optimize the structural
error between the desired function and mechanically developed
function, further used to optimize the generated design parameter
by four and five point synthesis.
3. The developed methodology increases the global accuracy in the
mechanisms by minimizing the maximum errors to the order of 10-
4
to 10-5.
In future work, the approach can be extended to cover five and six link
mechanisms and evolutionary optimization problems.
REFERENCES
Akcali, I., and Dittrich, G. (1989). Function Generation by Galerkin's

Method. Mechanism and Machine Theory, 24 1: 39-43.
Alizade, R., and Gezgin, E. (2011). Synthesis of Function Generating
Spherical Four Bar Mechanism for the Six Independent Parameters.
Mechanism and Machine Theory, 46 9: 1316-26.
Alizade, R. I., and Kilit, Ö. (2005). Analytical Synthesis of Function
Generating Spherical Four-Bar Mechanism for the Five Precision
Points. Mechanism and Machine Theory, 40 7: 863-78.
Ceccarelli, M., and Vinciguerra, A. (2000). Approximate Four-Bar Circle-

Tracing Mechanisms: Classical and New Synthesis. Mechanism and
Machine Theory, 35 11: 1579-99.
Cervantes-Sánchez, J. J., Medellín-Castillo, H. I., Rico-Martínez, J. M.,
and González-Galván, E. J. (2009). Some Improvements on the Exact
Kinematic Synthesis of Spherical 4r Function Generators. Mechanism
and Machine Theory, 44 1: 103-21.
Cossalter, V., Doria, A., Pasini, M., and Scattolo, C. (1992). A Simple
Numerical Approach for Optimum Synthesis of a Class of Planar
Mechanisms. Mechanism and Machine Theory, 27 3: 357-66.
Costarelli, D., and Spigler, R. (2014). Sigmoidal Functions Approximation
and Applications. Roma Tre” University, Rome, Italy,
Diab, N., and Smaili, A. (2008). Optimum Exact/Approximate Point
Synthesis of Planar Mechanisms. Mechanism and Machine Theory, 43
12: 1610-24.
Freudenstein, F. (2010). Approximate Synthesis of Four-Bar Linkages*
[1, 2]. Resonance,
George, N., and Arthur, G. Advanced Mechanism Design: Analysis and
Synthesis. Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 1984.
Hartenberg, R. S., and Denavit, J. Kinematic Synthesis of Linkages.
McGraw-Hill, 1964.
Huang, X., He, G., Liao, Q., Wei, S., and Tan, X. “Solving a Planar Four-
Bar Linkages Design Problem.” Paper presented at the Information
and Automation, 2009. ICIA'09. International Conference on, 2009.
Jaiswal, A., and Jawale, H. (2017). Comparative Study of Four-Bar
Hyperbolic Function Generation Mechanism with Four and Five
Accuracy Points. Archive of Applied Mechanics, 87 12: 2037-54.
Jaiswal, A., and Jawale, H. “Comparative Study of Structural Error in Four
Bar Mechanism for Hyperbolic Functions.” Paper presented at the
Advances in Mechanical, Industrial, Automation and Management
Systems (AMIAMS), 2017 International Conference on, 2017.
Lin, C.-C. (1998). Complete Solution of the Five-Position Synthesis for
Spherical Four-Bar Mechanisms. Journal of marine science and
Technology, 6 1: 17-27.
Mirmahdi, S., and Norouzi, M. (2013). On the Comparative Optimal

Analysis and Synthesis of Four-Bar Function Generating Mechanism
Using Different Heuristic Methods. Meccanica, 48 8: 1995-2006.
Norton, R. L., and Wang, S. S.-L. Design of Machinery: An Introduction to
the Synthesis and Analysis of Mechanisms and Machines. McGraw-
Hill Higher Education, 2004.
Peng, C., and Sodhi, R. S. (2010). Optimal Synthesis of Adjustable
Mechanisms Generating Multi-Phase Approximate Paths. Mechanism
and Machine Theory, 45 7: 989-96.
Rao, A. (1979). Optimum Design of Four-Bar Function Generators with
Minimum Variance Criterion. Journal of optimization theory and
applications, 29 1: 147-53.
Rose, R. S., and Sandor, G. N. (1973). Direct Analytic Synthesis of Four-
Bar Function Generators with Optimal Structural Error. Journal of
Engineering for Industry, 95 2: 563-71.
Saxena, A., and Ananthasuresh, G. (2003). A Computational Approach to
the Number of Synthesis of Linkages. Transactions-american society
of mechanical engineers journal of mechanical design, 125 1: 110-18.
Shariati, M., and Norouzi, M. (2011). Optimal Synthesis of Function
Generator of Four-Bar Linkages Based on Distribution of Precision
Points. Meccanica, 46 5: 1007-21.
Sheu, J.-B., Hu, S.-L., and Lee, J.-J. (2008). Kinematic Synthesis of a
Four-Link Mechanism with Rolling Contacts for Motion and Function
Generation. Mathematical and Computer Modelling, 48 5-6: 805-17.
Simionescu, P., and Beale, D. (2002). Optimum Synthesis of the Four-Bar
Function Generator in Its Symmetric Embodiment: The Ackermann
Steering Linkage. Mechanism and Machine Theory, 37 12: 1487-504.
Soong, R.-C., and Chang, S.-B. (2011). Synthesis of Function-Generation
Mechanisms Using Variable Length Driving Links. Mechanism and
Machine Theory, 46 11: 1696-706.
Sun, W. (1982). Optimum Design Method for Four-Bar Function
Generators. Journal of Optimization Theory and Applications, 38 2:
287-93.
Todorov, T. S. (2002). Synthesis of Four-Bar Mechanisms by

Freudenstein–Chebyshev. Mechanism and Machine Theory, 37 12:
1505-12.
Wilde, D. (1982). Error Linearization in the Least-Squares Design of
Function Generating Mechanisms. Journal of Mechanical Design, 104
4: 881-84.
Wu, T.-M. (2005). Non-Linear Solution of Function Generation of Planar
Four-Link Mechanisms by Homotopy Continuation Method. J. Appl.
Sci, 5 4: 724-28.
Zhou, H. (2009). Synthesis of Adjustable Function Generation Linkages
Using the Optimal Pivot Adjustment. Mechanism and Machine
Theory, 44 5: 983-90.
Chapter 5
IDENTIFYING TRANSFER VERTEX

FROM THE ADJACENCY MATRIX
FOR EPICYCLIC GEAR TRAINS
Mallu Chengal Reddy1,, Rudraraju Manish1

and Yendluri Daseswara Rao1
1
Mechanical Department, BITS-PILANI, Hyderabad, India
ABSTRACT
In this chapter, an algorithm to find the transfer vertex of a given

epicyclic gear train is given. The algorithm used searches for a gear pair
and finding the transfer vertex through corresponding turning pairs. It
also identifies cases where more than one transfer vertex, are possible.
This algorithm reduces the number of epicyclic gear trains to be checked
for rotational isomorphism after displacement isomorphism.
Keywords: epicyclic gear train, isomorphism, adjacency matrix, transfer

vertex

Corresponding Author Email:f20150428@hyderabd.bits-pilani.ac.in.
154 Mallu Ch. Reddy, Rudraraju Manish and Yendluri D. Rao
1. INTRODUCTION
The assumptions normally used while synthesizing Epicyclic Gear

Trains (EGT) are the mechanism should be planar and should have only
binary joints. Also, should obey the general degree of freedom equation,
have unlimited rotatability for all links. Each gear should have a turning
pair on its axis and each link should have at least one turning pair to
maintain constant center distance between each gear pair. There are three
ways to represent an epicyclic gear train
 Functional representation,
 Graphical representation, and
 Rotation graph
Functional Representation of a gearbox is shown in Figure 1.

In a graphical representation, the mechanism is illustrated using a
graph (Buchsbaum, F. and Fruedenstein, F. 1970).
Figure 1. Functional Representation of a gearbox.

Identifying Transfer Vertex from the Adjacency Matrix … 155
The links are represented as vertices and joints as edges, connecting

the vertices. A graphical representation of the gearbox is shown in Figure
2. In graphical representation, a turning pair is represented by a thin edge,
whereas a gear pair is shown as a thick edge. Graphs shown with thin and
thick edges are called bicolored graph and unlabeled when its edges are not
labeled (Buchsbaum, F. and Fruedenstein, F. 1970), (Dobrjanskyj, L. and
Fruedenstein, F. 1967).
To draw a rotation graph of epicyclic gear trains, graphical
representation of an EGT is divided into Fundamental Circuits (FC)
consisting of one gear pair and associated turning pairs.
Figure 2. A graphical representation of the mechanism.

Figure 3. Three fundamental circuits.
For example, if the EGT in Figure 2b is considered, the three FCs are
shown in Figure 3.
In a rotation graph of EGTs, the turning pair edges are removed and
each geared pair is labeled with the symbol of the associated transfer
vertex of the fundamental circuit (Tsai L. 1987). For example, for the
graphs of EGTs in Figure 2(a) and 2(b), rotation graphs are as shown in
Figures 4(a) and 4(b).
The fundamental rules followed (Buchsbaum, F. and Fruedenstein, F.
1970), (Ravisankar, R. and Mruthyunjaya, T. S.) while drawing a graph are
that, for n-links, epicyclic gear train having one degree of freedom, there
are n-vertices, n-1 turning pairs and n-2 gear pair edges. The subgraph
formed by deleting all the geared edges is a tree and a geared edge added to
the tree will form a fundamental circuit having one geared edge and several
turning edges.
The number of fundamental circuits will be equal to the geared edges
and a fundamental circuit containing only turning edges, is not permitted as
it violates the rotatability rule of an EGT. All vertices must at least be
connected to one turning edge. The level of a turning pair identifies its
location of its axis.
For example, for the graph of EGT shown in figure 2(b), the levels are
shown in Figure 3. For any fundamental circuit, the differential degree of
freedom should be the number of vertices minus two, with a minimum of
one.
Two turning pair edges at the same level must intersect at a common
vertex (Ravisankar R. and Mruthyunjaya, T. S.). All edges on one side of
the transfer vertex are at the same level and all edges on the other side are
at a different level.
Figure 4. Rotation Graphs of Figure 2(a) and 2(b).
1.1. Linkage Adjacency Matrix
Linkage adjacency matrix (Uicker, J. J. Jr. and Raicu A. 1975) is the

matrix, which represents the connectivity of links in the EGT. For an n-link
EGT, the adjacency matrix is a square matrix of order n obeying the
following rules (Fruedenstein, F. 1971), (Ravisankar, R. and
Mruthyunjaya, T. S.).
A(i, j) = 1 (if link i and link j are connected by a turning pair)

= 2 (if link i and link j are connected by a gear pair)
= 0 (otherwise, including the case i = j)
The adjacency matrix for the graph of EGT in figure 2 is given below.
Adjacency matrix of Figure 2(a):
0 1 1 1 2
1 0 2 2 0

1 2 0 0 1
 
1 2 0 0 0
2 0 1 0 0
Adjacency matrix of Figure 2(b):
0 1 1 2 1
1 0 2 1 0

1 2 0 0 0
 
2 1 0 0 2
1 0 0 2 0
1.2. Hamming Procedure
In the adjacency matrix, a hamming matrix is generated using the

Boolean algebra. To generate the hamming matrix, there are few rules to
be followed: from the elements in the adjacency matrix, the hamming
distance is measured between two links, which act as elements of hamming
matrix.
hij = ∑aik+ ajk if aik ≠ ajk

hii = 0 also hij = hji
and (aik + ajk) = 0 if aik = ajk
Hamming matrix and hamming value of each link (last column) of

graphs in Figure 2(a) and 2(b) are as follows,
 0 10 9 8 6 33
10 0 7 6 8 31
 
 9 7 0 1 7 24
 
 8 6 1 0 6 21
 6 8 7 6 0 27 
0 9 8 8 4 29
9 0 5 9 5 28

8 5 0 8 4 25
 
8 9 8 0 8 33
4 5 4 8 0 21
From the elements of hamming matrix, hamming value of a link is

calculated which is the sum of the elements of the specific link’s row. The
sum of hamming values of all links gives hamming number of graphs or
hamming value of the chain/graph. Hamming string is obtained by
concatenating the hamming number and the hamming values all the links.
The hamming string acts as an invariant and can be considered as a
parameter to check for structural isomorphism. Identical hamming strings
are considered as structurally isomorphic graphs. In doing this, we write a
computer code, which computes the hamming matrix, generates the
hamming string, and eliminates the matrices, which have identical
hamming strings. The hamming strings of the graphs are as follows:
136 33 31 27 24 21 136 33 29 28 25 21
From the hamming strings of the graphs, we can conclude that they are
not displacement isomers to one another.
1.3. Rotational Analysis
After the structural isomers are eliminated, the remaining graphs are
tested for rotation analysis. In this, a computer code is run to identify
fundamental circuits and find out the transfer vertex of the graph. If a
graph has a single transfer vertex, those graphs are directly taken to be
rotationally non-isomorphic. If a graph has more than one transfer vertex
possible, those graphs need to be manually iterated to find the transfer
vertex. After finding out the transfer vertex, rotation graphs are drawn and
the corresponding rotation matrices are generated. These rotation graphs
are now matched with the earlier adjacency matrices. On finding a match,
the graph is said to be rotationally isomorphic and is eliminated. Since the
rotation matrices that do not find a match, rotation-hamming strings are
generated and the string is iterated for a match to be eliminated as a
rotational isomer. The graphs, which are unique, are said to be rotationally
non-isomorphic.
1.4. Example
Graph in Figure 2(a) on rotation analysis shows that, it is a single

transfer vertex i.e., vertex 3 and can be said to be rotationally
nonisomorphic, which is not the case with Figure 2(b). Hence, Figure 2(b)
needs to be manually checked. The fundamental circuits in graph 2 are,
Corresponding to the rotational graph and fundamental circuit,
0 9 7 7 1 24
9 0 6 10 8 33
 
7 6 8 0 6 27  136 33 31 27 24 21
 
7 10 8 0 6 31
1 8 6 6 0 21
The rotation hamming matrix: Hamming string with hamming value

column.
We observe that, the hamming string is identical to the hamming string
of the graph in Figure 2 (a). Hence, we conclude that graph in Figure 2(b)
is rotationally not isomorphic to graph in Figure 2(a).
2. INTELLIGENT METHODS
The first graph of the given EGT is drawn and linkage adjacency
matrix is written using the rules mentioned above. Search for the gearing
pair in the adjacency matrix, in each FC. If a gear pair, denoted by
weightage 2, is in (i, j) position in the adjacency matrix in a FC, search in a
row i and row j for a set of turning pairs with weightage 1 having a
common vertex. The common vertex (link) is the Transfer Vertex (TV) for
that gear pair. In case there is no such common vertex (link), check for
other connecting pairs. If the number of turning pairs are more than two in
a FC, then the number of TVs are possible. In such cases where more than
one TV is possible, identify all the FCs in the EGT. Then assign levels to
each of the turning pairs without violating the fundamental rules of EGTs
(Ravisankar, R. and Mruthyunjaya, T. S.), (Tsai L. 1987). There by the TV
in that particular FC is identified.
2.1. Example
For example, in the adjacency matrix of Figure 2(a), one gear pair is at
position (1, 5). Then the search is for tuning pairs in rows 1 and 5. We find
that column 3 in each row has turning pair. That is, elements in the
adjacency matrix (1, 3) and (5, 3) are turning pairs (weightage 1).
Therefore, in the FC 1-5-3 as shown in Figure 5, it can be concluded that
vertex 3 is the TV for the Gear pair 1-5.
Considering another example, in the adjacency matrix of Figure 2(b),

gear pair with weightage 2 is at the element (4, 5).
So, rows 4 and 5 are searched for all turning pairs with an entry of 1 in
adjacency matrix.
Figure 5. In the FC 1-5-3.
Figure 6. Fundamental Circuit of Four Members.
However, in this case there is no common vertex among the turning

pairs in both the rows. That is, in the adjacency matrix, elements (4, 2) and
(5, 1) are 1, but no common vertex like (5, 1)-(4, 1) or (4, 2)-(5, 2).
However, since (2, 1) element is 1 (a turning pair), then this FC has more
than one TV as shown in Figure 6. That is, both the vertices 1 and 2 are
possible TVs.
The actual TV is ascertained by assigning the levels to the turning

pairs in the FC.
As levels for all the pairs in the EGT are assigned, the TV for the case
where ambiguity is there is identified. Graph with levels assigned is shown
in Figure 7. The rotation graphs for this EGT are as shown in Figure 4b.
Figure 7. Graph with levels assigned.
CONCLUSION
Structural isomorphism and rotational isomorphism are the two stages

in epicyclic gear trains synthesis. For rotational isomorphism transfer
vertex for each of the fundamental circuits in the epicyclic gear train is
essential. In this work, a simplified algorithm is given to find transfer
vertex of a given epicyclic gear train. This reduces the computational time
and effort largely. This algorithm also reduces the demand for manual
effort in rotational isomorphism test.
REFERENCES
Buchsbaum, F. and Fruedenstein, F. (1970), Synthesis of Kinematic

Structure of Geared Kinematic Chains and Other Mechanisms, Journal
of Mechanisms, 5, 357 - 392.
Cheng-Ho Hsu and Jin-Jhu Hsu (1996), An efficient methodology for the
structural synthesis of geared kinematic chains, Mech. Mach. Theory,
32(8), 957 - 973.
Dobrjanskyj, L. and Fruedenstein, F. (1967), Some Application of Graph
Theory to the Structural Analysis of Mechanisms, ASME Journal of
Engineering for Industry, 89(B), 153 - 158.
Fruedenstein, F. (1971), An Application of Boolean Algebra to the Motion
of Epicyclic drives, ASME Journal of Engineering for Industry, 93(B),
176 - 182.
Jae UK Kim and Byung man Kwak (1989), Application of edge
permutation group to structural synthesis of epicyclic gear trains,
Mech, Much. Theory, 25(5), 563 - 574.
Jae Kun Shin and Sundar Krishnamurthy (1992), Standard code technique
in the enumeration of epicyclic gear trains, Mech. Mack. Theory, 28(3),
347 - 355.
Rao YVD and Rao AC (2008), Generation of Epicyclic Gear Trains of One
Degree of Freedom, ASME J. Mech. Des., 130(5).
Ravisankar, R. and Mruthyunjaya, T. S. Computerized Synthesis of the
structure of Geared Kinematic Chains, Mechanisms and Machine
Theory, 20(5), 367 - 387.
Tsai L. (1987), An Application of the Linkage Characteristic Polynomial to
the Topological Synthesis of Epicyclic Gear Trains, ASME. J. Mech.
Trans. and Automation, 109(3), 329 - 336.
Uicker, J. J., Jr. and Raicu, A. (1975) A Method for the Identification and
Recognition of Equivalence of Kinematic Chains, Mechanisms and
Machine Theory, 10,375 - 383.
Chapter 6
3-D SIMULATION STUDIES ON

STRENGTHENING OF BEAM-COLUMN JOINT
Khan Mohammad Firoz, Garg Aman and H. D. Chalak

Department of Civil Engineering, NIT Kurukshetra, India
ABSTRACT
The beam-column joint is one of the most critical elements of a

structure. A slight rupture or loss in the strength of any beam column
joint in the building may lead to a decrease in its load carrying capacity,
which may further result in its progressive collapse. So it is important to
analyse for stresses and deflection in the vicinity of the beam-column
joint. However, the behavior of beam-column joint when subjected to
lateral loading is different as compared to when it is subjected to dead
load and live load. Over the years, the research in earthquake engineering
has been enhanced continuously leading to modification of codal
provisions for the design of structures. To incorporate these changes, the
structural members have to be reinforced adequately, and beam-column
joints need to be strengthened accordingly. In this research work, an
analytical study has been carried out with ABAQUS/CAE on the
strengthening of a beam-column joint under seismic conditions using

Corresponding Author Email: firozkhan2294@gmail.com.
166 Khan Mohammad Firoz, Garg Aman and H. D. Chalak
Carbon Fiber Reinforced Polymer (CFRP) sheets. A comparison analysis

is done for two different models created in ABAQUS/CAE namely
Model a: Beam column joint without the CFRP strengthening. Model B:
Beam-Column joint strengthened with CFRP sheets. Both the members
are subjected to the same seismic loading. Model A is subjected to a
seismic load of max amplitude of 30Mpa and Model B is subjected to two
different seismic loads, one with a maximum amplitude of 30 MPa and
the second one with a maximum amplitude of 60 MPa and later on the
behavior of beam-column joints of both models is observed in terms of
displacement and stress.
Keywords: beam-column joint, seismic analysis, strengthened beam-

column joint
1. INTRODUCTION
In RC building, a portion of columns that are common to beams at

their intersection is called beam-column joints. The complete knowledge of
it is of utmost importance if the building lies in the high seismic zone.
Over the past some decades, intensive research work has been reported on
reinforced concrete (RCC) beam-column joint. Improper design of
reinforced concrete structures gives rise to critical issues involving
technical as well as social aspects because these structures are designed for
gravity loads only, not for lateral loads. During an earthquake, columns
having a minimum cross-sectional area and longitudinal reinforcement will
not be able to meet the requirement of shear demand and flexure.
Secondly, construction of a strong beam-weak column under seismic loads
may lead to the formation of local hinges in the column. This associated
failure mode is characterized by catastrophic and brittle structural failure.
In interior joints where rebar is improperly anchored, bond failure in
longitudinal reinforcement has also been observed.
The strength of reinforced concrete beam-column joint plays a
significant role in the performance of the RC frame structure especially for
the cases when subjected to seismic loads. Inadequately detailed reinforced
3-D Simulation Studies on Strengthening of Beam-Column Joint 167
Concrete beam-column joints, especially exterior joints will be the most

critical element as they may fail due to excess amount of shear stresses.
Some researchers reported the work related to the strengthening of the
beam-column joint. In most of the cases, ferrocement composites were
used to strengthen the joint which led to the improvement of the joint in
terms of stiffness, energy dissipation and shear stress (Li et al., 2015). Use
of lightweight concrete was also done and tested, which led to joint getting
light weighted but it was useful for a certain limit of intensity of load
(Decker et al., 2015). Analysis of precast pinned beam column joint under
cyclic loading with the help of new expression (formula) was carried out
where the expressions were limited by geometrical configurations (Georgia
et al., 2017). An analytical study on a new moment resisting connection of
beam to the precast concrete column was carried out using ABAQUS. The
performance of that proposed connection was found similar to that of
monolithic connection (Bahrami.S et al., 2017). Various researches were
still being done in order to increase the efficiency of the joint.
In the present chapter, a study has been carried out on the reinforced
beam column joint analytically using ABAQUS (v.6.14). Following two
models are modelled for the analysis, and the comparison between the
results has been reported.
2. MATERIAL PROPERTIES
The metal properties used in the simulation are as follows

(Naveeena.N and Ranjitham.M, 2016). Longitudinal Rebar and stirrups
used in the column and beam are circular in profile. The materials, as well
as model properties, are given below in Table 1 and 2.
During analysis, concrete damage plasticity model is introduced to
predict the exact behavior of the structure.
From Figure 1 and 2 the location and orientation of the CFRP sheet put
on the joint can be determined. The interaction between beam and column
has been defined as a tie connection in ABAQUS, and the connection
between the CFRP sheet and beam-column joint in Model B is also defined
as tie connection. The interaction between reinforcement and concrete is
defined as an ideal, i.e., friction between the reinforcement and concrete is
infinite.
Table 1. Material properties
Material Modulus of elasticity (MPa) Poisson Density Strength

Ratio (tons/mm3) (MPa)
Concrete 22360 0.15 2.2 E-9 20 (28 days)
Steel 200000 0.3 7.85E-9 415 (Yield
Strength)
CFRP sheet (Ex) – 2.3x105 (Ey) – (Vxy) – 0.22 1.6E-9 3500
1.79x105 (Ez) – 1.79x105 (Vxz) – 0.22 (Tensile
(Gxy) – 1.179x105 (Gxz) – (Vyz) – 0.30 strength)
1.179x105 (Gyz) – 6.88x105
Table 2. Model properties
Element Length (mm) Cross section Main reinforcement Shear stirrups

(mm × mm)
Beam 600 200x150 8 bars of 12 mm φ Seven stirrups of 8 mm φ
Column 800 200x150 8 bars of 12 mm φ 6 stirrups of 8 mm φ
Figure 1. Model A. Figure 2. Model B.

Following finite elements are used for modelling:
1. C3D8R (explicit 8-node linear brick, reduced integration,

enhanced hourglass control): For beam, column.
2. S3: (A 3-node triangular general-purpose shell, finite membrane
strains) for CFRP sheet
3. T3D2 (Explicit 2-node linear 3-D truss): for longitudinal
reinforcement and shear stirrup.
3. MODEL VALIDATION
The present model is validated with the results obtained by N.

Naveeena.N et al. (2016). Two different cyclic loads of 30 kN and 60kN
have been applied at the tip of the beam of unstrengthened and
strengthened beam column joint respectively. Strengthening of the joint is
done by the wrapping of CFRP sheet up to 300mm on each side of the joint
region. The results obtained are presented in the form of table and graphs.
From the Figure 3 and 4, we can see that the stresses in the vicinity of
beam-column joint of Model B are comparatively less as compared to the
stresses in model A. The values by which the stresses differ are presented
in table 3 below. It is also noticed that the stresses generated in the
corresponding beam and column are comparatively less in model B.
Figure 3. Stress variation (Model A).

Figure 4. Stress variation when subjected to 60 MPa seismic loads (Model B)
Figure 5. Displacement variation (Model A).
Figure 6. Displacement Variation when subjected to 60 MPa seismic (Model B).

Table 3. Validation of Results
Beam-Column Cyclic Deflection Shear Stress

joint type Load ANSYS ABAQUS ANSYS ABAQUS
(kN) (Naveeena.N (Naveeena.N and
and Ranjitham.M,
Ranjitham.M, 2016)
2016)
Unstrengthened 30 6.2073mm 8.663mm 23.225 MPa 24.74
joint (Model-A) MPa
strengthened joint 60 5.9673mm 6.169mm 19.926 MPa 25.99
with CFRP MPa
wrapping up to
300mm on each
side of the joint
region (Model-B)
From Figure 5 and 6, we can easily compare the displacements

occurring at the vicinity of beam-column joint of both the Model A and
Model B. The displacement occurring in Model B is clearly very less as
compared to that of Model A. Given in Table 3 are the values of both
stresses and displacement of Model A and Model B.
4. CONVERGENCE STUDY
Convergence study has been carried out on unstrengthened reinforced

concrete beam-column joint subjected to 30KN cyclic loading at the tip of
the beam. Here the graph is plotted between the size of mesh and
deformation occur in the joint. Mesh size is on the x-axis and deflection on
the y-axis. The graph clearly shows that on converging the size of mesh,
the value of deflection is getting more refined and accurate. However, on
increasing the size of the mesh, the deflection value is decreasing, and after
50*50 mesh size, it is becoming constant. In the further model, the mesh
size of 50*50 is taken.
Figure 7. Graph showing convergence of deflection with mesh size.
5. RESULTS AND DISCUSSIONS
Comparative results are shown as follows in the form of table and

graphs. Figure 4 shows the variation of deflection and Figure 5 shows the
variation of shear stress with loading for the unstrengthened and
strengthened model.
From the Table 4, it is clear that Model B outperforms Model a
regarding deflection and stress generated. Given below is the graph
comparison of the analysis of both models.
We can see from the graphs plotted; Model B can give better results
than Model A even at the load double to that in the case of Model A.
Figure 8. Variation of deflection with load in different model.

Figure 9. Variation of Shear Stress with Loading for Different Model.
Table 4. Comparison of Unstrengthened and Strengthened Joint
Parameter Model-A at 30kN Model-B at 30kN Model-B at 60kN

Deflection 8.663mm 2.768mm 5.941mm
Shear stress 24.74 N/mm2 13.57 N/mm2 22.74 N/mm2
CONCLUSION
The following observation and conclusions can be drawn based on the

analytical results of the study.
 By doing CFRP strengthening on the beam-column joint, the joint

becomes more effective is withstanding the stresses and
deflections. This clearly states that strengthening by CFRP sheet
can be considered as one definite way of improving the working of
joint.
 The load carrying capacity of strengthened Model-B is 45.15%
more than the unstrengthened Model-A.
 Deflection in Model B is reduced by 68.04% when compared to
the Model A.
 On comparing the results, we can confirm that the load carrying

capacity of the strengthened Model-B is comparatively more than
that of the unstrengthened Model A.
 CFRP strengthened Model-B shows better ductility and strength as
compared to unstrengthened Model-A.
 This work can be further carried out by strengthening these joint of
a building frame, which can further lead to a comparison of
stresses generated in the corresponding beams and columns
meeting at a joint
REFERENCES
Bahrami. S, Madhkhan M, Shirmohammadi. F, Nazemi. N, (2017),

Behavior of two new moments resisting precast beam to column
connections subjected to lateral loading, Eng. Struct., volume 132:808-
821.
Decker. CL, Issa MA, Meyer KF, (2015), Seismic investigation of interior
reinforced concrete sand-lightweight concrete beam-column joints,
ACI Struct J. volume 112(3):287-297.
Georgia DK, Yasin. MF, Loannis LP, Ioannis NP, Spyros GT, (2017),
Numerical investigation of the resistance of precast RC pinned beam-
to-column joint under shear loading, Earthq. Eng. Struct. Dyn. Volume
46(9):1511-1529.
Li. B, Lam. ESS, Wu B, Wang YY, (2015), the Seismic behavior of
reinforced concrete exterior beam-column joints strengthened by
ferrocement composites, Earthq. Struct. Volume 9(1):233-256.
Naveeena. N, Ranjitham.M M., Numerical Study on Retrofitting of Beam-
Column Joint Strengthened with CFRP, (2016), Int. Res. J. Eng. Tech.,
volume 03(01):914-920.
Chapter 7
NANOMATERIALS, CERAMIC BULK AND

BIOCERAMICS: SYNTHESIS, PROPERTIES
AND APPLICATIONS
H. H. M. Darweesh
Department of Refractories, Ceramics and Building Materials,
National Research Centre, Cairo, Egypt
ABSTRACT
The overwhelming demand and interest of nanotechnology is one of

the most exciting disciplines and it almost incorporates into all applied
science branches as: physics, chemistry, materials sciences, biology,
agriculture, medicine, tissue engineering, bones caffolds, dentists, cement
and concrete, and all types of building materials particularly bricks,
ceramics and nanoceramics or in general advanced ceramics industries,
biomaterials, and many others. The author interests with using the
nanomaterials or accurately nanoparticles to prepare the ceramic batches
containing ultra fine and nano raw materials as nano-SiO2 and nano-
Al2O3 to indicate the importance of nano particles, to improve the

Corresponding Author Email: hassandarweesh777@gmail.com.
176 H. H. M. Darweesh
physical, chemical, mechanical properties and thermal properties as:

thermal shock, thermal expansion, firing resistance, and also durability of
the resulting bioproducts against aggressive environments. The
bioceramics can help people who are suffering from the deficiency in
their body bones as a result of accidents. These bioceramic products
resulting from biomaterials can compensate people for their lost or
broken bones, as arms, legs, fingers or even toes, and hence they feel
happy.
Keywords: nanomaterials, nanoparticles, bioceramics, firing, sintering,

expansion, shrinkage
ABBREVIATIONS
C: CaO S: SiO2 A: Al2O3

F: Fe2O3 H: H2O CH: Ca (OH)2
C3S: 3CaO.SiO2: Tricalcium silicate
C2S: 2CaO.SiO2: Dicalcium silicate
C3A: 3CaO. Al2O3: Tricalcium aluminate
C4AF: 4CaO. Al2O3. Fe2O3: tetracalcium aluminate ferrite
C2AS: 2CaO. Al2O3. SiO2 Dicalcium aluminate silicate (Gehlenite)
CAS2:CaO. Al2O3. 2 SiO2 calcium aluminate disilicate (Anorthite)
β-CS: CaO. SiO2 Calcium silicate (Wollastonite)
2C2S.CaCO3: 4CaO. 2 SiO2. 2 (Spurrite)
CaCO3
Al3 (Mg, Fe)2.(AS. SiO5. O13): (Cordierite)
2MgO. 2Al2O3. 5SiO2
A3S2: 3Al2O3. 2SiO2 (Mullite)
3 MgO. 4 SiO2. H2O montmorillonite
W.A. water absorption
B.D. bulk density
A.P. apparent porosity
B.S. bending strength
T.F. thermal features
Nanomaterials, Ceramic Bulk and Bioceramics 177
1. NANOMATERIALS
1.1. Introduction
Recently, nanomaterials are the backbone and basis of nanotrchnology.

Nanoscience is the most recent branch of the sciences, including vast or
nearly all specialties of researches. It is growing and increasing
continuously all over the world in the last two decades. It has the potential
to occur a revolution in the ways by which the materials and/or units could
be produced successfully. This already has a significant commercial effect
that will increase in the future. Nanomaterials are of a great interest due to
the fact that unique optical, rheological and electromagnetic properties, and
many others could be arised. These features are of a great importance in
nano-or-bioceramics, medicine, bones, etc. (Figure 1).
Figure 1. A carbon nanotube as a nanomaterial.
Nanotechnology differs from a field to a field and from a place to a

place and it is essentially used as a “catch all” to describe very small things
(Nalwa, 2000, Nalwa, 2011, Shackelford, 2004). Nanotechnology is
commonly defined as the understanding, control, and restructuring of
matters on the order of nanometers, i.e., less than 100 nm to make
materials with essentially new characteristics and functions (Poole, 2003,
Drexler et al. 1991). Nanotechnology comprises two main approaches: the
“top-down” approach, in which larger structures are minimized in size to

the nanoscale in a condition that it must maintain their original
characteristics without an atomic-level control, e.g., miniaturization in the
domain of electronics or deconstructed from larger structures into their
smaller, composite parts and the “bottom-up” approach that it can also be
called “molecular nanotechnology” or “molecular manufacturing,”
introduced by Drexler et al. (Drexler et al., 2003), in which materials are
engineered from atoms or moleculars through a process of assembly or
self-assembly(Figure 2). Most contemporary technologies depend on the
“top-down” approach. Molecular nanotechnology holds a great promise for
break throughs in materials and manufacturing, electronics, medicine and
healthcare, energy, biotechnology, information technology, and national
security (Askeland and Phule, 2006, Koch, 2002, Smith and Hashemi,
2005, Ashby and Jones, 2006, Sobole and Ferrada-Gutiérrez, 2005,
Darweesh, 1992, Tseng and Nalwa, 2009, Ashby, 2005).
Figure 2. The “Top-Down” and “Bottom-Up” approaches in nanotechnology.

Recently, the idea of creating the situations under which atoms or

molecules could be self assembly into beneficial structures was driven by
the reduction of its energy. The most important advantage of the self-
assembly is that the system concentrates to a definite configuration without
the need to a further control. Typically, the aggregates formed by the self-
assembly seemed to be bonded by relatively weak bonds with binding
energies only a few times higher than the thermal energy per atom. The
most important example of the self-assembly is the production of carbon
nanotubes. These nanostructures are composed of carbon atoms that
assemble into cylinders of about 1.4 nm in diameter (Figure 3). The use of
carbon nanotubes to make simple gears evolved by bonding ligands onto
the external surfaces of carbon nanotubes is to produce “gear teeth”
(Figure 4).
Figure 3. A single-wall carbon nanotube (as wrapped sheets of graphene) of 1.4 nm

diameter.
Figure 4. Nanogears made of carbon nanotubes with added teeth.

Figure 5. Self assembly of nanoclusters.
The efficiency of these gears depends on placing those gears teeth just
right in atomically precise positions. Also, the self assembling molecules
form micelles are very important. Micelles are aggregates of amphiphilic
molecules. Molecules with one end are soluble in water and the other end
that rejects it. These aggregates form spontaneously at a size that depends
on the concentration of the amphiphilic molecules in solution. The center
of the micelles acts as a chamber for chemical reactions (Darweesh, 1992;
Tseng and Nalwa, 2009; Ashby, 2005). Thus, it dictates the size of the
nanoparticles created (Figure 5).
1.2. Ultrafine Particles
Ultrafine particles are those particles that are having a grain size
distribution of nanometers (10-9 m). These are considered as nanostructured
materials (Nalwa, 2000; Drexler, 1991; Tseng and Nalwa, 2009), i.e., any
material comprised grains as clusters below 100 nm and/or layers of the
same dimensions had been seen as a nanostructure unit (Nalwa, 2000;
Poole and Owens, 2003; Ashby, 2005). The importance of those materials
is due to its too small size as: particles, grains or phases. Its high surface-
to-volume ratio, these materials shows unique mechanical, optical,
electrical and magnetic characteristics (Poole and Owens, 2003; Nalwa,
2000; Drexler, 1991; Ashby et al., 2009; Tseng and Nalwa, 2009). The
features of nanostructured materials mainly depend on the following

microstructural items:
i. Grain size distribution (< 100 nm).

ii. Chemical analysis and its ingredients.
iii. Existence of interfaces.
iv. Reactions between constituents dormant.
These features evidently and can easily determine the marvelous

characteristics of nanostructured materials.
1.3. Nanomaterials
Nanomaterial: It distinguishes between the external dimensions and its

internal structure of the whole object. Thereby, it can be >100 nm in all
three dimensions that are still being considered a nanomaterial if it has
structural features within the nanoscale range, i.e., between 1-100 nm
(Askeland and Phule, 2006; Koch, 2002; Smith and Hashemi, 2005; Ashby
and Jones, 2006; Sobolev and Ferrada-Gutiérrez, 2005; Darweesh, 1992;
Tseng and Nalwa, 2009; Ashby, 2005).
Sometimes, nanomaterials occur naturally. Engineered nanomaterials
are very important. To obtain useful advantages of extra fineness of
particles, they are designed to be applied in several commercial scopes as
sunscreens, sporting tools, pain-resistant clothes, cells, ceramics, bones,
dentists or teeth, and others (Nalwa, 2000; Ashby, 2009; Ashby, 2005).
The ratio of a fineness-to-a volume of nanomaterials is > its conventional
forms, which in turn, can lead to a greater chemical reactivity that affected
positively on their strength. At the nanoscale, quantum effects can become
much more important to identify many of materials properties leading to
optical, electrical and magnetic behaviors. Now, nanomaterials can be used
commercially in stain-resistant and wrinkle-free textiles, cosmetics,
sunscreens, electronics, paints, varnishes, windows, sporting machines,
bicycles and auto cars.
Nanoscale TiO2 and nano scale SiO2 are being used as a filler in
several products as bioceramics as: bones, teeth or dental fillings (Nalwa,
2000; Poole and Owens, 2003; Sobolev and Ferrada-Gutiérrez, 2005;
Tseng and Nalwa, 2009; Ashby et al., 2009; Ashby et al., 2007; Callister,
2007).
1.4. History and Advances of Nanomaterials
Nanomaterials started at once after the big bang when nanostructures

were formed in the early meteorites. Later, the nature evolved a lot of
nanostructures as: seashells and skeletons. Nanoscale smoke particles were
formed during the use of fire by early humans. However, the scientific
history of nanomaterials started much later. Nanostructured catalysts have
also been studied for over the last seventy years. By the early 1940’s, the
precipitated waste of silica fume nanoparticles were being manufactured
and could be sold as substitutes for ultrafine carbon black for rubber
reinforcements, and also for cement industry. Nanosized amorphous SiO2
could be applied in many consumer products as: automobile tires, optical
fibers and catalyst supports. During the period of 1960s to 1970’s, metallic
nanopowders for magnetic recording tapes were developed and improved.
In 1976, nanocrystals produced now by the most popular inert-gas
evaporation technique (Poole and Owens, 2003; Drexler et al., 1991;
Sobolev and Ferrada-Gutiérrez, 2005; Tseng and Nalwa, 2009).
Today, the menopause engineering has expanded very rapidly to
develop a number of organic and/or inorganic structural and functional
materials. This permits to impact mechanical, catalytic, electric, magnetic,
optical and electronic functions. The production of nanophase or cluster-
assembled materials is often based on the devoloping of small clusters
separately, which in turn are fused into a bulk-like material or on their
embedding into compact liquid or solid matrix materials, i.e., Si-nanophase
which differs from normal Si in physical and electronic properties that
could be applied to macroscopic semiconductors to develop new devices.
When a normal glass is adapted with colloids and/or semiconductors, it
becomes a high performance optical medium (Sobolev, K. and Ferrada-

Gutiérrez, 2005; Darweesh, 1992; Tseng and Nalwa, 2009; Ashby, 2005;
Ashby et al., 2009; Ashby et al., 2007; Callister, 2007).
1.5. Classification of Nanomaterials
Nanomaterials have very small size particles which are having 1-3
dimensions ≤ 100 nm or more. Nanomaterials can be nanoscales in one
dimension, e.g., surface films, two dimensions, e.g., fibers and three
dimensions, e.g., particles. They often exist in single, fused, aggregated
and/or agglomerated forms in spherical, tubular, and irregular shapes.
Conventional kinds of nanomaterials comprise nanotubes, dendrimers and
fullerenes. Nanomaterials have displayed different physical and chemical
characteristics from normal chemicals, i.e., nanosilver, carbon nanotube,
fullerene, photocatalyst, nanocarbon, nano-SiO2, nano-Fe2O3 and nano-
Al2O3 (Koch, 2002; Sobolev and Ferrada-Gutiérrez, 2005; Tseng and
Nalwa, 2009; Ashby, 2005; Ashby et al., 2009; Ashby et al., 2007;
Callister, 2007).
Figure 6. I: (0D: Spheres and clusters), II: (1D: Nanofibers, wires and rods), III: (2D:
Films, plates and network), V: (3D: Nanomaterials).
Nanomaterials are materials that characterized by an ultra fine grain

size (< 50 nm) or by a dimension limited to 50 nm. Nanomaterials can be
created with various dimensions. Nanostructured materials could be
classified into:- Zero dimension (0D) as: clusters and/or cluster assemblies
and filaments; one dimension (1D) as: nanofibers, wires, multi-layers and
rods; two dimensions (2D) as: ultrafine-grained over-layers or buried
layers, films, plates, and networks and three dimensions (3D)
nanostructures as: nanomaterials, nanophase materials consisting of
equiaxed nanometer sized grains (Figure 6).
1.6. Importance of Nanomaterials
In recent years, nanomaterials have taken an overwelming interest due

to the following reasons:
 Nanoceramics are flexible at high firing temperatures if compared

with those containing coarse size particles (Nalwa, 2000;
Darweesh, 1992; Sobolev and Ferrada-Gutiérrez, 2005; Tseng and
Nalwa, 2009; Ashby, 2005; Ashby et al., 2009; Ashby et al., 2007;
Callister, 2007).
 Structured nanosemiconductors did not display different lineal
optical characteristics. Always, they are used as solar cells window
layers.
 Nanosized minerals ashes often use to produce gas-tight materials
as: dense parts of porous linings.
 The very small or nanoparticles have special atomic structures with
distict electronic states, which give rise to special properties.
Nanomagnetic composites are used for mechanical strength
devolve as: magetoliquids, a high density storage knowledge and
magneto refrigeration.
 Structured nanomineral clusters and colloids of mono- or
crystalized mineral compositions have a particular action in
catalytic applications using as precursors for recent hetero-geneous
catalyses or cortex-catalyses. They tend to give essential
advantages regarding to activation, selectivity and lifetime in
chemical or electrocatalylic transformations or fuel cells (Nalwa,

2000; Tseng and Nalwa, 2009; Ashby, 2005; Ashby, 2009).
 Structured nanomineral-oxide thin films are having a continual
attention for inquiring gas sensors as: NO3, CO, CO2, CH4 and
some aromatic hydrocarbons. Structured nanomineral-oxide, MnO2
could be applied in re-chargeable batteries for cars or consumer
goods. Crystalline nono-SiO2 sheets for highly transparent contacts
in solar cells and also structured nano-TiO2 porous films for its
high transmission and significant fineness increasing going to a
strong absorption in dyesensitysed solar cells (Poole and Owens,
2003; Drexler et al., 1991; Askeland and Phule, 2006 Koch, 2002;
Tseng and Nalwa, 2009).
 Composites of polymers with high levels of inorganics with a high
dielectric constant are interesting materials for structured photonic
band gaps.
A variety of size-related effects can be incorporated by controlling the

sizes of ingredients (Nalwa, 2000; Sobolev and Ferrada-Gutiérrez, 2005;
Ashby, 2005; Ashby et al., 2009; Ashby et al., 2007; Callister, 2007), e.g.,
the mechanical strength of nanostructured minerals and nanoceramics are
get be better than that of the conventional materials which is mainly due to
the ultrafine microstructured (Nalwa, 2000; Sobolev and Ferrada-
Gutiérrez, 2005; Ashby, 2005; Ashby et al., 2009; Ashby et al., 2007;
Callister, 2007).
Nanostructured materials as the manufacturing of devices with large
magetoresistance enfluences are of magnetic applications (Nalwa, 2000).
Furthermore, nanostructured minerals and nanoceramics are tended to be
candidiates for new catalytic applications (Nalwa, 2000; Poole and Owens,
2003). The growth of semiconductors and nanoclusters is on the circle of
intense research works particles (Nalwa, 2000; Sobolev and Ferrada-
Gutiérrez, 2005). It is important because it is leading to electrical, optical,
optoelectronic, magnetic and magneto-optical properties, e.g., quantum
dots can be modified to emit and absorb a desired wavelength of light by
varying the particle diameters (Nalwa, 2000).
2. CERAMICS
The modern term ceramic was derived from an ancient Greek word
which referred to a burning or firing process. More recently, a ceramic
product was defined as an article made from clay with or without the
addition of either materials, shaped in the plastic state, dried and then fired
to give it the required strength and durability. Also, it may be defined as
one composed of inorganic, but non-metallic material (Jackson, 1969;
Kingery et al., 1975; Rayan and Radford, 1987). On this basis, ceramic
products are usually subdivided into four sections as follows:-
1) Structural products as bricks, roofing tiles, pipes, wall and floor

tiles.
2) Refractories, i.e., ceramics used mainly for their resistance to high
temperatures as kiln linings, steel and glass plants, and many other
products used at elevated temperatures.
3) Electrical and special ceramics: Modern ceramic materials, some
of which show exceptional qualities of refractoriness, hardness,
strength, chemical stability and specific electrical and inorganic
properties.
4) Whitewares as: Table and decorative wares, wall tiles and sanitary
ware products.
Nowadays, ceramics include a vast range of products, many of which

may be contain no clay whatsoever, so that a simple and definite definition
is impossible. Firing is still an essential feature in the manufacture of the
vast majority of ceramic articles, but there are few exceptions where the
product is classified as ceramic even in its nature, i.e., without firing
(kingery et al., 1975; Rayan and Radford, 1987). In the developing
countries, the average standard of living has been improved to a great
extent throughout the last two decades and is still improving continuously
with time. In Egypt, the policy of construction has been greatly increased
in order to face the overpopulation problem. So, there is an increasing
demand for ceramic tiles. Accordingly, new ceramic producing factories

were built and some others are still under construction so as to meet the
overwhelming demand on ceramic tiles. Though the world production of
ceramic floor and wall tiles is approximately 100 million square
meter/year, it cannot reply the world needs (Drews, 1983; Rayan, 1978).
Nanoceramics are composed of raw nanoparticles which can be classified
as inorganic, heat-resistant, nonmetallic solids compounds. In the last
decade, researchers have shown that nanoceramics have a lot of functions
than known materials (Khalil, 2012).
Nanoceramics were found out during early 1980s. It is fabricated by
sol-gel techniques, i.e., mixing raw nanomaterials in a gel solution to
produce nanoparticles. Recently, they involved by sintering through a
pressure and a heat due to many advances in fields of electronics (Njindam
et al.,2018; Martín-Márquez et al., 2008; Nuttawat et al.,2013; Kim and
Hwang, 2016; Ke et al., 2016). Increase the various solid wastes in the
countries particularly those do not resolve easily in nature by weathering
for long periods as glass bottle wastes that created ground pollution. The
recycling of these wastes is an important environmental and economical
solution. The glass is composed of high amounts of SiO2 (70–74) % and a
reasonable amount of alkaline earth oxides: Na2O (12–16) %, CaO (5–11)
% and MgO (1–3) % (Njindam et al., 2018; Martín-Márquez et al., 2008).
In Egypt, there are huge quantities of waste glass bottles of several shapes
and sizes. This evidently causes an environmental problem and acts as a
heavy burden on the environment. Therefore, it is very necessary to solve
this serious problem through scientific researches that could use these
priceless waste materials in traditional ceramics like as the wall and/or
floor tiles, table wares or porcelain stoneware tiles. These may be used for
out and/or indoor applications. The inert component glass waste can be
used as a fluxing agent in ceramic industry.
Clay is a product which is coming from the decomposition of granite
rocks. It is essentially composed of Al2O3 and SiO2 which represent the
major constituents of clay (Nuttawat et al., 2013; Kim and Hwang, 2016;
Ke et al., 2016). Clay is known to provide the plasticity of the whole
mixture. Granite often decomposes into feldspar that is composed of
Al2O3, SiO2, and a flux of an alkaline nature e.g., Na, K, Li and Ca. The
fluxing agents as feldspar, talc, pyrophyllite, etc. could produce the liquid
or glassy phase during sintering while SiO2 favours the dimension control
of the product during firing (Nuttawat et al., 2013; Kim and Hwang, 2016;
Ke et al., 2016). The nanoceramics have unique properties due to its
smaller size particles and molecular structure. To construct these
nanostructures using a “three-step process” to develop these complex
structures, the nanoceramic Greer is currently used Al2O3, or aluminum
oxide, and its maximum compression is about 1 micron from a thickness of
50 nanometers. After its compression, it can revert to its original shape
without any damage on the structure and was coated with Al2O3, different
molecules could yield a different result (Sonjida, 2011; Vorrada, 2009;
Darweesh, 2015).
2.1. Synthesis
For making a nanoceramic, it begins first with the sol-gel process. It

involves a chemical solution, made of nanoparticles in the phase, usually a
gel or polymer, made of molecules immersed in a solvent. The sol-gel is
then mixed to produce an oxide material which is generally a type of
ceramic. The rest of the excess liquid solvent is removed with a drying
method by evaporation. The desired particles are then heated through
densification to return a solid product (Jackson, 1969; Darweesh, 2016).
This method could also be applied to produce a nanocomposite by heating
the gel on a thin film to form a layer of nanoceramic on top of the film
(Sridhar, 1996).
2.2. Sintering
Sintering is a process of consolidating nanoceramic powders using

high temperatures to press the powder into a solid material. High
temperature sintering results in a rough material that deteriorates the
properties of ceramics and requires a lot of time to obtain an end product.

Microwave sintering has been developed to overcome the time and energy
problems of the previous technique. In microwave sintering, radiation is
produced from a magnetron to vibrate the powder and thus heating them.
This method allows for heat to be instantly transferred across the entire
volume of material instead of a graduate temperature gradient (Jackson,
1969; Darweesh, 2016). They first start to place the nanopowder in an
insulation box that is composed of low insulation boards. They place it in
an insulation box in order to increase temperature because nanoceramics
don't absorb microwave energy well in the room temperature. Inside the
boxes are suspectors that absorb microwaves at room temperature to act as
an initial heat source and initialize the sintering process. When the box is
placed in the microwave furnace, the microwaves heat up the suspectors, to
about 600 °C. Then, the nanoceramics start to absorb the microwaves and
bind the nanoparticles together.
2.3. History
In the early 1980s, Roy Komareni developed the first way to

synthesize nanoparticles, specifically nanoceramics (Julie, 2002). He used
a process called sol-gel and enabled researchers to test the properties of
nanoceramics. This process was later replaced by sintering in the early
2000s and continued to advance microwave sintering. (Nissan, 2014).
Seashells are very tough and yet are very similar to ceramics. They
discovered that the durability of seashells come from their
microarchitecture (Kimm, 2013). In 2012, California Institute of
Technology succeeded in replicating the skeletal structure of sea creatures
like sea sponge using ceramics (Sridhar, 1996). They developed the
nanoarchitecture that is known as nanotruss. It can be used for materials
other than ceramics to take on the properties of that materials, while
retaining its durable build. Their largest scale of nanoceramic is a 1mm
cube and they are still currently working on.
Nanoceramics have a large variety of functions and can be applied in

many fields including energy supply, and storage, communication,
transportation systems, construction, electrical and medical technologies.
There are also physical applications such as nanoceramics in nanotruss
architecture. Nanoceramic holds very flexible and durable properties
making the material very light and yet be as strong as current building
materials like concrete or steel. Medical technology has been using
nanoceramics for bone repair and developing materials that can enter the
body without conflict (Singer and Singer, 1971). Ceramic floor and wall
tiles are building materials which are designated for use as floor and wall
coverings both indoors and outdoors regardless of shape or size. They
undergo some processing such as milling, screening, blending and wetting.
Tiles are shaped by pressing methods at room temperature (ASTM;
Mostafa and Darweesh, 1992; Njindam et al., 2016). The possibility to
utilize the glass waste (10-40 %) as a flux with clay to produce porcelain
stoneware tiles was studied (Ashby et al., 2009). The firing shrinkage
inceased with glass content only up to 30 %, but more than that the product
started to melt and spread. It is increased also with firing temperature. The
water absorption decreased with glass waste content up to 30%, while the
bulk density increased. This reflected positively on the mechanical
strength.
2.4. Classification of Ceramics
The ceramic industries could be classified commonly into the

following main categories owing to the similarities in service,
manufacturing method and raw materials used (Ke et al, 2016; Sonjida et
al., 2011):
2.4.1. Heavy Clay Wares

Heavy clay wares or structural ceramics are the products used
extensively by the building trades and in civil engineering projects. They
include bricks of all types, commons, facing and engineering bricks,
hollow or perforated bricks, clay roofing tiles, draining and sewer pipes.
The used raw materials vary widely from easily worked highly plastic
clays to hard coal measure shales (Ke et al, 2016; Sonjida et al., 2011).
2.4.2. Refractories
Refractory products are those used in high temperature applications, in
kilns and furnaces of all types, in steel works and in other industries, in
glass works and other power stations. The raw materials used include fire
clays for lower temperature applications and oxides such as alumina
(Al2O3) and magnesia (MgO) for the higher temperature ones. The
manufacturing methods include extrusion and moulding, whereas dry
pressing techniques account for the great majority of the non-clay
refractories (Njindam et al., 2018; Sonjida, 2011).
2.4.3. Special Ceramics

Special or technical ceramics include many products which have little
or no resemblance to the conventional ceramics because they are always
made from pure synthetic raw materials to confirm technical specifications,
for example: electronics, aeronautics, nuclear engineering and electrical
power (ASTM, 1980; Mostafa and Darweesh, 1992).
2.4.4. Pottery (Whitewares)

The term pottery or white wares includes normal domestic tablewares,
wall tiles, sanitary wares and porcelain tills. The tableware also includes
earth wares, bone china and porcelains, other types include stonewares and
a number of proprietary brands. Ceramic products can be divided into
those used at normal and high temperatures. These divisions may be
subdivided into those products which have a porous body after firing and
those with non-porous ones (Chen-Chi and Jackie, 1999; ASTM, 1980) as
shown in Figure 7.
Each product must possess their properties in use such as: beauty,
mechanical strength, temperature resistance or any other property. The
properties of the finished ceramic article will broadly depend on:
i. Types, purity and relative amounts of materials from which it is

made.
ii. Methods and conditions used in its manufacture.
But, the two cannot be divorced since the materials used will dictate to
a considerable extent, the most suitable manufacturing processes which are
available. The composition of the traditional ceramic units lies, in general
in definite limits. They are controlled by some factors: 1-Shape of the ware
which needed to the inclusion of a sufficient plastic material (clay). After
shaping, the clay also holds the ware together during drying. 2-Avoiding of
cracking during firing due to the incorporation of the non-clay materials
which do not shrink during firing as: flint, quartz and tiles grog. 3-The
quantity of fluxes so that a sufficient glassy material formed to hold it
together when fired and so it does not warp (Njindam et al., 2018;
Darweesh, 2015).
Figure 7. The Classification of Ceramic products.

Figure 8. Triaxial diagram showing areas of commercial wares.
Now, the traditional bodies are essentially made up of clay, flint or

quartz, and feldspar that can be represented by a tri-axial diagram as shown
in Figure 8. The most natural method is by means of the batch
composition. But, it was shown from the raw materials analysis that the
traditional ones, namely clays and feldspars, are of available composition
and properties (Darweesh, 2016; Kimm, 2014). Clay is an essential
ingredient in the majority of white wares as well as ceramic tiles. It
possesses a number of characteristics which the ceramic technologist must
consider when choosing the most suitable clay for a particular application
(Njindam et al., 2018; Martín-Márquez et al., 2008; Nuttawat et al., 2013;
Kim et al., 2016; Ke et al., 2016; Sonjida et al., 2011).
2.5. Properties of Ceramics
2.5.1. Plasticity
When a dry clay is mixed with water, there are two water contents
which could be recognized easily and simply by working the mass of the
clay/water body in the hands. Above the higher value of water content, the
mass becomes sticky and difficult to handle. Below the lower one, it tends
to crack and crumble. Accordingly, between these two values (upper and
lower plastic limits), the clay can usually be worked without any difficulty.
Therefore, the plasticity can be expressed as workability which enables a
material to be deformed continuously under a force exceeds certain
maximum values. Plasticity enables a material to be changed in shape
without rupturing by the application of an external force and to retain that
shape when the force is removed or reduced below a certain value, but
there is no a quantitative definition or units for plasticity (Njindam et al.,
2018; Vorrada et al., 2009; ASTM, 1980).
Many trials were carried out to find a relationship between the stress
applied and the resultant measured strain. The stress may be considered as
a force per unit area and the strain as the resulting deformation of the body.
Basically, there are three types of stress which may be applied as tensile,
compressive or shear forces. The property of a body to return to its original
size and shape after having been stretched, compressed or deformed, is
simply that stress is proportional to strain (Rayan and Radford, 1987) as
follows,
Stress
K (1)
Strain
where, k is a constant and is called as “Modulus of elasticity.” When a

body is subjected to a simple tension (or compression), K is known as
“Young’s modulus” and the stress = force per unit area (F/A).
Strain =Extension per unit length
dI
S (2)
I
2.5.2. Factors Affecting Plasticity

Fineness: By separating clay samples into a number of size fractions,
Kalnin (kalnin, 1967; Kalnin, 1968) was able to relate measured plasticity
indexes to fineness expressed as surface area. His results indicated that the
relation was:
log S  AP  log B (3)
Where, S is the surface area of the clay sample (expressed as cm2/1000 g),
P is the measured plasticity index, A and B are constants.
The plotting of log S against B gave straight line graphs with intercept
on the log S axis giving the value of log B. In all samples, B was found to
be 1.8 x 106. Thus for P to have any value, S related to be greater than 1.8
x 106 cm2/100g (Rayan and Radford, 1987; Drews, 1983; Mostafa and
Darweesh, 1992). As the material becomes finer, not only does plasticity
increase, but also the amount of water needed to produce optimum
plasticity increases, and plasticity is exhibited over a wider range of water
content. Figure 9 indicates this behavior for various size fractions of clay.
Figure 9. Plasticity index as a function of water content for five size fractions of clay
sample.
2.5.3. Determination of Plasticity

There are many test methods can be used for measuring plasticity, but
its most are not accepted a definite or accurate test for plasticity. Tests
which depended mainly on the moisture content are the optimum (Rayan
and Radford, 1987; Chiang, 1997). So, the most used tests are:
2.5.3.1. Pfefferkorn Test

A standard weight is released from a fixed height and falls on to a
cylindrical sample of the material under test. The cylinder is deformed.
The deformation ratio is defined as:
Original length of the cylinder

Deformation ratio  (4)
Deformed length of the cylinder
Pfefferkorn number is calculated as the moisture content at which the

deformation ratio lies in the range of 3:1. The higher the Pfefferkorn
number is the more plastic material. The test is repeated several times and
then the mean value is considered (Rayan and Radford, 1987; Drews,
1983; Rayan, 1978; Khalil, 2012).
2.5.3.2. Atterberg Plasticity Test

Two moisture contents are measured corresponding to the plastic and
liquid limits. The plastic limit is the moisture content below which the
material stops to behave plastically and become crumbly. While the liquid
limit is the moisture content above the material acts as a fluid. The material
becomes as a paste if it is being in contact with water and placing it in a
standard cup, a standard groove is cut in the sample and the cup is given a
number of standard bumps to close the groove. The liquid limit can be
defined as the moisture content at which about 25 bumps are needed to
close the gap (Rayan and Radford, 1987; Drews, 1983; Rayan, 1978;
Khalil, 2012).
The plasticity index is the numerical difference between the moisture
content of the liquid and plastic limits. This method is looking for the
range of a moisture content over which the material can behave plastically,
where the greater the range is the higher plasticity. Figure 10 shows the
BCR compression plastometer by which the plasticity is measured. The
motor moves the lower plate upwards at a constant speed and when the
cam scale reads zero deformation, the upper plate just come into contact
with the top of the sample. As the cam turns further and the lower plate
rises, the sample is compressed between the upper and lower plates
(Drews, 1983; Chiang, 1997).
Figure 10. A diagrammatic representation of the BCR compression plastometer.
Readings on the dial gauge caused by flexing of the beam

(proportional to the stress causing deformation) are taken at 10 % and 50
% deformation as registered on the cam scale (proportional to the strain).
The plasticity index P can be deduced by the following relation:
Dial reading at 10 % deformation

P  1.8  (5)
Dial reading at 50 % deformation
where, 1.8 is a factor used to permit for a change in the cross-sectional area
of the cylinder during deformation and is being equal to:
Cross-sectional area of sample at 50 % deformation

Cross-sectional area of sample at 10 % deformation
Care must be taken to ensure that the samples are of the correct length
and have parallel ends. Deformation of the sample cylinder must therefore
be avoided when unmoulding it from metal die. Often sex samples are
tested and the mean value is considered. Shrinkage of the shaped ceramic
articles takes place during drying and firing stages of manufacture. The
extent of these contraction is very important where not only will the size of
the finished product depend on their values, but also if they are too large,
or if they are occurring too rapidly, cracking or distortion of the article is
likely. So, the ceramist must recognize the wet-to-dry, dry-to-fired and
wet-to-fired processes. Each may be expressed on either a linear or a
volume basis depending on the method of measuring (Rayan and Radford,
1987; Drews, 1983; Rayan, 1978; Khalil, 2012) as the wet-to-Dry
contraction. In most pottery bodies, the clay content is primarily
responsible to determine the extent of the contraction though the ratio, size
and size distribution, particle shape of the non-clay component also play a
vital role. The fineness and plate-like character of the clay crystallites lead
not only to the desirable properties of the high plasticity and dry strength,
but also to high drying contraction. It is well known that a small drying
shrinkage is often desired in pottery products, where it permits the formed
product to shrink away from the plaster mould and facilitates its removal
from the mould. However, the excess shrinkage, specially anisotropic
shrinkage, i.e., different shrinkage values in different directions due
usually to a particle alignment that taking place during pugging, plastic-
making or slip casting, is responsible for a considerable rate of
manufacture losses. High drying shrinkage is also resulting in a loss of
dimensional accuracy and variations, which may give problems in
decoration, e.g., a lithograph is applied to the circumference of a plate, a
high dimensional accuracy and the moisture content of the body may be
kept low as possible (Rayan, 1978; Khalil, 2012), ASTM, 1980).
When a plastic body dries, water is removed from the surface and
replaced by water from the interior. As water content is reduced, particles
are drawn closer together. The particles eventually will touch each other
and though water still exists in the voids between particles, no further
shrinkage can take place on the removal of this water (Figure 11). If a
piece of plastic clay or body lets to dry, values of the volume and the
corresponding moisture content are given at intervals, a plot of volume as a
function of moisture content is as shown in Figure 12. On drying, the
shrinkage takes place as water is removed. During this stage, water still
separate the particles, but the particles are moving closer together if water
is removed. The point at which there is no further removal of water is
known as the critical moisture content “CMC.” Its value often varies from
clay to clay and also from a ceramic body to another. This depends
essentially on clay to non-plastic ratio, particle size, distribution and
particle shape. So, this value is very important where drying from water
content above it will result in a shrinkage involving a risk of distortion or
cracking, and if below it, no shrinkage is involved.
Figure 11. Shrinkage of a plastic body during drying.
Figure 12. The volume as a function of moisture content for a plastic body during
drying.
At moisture content higher than the CMC, the rate at which water is
lost will be constant under constant conditions. Since, drying involves only
evaporation of water from the surfaces of samples which is replaced by
water flowing from the interior of the surface. This is known as “Constant
rate period” (Rayan and Radford, 1987; ASTM, 1980; Konta, 1972). The
interval between the CMC and the dryness is known as “Falling rate
period” (Figure 13). Clay particles are a plate-like in nature and when
mixed with water to form a paste or slip, the particles are free to orientate
themselves when a force is applied. This orientation occurs to present the
least resistance to the applied force, e.g., in a pug, the particles will align
themselves in the direction of the force causing extrusion. The extruded
column of the clay will contain particles oriented so that their major axes
are parallel to the length of the column (Figure 14). Similar orientations
occur during plastic making and slip casting processes.
Figure 13. A plot of water removal rate as a function of moisture content for a plastic
body during drying.
Figure 14. Preferable orientation of the produced particles resulting from anisotropic
drying shrinkage.
2.5.3.3. Test Methods
a) Clays
About 0ne kg of the clay is mixed with water to give a slurry which is
sieved through 120 mesh (China clay) or 80 mesh sieve (plastic clay). The
resulting suspension is poured on to a plaster of Paris batt and then let to
dry to the plastic condition. When sufficiently dry, the clay is removed
from the batt and thoroughly wedged by the hand till all air inclusions have
been removed, and a freshly cut surface (cut with a cheese wire) has a
homogeneous appearance. The test samples made for measuring its
shrinkage may be either slabs or, a hand-moulded in a plaster of Paris
moulds, or rods of a circular cross –section of about ½ in diameter. Rods
are generally extruded with the help of wad box, and may be used for
shrinkage and modulus of rupture. When the mechanism is operated, a
plastic clay is forced through a tapered section and ultimately through a ½
in diameter orifice to produce rods. Soon after extrusion, the rods are
placed in the grooves of a wooden pallet. The rods are then let to dry in air
for 24 hours after which drying is completed in a drier at 110 ºC. The
distance between the marks is then measured on each of the five rods and
the mean dry length is obtained (kingery, 1975; (Rayan, 1978, Todor,
1972). The linear wet-to-dry shrinkage is reported on the wet basis, i.e., the
% linear drying shrinkage (wet basis) is given by:
Wet length - Dry length

Linear drying shrinkage, %= (Average) 100
Wet length
The moisture content of the rods at the beginning of drying should also
be quoted with these results. The rods should be then fired through the
production kiln in such a position as to receive an average firing
temperature, and a pyroscope (cone or ring) should be fired with the
samples. Again, the distance between marks is measured on the rods, and
the average fired length is calculated. It is suggested that the linear dry-to-
fired shrinkage is reported on the dry basis, i.e., the % linear firing
shrinkage (dry basis) is given by:
Dry length - Fired length

Linear frying shrinkage, %= (Average) 100
Dry length
The pyroscope reading should be reported together with the result of

firing shrinkage (kingery, 1975; Khalil, 2012; ASTM, 1980; Mostafa and
Darweesh, 1992).
b) Ceramic Body
Concerning the shrinkage measurements on the body, the production
of a casting slip or a plastic body should be used as the starting material,
where the same procedures as for clay are used.
c) Volumetric Contraction
The more accurate measurement of both drying and firing shrinkages
could be obtained on the basis of volumes by using Doulton Densometer
(Figure 15). Two weights are often required during drying of sample, the
weight of the sample (w1) and the down thrust force (w2), which is equal
to the up thrust of mercury + the sample weight when the sample is
immersed in mercury and held immersed by the cradle.
Figure 15. Doulton Densometer for bulk density and volume shrinkage.
An egg-shaped sample is moulded from the plastic clay or body using

a special plastic mould. Then, it is weighed to an accuracy of 0.01 g on the
top-pan balance. A sample beaker containing 1 g of mercury is placed on
the balance pan, the cradle is lowered into the mercury using the coarse
hand wheel adjustment. The bridge is then locked in a place with the
locking screw. Using the micrometer adjustment, the needle pointer is
adjusted to just touch the surface of the mercury. The balance is then
turned back to zero. The cradle is then raised using the coarse adjustment
and the sample placed on the surface of the mercury under the cradle. The
cradle is lowered using the coarse adjustment till the sample is immersed.
The bridge is locked and the pointer is once again adjusted to touch the
surface of the mercury using the micrometer adjustment. The balance is
read to 0.01 g. this weight represents w2 the up thrust due to mercury
displaced + the weight of the sample (Rayan and Radford, 1987; Drews,
1983; Rayan, 1978; Khalil, 2012).
Several readings of w1 and w2 at intervals are taken, while the sample
is drying. The sample is fully dried in an oven and the measurements of w1
and w2 are determined. The moisture contents of the sample at different
stages of drying can be obtained by subtracting the dry weight of the
sample from the various values of w1 that obtained during the drying
process. The volumes of the samples corresponding to these moisture
contents can be calculated as w2/D, where D is the density of mercury at
the room temperature. So, w2 is measured to an accuracy of 0.01 g, the
volume is obtained to an accuracy of 0.001 cm3. The volume of a sample is
then plotted against to a moisture content.
d) Adsorbed Water
Plasticity is associated with adsorbed water films around each clay
particle or micelle. The electrical charges on clay particles enable them to
adsorb polar micelles like water at their surfaces. These adsorbed films
appear to act as lubricants, and are responsible for the plastic behavior. If
water in a clay paste is replaced by a liquid of lower polarity, e.g., alcohol,
the plasticity is much reduced since adsorption of the liquid is lessened.
Non-polar liquids as benzene give almost no plasticity when mixed with

clays (Jackson, 1969; kingery et al., 1975; Rayan and Radford, 1987).
e) Ionic Effects
All clay particles carry a negative charge, and hence have the ability to
attract and adsorb cations at their surfaces. These cations are not firmly
held and on treatment with suitable electrolyte solutions one type of
cations can be replaced by another, g. a clay having adsorbed Ca2+ ions at
its surface can have these replaced by Na+ ions on treatment with Na2CO3,
Ca-Clay + Na 2CO3  Na-Clay + CaCO3 (6)
The Na-clay produced has very different physical properties than those
of Ca-clay. In a flocculated clay or body, the particles attract to each other
giving open packed particle arrangement requiring such water to produce
an optimum plasticity. Therefore, flocculation is used to increase the
workability of industrial tableware units. Such treatment suffers from a
disadvantage that unfired strength is reduced due to open particle packing
and reduced its bulk density (Rayan and Radford, 1987; Drews, 1983;
Rayan, 1978; Khalil, 2012).
f) Hand Wedging (Pugging)

This process increases the workability of the material. This is mainly
attributed to the improvements which can take place in the body texture
which is resulting from the elimination of air inclusions and reduction of
moisture distribution throughout the material.
g) Ageing of Souring
It is the practice of storing plastic clay or body for a time period prior
to its use, which commonly employed in the heavy clayware production,
but less so in white wares. The improvements in workability are essentially
due to two factors:
1) The penetration of water into hard aggregates to wet not previously

wetted outer surfaces. This increases the effective surface area of
the material.
2) Some flocculation of the material caused by dilute acids formed by
the decomposition of organic matter by a bacterial action.
When the clay dries from the plastic state or the dried unfired ware, its
strength has been developed. The dry strength of the clay depends on the
nature of the clay and the size of the clay particles. As a general rule, the
more plastic clays are the strongest. The construction usually takes place as
the clay dries due to the loss of water of plasticity. A little shrinkage is
desirable and excessive construction leads to problems such as warping
and cracking. A further construction always takes place during the firing of
clay article due to the melting of fluxes which are present in varying
amounts in the different clays (kingery, 1975; Mostafa and Darweesh,
1992). Darweesh, 2015 studied the dry and firing shrinkage of some
ceramic composites containing 0, 5, 10, 15, 20 and 25 wt. % cement kiln
dust waste (CKD) at 0, 1000, 1050, 1100 and 1150ºC to throw light on the
behavior of the samples on heating according to ASTM-Specification,
1980. Table 1and Figure 16 illustrate the dry and firing shrinkage as a
function of CKD content. He found that the dry and firing shrinkage
slightly increased with the increase of CKD content up to 25 wt. %, and
also with the increase of firing temperature (kingery, 1975; 1978; Khalil,
2012; ASTM, 1980; Mostafa and H. H. M. Darweesh, 1992; Chiang,
1997).
Table 1. The dry and firing shrinkage of ceramic products with

various CKD content at different firing temperatures
CKD content, wt %
Firing temperature, 0 5 10 15 20 25
ºC 0 0 0 0 0 0 0
1000 0.022 0.043 0.105 0.165 0.189 0.251
1050 0.106 0.152 0.286 0.482 0.535 0.603
1100 0.313 0.406 0.654 1.108 1.185 1.223
1150 0.473 .518 0.654 1.312 1.342 1.363
Figure 16. Dry and firing shrinkage of ceramic products with various CKD content.
h) Fired Color
The unfired color of the clay becomes another color after firing. For
example: many pottery clays in the dry state are brown or dark grey and
become almost white color by firing. A slip is a suspension of clay in
water. When 9 parts by weight of dry constituents are mixed with 10 parts
by weight of water, the resulting slip will have the consistency of a stick
cream. If we now add a small amount of two common alkaline materials
(CaCO3 and Na3SiO3), the slip will become quite fluid and it will be
possible to add a further 17-18 parts by weight of the dry materials before
the slip starts to thicken again without adding any more water. The process
of rendering a thick slip fluid in this manner is known as deflocculation
and the materials used to bring it about are called deflocculents.
Deflocculation of a pottery body is dependent on the clays present and is
very important in the production of a slip for the making process known as
casting. This method is used for hollowware, tea ports, jugs, etc., and in the
manufacture of sanitary wares (Jackson, 1969; Mostafa and Darweesh,
1992).
Clay and Loams are sedimentary species; clay; silt and sand with low
alumina content from 16-23% and high proportion of iron compounds, are
used in the manufacture of colored ceramic tiles. They are usually divided
into two catigories: not-easily fusible pottery clays and loams, and
common pottery ones. It is subdivided by konta (Konta, 1981) into:
1) Pottery clays and/or loams containing 3.5-16% Fe2O3 and a slight

to nearly zero amount of CaO.
2) Pottery clays and loams, calcareous clays (or lime clays),
containing sometimes up to 40% calcite, which is suitable for the
production of stone tiles at low fusion temperatures (1200-
1300ºC).
2.6. Raw Materials
The chemical composition of raw materials used in the ceramic

industry is shown in Table 2 and are frequently classified to:
Table 2. The chemical composition of ceramic raw materials, wt. %
Materials Oxides Clay Felspar Quartz Limestone

SO2 53.47 75.37 93.63 0.08
Al2O3 26.78 13.62 3.64 0.03
Fe2O3 3.99 0.41 0.08 0.04
CaO 0.60 0.63 0.18 56.84
MgO 1.38 ---- ---- 0.10
MnO 0.03 0.03 0.02 ----
Na2O 1.15 3.44 0.17 0.12
K2O 1.18 5.84 0.14 0.05
TiO2 1.12 0.05 0.16 0.01
SO3 ---- 0.02 0.14 0.02
P2O5 0.51 ---- ---- ----
Cr ---- 0.02 0.06 0.08
LOI 9,72 0.67 1.78 42.63
2.6.1. Fillers
This is a rather loose term used to describe non-plastic materials used
in body compositions, which are not fluxes. In practice, the filler of white
ware bodies is always silica, either in the form of flint of sand though few
bodies often used alumina. Fillers serve to open up the body facilitating the
drying process, reducing the drying shrinkage and plasticity. Moreover,
they are controlling the thermal expansion and contributing the whiteness
of the fired bodies. Fillers are often representing a large ratio of the total
body composition. So, the particle size distribution is very important in the
determination of the particle packing in the body. This is in turn affects
drying and firing shrinkage. Therefore, the fineness of the filler affects the
thermal expansion of the fired body. Fillers are high melting, chemically
resistant inorganic materials whose main function is to reduce a ceramic
body's tendency to warp or distort when fired to high temperatures. They
also play an important role in the determinaton of thermal expansion.
Quartz and/or flint are the most common fillers used in white wares and
other ceramic bodies. Though Al2O3 is an important ingredient in many
refractory and technical ceramic bodies, it is used in a few tableware
bodies (Jackson, 1969; kingery, 1975; Rayan and Radford, 1987; Mostafa
and Darweesh, 1992; Grim, 1962).
2.6.2. Quartz Sand

Quartz or silica is the most abundant oxide in the earth's crust. It is
estimated that silica occupies approximately 19 wt. % of the outer 25 miles
of the earth. White wares often contain siiica in the form of silicates, i.e.,
silica can be combined with other elements to form the silicate minerals,
but a large proportion occurs as free silica, mostly in the form of quartz.
Since, this is the mineral form which is stable under the normal
atmospheric conditions. Earthen wares for an example are containing 72-
76 % silica. In many countries like UK, flint is the popular form of the
silica component of white wares, though the continental countries have
used quartz sand for many years. Recently, however and mainly for
economic purposes, many British manufacturers have taken to the
continental practice and replaced flint with sand in their formulations.
Sandstones are common rocks, but the sandstone pure enough to use in
white ware bodies are infrequent. Many are containing iron compounds,
which stain them brown or red colors (kingery, 1975; Mostafa and
Darweesh, 1992; Chiang, 1981).
2.6.3. Flints
Flints contain small amounts of water, minute air spaces and interstitial
amorphous SiO2 due to its mode of formation from the skeletons of
sponge-like organisms which dissolved in sea water and were later
deposited in chalky deposits. The fine microstructure of flints makes it
more reactive than quartz, and in particular, it converts during firing more
readily to cristobalite (another crystalline form of silica) which is
profoundly affects the thermal expansion of the body (kingery, 1975;
ASTM, 1980). Flint occurs as hard nodules in the middle and upper parts
of the chalk deposits often arranged in layers along bedding planes
separating the beds rather than evenly distributed throughout the chalk.
Therefore, flints were formed by precipitation from silica in solution. This
is confirmed by the fact that the white skin of flint gradually varies in
composition from CaCO3 to silica. In order to prepare flints for use, flints
pebbled from the sea shore or from chalk deposits are calcined up to
1100ºC to make it friable, and to facilitate its grinding to the required
fineness (45-50%less than 10 µ). During calcinations, flints easily shatter
due to expansion of air and water pockets within the structure. This
changes its color from black to white as an organic material is burnt off.
The specific gravity of flint is 2.62 reduced to 2.50 after calcination due to
the exfoliation of the flint on heating and to a lesser extent to the formation
of cristobalite (spec. gr. 2.3). About 18% of quartz is converted to
cristobalite during calcination process of flint. The calcined flint is wet
ground to the required size and used in a slop form in the body mix.
Grinding to the correct fineness is very important since it affects the extent
of conversion of quartz to cristobalite during firing of the body (kingery,
1975; ASTM, 1980; Darweesh, 1992). At all, flints of interest to the potter
are divided to:
2.6.3.1. Chalk Flint

These are generally large lumps with a thick chalk crust.
2.6.3.2. Wash Mill Flints

These are a by-product of the cement industry and by far most
important source of free silica for earthenware and tiles.
2.6.3.3. Beach Flints

These are smooth flint pebbles that occur on a number of beaches at
the southern coasts of England, and at the northern coasts of Belgium and
France, where the surface chalk is completely washed away (Rayan and
Radford, 1981; Sonjida et al., 2011; Vorrada et al., 2009; Darweesh, 2015).
The firing temperature of all pottery bodies is several hundred degrees
below the melting point of the filler. SiO2 existed in a variety of crystal
forms which differ in density and thermal expansion characteristics from
one another. The three most important crystal forms of SiO2 are quartz,
cristobalite and tridymite.
Quartz exists in two forms differing slightly in density. Below 573ºC,
α -quartz, the denser form, is the stable one, while above this temperature,
β-quartz is stable. The change between the two occurs rapidly as the
temperature changes through 573ºC and since only a comparatively small
displacement in the relative positions of the SO4 tetrahedra is removed.
This type of change between high and low temperature modifications of
silica is referred to as a displacive change. When quartz is subjected to
temperature above approximately 1000ºC, bonds between atoms are
broken and a new crystal structure results. The mineral cristobalite has
such lower density than quartz. Since the change involves a major
reconstruction of the crystal. Cristobalite also exists in three modifications
(Khalil, 2012; ASTM, 1980; Darweesh, 1992).
 Inversion and conversion of silica
The conversion of SiO2 to cristobalite is slow depending on the degree

of temperature, soaking of firing and the presence of mineralizers the
thermal reactions. Thermal reactions representing (kingery, 1975; ASTM,

1980; Konta, 1981) of the conversion of SiO2 at different temperatures are
as follows,
5730 C
 -Quartz  -Quartz (7)
8700 C
 -Quartz  2 -Tridymite (8)
14100 C
2 -Tridymite   -Cristobalite (9)
220 2800 C
 -Cristobalite  -Cristobalite (10)
1630 C
2 -Tridymite 1 -Tridymite (11)
1170 C
1 -Tridymite  -Tridymite (12)
Flint is composed of extremely small crystals of quartz. The surface of

each minute crystal is available for conversion. Consequently, more
cristobalite is formed in a body, whereas the flint filler is more than in the
same body in which the ground sand is substituted for flint. Therefore, flint
is preferred in the earthenware industry. Since, the contraction which
accompanies the change as the cristobalite cools is an important factor in
preventing crazing (kingery, 1975; Khalil, 2012; ASTM, 1980; Mostafa
and Darweesh, 1992). They are added and usually used in ceramic tiles or
any ceramic masses as fillers. It is a common accessory constituent in clays
used for the production of tiles (Todor, 1972). Quartz relics act as a source
of flaws in ceramic products. To get over it, the grain size of the quartz
should be greatly reduced. In this case, it participates readily in the liquid
or glassy phase formation.
2.6.4. Fluxes
A flux is a material which reduces the fusion temperature of the
materials or mixtures to which it is added. It is not strictly possible to
divide material to fluxes and refractory materials, since whether or not a
material acts as a flux depends not only on the material itself, but also on
the material to which it is added. It is good mention that as the alkali
content increases in a fluxing material, its fluxing action increases. The
F2O3 content of the fluxing material must be in its lowest content, or it will
impart its color to the fired ceramic body. Also, the grain size of the fluxes
is of great and vital importance (kingery, 1975; Rayan and Radford, 1987;
ASTM, 1980; Mostafa and Darweesh, 1992). Fluxes melt when the ware is
fired. On cooling, the glass solidifies and provides the bond that holds the
whole mass together.That is responsible for the strength of the pottery
ware. Cornish stone, feldspar and nepheline syenite are fluxes (kingery,
1975; Grim, 1962; Schuller and Jager, 1979).
2.6.4.1. Feldspars
It is essentially a mixture of two feldspars namely, orthoclase-potash
feldspar, KAlSi3O8 or K2O. Al2O3. 6SiO2 and albite-soda feldspar,
NaAlSi3O8 or Na2O. Al2O3. 6SiO2. Small amounts of mica and quartz are
usually present. The feldspar is often extracted by a normal quarrying or
mining with visual control over the extracted material at the quarry face.
Briefly, the method starts by crushing the rock to less than one inch by
means of jaw and gyratory crushers which followed by wet milling to pass
a 30 mesh sieve. The three stage froth flotation process which follows, first
removes the mica, garnet and other similar materials. Then, the feldspar is
frothed from the quartz. Finally, the separated feldspar undergoes a further
flotation to reduce the quartz as much as possible. The concentrated
feldspar and quartz which are separated by this method are then dried and
stored. Though feldspar contains about 5% quartz, quartz contains 6-8%
feldspar (kingery, 1975; Mostafa and Darweesh, 1992; Schuller and Jager,
1979).
Table 3. Comparative analyses of some commercial fluxes, wt. %
Materials Oxides Feldspar Cornish stone Neph. Syen. 1 Neph. Syen. 2

SiO2 67-70 72-74 61.0 56.0
Al2O3 10-18 14-16 32.0 25.0
Fe2O3 0.1-0.3 0.1-0.3 0.07 0.08
CaO 0.3-0.5 1.5-1.9 0.6 0.75
MgO 0.03-0.15 0.05-0.20 0.10 ----
K2O 9.0-11.0 4.0-4.5 4.5 9.2
Na2O 2.5-3.3 3.0-4.0 10.0 8.1
TiO2 0.02-0.07 0.05-0.15 ---- 0.1
LOI 04-0.7 1.0-2.0 0.6 0.75
The temperature at which fluxes start to melt is largely depending on

the total alkali content and the particle size of the material. Mica,
particularly in the extremely finely divided form in which it occurs in the
clays, is one of the first of glass-formers to melt. The outer fluxes all
resemble one another at the lower temperature, e.g., pressed discs of
Cornish stone, feldspar and nepheline syenite when fired up to 1100ºC
have very much the same appearance-hard and with sharp well defined
edges. As the temperature is raised, the expected variation becomes
apparent with the nepheline syenite showing the greatest tendency to flow.
So, increasing the amount of flux in a body will not necessarily reduce the
fired porosity, particularly at relatively low temperatures (Grim, 1962;
Schuller and Jager, 1979).
Schuller and Jager, 1979 showed that the viscosity of feldspar melts
increased with the rise in K2O/Na2O ratio. Schuller, 1964found that soda
feldspar seems to increase the reactivity of the melt and promotes the
attack on the aggregates of primary mullite (3A2O3.2SiO2 or A3S2) at a
higher temperature. Table 3 illustrates the chemical analysis of some
commercial fluxes.
2.6.4.2. Nepheline Syenite

It contains a high alkali ratio (9-10%) and therefore, it is considered as
a powerful flux. So, it can be used in a body composition of glazes and
porcelain enamels have increased considerably over the last 20 years. It is
claimed that the substitution of nepheline syenite instead of feldspar in the

ceramic body reduces the firing temperatures resulting in saving time and
fuel. Also, it increases the firing ranges. There are two main sources for
nepheline syenite in Canada known as “Lakefield variety” and in Norway
known as “North Cape variety.” The second variety has a total alkali
content and K2O/Na2O ratio greater than those of the first. Therefore, the
Norway variety has a higher fluxing action than the Canadian (Jackson,
1969; kingery, 1975; Rayan and Radford, 1987, Drews, 1983; Rayan,
1978).
2.6.4.3. Cornish Stone

It is clear from Table 3 that Cornish stone is not a powerful flux as
either feldspar or nepheline syenite, where its total content is considerably
less, but it was however very widely used as a flux. In this country is being
the only native deposit of fluxing material commercially viable. Therefore,
its use has declined recently, and feldspars and nepheline syenite are being
imported to replace it. In fact, the Cornish stone is effectively a mixture of
kaolin and feldspar. Generally, it is graded on the basis of its color. There
are three Cornish stone varieties known as “Hard purple,” “Mild purple
“and “Dry white.” All varieties contain feldspar, quartz, clay minerals,
mica, fluorspar (CaF2), and other minor impurities. The purple varieties are
the richest in feldspar, while the white variety is the poorest. Hence, the
hard purple is the strongest flux and the dry white is the weakest. The color
is mainly due to the impurity content exist in the fluorspar of the stone.
During firing, as the feldspar and fluxing ability increases, the fluorspar
content increases too. This often accompanied with the emission of
fluorine. This is usually detrimental to both environment and human health
(Jackson, 1969; kingery et al., 1975; Rayan and Radford, 1987; Drews,
1983; Rayan, 1978; Darweesh, 1992).
2.6.4.4. Talc
It is a magnesium silicate and a member of montmorillonite group
having the formula of 3MgO. 4SiO2. H2O. It is a very cheap source of
magnesia which is acting as a flux. The presence of Talc in bodies that
fired at high temperatures results in the formation of cordierite. This

imparts low thermal expansion and therefore good thermal shock
resistance. This makes it to be more suitable for the production of table
wares. The only disadvantage is the very narrow range of firing which
makes it easily distort on firing. Furthermore, according to its lower
expansion, they become more difficult to match with the glaze of a suitable
expansion or decorate to an acceptable standard. It is believed that lime-
bearing talc in particular gives bodies of low shrinkage the high strength
and good crazing resistance. They are successfully used in the manufacture
of both table wares and wall tiles (Jackson, 1969; kingery, 1975; Rayan
and Radford, 1987; Drews, 1983; Rayan, 1978; ASTM, 1980).
2.6.5. Clyas
Clay minerals are often formed from the decomposition of igneous
rocks as granite which is composed mainly of roughly equal proportions of
potash mica (K2O. 3 Al2O3. 6SiO2. 2H2O), quartz (SiO2) and potash
feldspar (K2O. Al2O3. 6SiO2). Kaolinitic clay was the decomposition or
kaolinization of feldspar in the presence of air and water (Jackson, 1969;
kingery, 1975; Singer and Singer, 1971; ASTM, 1980; Mostafa and
Darweesh, 1992) as follows:
K2O.3 Al2O3 .6SiO2  H 2O  Al2O3 .2SiO2 .H 2O  K 2O  4SiO2 (13)
Kaolinite is a crystalline material in which the crystals are being flat

and extremely small hexagonal in shape. The crystal size may vary from 5
µ to 1 µ 0r 10-4 meters, which in turn responsible for their extreme
properties (Jackson, 1969; kingery, 1975; Rayan, 191987; Drews, 1983;
Rayan, 1978). Clays provide the plasticity which simplifies the
manufacture of clay wares and reduces handling losses. When clay is
formed from the parent rock, it may have been deposited at its origin site.
Hence, it is known as “Primary or residual clay,” but if it is transported by
means of water or winds or any other means and deposited into a distance
away is known as “Secondary or sedimentary clay” (ASTM, 1980;

Mostafa and Darweesh, 1992).
At 100-200ºC, the volume of certain argillaceous minerals shrinks as a
result of the water loss which causes a dimensional change. At 200-300ºC,
the oxidation of certain organic materials present in argillaceous minerals
begins. The degree of oxidation depends on the nature of the organic
materials, the amount of oxygen available in the furnace and the readiness
with which this can penetrate into the mass of the argillaceous material in
order to promote the oxidation. In general, the degree of oxidation
increases with rising temperatures (ASTM- Specification, 1980; Mostafa
and Darweesh, 1992).
The main clay mineral for potters is kaolinite (Al2 Si2 O5 (OH)4 or
Al2O3. 2SiO2. 2H2O or AS2H2), which is seen in the SEM to crystallize into
minute hexagonal plates. Kaolinite is a decomposition product of feldspar
450-500ºC to form metakaolin (Al2O3. 2SiO2 or AS2) and water vapor
(Khalil, 2012; Mostafa and Darweesh, 1992).
450500 C
Al2O3 .2SiO2 .H 2O   Al2O3 .2SiO2  H 2O
0
(14)
At this moment, the ceramic units suffered from a slight expansion,

where it becomes enlarged than before. At 980ºC, some heat was released
followed by the total destruction of the whole structure to form mullite
(3Al2O3. 2SiO2 or A3S2) and quartz minerals.
Al2O3 .2SiO2   3 Al2O3 .2SiO2  SiO2

0
980 C
(15)
2.6.6. Limestone
It is an important ingredient in wall and floor tile compositions due to
its capacity to redduce the moisture expansion of the resulting ceramic
units. This is a very critical charater specially in floor tiles (Kingry, 1975,
Mostafa and Darweesh, 1992). Limestone is often used at a range of 10-12
wt. % addition in tile products to lower the quantity of melting component
during firing.This is the feature of the body which is prone to moisture

expansion which is more complecated. Accordingly, it cannot be
considered as a filler or flux. Inside the kiln, limestome starts to break
down into CaO and CO2↑ as shown in the following equation:
700900 C
CaCO3   CaO  CO2
0
(16)
The formed CaO tends to react with the surrounding ingredients to

form anorthite (CaAl2Si2O8) instead of going to the melting minerals. This
reduces in the melting condition and then prevents the firing shrinkage.
Limestone or ground chalk or CaCO3 is commonly known as “Whiting”
could also be used as an important glaze material (Kashlyak and Kareev,
1978; Kisel, 1980). Tiles (Rayan and Radford, 1987; Danto, 1979) are
generally classified according to their degree of water absorption (W.A.)
into three main groups:
i. Low water absorption (W. A. < 3%).

ii. Medium water absorption (W. A. = 3-6%).
iii. High water absorption (W. A. > 10%).
 Floor tiles are belonging to the 1st and 2nd groups with W. A. only
up to 6% (Rayan and Radford, 1987; Drews, 1983; Rayan, 1978).
There are two different types of floor tiles:
a) Floor Tiles
They are dense bodies, fully vitrified, colored frequently by the color
of the natural clays used in the batch or by colors, and can be obtained in
various colors as well as different textures. The extenal surfaces of floor
tiles are fine and soft. In some cases, surfaces are self-glazed. These are
characterized by their high resistance to abrasion and weather stains.
b) Quarry Tiles
The quarry tiles are thicker and extremaly bigger if compared with
floor tiles and are having a rough surface appearance with W. A. between
2-5%. These tiles are used in floor of public buildings, silos and wall
linings.
c) Decorative Floor Tiles

These are ready dense unglazed tiles. There is a wide range of bright
colors in these colored types. The floor tiles production all over the world
is based on the available local raw materials and it is usually established
near the sites of these raw materials. Since, it is considered that the
transportation of raw materials over large distances is uneconomic. In
addition, the simple shape of the floor tiles and the conventient size of each
tile in the production, make it possible to mechanization and automization
of the industry (kingery, 1975; Kashlyak and Kareev, 1978). Many
attempts were made in the world, each to use local raw materials in the
production of floor tiles (Danto, 1979; Vauequelinet et al., 1980; Fiori and
Fabbri, 1980; Darweesh and El-Din, 2000). Accordingly, it was found that
tiles with water absorption of 7-9% can be prepared by replacing the
expensive refractory clay type with local clay mixtures (Danto, 1979). On
the other hand, the use of granite or kaolinized granite for the manufacture
of red tiles by rapid firing was done giving a firing temperature 40ºC lower
than the best comparable clay body one (Schuller, 1964; Kashlyak and
Kareev, 1978). In some compositions, the firing temperature was found to
reach 1000ºC (Bacanac and Babic, 1984; Belous, et al. 1987).
Belous et al., 1989 [57] studied the influence of chemical and structure
composition on the characteristics of floor tiles. They found that the
presence of ZrO2 had no a significant effect, but the bending strength was
increased by increasing the Al2O3 content and was lowered by more SiO2
content. The compressive strength reached a maximum value 3100 MPa in
presence of 95% Al2O3. Ceramics, in general, are distinguished by their
high mechanical strength that enables them more convenient for many
applications. So, Watchman, 1969 had classified ceramics into different
categories as follows:
1) Single crystal types.

2) Glass fibers and film types.
3) Polycrystalline aggregates in a glassy matrix types.
4) Polycrystalline glass-free ceramic types.
5) Devitrified ceramic types.
6) Composites types.
Traditional whiteware ceramics specially floor tiles fall into the

category of polycrystalline aggregate in a glassy matrix, e.g., multiphase
body with phases having too different coefficients of thermal expansion
which may tend to develop some defects. So, for bodies made of
polycrystalline aggregate in a glassy matrix, the physicomechanical
properties, i.e., the ceramic parameters, are highly affected by stresses
arised in the melting condition and corroded relics of quartz grains and
mullite crystals (Warshow and Seider, 1967). Quartz, which is one of the
ingradients in the floor tile composition, is characterized by a relatively
higher thermal expansion than other components, that reaches a value of
26.2 x 10-6 between room temperature and 1000ºC (Genin, 1958). So,
quartz controls the total thermal expansion of tile units. Therefore, the
overall cracking is resulting from expansion changes owing to phase
conversions of SiO2 in the form of quartz. In addition, the difference in
expansion between the glassy phase and SiO2 was proved in the work done
by Genin, 1958. Kalnin, 1967 and Kalnin et al., 1968. They found that both
young's and their modulus as well as the strength of white wares were
increased with the total amount of mullite which was also claimed by
Mortel, 1977 in case of the fast fired bodies and by Giamlem and Lyng,
1974 with the addition of anorthothite as a flux to rise the bending strength
of the produced articles. Schamberg et al., 1984 stated that the bending
strength of the tiles can be increased with more than 20 N/mm2 by
preventing the formation of gehlenite compound by the addition of about
8% barium compound in the tile composition and firing at 800-1200ºC.
On another side, strengthening of glassy phase of tile bulk was by

rising its Al2O3 and lessening its K2O contents (El-Alfi et al, 2004;
Darweesh et al., 2011; Darweesh and M. G. El-Meligy, 2014; Aggarwal, et
al. 1980), whereas the MgO, K2O and Na2O are reasons that controlling the
maturing temperature, liquid phase and crystallization which can develop
the different properties.
In recent trends, to save raw materials and to reach good mechanical as
well as ceramic properties of floor tiles and moreover to use both industrial
byproducts and wastes, the industrial waste materials containing oxides
suitable for tile composition were introduced in the tile mix composition.
The most important waste materials among this field of application are
both FA and cement kiln dust. Aggarwal et al., 1980 found that FA from
power stations and aluminium industry waste can be used in the
manufacture of wall tile up to 42 wt.%. The bending strength of ceramic
tiles was increased and their water absorption decreased by the addition of
fly ashes and slags from coal-fired power plants. The sintering effect of
slags was due to the increased amount of FeO and that of ashes to the
amount of alkalies in the composition (Khrundzhe and Babushkin, 1983;
Ibrahim et al., 1980; Pinalova et al., 1984).
Pinalova et al., 1984 used the tailing from a coal benification for the
manufacturing of ceramic materials. A typical waste 16-20 mm diameter
contained 7-16% moisture and 75-85 wt. % ash was examined. The waste
was added to the ceramic batch and ground to a particle size 0.5 mm,
mixed with water and plasticizers, and then pressed or extruded at 20-30
kg/cm2. The products were dried, fired at 800-1110ºC, and then optionally
glazed at 980ºC to give tiles and facing materials. On the way, to increase
the capacity of tile firing furnaces and decrease the energy consumption in
tiles manufacture, silica flour was used in the ceramic batch composition to
produce tiles glazed and fired by glost firing at 1050-1250ºC for 20-60
minutes only (Pitskhelauri et al., 1988). Also, tiles had 1.0-1.5% shrinkage
and 13.8-14.5 MPa bending strength were produces at the Slavyanshoe
plant (ASTM, 1980) by using the waste slurry obtained during the
production of WG. This waste has a high SiO2 content (66.67%) and
alkalies (Na2O, 2.5-3.0%) but a low content of Al2O3 (0.5-0.6%). Also, to
decrease volume and weight with retained strength and decrease the firing
temperature of the tiles (ASTM, 1980), the raw mixture of tiles contains
wastes from asbestos cement production was used.
2.7. Methods of Investigation
2.7.1. Densification Parameters

White ware products are known as “Vitreous” in which water
absorption is less than 1% or “Porous” in which water absorption is more
than 1% and may be reached to 18% as in wall tiles. Open pores are voids
which allow the ingress of a penetrating fluid from the surface of the
article. The voids are created due to the migration of gases during the
drying and firing. Sealed pores are often formed during firing when
bubbles of gases are frozen into the glassy matrix or when open pores are
sealed by the molten material. Some clays and bodies are bloated when
over fired due to the melting action of the fluxes together with the
evolution of gases from such impurities as calcium sulphate, CaSO4
(kingery, 1975; Rayan and Radford, 1987; Drews, 1984; Rayan, 1978;
ASTM, 1980; Mostafa and H. H. M. Darweesh, 1992).
The density of a ceramic material or as “apparent volume,” since the
terms are also appropriate to porosity and water absorption evaluations.
The density of a material has been defined as the relationship between its
mass (weight) and its volume (Rayan and Radford, 1987; Rayan, 1983;
Rayan, 1978) or,
Mass
Density 
Volume (17)
For a vitreous object, there is only one weight and one volume
involved, but for a porous solid, there are three common volume
expressions as follows:
1) Apparent volume: This is the envelope volume of the porous solid

which is usually called “bulk volume.” The material often
comprises both open and sealed pores. It can be determined by: A
physical method, a mercury displacement method, the difference
between the soaked (S) and immersed (I) weights, since the water
is used, the numerical value (S - I) g, gives the apparent volume
(Rayan, 1978; Khalil, 2012; Njindam, 2018).
2) True volume: This refers only to the volume of the solid
component. It can be determined by the “density bottle method,”
where the material is crushed to be in the powder form and then all
pores are disappeared.
3) Apparent solid volume: This lies between the two previous
volumes, where it can be determined by the difference between the
dry weight (D) and the immersed weight (I) of the material. At all,
S - I, S - D and D – I are the volume of the open pores + sealed
pores + solid. and therefore by subtraction, respectively (kingery,
1975; Rayan, 1978; Ke et al., 2016; Sonjida et al., 2011; Vorrada
et al., 2009; Kimm, 2014; Kimm, 2013; Nissan, 2014). There are
three expressions of density corresponding to the three previous
volumes discussed above as follows,
Weight D
Appearent (bulk ) Density   (18)
Appearent volume S  I
Weight
True Density 
True volume (19)
Weight D
Appearent solid Density   (20)
Appearent solid volume D  I
where, D, S and I are the dry, soaked and immersed weights of the
material, respectively.
This property is measured by comparing the volume of the pores to the

weight of the material itself. There two widely known expressions for
porosity are “Apparent porosity or AP” and “Water absorption or WA.”
The AP is a ratio of open pore volume to the total volume, whereas the
WA is a ratio of the open pore volume to the weight of the test material.
The 1 cm diameter/1 cm thickness disc-shaped specimens were prepared
under 20-22 KN loading using a suitable piston. At first, the prepared
specimens are left to dry on air for 24 hours and then at 105ºC for another
24 hours. The dried specimens are then fired at different firing
temperatures using soaking time of 1-2 hours. The used furnace must leave
to cool slowly over night. The fired ceramic bodies after cooling are then
subjected to densification parameters (ASTM, 1980; Darweesh et al.,
2012), in terms of water absorption, %; bulk density, g/cm3 and apparent
porosity, % which could be calculated from the following relationships,
W1  W2
W . A, %  100 (21)
W3
W3
B.D, g / cm3  (22)
W1  W2
W1  W3
A.P, %  100 (23)
W1  W2
where, W1, W2 and W3 are the saturated weight in air, suspended weight
and the dry weight, respectively. Darweesh et al., 2012 studied and
discussed W.A, B.D and A.P of some ceramic batches containing various
contents of CKD waste up to 25 wt. % (Figures 17-19). Table 4 shows the
chemical analysis of the starting raw materials used in this work. They
reported that the W.A and A.P decreased with firing temperature up to
1100ºC, but only decreased with CKD content up to 10 wt. %, and then
decreased with more addition of CKD content as shown in Figures 17 and
18, respectively, while the bulk density increased as shown in Figure 19.
Table 4. The chemical composition of raw materials, wt. %
Materials T-Clay Feldspar Sand Limestone Homra (H) Leached

Oxides (TC) (F) (S) (L) C. Dust
(LCD)
L.O.I 9.72 0.67 1.78 42.63 - 24.51
SiO2 53.47 75.37 93.63 0.08 58.22 12.84
Al2O3 26.78 13.62 3.64 0.03 28.25 1.86
Fe2O3 3.99 0.41 0.08 0.04 8.16 1.53
CaO 0.60 0.53 0.18 56.84 0.79 52.51
MgO 1.38 --- --- 0.10 0.46 1.84
MnO 0.03 0.03 0.02 --- --- ---
K2O 1.18 5.84 0.14 0.05 1.46 1.65
Na2O 1.15 3.44 0.17 0.12 1.32 0.83
TiO2 1.12 0.05 0.16 0.01 1.34 ---
SO3 --- 0.02 0.14 0.02 --- 2.43
P2O5 0.51 --- --- --- --- ---
Cl- --- 0.02 0.06 0.08 --- ---
Figure 17. Water absorption of ceramic products with different CKD contents up to 25
wt. %.
Figure 18. Apparent porosity of ceramic products with various LCD contents up to 25
wt. %.
Figure 19. Bulk density of ceramic products with various LCD contents up to 25 wt.%.
2.7.2. Mechanical Properties

Rod-shaped specimens of the dimensions 1 x 1 x 7 cm3 are prepared
using water as a binder under 20-22 KN loading by a suitable piston. At
first, the prepared specimens are left to dry on air for 24 hours, and then at
105ºC for another 24 hours. The dried specimens are then fired at different
firing temperatures using soaking time of 1-2 hours. The used furnace must
leave to cool slowly over night. The fired rod-shaped specimens after
cooling are then subjected to flexural or bending strength measurements
(Khalil, 2012; Mostafa and Darweesh, 1992) by three point adjustments
system (Figure 20), where, S is the span (cm), W and T are width and
thickness of the sample (cm), respectively. The beam load was applied
perpendicular to the axis of the sample. The F.S. or B.S.
(Mukhamedzhanovet al., 1990) could be calculated from the following
relation,
3PL
F .S , MPa  (24)
2bd 10.2
where, δ: flexural strength, MPa, P: the load of rupture, kg, L: span or the
distance between the two lower beams (5 cm), b: width of sample, cm and
d: thickness of sample, cm.
Figure 20. Schematic diagram of the bending strength, B: beam, S: span, W: width and
T: thickness.
Figure 21. Flexural or bending strength of ceramic products with different contents of
CKD waste, %.
Darweesh et al., 1992 studied B.S. or F.S. of some ceramic batches

incorporating different contents of CKD waste (Figure 21). They found
that the B.S. improved and enhanced with firing temperature up to 1100ºC,
but only increased up to 10 wt. % CKD content and then decreased.
2.8. Body Glaze
Generally, the glaze is a glass layer covered the ceramic body. It is a

vitreous material which is a super-cooled liquid below the point at which it
might have crystallized. It is a liquid of very high viscosity due to the
following facts:
1) A glass rod ultimately deformed and bends when suspended

between two supports.
2) On heating, the glass softens and liquifies over a temperature range
of 1000-1150ºC, but it does not recorded a limited melting point.
Figure 22. Structure of crystalline silicate and silicate glass.
Normal solids are crystalline and have a definite arrangement of atoms

and molecules. This can be shown by X-ray diffraction patterns (XRD).
Glazes however, have no a definite structure, the atoms are oriented in a
random network. This may be appreciated by comparing the structure of
crystalline silicate minerals with glazes (Rayan and C. Radford, 1987;
Mukhamedzhanov et al., 1991; Keijiro et al., 1991; Moroz et al., 1991).The
unit of structure is the silicon tetrahedron. This shape is dictated by the size
of the atoms, on a comparative scale oxygen = 1.32 and silicon = 0.39. An
example of the structure of a crystalline silicate and a silicate glass is
shown in Figure 22, which clearly indicates the difference in their
structures.
They have a random three-dimentional network, but no units repeat
itselves at fixed and regular intervals. On heating, the glaze always softens
over a given temperature range due to the varying amounts of energy
required to detach different parts of the network, which are not structurally
equivalent. The interstices within the network may be filled with atoms of
other elements which modify the physical properties as color, brilliance
and so on. Glazes could be classified to:
a) Lead and Leadless Glazes

i. Lead glazes have a brilliant appearance and can be used to glost
firing temperature of 1150ºC above which the lead tends to
volatilize.
ii. Leadless glazes have been developed and improved. Now, they
have a reasonable brilliancy and firing range. However, glazes that
mature below 1000ºC often have poor craze resistance. This is
mainly due to the fact that high proportions of soda and potash
required to give the glaze a low melting point, and also confer a
relatively high expansion, but the glaze must have a lower thermal
expansion compared to the ceramic body.
b) Fritted and Raw Glazes

A fritted glaze often has one or more material used in its formulation,
which have been subjected to the fritting operation, i.e., the heating
together of components to form a glass which is subsequently ground to a
given particle size. The main reasons for fritting are:
i. When soluble compounds fritted with other materials in the right

ratios, it become insoluble. This means that glaze components
should be insoluble, otherwise they tend to immigrate into the
pores of glass wares. When the glaze fired, it may then have a
starved appearance. Even, if a vitreous ware is used with a glaze of
slightly soluble components, segregation will take place as the
drying process progresses.
ii. Though lead compounds are available to ceramic industry, they are
generally insoluble in water, but they may have a considerable
solubility in dilute acids. For an example, if the compounds are
soluble in dilute HCl acid, they become toxic, where the gastric
juices are highly acidic.
iii. If glazes contain high ratios of clay materials, it may crack on
drying. The chemically bound water content tends to release
during the glost firing. This may lead to a poor glost finishing.
iv. If any compound evolved gases on heating like limestone (CaCO3)

used in a glaze with a low maturing temperature, is fritted so that
the used glaze is not likely to suffer from pin-holing and bubbling.
For a common practice, it often used two frits, i.e., Lead and Borax
frits or its oxides (PbO and B2O3). When these oxides are fritted together,
the lead does not generally reach satisfactory limits of insolubility.
However, lead borosilicate can be produced satisfactory if the PbO:B2O3
ratio is carefully selected (Rayan and Radford, 1987; Mostafa and
Darweesh, 1992; Mukhamedzhanov et al., 1991).
The glaze could be classified also due to the type of ware on which the
glazes applied as for Earthern wares, Sanitarywares and Porcelain glazes.
There is a wide range of glazes available within each group. This
classification has been broadened to include firing and maturing
temperatures noticing that in the following examples, glaze molecular
formulae have been given, where the oxides have been categorized as
basic, amphoteric, acidic. The sum of the basic oxides is one, where the
temperatures are referred to the maturing range (Jackson, 1969; Kingery,
1975; Rayan and Radford, 1987; Njindam et al., 2018; ASTM, 1980).
1- Majolica (900-1050ºC)
[0.7 PbO +0.3 CaO] +0.15 Al2O3 + [2.0 SiO2 + 0.3 B2O3]
2- Earthenware (1000-1150 C-Lead glaze
[0.4 PbO+0.3 CaO+0.3(Na,K)2O] 0.25 Al2O3 [2.5 SiO2+0.5 B2O3]
3- Leadless glaze
[0.55 CaO 0.3 (Na,K)2O] 0.30 Al2O3 [3.0 SiO2 + 0.8 B2O3]
4- Sanitaryware (1200-1250ºC)
[0.6 CaO+0.2 (Na,K)2O+0.2 ZnO] 0.35 Al2O3 [3.0 SiO2]
5- Hard paste porcelain (1400ºC)
[0.68 CaO+0.20 MgO+0.12 ZnO] 1.0 Al2O3 [10.0 SiO2]
There is a wide range of commercially coloring agents or stains which

are used extensively in the ceramic industry. White wares may be
decorated using coloring oxides in different ways as follows:
- Colored bodies. - Colored slips (engobes).

- On glaze.- Under-glaze. - In glaze.
Colored bodies and slips have a few percent of the desired color added
to the bulk mix during preparation in the slip house. The major
disadvantage of using colored bodies is that shading of wares may take
place. A ceramic color seldom consists simply of a coloring oxide or
compound. This fact is understood by considering the following cobalt
blue color known as “Royal blue” or “Maz blue” (Jackson, 1969; Rayan,
1978; ASTM, 1980).
Cobalt oxide 45%, Whiting 10%, Flint 18%, Alumina 5%, Feldspar
22%. The color may be analyzed in terms of the function of its
components. Cobalt is the basic coloring compound giving the intense blue
color. CaCO3 acts as a modifying agent that modified the blue color to give
new blue tint and stability. Flint and Al2O3 act as a diluents, i.e., as filler
giving a slightly faint but more stable color. Feldspar which uses as a flux,
is always assisting in the sintering of color mixes through the calcination
process noticing that the underglazed colors need about 5% flux, but on
glaze colors need about 70%. Ceramic colors are usually prepared to fix
and carefully control specifications. Even so, it is difficult to produce
decorative wares of consistent quality on a commercial scale. It must
always remember that the final color depends mainly on the composition
of the glaze, firing temperature and kiln type. These are very important
particularly in sanitary ware industry, where the finished product can be
sold to an exact standard color, while the vitreous sanitary ware is
generally colored with opaque pastel glazes which have to match other
colored fittings, in the cloak or path rooms manufactured from quite
different materials. The metamerism phenomena must be taken into
consideration which is two color samples may appear to match, under a
given illumination, even though their spectrophotometric curves are not the
same (Figure 23). So, the sanitaryware colored glazes are formulated from
a color and compatible base glaze to an exact specification. If a different
illuminant is used, the samples may no longer match where the perceived
color depends essentially on the illuminant and the reflection curve at the
surface (Drews, 1983; Danto, 1979, Vauequelinet et al., 1980; Fiori and
Fabbri, 1980; Darweesh and El-Din, 2000). The color specification by CIE
system gives the chromaticity values or coordinates x, y and Y where these
values may be used in conjunction with the CIE chromaticity diagram
(Figure 24).
Figure 23. The metamerism curves of ceramic products.
Figure 24. The CIE chromaticity diagram.

2.9. Thermal Expansion
The difference between the contraction of the ceramic body and the
glaze during cooling often causes some faults. The thermal behavior or
expansion of ceramic products during firing can be measured by using
Ortom Automatic Dilatometer (Figure 25). In this type of instruments, the
temperature can be controlled in the range of 0.2-10ºC/min. and the rate of
heating is 3-5ºC/min. (Darweesh, 2016) investigated the thermal behavior
or expansion of some ceramic batches as 0, 5, 10, 15, 20 and 25% cement
kiln dust waste (CKD) using Orton Automatic Dilatometer to throw light
on the behavior of the samples on heating.
Figure 25. Ortom Automatic Dilatometer for thermal expansion.
Table 5 and Figure 26 illustrate the coefficient of linear thermal

expansion versus CKD content and its effect on the thermal expansion of
ceramic units incorporating it. The coefficient of linear thermal expansion
decreased with the increase of CKD content up to 25 wt. %.The equipment
(Figure 27) in its fully automated form is a direct reading dilatometer
which plots the percentage of expansion against temperature. Specimens
between from 55 to 80 mm can be accommodated. The silica correction is
added automatically and after the initial setting up the instrument requires
no supervision. The Dilatometer is also available in a hand-operated form
and the following test procedures relate to this less sophisticated
instrument. The test piece with parallel ends is preheated to 900ºC, cooled
and then pushed along the silica tube using thrust rod till it contacts the
silica stopping disc. The dial gauge is firmly clamped into position and the
needle adjusted so that a slight pressure is exerted on the thrust rod. The
outer ring on the dial gauge is moved to zero reading. The thrust rod
should move quite freely within the silica tube. The base of the apparatus
should be gently tapped till the dial gauge gives a consistent zero reading.
Figure 28 shows the typical thermal expansion of earthenware body and
low sol glaze. At 500ºC, the thermal expansion is 0.38% and 0.32% for
body and glaze, respectively. The difference of expansion is 0.06%.
Table 5. The coefficient of linear thermal expansion as a

function of CKD content and its effect on the thermal
expansion of ceramic products
CKD, wt. % Coefficient of Linear thermal Expansion, α/K

0 8.731 x 10-6
5 8.492 x 10-6
10 8.049 x 10-6
15 7.145 x 10-6
20 6.382 x 10-6
25 6.120 x10-6
Figure 26. The coefficient of linear thermal expansion as a function of CKD content
and its effect on the thermal expansion of ceramic products.
Figure 27. Thermal expansion apparatus (A: Slate base, B: Silica tube, C: Stopping
disc, D: Fused silica rod, E: Invar extension rod, F: Dial gauge, G: Furnace, H: Test
specimen.
Figure 28. Thermal expansions of earthen ware body and glaze.
2.10. Applications
Some investigators (Mukhamedzhanov et al., 1991; Moroz et al., 1991)

used mining tailings and metallurgical wastes in the production of ceramic
floor tiles by dewatering, grinding, pressing and firing at 1000-1100ºC.
Mining tailings consist of argillaceous shale and sandstone, where the
argillaceous shale serves as a binder, while sandstone as a flux substituted
for pegmatite. Ceramic tiles of a low porosity and good mechanical
properties were obtained from mixtures containing argillaceous shale 70-
76, sandstone 15-20, chamotte 3-5 and bentonite 1-5 wt. % (Keijiro et al.,
1991).
The CKD waste is the main source of air pollution in many countries,
which resemble the present day problem, be a matter of great importance
due to the precipitation of huge amounts of it behind the kilns and its
characteristics such as: light weight and fine grains so that it can be
transported easily by air storms to many agriculture and population areas.
Keijiro et al., 1991 manufactured high strength ceramic tiles from slurry
containing cement and mineral materials, e.g., feldspar, siliceous stone and
SiO2 by moulding the slurry followed by drying and firing. The cement is
selected from Portland and/or white Portland cements. Abdel-Fattah and
Nour, 1981; Abdel-Fattah et al., 1982; Abdel-Fattah and Eh-Didamony,
1981) studied the CKD supplied by the factory. The dust was fired
between 1000-1250ºC, then two calcines (70:30) and (50:50) were
prepared by blending the raw dust and kaolin. It was stated that it consisted
mainly of dolomitic limestone, minor amount of Na2O, K2O, SiO2, 2
C2S.CaCO3 and 2 C2S.CaSO4. The formed phases in the 1st calcined mix,
was found to be mainly of gehlenite (C2AS), but in the 2nd calcined mix, it
was mainly of β-C2S.
Using of CaO sources in ceramics fields and tile production is
recommended as a whiting and a flux (Darweesh, 2001). The cement kiln
dust (CKD) is a mixture of finely divided particles and partially calcined
raw materials with condensed volatile salts (Anlagenbau et al., 1995;
Saltyvskaya, 1985). So, the use of dust along with kaolin in the form of
calcined mix proved also to be used successfully in replacing feldspar in
porcelain bodies. Darweesh, 2010 also studied and utilized the alkaline
wastes obtained from cement kilns to produce porcelaineous bodies.
Darweesh, 2010 exploited the CKD waste from cement factories as a
source of CaO in ceramic industry by 25% to produce wall and floor tiles.
W.A, B.D. and A.P. as well as B.S. and T.F. were researched (Figures 29-
31). The addition of the waste improved and enhanced all properties of the
prepared units. Darweesh, 2001; Darweesh and khalil, 2001 demonstrated
that adding 5 up to 10% waste to the aumina cement improved the specific
characters of the cement hydrated up to 28 days. More than 10% waste,

adversely reflected on these properties (Figures 32-34).
Figure 29. Water absorption, bulk density and apparent porosity of the ceramic bodied
of the base batch and the batches containing CKD waste.
Figure 30. Compressive and bending strengths as well as abrasion resistance of the
base batch and the batches containing CKD waste.
Figure 31. Firing shrinkage and linear thermak expansion of the base batch and the
batches containing CKD waste.
Figure 32. Chemically-combined water content of the various alumina cement pastes
containing cement kiln dust waste hydrated up to 28 days.
Figure 33. Free lime content of the various alumina cement pastes containing cement
kiln dust waste hydrated up to 28 days.
Figure 34. The compressive strength of the various alumina cement pastes containing
cement kiln dust waste hydrated up to 28 days.
3. BIOCERAMICS
3.1. Introduction
There is an advanced branch of ceramics known as “Bioceramics”

including the ceramic products which can be used mainly in the human
bodies as bioactive bone scaffolds. Ca2 (PO4)2 was used in bioceramics

which recently have received a great attention as bone-scafolds substitutes
due to its good biocompatility and osteoconductive properties if compared
to other materials. Hydroxyapatite (HA) has a similar composition to the
major inorganic constituents of the skeletal system of vertebrates. Using of
ceramic units in the human body, i.e., to exchange the broken or expired
bones with a ceramic material that can function the remaining years of the
patient’s life is very important. Moreover, the overwhelming demand for
spare parts of bones for human beings took a particular interest nowadays.
The cronical bone is the solid part of bones. Trabecular bones like
scaffolds or a honey-comb (Figure 35). Spaces among bones are filled with
fluid bone narrow cells which are making the blood and some fat cells.
Figure 36 shows a simple structure of the bone. Bones have different
shapes, sizes and structure tissues. The common bioactive ceramic
materials are Al2O3 and ZrO2 due to its excellent biocompatibility. The
main advantages of Al2O2 are its higher hardness and wear resistance,
while ZrO2 has a higher strength and fracture toughness, in addition to the
lower Young’s modulus (Taha, 2011; Mohamed, 2012; Kokubo, 1991;
Iarry, 1991; Klein,1994; Kong et al., 1999; Susan, 2007).
Figure 35. The outside (cortical) and Inside (trabecular) of a bone.

Figure 36. A simple structure of a bone.
Figure 37. Chemical structure of Gelatin.
Figure 38. Chemical structure of Agarose.

The sol-gel produces porous Al2O3 and ZrO2 toughened alumina (ZTA)
substrates that can be prepared and impregnated with a synthetic body fluid
(SBF) and dicalcium phosphate (DCP) solutions for different periods to
prepare the bioactive composites. The resulting porous ceramic substrates
are then characterized by measuring their physical properties in terms of
B.D and A.P. The phase composition and the microstructure of the
prepared scaffolds or porous ceramic substrates can be determined by
using XRD and SEM techniques (Mohamed, 2012; El-Hady et al., 2015;
Hench and West, 1990).
El-Hady et al., 2015; Hench and West, 1990 used a certain polymer
based on the gelatin from bovine skin (Figure 37) and agarose (Figure 38)
to prepare composite scaffolds using the sol-gel technique. They also used
ciprofloxacin (Figure 39) to avoid bacterial infections (Daniel and Vallet-
Regi, 2010; Balamurugan et al., 2008; Barba-Izquierdo et al., 2000).
Figure 39. Chemical structure of ciprofloxacin.
3.2. SOL-GEL Techniques and Applications
The sol–gel process is often utilized to prepare glasses and ceramics

(Figure 40). The process is based on inorganic polymerization reactions of
metal alkoxide precursors. These precursors undergo hydrolysis and
condensation reactions if dissolved in a solvent to form soluble metal
hydroxides (Mohamed, 2012; Saravanapavanand Hench, 2003; Phulé and
Wood, 2001; Julian et al., 2006; Balamurugan et al., 2006; Newport et al.,
2007; Gupta et al., 2005).
Figure 40. The Sol-gel processing could be used to produce nanostructured layers
and/or coatings as well as nanoporous membranes. 1- Hydrolysis. 2: Condensation. 3:
Gelation, 4: Evaporation and Drying. 5: Nanoporous membranes.
The main characters of a multi-component sol–gel process are a

homogeneous solution that formed before polymerization at lower
temperatures, but also phase separation, crystallization and chemical
decomposition can be eliminated (Saravanapavan and Hench, 2003).
Briefly, the advantages of the sol-gel processes include lower processing
temperatures, high levels of purity, control of concentrations, and the
capacity to synthesize multicomponents in various forms (Phulé and
Wood, 2001). In addition, the sol–gel derived bioactive glasses tend to
have more simple compositions and enhanced bioactivity as well as
resorbability due to the mesoporous texture (2–50 nm) inherent to the sol–
gel process (Julian et al, 2006). Enhanced bioactivity is due to their
residual hydroxyl ions, micro pores, and a large specific surface area
(Balamurugan, et al., 2006; Newport et al., 2007). The dynamics of the
sol–gel processes depend on various physicochemical properties of the
sol–gel composition, water to precursor ratio, the type of catalyst, choice of

precursors, pH value, temperature and solvent (Gupta et al., 2005).
Now, calcium silicate could be analysed by the sol–gel process. It is
composed mainly of SiO2 (50-53%), and CaO (48-50%) by the chemical
reactions between tetraethyl orthosilicate and calcium nitrate tetrahydrate.
Tetraethyl orthosilicate was used as metal alkoxide precursors of the
corresponding oxide SiO2. Metal alkoxides represent the precursors of
choice for gel processing of oxide ceramics and glasses (Phulé and Wood,
2001). Metal salts are in the form of NO3- because they are less thermally
stable if compared to other anions. This is due to that it is easy to
decompose. Acetate salts are highly basic leading to a rapid gelation in
silicate systems (Saravanapavanand Hench, 2003). CaCO3 did not use for
biomedical applications, where CO32- ions incorporated in the hydrated
layer slow down the growing of the apatitic (Siriphannon et al., 2002;
Drouet et al., 2007). Moreover, when used the salt of calcium nitrate
tetrahydrate,a problem of differential hydrolysis and polycondensation of
various alkoxides arised (Phulé and Wood, 2001). Hence, Ca2+ in the form
of calcium nitrate tetrahydrate could be incorporated.This explains the
cause to select calcium nitrate tetrahydrate as a starting material as well as
the percentage of them to prepare wollastonite.
So, HNO3 acid must use to catalyze hydrolysis during sol-gel process.
Where, the use of acids as catalysts allows the hydrolysis of all compounds
outcomes due to the partial charge distribution in hydrolyzed alkoxides
(Saravanapavanand and Hench,2003; Meiszterics and Sinkó,2007;
Crayston, 2003).The choice for using HNO3 was depending on using it by
several investigators (Saravanapavan and Hench,2003; Phulé and Wood,
2001; Balamurugan et al., 2006; Meiszterics and Sinkó, 2007; Alemany et
al., 2005; Lukito et al., 2005; Olmo et al., 2003; Barba et al., 2006; Mila
and Vallet- Regm, 2001; Wei-Hong et al., 2006; Xia and Chang, 2006;
Brinker and Scherer,1990). In the present study the procedure was carried
out also without catalyst and this leads to delay in hydrolysis and
condensation of TEOS and gelling time to produce powder. The powder
was obtained after 7 days instead of 10 hr in case of the use of HNO 3 as a
catalyst. Meiszterics and Sinkó, 2007 reported that gelling time of the
powder be based on the kind of the catalyst and the porosity of the gel
(Saravanapavanand Hench, 2003).On the other hand, Brinker and Scherer,
1990 considered that the gelation is essentially depends on the ratio of
H2O/TEOS. At lower ratios, the gelation time became quicker.
Water is added as an initiator and it can be generated by condensation
reactions, e.g., esters by a condensation of carboxylic acids and alcohols
(Iarry, 1991). In some researches, ethyl alcohol could be used as a solvent,
where hydrolysis takes place by water into a non-aqueous solution as
alcohols, whilist Saravanapavan and Hench, 2003 reported that CaO ratio
is < the theoretical value. This is attributed to the variation in the leaching
of cations during aging and drying stages when the process was carried out
in presence of the pore liquor (ethyl alcohol). In the present study ethyl
alcohol did not used.
After a transparent green sol was obtained, The sol was held at 60°C
(aged) to reach to a viscosity close the gel point, where transparent gel was
formed (in the present procedure two hours at 60°C, the gelation began.
This temperature was in agreement with other research (Saravanapavan
and Hench, 2003; Balamurugan et al., 2006; Crayston, 2003; Lao et al.,
2006; Balamurugan et al., 2005; Liu and Miao2004; Wang et al., 2008;
Lao et al., 2007), and it was chosen according to observation of
temperature at which the solution become completely miscible or in other
word when the solution become completely miscible the temperature
already reach to 60°C. So, we kept the solution at this temperature to
obtain transparent gel system. Hydrolysis and condensation continue
giving an increase in viscosity (Chen et al., 1999; Saravanapavan and
Hench, 2003). Minimizing of gelling time and rising of Ca2+ ions are due
to a lack of ionic charges on sol-gel by the salt (Saravanapavanand Hench,
2003; Wang et al., 2003). Gupta et al., 2005 reported that gelation
increases the viscosity, shrinkage, weight loss, but decreases the pore size
distribution. Scherer, 1999 also reported that after the gelation, the
chemical reactions still proceeded, increased the rigidity and shrinkage.
Gel was dried slowly up to 110°C to remove water and also adsorbed
moisture from pores. Heat will densify the gel by removing of water.Then,
discs transformed from transparent to translucent opaque. The drying
process yields a powder-like product. Drying of the gel was achieved by

the gradual increase of temperature 110°C in order to avoid cracking.
Viitala et al., 2002 reported that drying process is very important to obtain
homogeneous and free crack products. Crayston, 2003 reported that drying
determines the nature of final products.
Scherer, 1999 reported that on drying, a capillary pressure develops a
higher shrinkage and so, the modulus increases and shrinkage stops, where
the slow drying reduces gradients and damage. Saravanapavan and Hench,
2003 noted that at lower Ca2+ ions, the gels were white in colour, whereas
the other compositions ranged from light yellow to yellow due to the
increase of NO3-1 ions in the gels. When wet gels were exposed to the
atmosphere, the discoloration was stronger and a light yellow was
obtained.
After drying, the gel networks were converted to white powder. The
dried gels were calcined at 500, 600 and 800˚C for two hours soaking with
a heating rate of 5°C/min. Objects are to eliminate the organic content, and
to achieve nitrate removal and further densification. Calcination
temperature 800°C gives no weight loss. Therefore, it completes
elimination of nitrate and the calculated amount of wollastonite was
obtained.The holding time was two hours calcination time that provides a
sufficient time to create additional nucleation sites in glass powders.
Wollastonite powder was calcined at 500°C in some researches
(Siriphannon et al., 2002; Viitala et al., 2002; Binnaz and Hazar, 2007).
Other researches calcined wollastonite at 600°C (Alemany et al., 2005;
Hayashi et al., 1999), whereas other researches calcined wollastonite at
700°C (Wang et al., 2008; Padilla et al., 2005; Long et al., 2006), in
addition to some researches (Hayashi et al, 1999; Lin et al, 2005; Longet
al, 2006) that calcined it at 800°C. The holding time used in this study for
calcination was used by some researches (Siriphannon et al., 2002, 120,
126, Viitala et al., 2002, Padilla et al., 2005; Long et al., 2006; Lin et al.,
2005; Longet al., 2006); Haiyan and Chang, 2004).
Saravanapavan and Hench, 2003 noted that there is an initial
endothermic process around 100°C which can be attributed to loss of
residual water and ethanol. This weight loss over 450°C is due to the loss
of organics. The second endothermic peak occurs at 55°C with a weight

loss of 35 % which is attributed to the loss of nitrates. Alemany et al., 2005
found that the TG curve indicated that a weight loss was observed at 0-
150°C and 150-600°C. Then, it became stable to 1350°C. DTA curve
showed that four endothermic peaks at 150, 570, 560 and 1200°C,
respectively.
To achieve NO3- elimination and more densification, Lao et al., 2006
heated the gel powder at 700°C for 24 hours soaking. Liu and Miao, 2004
reported that at 800°C, the bioglass powder still kept the amorphous
crystals, but on further increase of temperature, crystallization occurs.
Padilla et al., 2005 stated that materials fired at 700°C and 800°C, the
XRD showed typical amorphous materials. Meiszterics and Sinkó, 2007
showed that according to the TG curves, heating of Ca silicate gel systems
should be carried out ≥ 600°C.
After calcinations step, powders were ground and sieved to obtain 63
µm particulates. This particle size was chosen by other researches (Olmo et
al., 2003; Hayashi et al., 1999; Long et al., 2006; Haiyan et al, 2004),
where the range of particle size was from 32 to 63 µm. Sieved powder that
pressed into a cylinder with 15 mm in diameter and under different
compaction loads to obtain the optimum load. A pressure of 60 KN was
chosen on the basis of some experiments which will be discussed later. To
obtain a hard unit, cylindrical specimens of CaSiO3 were sintered for 3
hours in an electric furnace at 900-1250°C, where samples were let to cool
to room temperature. Lin et al., 2004 and 2005 confirmed that the increase
of holding time from 1-3 hours increased mechanical properties and
remained constant with further increase. So, 3 hrs were the optimum
condition to achieve the best mechanical strength of CaSiO3 ceramics.
Also, other investigators (Mila and Vallet- Regm, 2001; Xia and
Chang, 2006; Hayashi et al, 1999; Haiyan and Chang, 2004; Meseguer-
Olmo et al, 2008; Lin, 2004; Wu et al, 2007; Kokubo et al, 2003; Wan et
al, 2008; Arstila et al, 2007; Liu, and Ding, 2002) chose 3 hrs as holding
time and this was the basis for our choice, while the sintering temperatures
were chosen on the basis that there is no weight loss in the produced
powder after 800°C. In addition, from the CaO/SiO2 phase diagram, the
phase transition temperature of β-CaSiO3 and α-CaSiO3 is 870°C (from

amorphous to β-CaSiO3) and 1125°C (from β-CaSiO3 to α-CaSiO3;
Hayashi et al., 1999), respectively. Furthermore, the XRD analysis was
done to measure crystalline phases present at these temperature and to be
sure that wollastonite is the major phase without any traces of impurities.
Finally these range of temperature (900-1300°C) was the range which used
by other researches (Balamurugan, 2006; Crayston, 2003; Lukito et al.,
2005; Lin et al., 2005; Long et al., 2006; Haiyan and Chang, 2004;
Meseguer-Olmo et al,, 2008; Lin, 2004; Wu et al., 2007; Kokubo et al.,
2003). Xia and Chang, 2006 stated that the DSC showed a sharp
exothermic peak at 832°C, which is belongs to β-CaSiO3, and a broad
exothermic peak at 1142°C, which is related to α-CaSiO3.
Saravanapavan and Hench, 2003 reported that gel-glasses produced are
confirmed to be amorphous even after stabilization at 600°C. Arstila et al.,
2007 stated that wollastonite type glasses crystallized at 900°C. Liuand
Ding, 2002 stated that the sintering did not < 1300°C due to CaO–SiO2
binary phase diagram, the tridymite and metastable phases of quartz are
easy to precipitate from the wollastonite that melts at 1436°C.
The lower sintering temperature in sol-gel process was explained by
Phulé and Wood, 2001. The conversion of a sol into a crystalline ceramic
or glass can be often accomplished at lower temperatures than those in
traditional units. Crayston, 2003 attributed the Lower temperature
synthesis to the homogeneity and the smaller particle size, nucleation and
growing of crystalline phases of dried powder/gel units that can occur at
lower temperatures. While Saravanapavanand Hench, 2003 suggested that
the excess free energy of a gel to a glass of the same composition, a gel
might be changed to glass at lower temperatures, i.e., below the liquid
phase temperature of the same composition.
Bioactive glasses and glass-ceramics represent wollastonite in the
study which suggested some biomaterials for a bone tissue regeneration.
The use of Al2O3, ZrO2 and TiO2 could widen their application fields
which represented a bioinert material with the bioactivity of wollastonite as
implant and dental implants. Materials which used as additives to improve
mechanical properties and used it in load-bearing bone substitution were
chosen due to its high wear resistance, their excellent scratch resistance,
good frictional properties and fracture toughness (Nalwa, 2000; Tseng and
Nalwa, 2009; Ashby, 2009).
The high stability of oxide ceramics as low corrosion and low ion
release is due to the high heat of formation of their molecules, in addition
to its biocompatibility with the physiological environment, that make these
materials reliable for a variety of applications as biomedical applications
including knee-prostheses and dental implants. Hence, ceramic oxides of
Al2O3, TiO2 and ZrO2 have been introduced as reinforcing agents (Kokubo
et al., 2003; Wan et al., 2008; Arstila et al., 2007; Liu, and Ding, 2002;
Wei-hong et al., 2006).
Sol-gel helps the formation of nano-particles, nano-films, and nano-
porous membranes (Figure 40). A precursor solution in a solvent, e.g.,
alkoxides and cations can form a colloidal suspension (SOL-GEL) due to
some polymerization reactions, which by adding a surfactant the dispersed
particles could be kept in a suspension used to extract the particles for
more processings to be a substrate. Solvent evaporation creates dense or
nano-porous films. Sol-gel techniques can develop many new materials as
paints, nano and/or bioceramics (Wang et al., 2008; Lao et al., 2007;
Binnaz and Hazar, 2007; Padilla et al, 2005; Long et al., 2006; Long et al.,
2006; Meseguer-Olmo et al., 2008; Wu et al., 2007; Wan et al., 2008;
Arstila et al., 2007; Peng et al., 2004).
CONCLUSION
 It is well known recently that nanotechnology is one of the most

exciting disciplines, and it incorporates physics, chemistry,
materials sciences, biology, cement and building materials,
ceramic and bioceramics industries, biomaterials, medicine and
many others. In this chapter, the author interests with using the
nanomaterials to prepare the ceramic batches containing ultrafine
and nono-raw materials to indicate the importance of
nanomaterials and/or nanoparticles for improving the physical and

chemomechanical properties of the resulting bioproducts.
 It could be concluded that the use of nanoparticles of ceramic
materials as clay, limestone, quartz, feldspar and many others
achieved better properties particularly the mechanical strength like
tensile and bending strengths than those of the traditional particles.
The crystals of the resulting ceramic products are sharp and well
defined. The morphology or external appearance of the fired units
is well developed.
 The nano - and/or biomaterials are recently used to prepare the
ceramic batches containing ultrafine or nanoparticles to produce
different shapes and sizes of bone scaffolds to place them inside
the human body to lighten the pains of patients subjected to traffic
accidents and lose a part or more of their body bones. In the future,
the scientists must interest to increase and look for new and more
effective materials suitable to produce artificial bone scaffolds.
REFERENCES
Abdel-Fattah, W. A. & Nour, F. A., Utilization of electrostatic precipitator

dust in borcelain, XIII- Silicon, for silicate Industry and Science
Conferences, Budapest, June, 1981.
Abdel-Fattah, W. I. & Ei-Didamony, H., Thermal investigation on
electrostatic precipitator kiln dust,” Thermochimica, Acta, 51, 1981,
297-306.
Abdel-Fattah, W. I.; El-Didamony, H. & Taha, A. S., Synthesis and
thermal characteristics of Ca rich phases,” TIZ-Fachberichte, 106 (5),
1982, 351-354.
Aggarwal, P. S.; Lele, R. V. & Sen, S. K., Utilization of fly ash for making
wall tiles, Central Glass and Ceramic Res. Inst. Calcotta, 700 012
India, Cent. Glass Ceram. Res. Inst. Bull., 27 (1-2), 1980.
Alemany, M. I.; Velasquez, P.; de la Casa-Lillo, M. A. & De Aza, P. N.,

Effect of materials processing methods on the in vitro bioactivity of
wollastonite glass-ceramic materials, J. Non-Crystalline Solids 351,
2005, 1716–1726.
Anlagenbau, A., Ceramic tiles, GmbH, Ger. Offen. DE 3 313, 594 (CI. CO
4B33/34), Oct., 18, 1984. Chem. Abst. 102:49975 k, 1995.
Arstila, H.; Vedel, E.; Hupa, L. & Hupa, M., Factors affecting
crystallization of bioactive glasses,” Journal of the European Ceramic
Society 27, 2007, 1543–1546.
Ashby, M. F. & Jones, D. R. H., Engineering materials 1, 3rd ed., Elsevier
Butterworth Heinemann, ISBN 0-7506-6380-2, 2006.
Ashby, M. F. Ferreira, P. J. & Schodek, D. L. “Nanomaterials,
Nanotechnology and Design,” Text Book, Elsevier Ltd, ISBN: 978-0-
7506-8149-0, 2009.
Ashby, M. F., Materials selection in mechanical design, 3rd ed.,
Butterworth-Heinemann, Elsevier, 2005.
Ashby, M. F., Shercliff, H. R. & Cebon, D. Materials: engineering,
science, processing and design, Butterworth Heinemann, ISBN-13:
978-0-7506-8391-3, 2007.
Askeland, D. R. & Phule, P. P., The science of engineering materials, 5th
ed., Thomson Publishing, 2006.
ASTM-Specification, C326-32-76, “Standard Test Method for shrinkage of
ceramic white ware clays after drying and firing,” Part, 17: 1980, 266-
267.
ASTM-Specification, C373-72, “Standard Test Method for water
absorption, bulk density, apparent porosity and specific gravity of
whiteware products,” Part, 17: 1980, 308-309.
ASTM-Specification, C674-71, Standard Test Method for flexural
properties of ceramic white ware Products,” Part, 17: 1980, 668-671.
Bacanac, M. & Babic, M., Materials for ceramic tiles with a backing
temperature of 1080C and use of Trokyte, Technika, Belgrade, 39 (4),
1984, 1-5.
Balamurugan, A.; Balossier, G.; Laurent-Maquin, D.; Pina, S.; Rebelo, A.
H. S.; Faure, J. & Gerreira, J. M. F., An in vitro biological and
antimactobial study on a sol-gewl derived silver-incorporated bioglass

system,” Dental Materials, 24, 2008, 1343-1351.
Balamurugan, A.; Sockalingum, G.; Michel, J.; Fauré, J.; Banchet, V.;
Wortham, L.; Bouthors, S.; Laurent-Maquin, D. & Balossier, G.,
Synthesis and characterization of sol gel derived bioactive glass for
biomedical applications, Materials Letters 60, 2006, 3752–3757.
Barba, I.; Conde, F.; Olmo, N.; Lizarbe, M. A.; García, M. A. & Vallet-
Regí, M., Vitreous SiO2–CaO coatings on Ti6Al4V alloys: Reactivity
in simulated body fluid versus osteoblast cell culture, Acta
Biomaterialia, 2, 4, 2006, 445-455.
Barba-Izquierdo, I.; Salinas, A. J. & Vallet-Regi, M., “In vitro calcium
phosphate layer formation on sol-gel glasses of the CaO-SiO2, J.
Biomed. Mater. Res., 47, 2000, 243-250.
Belous, K. P.; Kushel, M. I.; Evplov, Yu. N. & Ohkrimenco, G. M.,
“Effects of technological factors on the strength of ceramic materials,”
Inst. Probl. Prockn., Kiev, USSR, 2, 1989, 42-49.
Binnaz, A. Hazar, Y., Preparation and in vitro bioactivity of CaSiO3
powders, Ceramics International, 33, 2007, 687–692.
Brinker, C. J. & Scherer, G. W., The physics and chemistry of sol-gel
processing, Academic Press Inc., San Diego, CA, USA, 1990.
Budnikov, P. P., The technology of Ceramics and Refractories, Edward
Arnold Publishers Ltd., London, 1964.
Callister, W. D. Jr., Materials science and engineering, an introduction, 7th
ed., Wiley, ISBN 0-471-73696-1, 2007.
Chen, Q.; Miyaji, F.; Kokubo, T. & Nakamura, T., Apatite formation on
PDMS-modified CaO-SiO2-TiO2 hybrids prepared by sol-gel process,
Biomaterials 20, 1999, 1127-1132.
Chen-Chi, W. & Jackie, Y., “Sol−Gel Synthesis and Hydrothermal
Processing of Anatase and Rutile Titania Nanocrystals.” Chemistry of
materials, 11 (11): 1999, 3113–20.
Chiang, Y. M.; Birnie, D. P. & Kingery, W. G., “Physical Ceramics -
Principals for Ceramic Science and Engineering,” 3rd Edn., John
Wiley and Sons, Lehigh Press. Inc., USA, 1997.
Crayston, J. A, Sol-gel, Comprehensive Coordination Chemistry II 1,

ISBN 0-08-0443230, 2003, 711–730.
Drexler, K. E., Peterson, C. & Pergamit, G., Unbounding the future: the
nanotechnology revolution William Morrow, New York, 1991.
Daniel, A. & Vallet-Regi, M., Sol-gel silica based biomaterials and bone
tissue regeneration, Acta Biomaterialia, 6, 2010, 2874-2888.
Drouet, C. R. C.; Sfihi, H. & Barroug, A., Physico-chemical properties of
nanocrystalline apatites: Implications for biominerals and biomaterials,
Materials Science and Engineering C 27, 2007, 198–205.
Darweesh, H. H. M., Utilization of Cement Kiln Dust in Ceramics to
minimize Environmental Pollution, M. Sc. Thesis, Inst. Environmental
Studies and Researches, Ain Shams Univ., 1992.
Darweesh, H. H. M., “Ceramic Wall and Floor Tiles Containing Local
Waste of Cement Kiln Dust - Part I: DensificationParameters,”
American Journal of Environmental Engineering and Science, 2, 5,
2015, 35-43. (http://www.aascit.org/journal/ajees), ISSN: 2381-1153.
Darweesh, H. H. M., “Ceramic Wall and Floor Tiles Containing Local
Waste of Cement Kiln Dust- Part II: Dry and Firing Shrinkage as well
as Mechanical Properties,” American Journal of Civil Engineering and
Architecture, 4, 2, 2016, 44-49. DOI: 10.12691/ajcea-4-2-1, (http://
pubs.sciepub.com/ajcea/4/2/1).
Darweesh, H. H. M. & El-Din, M. N., Characterization of some Egyptian
Granites and Marbles as dimensional stones, Bull. NRC, Egypt, 25, 4,
2000, 323-339.
Darweesh, H. H. M.; Awad, H. M. &Tawfik, A., Red Bricks from Dakhla
Formation Clay - Tushka area-Incorporated with some Ind. Wastes or
byproducts” Industrial Ceramics, Italy, Vol. 31(3) 2011, 201-207.
Darweesh, H. H. M. & El-Meligy, M. G., Non-conventional light-weight
Clay bricks from homra and kraft pulp wastes, J. Chemistry and
Materials Research, 1(4) 2014, 1-7.
Darweesh, H. H. M.; Wahsh, M. M. S. & Negim, E. M., Densification and
Thermomechanical Properties of Conventional Ceramic Composites
Containing Two Different Industrial Byproducts,” Amer.-Eurasian
Journal of Scientific Research, 7 (3), 2012, 123-130.
Darweesh, H. H. M., Building materials from siliceous clay and low grade
dolomite rocks, Ceramics International, 27, 2001, 45-50.
Darweesh, H. H. M., “Utilization of Cement Kiln Dust Waste as a source
of CaO in Ceramic Industry,” Sil. Industrials (Ceramic Science and
Technology), 66 (3-4) 2001, 47-52.
Darweesh, H. H. M. & Khalil, N., Effect of cement bypass dust waste on
the physico-mechanical properties of alumina cement, Ind. J. Chem.
Techn. India, 7, 2001, 287-291.
Danto Co. Ltd. “Decorative tiles,” Jpn. Kokai, 79/70, 250, Jan. 14, 1979.
Drews, M., “Wall and floor tiles,” Ceramic Monographs 2.4.1. Verlag
Schmid GmbH, 1983.
El-Alfi, E. A.; Radwan, A. M. & Darweesh, H. H. M., Effect of sand as
non-plastic material on ceramic properties of clay bricks, Inter Ceram
(Intern. Cer. Review), 53 (5) 2004, 330-333.
El-Hady, B. I.; El-Kady, A. M. & Hassan, M. M. A., Preparation and
characterization of biocomposites for localized bone treatment based
on Agarose and Gelatin, Ph. D. Thesis (Biophysics), University
College of Women, Ain Shams University, 2015.
Fiori, C. & Fabbri, B., Granite containing bodies for the production of
Stoneware tiles, Res. Inst. Ceram. Techn. National Res. Council,
Faenza, Italy, Inter Ceram 32 (1) 1980, 21-22.
Genin, G., Influence of Quartz on the strength of porcelain, Stekelo.
Keram., 15, 1958, 35-41.
Giomlem, K. & Lyng, S., Anorthosite as a ceramic raw material-Part II;
Anorthosite for tile manufacture, Trans. Brit. Ceram. Soc. 73, 1974,
139146.
Grim, R. E., Clay Mineralogy, McGraw Hill, NewYork, 1962.
Gupta, R.; Mozumdar, S. & Chaudhury, N. K., “Fluorescence
spectroscopic studies to characterize the internal environment of
tetraethyl-orthosilicate derived sol–gel bulk and thin films with aging,”
Biosensors and Bioelectronics 20, Issue 7, 2005, 1358-1365.
Haiyan, L. & Chang, J., Fabrication and characterization of bioactive
wollastonite /PHBV composite scaffolds, Biomaterials, 25, 2004,
5473-5480.
Hayashi, S.; Nakagawa, Z.; Yasumori, A. & Okada, K., Effects of H2O in
EtOH-H2O disperse medium on the electrophoretic deposition of
CaSiO3 Fine Powder,” Journal of the European Ceramic Society, 19,
1999, 75-79.
Hench, L. L. & West, J. K., The sol-gel process, Chemical Review, 1990,
90-133.
Iarry, H. L., Bioceramics from concept to clinical, Amer. Cer. Soc., 74,
1991, 1487-1491.
Ibrahim, D. M. M; Abdel-Fattah, W. I. & Nour, F., Air-cooled blast
furnace slag as a fluxing material in the Ceramic Industry, Ber. Dt.
Keram. Ges. 57, 9-10, 1980, 187-194.
Jackson, G., “Introduction to white wares,” Elsevier Publishing Co., 1st
Edn., 1969.
Joachim, S.; Manfred, S. & Peter, T., Ceramic composition for light-
colored gehlinite-free glazed earthenware tiles, VEB kombinat Fliesen
and Sanitaerkeramic “Kurt Buerger,” Ger. (East), DD 205, 883, CI
CO 4B33 /00, Jpn. Jan, 11, 1984.
Julian, R. J.; Lisa, M. E. & Larry, L. H., Optimising bioactive glass
scaffolds for bone tissue engineering, Biomaterials, 27, 7, 2006, 964-
973.
Julie, C. D., “R & D 2002: Nano Ceramics.” MIT Technology Review.
Retrieved Dec. 1, 2002.
Kalnin, I. L., Strength & elasticity of Whitewares-Part I; Relation between
Flexural strength and elasticity, Ceram. Bull., 46, 1174, 1967,
Kalnin, I. L.; Cucka, P.; Warshow, S. I. & Seider, R., “Strength and
elasticity of Whitewares,” Ceram. Bull., 47, 1968, 498-503,
Kashlyak, L. L. & Kareev, Yu. B., Production of tiles for floor and Façade
tiles in the Leningrad Ceramic Products Plant, Keram from St. 8, 10,
1978.
Ke, S.; Wang, Y.; Pan, Z.; Ning, C. & Zheng, S., Recycling of polished tile
waste as a main raw material in porcelain tiles, J. Cleaner Prod. 115,
2016, 238–244.
Keijiro, S.; Takashi, o.; Yoshifumi, K.; Yoshiyuki, O; Masaru, T. & Kozo,
M., Manufacture of ceramics containing Portland cement, Sumitomo
Cement Co. Lmt., Jpn. Kokai, Tokyo Koho JP, o3, 215, 350, 1991,
[91, 215, 350] (C1. CO 4B35/16). Chem. Abst. 115: 285800p.
Khalil, A., “Advanced Sintering of Nano-Ceramic Materials.” Ceramic
Materials - Progress in Modern Ceramics, 2012.
Khrundzhe, A. V. & Babushkin, V. I., State Reginal electric power plant
wastes used for the manufacture of ceramic tiles, USSR, Steklo,
Keram. 3, 1983, 6-8.
Kim, K, and Hwang, J., Characterization of ceramic tiles containing LCD
waste glass, Ceram. Int. 42, 2016, 7626–7631.
Kimm, F., “Ceramics Don't Have to Be Brittle.” Cal Tech. Retrieved 11,
2014.
Kimm, F., “Made-to-Order Materials.” Cal Tech. Retrieved 5, 2013.
Kingery, W. D.; Bowen, H. K. H. K. & Uhlmann, D. R, Introduction to
Ceramics, 2nd edn., John Wiley and Sons, New York, London, Sydney,
Toronto, 361, 594, 1975.
Kisel, I. I.; Strodyreva, G. V. & Demidovisch, V. B., Possibilities of using
local clay in manufacturing floor tiles, Steklo, Sitally Silik, 1980.
Klein, I. Sol-gel optics: Processing and applications, Springer Verlag,
ISBN 0792394240, 1994.
Koch, C. C., Nano structured materials: Processing properties and
applications, Noyes Publications, 2002, ISBN 0-7155-1451-4.
Kokubo, T., Bioactive glass ceramics, Properties and applications,”
Biomaterials, 12, 1991, 155-163.
Kokubo, T.; Kim, H. & Kawashita, M., Handbook of Advanced Ceramics,
2003, 385-416. Chapter 14 - 14.1 Ceramics for Biomedical
Applications.
Kong, Y. M.; Kim, S. & Kim, H. E., Reinforcement of hydroxyapatite
bioceramics by addition of ZrO2 coated with Al2O3, Amer. Cer. Soc.,
82, 1999, 2963-2968.
Konta, J., Properties of ceramic raw materials, Ceramic Monograph 1.1.4.
Verlag Schmid G.m.b.H, 1981.
Lao, J. Nedelec, J. M.; Moretto, Ph. & Jallot, E., Biological activity of a
SiO2–CaO–P2O5 sol–gel glass highlighted by PIXE–RBS methods,
Nuclear Instruments and Methods in Physics Research B 261, 2007,

488–493.
Lao, J.; Nedelec, J. M.; Moretto, Ph. & Jallot, E, Micro-PIXE
characterization of interactions between a sol–gel derived bioactive
glass and biological fluids, Nuclear Instruments and Methods in
Physics Research B 245, 2006, 511–518.
Lin, K., Preparation of macroporous calcium silicate ceramics, Materials
Letters 58, 2004, 2109–2113.
Lin, K.; Zhai, W.; Ni, S.; Chang, J.; Zeng, y. & Qian, W., Study of the
mechanical property and in vitro biocompatibility of CaSiO3 ceramics,
Ceramics International, 31, 2005, 323–326.
Liu, J. and Miao, X., Sol–gel derived bioglass as a coating material for
porous alumina scaffold, Ceramics International 30, 2004, 1781–1785.
Liu, X. & Ding, Ch., Characterization of plasma sprayed wollastonite
powder and coatings, Surface and Coatings Technology, 153 (2-3)
2002, 173-177.
Long, L. H.; Chen, L. D. & Chang, J., Low temperature fabrication and
characterizations of β-CaSiO3 ceramics” Ceramics International, 32,
2006, 457–460.
Long, L. H.; Chen, L. D.; Bai, S. Q.; Chang, J. & Lin, K. L., Preparation of
dense β-CaSiO3 ceramic with high mechanical strength and HAp
formation ability in simulated body fluid, J. European Ceramic Society
26, 2006, 1701–1706.
Lukito, D.; Xue, J. M. and Wang, J., In vitro bioactivity assessment of 70
(wt.) %SiO2–30 (wt.) %CaO bioactive glasses in simulated body fluid,
Materials Letters 59, 2005, 3267 – 3271.
Martín-Márquez, J.; Rincón, J. Ma & Romero, M., Effect of firing
temperature on sintering of porcelain stoneware tiles, Ceram. Int. 34,
2008, 1867–1873.
Meiszterics, A. & Sinkó, K., Sol–gel derived calcium silicate ceramics,
Colloids and Surfaces A: Physicochem. Eng. Aspects, 2007.
Meseguer-Olmo, L.; Bernabeu-Esclapez, A.; Ros-Martinez, E.; Sańchez-
Salcedo, S.; Padilla, S.; Martìn, A. I.; Vallet-Regì, M.; Clavel-Sainz,
M.; Lopez-Prats, F. & Meseguer-Ortiz, C. L., In vitro behaviour of
adult mesenchymal stem cells seeded on a bioactive glass ceramic in

the SiO2–CaO–P2O5 system, Acta Biomaterialia, 4, 2008, 1104-1113.
Mila, R. A. & Vallet- Regm, M., Static and dynamic in vitro study of a sol-
gel glass bioactivity, Biomaterials 22, 2001, 2301-2306.
Mohamed, E. M., Inorganic –organic hybrids based on PVA and Silica
with added Titania, M. Sc. Thesis (Biophysics), University College of
Women, Ain Shams University, 2012.
Moroz, B. I.; Gumenyuk, A. E.; Mel’nekov, V. M. &Trubachov, V. I.
“Bodies for the manufacture of ceramic tiles,” USSR, Stroit. Mater.
Konstr., 2, 17, 1990, 2, 17. Chem. Abst. 114: 10710y, 1991.
Mortel, H., Porcelains for fast firing, Inter Ceram, 3, 1977, 65-69.
Mostafa, M. Z. & Darweesh, H. H. M., “Mechanical properties of ceramic
floor tiles containing local waste of cement kiln dust,” The 8th SIMCER
Intern. Symposium on Ceramics, Rimini (Italy)-Exhibition Centre, Nov.
10-12, 1992.
Mukhamedzhanov, M. T.; Palvanov, V. P.; Irkahodzhaeva, V. P. &
Toirova, M. Kh., The use of mining tailings in the manufacture of
ceramic tiles, Inst. Sredazuiprotsvtmet, Almalyk, USSR, Kompleksen.
Ispol’z. Miner. Syr’ya, 11, 1990, 82-84. Chem. Abst. 114: 212495n,
1991.
Nalwa, H. S. Ed., Encyclopedia of Nanoscience and Nanotechnology,
Volumes 1-25, American Scientific Publishers, Los Angeles, CA
(2004/2011).
Nalwa, H. S. Ed., Handbook of Nanostructured Materials and
Nanotechnology, Volumes 1-5, Academic Press, San Diego, CA, 2000.
Newport, R. J.; Skipper, L. J.; FitzGerald, V.; Pickup, D. M.; Smith, M. E.
& Jones, J. R. “In vitro changes in the structure of a bioactive calcia–
silica sol–gel glass explored using isotopic substitution in neutron
diffraction,” Journal of Non-Crystalline Solids 353, 2007, 1854–1859.
Nissan, B., “Nanoceramics in Biomedical Applications.” MRS Bulletin 29
(1): 2014, 28–32.
Njindam, O. R.; Njoya, D.; JMache, R.; Mouafon, M. Messan, A. &
Njopwouo, D., “Effect of glass powder on the technological properties
and microstructure of clay mixture for porcelain stoneware tiles
manufacture,” Construction and Building Materials 170, 512–519,

www.elsevier.com/locate/conbuildmat, 2018.
Nuttawat, K.; Jaimasith, M. & Thiemsorn, W., Fabrication of ceramic floor
tiles from industrial wastes, Suranaree J. Sci. Technol. 21, 2013, 65–
77.
Olmo, N.; Ana, I. M.; Antonio, J. S.; Turnay, J.; Vallet-Reg, M. & Lizarbe,
M. A., Bioactive sol–gel glasses with and without a hydroxycarbonate
apatite layer as substrates for osteoblast cell adhesion and proliferation,
Biomaterials 24, 2003, 3383–3393.
Padilla, S.; Román, J.; Carenas, A. & Vallet-Regì, M., The influence of the
phosphorus content on the bioactivity of sol–gel glass ceramics,
Biomaterials, 26, 2005, 475–483.
Peng, F.; Liang, K.; Hu, A. &Shao, H., “Nano-crystal glass-ceramics
obtained by crystallization of vitrified coal fly ash,” Fuel 83, Issues 14-
15, 2004, 1973-1977.
Phulé, P. P. & Wood, T. E., Ceramics and Glasses, Sol-Gel Synthesis of”
Encyclopedia of Materials: Science and Technology, ISBN: 0-08-
0431526, 2001, 1090-1095.
Pinalova, D.; kukushiva, M. & Belchev, K., Wastes from Coal benefication
for effective manufacture of ceramic products, Stroit. Mater. Silik., 25,
6, 1984, 14-15.
Pitskhelauri, I. A.; Lomidze, N. M.; Chkheidze, K. S.; Sakandlidze, A. A.
& Zambakhidze, N. P., Raw mixtures for producing ceramic wall
articles,” Tbilisi State Academy of Arts, USSR, Su 1, 357, 392 (C1./
CO 4B33/00), Dec. 7, 1987. Chem. Abst. 108: 100003v, 1988.
Poole, C. P. Jr., & Owens, F. J., Introduction to nano technology, Wiley-
Interscience, ISBN 0-471-07935-9, 2003.
Rayan, R. W. & Radford, C., “White wares production, testing and quality
control,” 1st Edn. The Institute of Ceramics, Pergamon Press. Oxford,
New York, Beijing, Sedney, Frankfurt, Sao Paulo, Tokyo, Toronto,
1987.
Rayan, W., “Properties of ceramic raw materials,” 2nd Edn. Pergamon
Press, 1978.
Saltyvskaya, I. M., Use of chemical industry byproducts for the

manufacture of ceramic tiles, (Khar’k. Inzh.-Ekon. Inst. Khakov,
USSR). Steklo Keram, 9, 19-20. Chem. Abst. 102: 30662j, 1985.
Saravanapavan, P. & Hench, L. L., Mesoporous calcium silicate glasses. I.
Synthesis, J. Non-Crystalline Solids 318, 2003, 1-13.
Scherer, G. W., Structure and properties of gel,” Cem. Concr. Res., 29, 8,
1999, 1149-1157.
Schuller, K. H. & Jager, H., The chemistry and properties of feldspar and
their effects in porcelain, Ber. Dt. Keram. Ges., 56, 29, 1979.
Schuller, K. H., Reactions between Mullite and glassy phase in Porcelains,
Trans. Brit Ceram.Soc., 63, 1964, 103-110.
Shackelford, J. F., Introduction to materials science for engineers, 6th ed.,
Prentice Hall, 2004.
Singer, F. & Singer, S. S., “Industrial Ceramics,” 2nd Edn. Leipziger,
Druchkaus, GDR, 1971.
Siriphannon, P.; Kameshima, Y.; Yasumori, A.; Okada, K. & Hayashi, S.,
Formation of hydroxyapatite on CaSiO3 powders in simulated body
fluid, J. European Ceramic Society 22, 2002, 511–520.
Smith, W. F. & Hashemi, J., Foundations of materials science and
engineering, 4th ed., McGraw Hill, 2005.
Sobolev, K. & Ferrada-Gutiérrez, M., How nanotechnology can change the
concrete, world: Part 1, Am Ceram Soc Bull, 84 (10), 2005, 14–17.
Sonjida, M; Ahsan, M.; Hamid, A. & Ahmed, S., Effect of waste glass
powder on physico-mechanical properties of ceramic tiles, Bangladesh
J. Sci. Res. 24, 2011, 169–180.
Sridhar, K., “Some significant advances in sol-gel processing of dense
structural ceramics.” Sol-Gel Science and Technology 6 (2): 1996, 127-
132.
Susan, H., Basic biomaterials, 5th Edn., 88 ISBN 0071260412, 2007.
Taha, R. T. A., Bioceramic composites suitable for bone graft, M. Sc.
Thesis (Biophysics), University College of Women, Ain Shams
University, 2011.
Todor, D. N., Thermal analysis of minerals, English Edn., Abacus Press,
1972.
Tseng, T. Y. & Nalwa, H. S., Handbook of Nanoceramics and Their Based

Nanodevices, Volume 1-5, American Scientific Publishers, Los
Angeles, CA, 2009.
Vauequelinet, M; Gault, C & Boch, B., Reheology of clay mixture ceramic
bodies, Ind. Ceram. 744, 1980, 736-740.
Viitala, R.; Jokinen, M.; Peltola, T.; Gunnelius, K. & Rosenholm, J. B.,
Surface properties of in vitro bioactive and non-bioactive sol–gel
derived materials, Biomaterials 23, 2002, 3073–3086.
Vorrada, L.; Panyachai, T.; Kaewsimork, K. & Siritai, C., Effects of
recycled glass substitution on the physical and mechanical properties
of clay bricks, Waste Manage. (Oxford) 29, 2009, 2717–2721.
Wan, X.; Hu, A.; Li, M.; Chang, Ch. & Mao, D., Performances of CaSiO3
ceramic sintered by Spark plasma sintering, Materials
Characterization 59, 2008, 256-260.
Wang, E.; Chow, K.; Kwan, V.; Chin, T.; Wong, C. & Bocarsly,A., Fast
and long term optical sensors for pH based on sol–gels, Analytica
Chimica Acta, 495 (1-2) 2003, 45-50.
Wang, H.; Sockalingum, G.; Michel, J.; Fauré, J.; Banchet, V.; Wortham,
L.; Bouthors, S.; Laurent-Maquin, D. & Balossier, G. Synthesis and
microwave dielectric properties of CaSiO3 nanopowder by the sol–gel
process,” Ceramics International 34, 2008, 1405-1408.
Warshow, S. I. & Seider, R., Comparison of strength of triaxial porcelains
containing alumina and silica, J. Amer. Ceram; Soc., 50, 1967, 337-
342.
Watchman, J. P., Mechanical properties of ceramics-An introductory
Survey, Ceram Bull., 46, 1967, 756-762.
Wei-hong, Z.; Jin-shu, C.; Jian, Q.; Xian-chun, L. & Jian, L. I.,
Crystallization and properties of some CaO-A12O3-SiO2system glass-
ceramics with Y2O3 addition, Trans. Nonferrous Met. SOC. China, 16,
2006, 105-108.
Wu, Ch.; Ramaswamy, Y.; Kwik, D. & Zreiqat, H., The effect of strontium
incorporation into CaSiO3 ceramics on their physical and biological
properties” Biomaterials 28, 2007, 3171–3181.
Xia, W. and Chang, J., Well-ordered mesoporous bioactive glasses (MBG):

A promising bioactive drug delivery system, J. Controlle Release, 110
(3) 2006, 522-530.
Chapter 8
SIMULATION OF LOW COST AUTOMATION

AND LIFE CYCLE COST ANALYSIS FOR A
SPECIAL PURPOSE MACHINE
Pattanayak Satyajit and Hauchhum Lalhmingsanga

National Institute of Technology Mizoram,
Aizawl, India
ABSTRACT
The need for simulation of low cost automation and life cycle cost
analysis for a special kind of machine is a promising research area in
industries. This machine is capable of machining (undercut, head
diameter, fillet, chamfering) operation of engine valve, used in
automobile industry. The objective of this work is to reduce rejection,
lower the cost, increase productivity and quality of performance. In this
chapter, a methodology to design, develop and simulate a twin spindle
turning Special Purpose Machine is presented based on the data collected
from hydraulic, pneumatic, electro pneumatic which will serve as low

Corresponding Author Email: lalhmingsanga.mech@nitmz.ac.in
264 Pattanayak Satyajit and Hauchhum Lalhmingsanga
cost automation. The electro pneumatic circuit design presented here is

simulated using commercially available Fluid SIM software. Special
purpose machine tool development based on special product
manufacturing (engine valve) requirements is also presented in this
chapter. The design is based on Low Cost Automation principles rather
than pneumatic and Programmable Logic Controller principles. A 3D
model of the machine is also presented using Pro-E software to represent
the implemented components after the design and development process.
Keywords: low cost automation, special purpose machine, productivity,

cost reduction, life cycle cost, simulation
NOMENCLATURE
Ph Parts per hour

S Single part time hours
L Cost of labour (USD)
LS Lot size
W Wage rate dollar
TC Tool cost dollar
Cp Cost per part dollar
BP Break-even point
ABBREVIATIONS
SPM Special Purpose Machine

LCA Low Cost Automation
LCC Life Cycle Cost
PLC Programmable Logic Controller
CPU Central Processing Unit
NC Numerical Control
DCV Directional Control Valve
Simulation of Low Cost Automation and Life Cycle Cost … 265
1. INTRODUCTION
Low Cost Automation (LCA), is the introduction of simple pneumatic,

hydraulic, mechanical and electrical devices into the existing production
machinery, with a view of improving their productivity (Sulivan, 1989;
Albarlos, 1998). This will also enable the operation of this equipment by
semi-skilled and unskilled labour, with a little training. This will involve
the use of standardised parts and devices to mechanise or automate
machines, processes and systems (Esposito, 1994; Freund et al., 2000).
In the context of globalisation and liberalisation, quality improvement
and cost reduction are two major steps to increase the productivity.
Moreover, making any heavy investment is not possible considering many
factors like economic recession, slump in demand, risk and lack of funds
(Erbe, 1995,1996,2000,2003; Inagaki, 2000; Lay, 2002). One of the very
practical, safe, economical and rewarding strategies is the application of
Low Cost Automation. LCA should not be regarded in terms of a specified
maximum capital outlay, but as an approach to automation using
equipment and control devices that are, in general, both technically and
economically, within the scope of the company concerned (Legierski et
al.,1998; Hohwieler et al., 2002). The lower level technologies can be
made highly productive by automation at low cost and in simple form
(Ollero, 2002).
Life Cycle Cost (LCC) accounts for the total costs from starting point
of equipment and projects to their disposal which is derived analytically
through an estimation of the total costs for their life time (Carvalho, 1986;
Chuang et al. 1992;1993). Usually Engineers used LCC based analysis to
justify the selection of equipment and process. This approach helps in
choosing the most cost-effective approach with the aim of achieving
minimum long-term cost. This is because the initial purchase price is
usually much less than the cost of maintenance operation and disposal
(Barringer et al., 1996). The total costs anticipated to be incurred in the
design, production, development, maintenance, operation, support and
disposition over the life span of a major system constitute its life cycle
cost. A minimum total of LCC correspond to the optimal balance of the
cost elements (Swider, 2002; Swider et al., 2003;2004). Automation

demands replacement of conventional Radial Drilling Machines by Special
Purpose Horizontal Multi Spindle Drilling Machine by application of LCA
(Gajmal et al., 2014). The machining areas in cellular manufacturing have
been proven to be an economic, efficient and lean approach by bringing
flexibility. After investigating existing analytical methods for measuring
work, the LCA is useful for identifying and quantifying the different
manual tasks of cellular manufacturing line (Soloman, 1996; Seifermann et
al., 2014).
Three machines viz. conventional lathe machine, NC lathe machine
and special purpose machine which performs machining works like turning
head diameter, chamfering, fillet and undercut or grooving are taken into
consideration. In these three machines; select one machine, which is
greater as per LCC principle. Hence, economic analysis should be done in
these three machines by selecting one machine which is best as per LCC
principle and our required objectives like increasing productivity, number
of quality products, production cost reduction, reduction of time and
avoiding skilful labour can be fulfilled.
2. METHODOLOGY
2.1. Key Main Product Features of Automation
Designs of low cost automation in special purpose turning machines

have some key features. Following are the main components of Low Cost
Automation:
1) Chute on/off
2) Loader on/off
3) Pusher on/off
4) Ejector on/off
5) Collet on/off
1) Chute is nothing but manually conveying systems of material as

shown in Figure 1. Chute is fitted in a machine at certain angle. So, the part
is moving down through gravitational forces. Here, chutes door is closed
which is operated by pneumatic cylinder. When the cylinder is retracting at
that time, the part is moving down to loader plate.
Figure 1. Chute assembly.
2) Loader is loading the product, which comes from the chute as

shown in Figure 2. It has one cylinder which is moving down and up by
using pneumatic cylinder. When the cylinder is extending at that time,
loader is down and set the same line of spindle axis.
Figure 2. Loader assembly.
3) Pusher as shown in Figure 3 is pushing the product from loader to

spindle collet. Pusher is also fitted with the same line of spindle axis. So,
pusher pushes the product easily in collet in the spindle.
Figure 3. Pusher.
4) Collet is a standard product as shown in Figure 4. The collet end

side has thread which holds the spring. Another spring is attached in
pneumatic cylinder rod. When cylinder is retracted, spring is pulled to the
collet at the same time and collet comes on to hollow spindle and fixed the
work part.
Figure 4. Collet.
5) After the machining process, the spindle collet cylinder given in

Figure 5 moves forward and pushes the collet. Hence, the part of the collet
is loose and at the same time, ejector move forward and eject the part from
collet. Ejector, spindle and collet assembly is given in Figure 6.
Figure 5. Spindle collet assembly.

Figure 6. Ejector, spindle and collet assembly.
Figure 7. Assembly of Special Purpose Machine.
2.2. Purpose of Special Purpose Turning Machine
The Special Purpose Turning Machine, having two spindle head unit,
is used for carrying out the turning head diameter, undercut, chamfering
head and neck operations during manufacturing of engine valves.
Assembly of Special Purpose Turning Machine is shown in Figure 7.
The engine valve manufacturing involves a sequence of operations
starting from casting to final finishing, out of which turning head diameter,
undercut, chamfering and fillet form four intermittent operations. These
operations are required for turning head diameter as per dimension,
undercut operation for holding the spring, chamfering operation for good
fittings.
The Special Purpose Turning Machine will have the combined
advantages of a CNC Machine and Special Purpose Machine. This
machine is built only for the purpose of turning head diameter, undercut
and chamfering generation in valve manufacturing with computerized

controlled.
2.3. Circuit Design
2.3.1. Sequence of Operation
A-A+B+C+D-C-B-D+E+E-
A- Open Chute door cylinder A+ Close Chute Door Cylinder

B- Upward Loader Cylinder B+ Downward Loader Cylinder
C- Retraction Pusher Cylinder C+ Forward Pusher Cylinder
D- Close Collet Cylinder D+ Open Collet Cylinder
E- Retraction Ejector Cylinder E+ Forward Ejector Cylinder
2.3.2. Position Step Diagram
Figure 8. Position step diagram.

2.3.3. Flow Diagram of Step Sequence Operation
Figure 9. Flow diagram of step sequence operation.
2.3.4. Drawing of Pneumatic Circuit

The typical pneumatic circuit is shown in Figure 10. This pneumatic
circuits have five cylinders, ten flow control valves, eleven 5/2 directional
control valves, ten roller type 3/2 DCV for limit switches, one AND valve,
One timer etc. This circuit is very complicated and require a big space for
assembling.
The main disadvantages of this circuit are noted below:
• Very complicated
• Large space is required
• Difficult in assembling and dissembling
• Difficulty in maintenance.
• One labour is always required when process is going on, for
operating the start button at every cycle.
For these reasons, it is necessary to design electro pneumatic circuit

design.
Figure 10. Pneumatic circuit.

3. RESULTS AND DISCUSSION
3.1. Circuit Simulation
As per the sequence of operation in Figure 11, the electro pneumatic

circuit simulation as per the required sequence of operation in various steps
is presented. Dark colour shows air flow to the cylinder from compressor
and light colour shows return air from cylinder to exhaust as shown in
Figure 12.
Figure 11. Electro pneumatic circuit design.
Figure 12. Simulation of home position of electro pneumatic circuit.

Figure 13. Simulation of second step.
Step 1: Home position of the circuit, (A+ B- C- D+ E-). Figure 12

Shows home position of the circuit.
Step 2: The opening of chute cylinder door indicate Y1 solenoid is
actuated and cylinder retraction position is (A- B- C- D+ E-). Figure 13
shows that chute cylinder A is retracted.
Step 3: Chute Cylinder door is closed as shown Figure 14. When
piston is at A- position, at the same time sensor is sensed and Y2 solenoid is
actuated. Hence, air is going from port 1 to port 2 through blank end of
cylinder and cylinder is forwarded, (A+ B- C- D+ E-).
Figure 14. Simulation of third step.

Figure 15. Simulation of fourth step.
Step 4: A loader cylinder is downward (B+) as shown in Figure 15.

When chute cylinder is at A+ position and at the same time sensor is
sensed, then Y3 solenoid is actuated. Hence, air goes from port 1 to port 4
through blank end of loader cylinder. So, cylinder is moving forward, (A+
B+ C- D+ E-).
Step 5: Fifth step shows a pusher cylinder push the product in collets
as given in Figure 16. When loader cylinder is coming to B+ position and
at the same time sensor is sensed, then Y3 solenoid is actuated. Air goes
from port 1 to port 4 through blank end of pusher cylinder. Hence, pusher
cylinder is moving forward (A+ B+ C+ D+ E-).
Figure 16. Simulation of fifth step.

Figure 17. Simulation of sixth step.
Step 6: A collet is clamped to the part by gripper cylinder as shown in

Figure 17. When pusher cylinder piston comes at C+ position, Y2 solenoid
is actuated at the same time. Air goes from port 1 to port 4 through rod end
area of gripper cylinder. So, gripper cylinder is retracted. Here, we can see
that the pipe line from valve 4 to cylinder is different to another loader,
pusher and ejector cylinder as our requirement is to open collet at home
position. It is important to design the port 2 connected with blank end area
of gripper cylinder and port 4 connected with rod end area of gripper
cylinder, (A+ B+ C+ D- E-).
Step 7: Pusher cylinder is retracted as shown in Figure 18. When
gripper cylinder is at D- position and Y6 solenoid is actuated at the same
time. Air goes from port 1 to port 2 through rod end area of pusher
cylinder, (A+ B+C- D- E-).
Step 8: In this step, loader cylinder is retracted as shown in Figure 19.
When pusher cylinder is at C- position and Y4 solenoid is actuated at the
same time. Air goes compressor from port 1 to port 2 (V2) through rod end
area of loader cylinder. Then, the loader cylinder is retracted,
(A+ B- C- D- E-).
Step 9: The gripper cylinder unclamped the part as shown in Figure
20. When loader cylinder is at B- position and timer is there at the same
time, then the time will be set as per requirement of machining operation,
thereby actuating Y8 solenoid. Air goes to compressor from port 1 to port 2
(V4) and to blank end area of gripper cylinder. therefore, gripper cylinder
moves forward, (A+ B- C- D+ E-).
Step 10: Ejector cylinder moves forward as shown in Figure 21. When
gripper cylinder is at D+ position and Y9 solenoid is actuated at the same
time. Air goes to compressor from port 1 to port 4 (V5) and to blank end
area of ejector cylinder. Therefore, ejector cylinder is ejecting the part from
collet, (A+ B- C- D+ E+).
Figure 18. Simulation of seventh step.
Figure 19. Simulation of eighth step.

Figure 20. Simulation of ninth step.
Figure 21. Simulation of tenth step.
Figure 22. Simulation of eleventh step.

Step 11: Ejector cylinder is retracted as shown in Figure 22. When

ejector cylinder is at E+ position and Y10 solenoid is actuated at the same
time. Air goes to compressor from port 1 to port 2 (V5) and to rod end area
of the ejection cylinder. Therefore, ejector cylinder is retracted, (A+ B- C-
D+ E-) which means it return back to home position. This completes one
cycle of the feeding to ejection automation process.
3.2. Programmable Logic Controller Ladder Programming
1) The sequence is A-A+B+C+D-C-B-D+E+E-. This can be

performed by using Programmable Logic Controller (PLC)
programming. Performance of this sequence is shown in Figure 23.
In the programmed, one side is connected with 24V and the other
side is connected with 0V. Here, only one input is given, i.e.,
START button. When the start button is started, it gives an output
i.e., CR2 activated which means Y2 solenoid in 5/2 DCV1 is
actuated (A-).
2) When A- is actuated which means chute door is open, timer is
activated and set at 01 s. After 01 s, CR1 output is activated which
means Y1 solenoid in 5/2 DCV1 is actuated (A+).
3) When A+ is actuated at the same time, the timer is set at 01 s.
After 01 s, CR3 is activated which means Y3 solenoid in 5/2 DCV2
is actuated (B+).
4) After setting the timer at 01 s, CR5 is activated which means Y5
solenoid in 5/2 DCV3 is actuated (C+). Then, cylinder pusher is
extended.
5) When the pusher is extended, the sensor sensed to activate CR8
which means Y8 solenoid in 5/2 DCV4 is actuated (D-). Then,
Gripper cylinder is clamping the part. Then, one timer is set again
at 01 s. After 01 s, CR6 is activated which means Y6 solenoid in
5/2 DCV3 is actuated. After which the pusher is retracted.
Figure 23. PLC ladder diagram.

6) When the pusher is retracted and at the same time, loader is also
retracted. After 01 s, CR4 is activated which means Y4 solenoid in
5/2 DCV2 is actuated (B-).
7) When loader is retracted, the required machining work begin. After
completing machining work, ejector cylinder is activated. Then,
CR7 is activated which means Y7 solenoid in 5/2 DCV4 is actuated
(D+). Before ejecting the part, the gripper cylinder should unclamp
the part from collet.
8) After unclamping the part, the ejector cylinder is extended and
eject the part from collet. Then, the ejector cylinder is retracting
and one cycle is completed. Then, CR9 is activated which means
Y9 solenoid in 5/2 DCV5 is actuated (E+). The next round has one
timer set at 01 s. After 01 s, CR10 is activated which means Y10
solenoid in 5/2 DCV 5 is actuated (E-). This complete one cycle of
automation by using PLC programming.
3.3. Life Cycle Cost Analysis
3.3.1. Economic Analysis

The tool designer must furnish management with an idea of how much
the tooling will cost and how much the production method saves over a
specific run. This information is generally furnished in the estimated cost
of the tool and projected savings over alternate methods. The estimate also
includes any special conditions that may justify the cost of tooling, such as
close tolerances or high-volume production. For a valid estimate, the tool
designer must accurately estimate the cost and productivity of the design in
terms of materials, labour, and the number of parts per hour the tool will
produce. Edward G., 2004 had determined calculation of economic
analysis.
3.3.1.1. Economic Analysis of Conventional Lathe Machine

Machining Operation:
a) Turning Head Diameter

b) Chamfering
c) Fillet
d) Undercut/Grooving
First step: Calculation of parts per hour

The first step in estimating is calculating the number of parts per hour
the machine will produce simplest method is to divide 1 hour by the single-
part time or the time takes to load machine and unload each part. It can be
expressed as given in Equation 1:
1
𝑃ℎ𝑐𝑜𝑛𝑣 = 𝑠 (1)
Second step: Calculation of labour cost

Labour cost is the most expensive factor in manufacturing. If labour
expenses can be reduced, so can overall production costs. The cost of
labour can be calculated as given in Equation 2:
𝐿𝑆
𝐿𝐶𝑜𝑛𝑣 = 𝑃ℎ × 𝑤 (2)
Third step: Calculating the cost per part

For accuracy, it is a must to calculate how much the design is worth in
terms of total production and cost per part. Equation 3 gives the formula
for finding this value:
𝑇𝐶+𝐿
𝐶𝑝 (𝐶𝑜𝑛𝑣) = 𝐿𝑆
(3)
3.3.1.2. Economic Analysis of NC Turning Machine


b) Chamfering
c) Fillet

the machine will produce with simplest method is to divide 1 hour by the
single-part time, or the time taken to load machine and unload each part. It
can be expressed as given in Equation 4:
1
𝑃ℎ𝑁𝐶 = 𝑠 (4)

Labour is the single most expensive factor in manufacturing. If labour
expenses can be reduced, so can overall production costs. The cost of
labour can be calculated as in Equation 5:
𝐿𝑆
𝐿𝑁𝐶 = ×𝑤 (5)
𝑃ℎ

For accuracy, it is very important to calculate how much the design is
worth in terms of total production and cost per part. Equation 6 gives the
formula for finding this value as follows:
𝑇𝐶+𝐿
𝐶𝑝 (𝑁𝐶) = 𝐿𝑆
(6)
Fourth step: Calculating the Break-Even Point

The break-even point is the minimum number of parts a tool must
produce to pay for itself. Any number less than this minimum results in a
loss of money; any number more results in a profit. It is logical to assume
that the lower the break-even point, the higher the profit potential. The
formula to find the break-even point is given in Equation 7:
𝑇𝐶
𝐵𝑃 = (𝐶 (7)
𝑝(𝐶𝑜𝑛𝑣) −𝐶𝑝(𝑁𝐶)
Fifth Step: Calculating total savings

To determine the most economical production method, it is required to
compare production alternatives. The following formula given in Equation
8 assumes that both alternatives are being considered which require special
tooling to produce the part is:
𝑇𝑆 = 𝐿𝑆 × {𝐶𝑝(𝐶𝑜𝑛𝑣) − 𝐶𝑝(𝑁𝐶) } (8)
3.3.1.3. Economic Analysis of Special Purpose Machine


b) Chamfering
c) Fillet

the machine will produce by simplest method is to divide 1 hour by the
single-part time, or the time taken to load machine and unload each part. It
can be expressed as given in Equation 9:
1
𝑃ℎ𝑆𝑃𝑀 = 𝑠 (9)

Labour is the single most expensive factor in manufacturing. If labour
expenses can be reduced, so can overall production costs. The labour cost
can be calculated as per Equation 10:
𝐿𝑆
𝐿𝑆𝑃𝑀 = 𝑃ℎ × 𝑤 (10)

For accuracy, it is important to calculate how much the design is worth
in terms of total production and cost per part. The formula for finding this
value is given Equation 11:
𝑇𝐶+𝐿
𝐶𝑝 (𝑆𝑃𝑀) = (11)
𝐿𝑆
Fourth step: Calculating the Break-Even Point

The break-even point is the minimum number of parts a tool must
produce to pay for itself. Any number less than this minimum results in a
loss of money; any number more results in a profit. It is logical to assume
that the lower the break-even point, the higher the profit potential. The
following formula given in Equation 12 is used to find the break-even
point:
𝑇𝐶
𝐵𝑃 = (12)
(𝐶𝑝(𝑁𝐶) −𝐶𝑝(𝑆𝑃𝑀)
Fifth Step: Calculating total savings

To determine the most economical production method, it is required to
compare production alternatives. The following formula, which assumes
that both alternatives which are being considered required special tooling
to produce the part is given in Equation 13:
𝑇𝑆 = 𝐿𝑆 × {𝐶𝑝(𝑁𝐶) − 𝐶𝑝(𝑆𝑃𝑀) } (13)
3.3.2. Comparative Analysis of These Three Machines

A comparative analysis of these three machines are shown in Table 1.
By comparing each method, the tooling requirements in terms of costs
versus savings can be obtained. Then, the method that returns the most for
each dollar spent can be selected. When preparing this comparison, it is
important to weigh all the economic factors in relation to expenses and
productivity.
Table 1. Comparison work sheet of three machines
COMPARISON WORK SHEET

Economic & Productivity Conventional NC Special Purpose
Factors Lathe Lathe Turning
Machine Machine Machine
Lot Size (Ls) 80 80 80
Tool Cost (Tc) $312.77 $312.77 $312.77
Parts Per Hour (Ph) 7.49≠8 29.81≠30 148.14
Labour/Hour(W) $1.50 $0.744 $0.5585
Labour/Lot (L) $16.02 $1.99 $0.3015
Cost Per Part (Cp) $4.10 $3.93 $3.91
Quality of The No. of Part/Lot 8 32 76
Maintenance Cost of The $223.41 $670.24 $446.82
Machine Per Annum
Electricity Charge $16.37 $27.37 $22.34
Coolant Tank Size 0 63 Litres 63 Litres
Figure 24. Comparison graph.
From Table 1, it can be seen that the first machine is not suitable, when
cost and quality are the only factor. The second machine also has slow
production rate and low labour cost. But number of quality of the product
is less and electricity charge is higher than other machines. This machine’s
total cost of savings is $13.6. When quality is the only factor, the second
machine is not suitable.
The third machine produces the parts at a higher production rate and a
lower labour cost. The number of quality product is high and electric
charge is minimum as compared to second machine. The savings are again
offset, this time by the tool cost. If the production run were greater, this
method would have been the least expensive. Our requirements are high
volume production rate, the number of quality product, less rejection and
reduction of cost of production rate. In this machine skilled labour is not
required. So, the third machine is selected for production (Graphically
shown in Figure 24).
CONCLUSION
This work is carried out based on the industry requirements to achieve

low cost in production, maintaining quality with reduced rejections. LCA
techniques, use of simulation software and PLC programming, have been
considered to achieve the objective of this work. At first, design has been
carried out by using the pneumatic circuit for the required operation
sequence, but this pneumatic design technique has some disadvantages
such as large space requirement, high maintenance cost and labour cost. To
overcome these disadvantages, it is decided to adopt electro pneumatic
circuit design, because this is very easy to implement and also nullifies the
disadvantages of the pneumatic circuit. The electro pneumatic design is
simulated using FLUID-SIM software and PLC programming and the
required operation sequence is presented by simulation. Life Cycle Cost
principle is used for the economic analysis and comparison of three
machines, namely conventional lathe, NC lathe and the Special Purpose
turning Machine had been investigated. It can be concluded that the
Special Purpose Turning Machine will be profitable for long run mass
production.
REFERENCES
Albarlos, P. (1998). Trends in Low Cost Automation, In: Proceedings of

the 5th IFAC Symp. On Low Cost Automation, Elsevier Science Ltd.,
Oxford.
Barringer H. P., Barringer P. E, and Weber P. D. (1996). Life Cycle Cost
Tutorial. 5th International Conference on Process Plant Reliability,
Gulf Publishing Company and Hydrocarbon Processing, Marriott
Houston Westside Houston, Texas, 1-58.
Carvalho, A. J. P.(1986). SYSSIM- interactive systems for real time
control simulation, IFAC Proceedings, 19 (13), pp.215-218.
Chuang, S. H., Du, C. Y., Chao, C. H., Lee, C. E., and Tsai, B. Y. (1992).
Software Development of Pneumatic Loop Design and Simulation,
PLDS 2.0, Technical Report, NCHU.
Chuang, S. H., Du, C. Y., Chao, C. H., Lee, C. E., and Tsai, B. Y. (1993).
Software Development of Pneumatic Loop Design and Simulation,
PLDS 2.1, Technical Report, NCHU.
Edward G. (2004). Jigs and Fixture Design, Hoffman Published.
Erbe, H. H. (1995). Refitting conventional machine tools to numerical
control. In: Preprints, 4th IFAC Symp. on Low Cost Automation,
Buenos Aires, Argentina, pp. 16-20.
Erbe, H.-H. (1996). Technology and Human Skills in Manufacturing, In:
Balanced Automation Systems IT, Chapman & Hall, London, pp 483
490.
Erbe, H. H. (2000). Technologies for Cost Effective Automation in
manufacturing (Low Cost Automation). Center for Human-Machine
Systems, Technische Universitat, Berlin, and Germany, IFAC
Professional Brief, Berlin, 1-32.
Erbe, H. H. (2003). Manufacturing With Low Cost Automation IFAC
Automatic Systems for Building the Infrastructure in Developing
Countries, Elsevier IFAC Publications, Istanbul, Republic of Turkey,
95-100.
Esposito, A. (1994). Fluid Power with Applications, Prentice-Hall,

Englewood Cliffs, NJ, 3rd ed, 1-589.
Freund, E., A. Hypki, F. Heinze and R. Bauer (2001). COSIMIR PLC - 3D
simulation of PLC programs. In: Proceedings of the 6th IFAC
Symposium on Cost Oriented Automation, Berlin.
Gajmal. S. S. and Bhatwadekar. S. G., (2014). Low Cost Automation
(LCA): A Case Study, IJAGET, 2 (12), 172-181.
Hohwieler, E. and Berger, R. (2002). New Electronic Service for the
Production Stor, In: Proceedings of the 6'h IFAC Symp. On Cost
Oriented Automation, Elsevier Science Ltd. Oxford, pp. 237-241.
Inagaki T. (2000). Automation and the cost of authority. Institute of
Information Science, center for Taka, University of Tsukuba, Tsukuba,
Japan, 8305-8573.
Lay, G. (2002). Is High Automation a Dead End? Cutbacks in Production
Over engineering in the German Industry, In: Proceedings of the
6thIFACSymp. Cost Oriented Automation, Elsevier Science Ltd.
Oxford, pp. 225-230.
Legierski, T., and Wyrwa, J., (1998). PLC Programming, Jaska
Skalmierski’s Publishing Company, Gliwice (in polish).
Ollero, A. (2002). Low Cost Automation in field Robotics In: Proceeding
of the 6th IFAC Symp. Cost Oriented Automation, Elsevier Science
Ltd. Oxford.
Seifermann. S., Böllhoff. J., Metternich. J., and Bellaghnach. A. (2014).
Evaluation of Work Measurement Concepts for a Cellular
Manufacturing Reference Line to enable Low Cost Automation for
Lean Machining, Procedia CIRP 17, 588 – 593.
Soloman, S. (1996). Affordable Automation, McGraw-Hill, New York.
Sullivan, J. A. (1989). Fluid Power Theory and Application, Prentice-Hall,
Englewood Cliffs, NJ, 3rd ed.
Swider, J. (2002). Control and Automation of Technological Process and
Mechatronics Systems, Silesian University Publishing Company,
Gliwice. pp. 546.
Swider, J. and Wszoek, G. (2003). The methodical collection of
labouratory and project tasks of technological process control.
Pneumatic and electropneumatic systems with logical PLC control,

Silesian University Publishing Company, Gliwice, pp. 270.
Swider, J. and Wszoek, G. (2004). Analysis of complex mechanical
systems based on the block diagrams and the matrix hybrid graphs
method, Journal of Materials Processing Technology, Elsevier, the
Netherlands, 157-158.
Chapter 9
ESTIMATION OF AXIAL FORCE IN

INCREMENTAL SHEET METAL FORMING
Bohra Murtaza and Y. V. D. Rao

Birla Institute of Technology and Science Pilani, Hyderabad, India
ABSTRACT
In sheet metal incremental forming, the sheet blank edge rigidly
clamped deforms using a spherical tool, assisted by a CNC controlled
incremental tool movement. Limitation in Single Point Incremental
Forming (SPIF) is that it is limited to the production of parts with wall
angle between 70-80 degree. The formability of a metal sheet differs
according to the direction of the tool movement because of the plane
anisotropy. The formability of a sheet metal is better accounted for using
Forming Limit Diagrams. A considerable amount of knowledge about the
forming forces is required for the preservation of tooling and machinery.
A dynamical model is to be developed and using numerical techniques
and FEM simulations, an approximated value of axial force will be
computed. Based on these results a decision is to be made regarding a
servo-controlled motor for the robot used for base rotation.

Corresponding Author Email: murtazabohra17@gmail.com.
292 Bohra Murtaza and Y. V. D. Rao
Keywords: incremental sheet forming, CNC technology, servo controlled,

forming limit diagrams, axial force
1. INTRODUCTION
Metal Forming is a manufacturing process which involves plastic

deformation of metal to change the shape of the sheet. In metal forming a
tool applies stress on the sheet which exceeds the yield stress limit and the
metal takes a shape determined by the shape of the die. Stresses applied to
plastically deform the metal sheet can be of any type like Compressive,
Tensile, and Shear. Some of the examples are rolling, extrusion, die
forming, forging, stretching, deep drawing, spinning, etc. Our main focus
on this paper is Single Point Incremental Forming (SPIF). The incremental
forming process is originated from stretch forming and metal spinning
process. The SPIF have the combined advantage of both Stretching and
Spinning. In SPIF a single point tool is made to move on a metal sheet.
Recently, research has shown that it is also possible to produce low-cost,
small-batch, polymer sheet components by means of single point
incremental forming at room temperature (Martins et al., 2009) (Yonan
et al., 2014). The tool generally moves in a spiral way either inwards or
outwards in the plane of the sheet taking small incremental steps in the
downward direction forming a 3-D dimensional shape from a 2-D sheet.
In SPIF the stresses are localized near the point of contact of the tool
with sheet surface. This increases the formability of the process also the
quality of the metal sheet after forming is better than the conventionally
formed sheets, therefore, ISF is primarily used in Aerospace and
Automobile industry where high-quality surface finish and precision is
required.
SPIF uses modern CNC technology to produce complex 3D shapes
although it produces complex geometries it is not suitable for mass
production because it is a relatively slow process. However, the limitation
in SPIF is, it can produce parts with maximum wall angle between 70-80
degree (Tisza, 2012) (Crina, 2011). To overcome this problem another
Estimation of Axial Force in Incremental Sheet Metal Forming 293
technique known as Multi-Stage Single Point Incremental Forming

(MSPIF) is used. MSPIF was proposed to overcome the abovementioned
difficulty of conventional SPIF related to the production of complex sheet
metal parts with vertical walls (Skjødt et al., 2010). MSPIF is donemainly
by alternating or sometimes repeating the tool movement from upwards to
downwards, and the wall angle is increased gradually in consecutive
stages. The number of intermediate stages is kept minimum to avoid
significant surface wear and to limit the overall forming time. The risk of
wrinkling is said to increase with the decrease in the intermediate number
of stages (Hirt et al., 2004) (Bambach et al., 2004).
2. THE BASIC SETUP FOR SPIF
In SPIF, the sheet blank’s edges are usually rigidly clamped and the
sheet is then deformed usually by a hemispherical tool which moves as per
the numerical computer program feed, following specific contours which
result in the required shape of the sheet blank. Incremental sheet forming
can be performed on any universal milling machine which is having at
least 3-axis CNC control system. Thus, the basic elements of incremental
forming processes are the material to be formed, a sheet blank holder
clamping the blank, universal forming tool and the forming machine with
the CNC control.
Figure 1. Basic Setup for SPIF showing all the Perimeters.

A basic setup of SPIF is shown in Figure 1, here 𝜓 = Wall Angle, and

Δz = Incremental step height in the downward direction. It’s known from
the theory of spinning that the thickness of the part changes following the
sine-law.
tf = tO𝑆𝑖𝑛(𝛼) = tO𝑆𝑖𝑛(90 − 𝛼)= tO𝐶𝑜𝑠(𝜓), here tO = Initial thickness of

sheet, tf = final thickness of sheet
3. TYPES OF ASYMMETRIC INCREMENTAL SHEET

METAL FORMING TECHNIQUE
 Single point incremental forming - It uses a single point tool to

form the elements into desired 3-D shapes as shown in Figure 2
(a).
 Multi-point Incremental sheet forming - It uses two single point
tools one main which is in contact with the metal sheet on one side
and another supporting tool which is in contact on the other side of
the metal sheet. Instead of supporting tool partial die or fully
developed dies are also used as a supporting structure, as shown in
Figure 2 (b).
 ISF using Hydraulic Fluid - In this type of incremental forming
one side of the metal sheet is in contact with the single point tool
and the other side is supported by pressurized fluid, as shown in
Figure 2 (c).
 Hybrid Incremental Forming - This type of ISF uses both stretches
forming as well as the single point incremental forming. First, the
metal sheet is stretched over a die and then a single point tool is
moved from giving it a perfect shape and for making grooves. This
type of technique is primarily used in making body panels for
Automobiles as shown in Figure 2 (d). (Khare and Pandagale,
2014) (Tisza, 2012).
(a) Single point incremental forming
(b) Multi-point Incremental sheet forming
(c) ISF using Hydraulic Fluid
(d) Hybrid Incremental Forming
Figure 2. Types of sheet metal forming techniques.

3.1 Clamping
Clamping is done to fix the metal sheet onto the blank holder. There
are two types of clamping methods used. In the first one, Solid or Rigid
clamping the sheet is fixed on the blank holder using nuts and bolts. And in
the second case, Ball clamping the sheet is clamped or fixed by forcing
free rotating balls along its perimeter. The ball clamping method is more
effective than solid clamping as it provides a relatively freer flow of the
sheet during the ISF process (Khare and Pandagale, 2014). There is a lot of
scope for research in ball clamping method.
3.2 Types of Tools Used in ISF
 Hemispherical Tool - With this tool, the contrasting point of the

tool has a hemispherical shape.
 Ballpoint tool - In this tool at contacting tip a hardened steel ball is
placed which is free to rotate just like the nib of a ballpoint pen.
Experiments show that spherical tool gives more depth and provides
better surface roughness when compared to a hemispherical tool, this is
because of its smooth rolling action over the blank surface without
rubbing. Kim and Park, 2002 found that the formability of SPIF process is
improved when a ball tool of a particular size is used with a small feed rate
and a little friction. The feed rate is the speed the forming tool with which
moves around the mill bed. Decreasing the feed rate also increases the time
of the process, therefore it should be wisely decided according to the shape
of the part and surface finish required. The formability also differs
according to the direction of the tool movement because of the plane
anisotropy.
Effect of Lubrication
Lubrication is very important in incremental sheet forming. It is used
to decrease the wear of tool and sheet. Also, lubrication helps in getting a
better surface finish and helps in dissipating heat generated at the contact
area between tool and metal sheet.
It is observed that spherical tool without lubrication gives better
formability because the friction at the tool and sheet interface increases the
tool pressure lowering state of stress in the sheet (Khare and Pandagale,
2014).
3.3 Forming Limit Diagram
All the metals have a maximum limit up to which it can be stretched or

deformed plastically. If a particular metal is stretched beyond its Ultimate
tensile strength necking will occur which will ultimately lead to tearing or
fracture of the part. Due to this reason, formability of a particular metal
part is restricted to a certain limit. To account for the formability of a Sheet
Metal, Forming Limit Diagrams (FLD) are widely used as an appropriate
solution. The amount of strain that a sheet metal can tolerate just before
localized necking is called as limit strain. Based on this principle, FLDs
represent limiting major and minor available principal strains in the plane
of the deformed sheet that can be achieved. Due to this FLDs becomes a
valuable tool for analyzing sheet metal forming. FLDs were initially
introduced by
Keeler (Keeler, 1968) and Goodwin (Goodwin, 1968) popularly
known as Keeler-Goodwin diagram. FLDs represent comprehensively
sheet metal formability and has been used widely as one of the criteria for
optimizing Forming processes, in the designing of dies and selection of
tools for forming. To get FLDs basic strategy used is that a grid of circles
is etched on the surface of a sheet metal. The sheet metal is then stretched
and subjected to deformation. The circles get deformed into elliptic shapes
Figure 3. The strain along two principal directions could be expressed as
the percentage change in length of the major and minor axes. Negative
minor strain results when minor axis contracts i.e., if Minor axis length is
less than original circle’s radius. The strains measured near necks or
fracture are called failure or limiting strain. A plot of the major strain
versus minor strain is then made. This plot is called Keeler-Goodwin

forming limit diagram Figure 4. This plot gives limiting strains
corresponding to safe deformations. The FLD is thus a plot of the
combinations of major and minor strains which lead to fracture. Any
combination of strains which lies in the region above the limiting curves in
the Keeler-Goodwin diagram represents failure, while those in the region
below the curves represent safe deformations. Such diagrams indicate both
of the principal strains ε1 and ε2 at diffuse or localized instability in the
plane-stress state for different strain paths. Factors which influence the
property of FLDs are strain hardening exponent, anisotropy constant
(Sowerby and Duncan, 1971), tool geometry, friction effect, strain rate
sensitivity (Ghosh and Hecker, 1974), grain size (Stachowicz, 1989) and
strain path changes(Graf and Hosford, 1994) all these factors affect the
property of FLDs.
Figure 3. Geometrical transformation of circle to ellipse due to stretching of sheet.
Forming limit analysis plays a major role in SPIF as it helps in

deciding or choosing a particular metal for sheet, type of tool and design of
the part. FLDs in ISF can also be useful in separating safe and unsafe
areas. Extensive research has shown that FLD for ISF differs a lot from
FLDs of conventional forming (Tisza, 2012) although the method to obtain
FLD remains same. Figure 5 shows that the FLD for SPIF lies above the
FLD obtained for conventional forming. This shows the higher formability
of SPIF compare to conventional forming. The main reason for the higher
formability of SPIF is because the strain here is localized near the forming
tool.
Figure 4. FLD for conventional sheet metal forming.
Figure 5. FLC for conventional and incremental forming (Tisza, 2012).

4. FORCE PREDICTION IN SPIF DEDUCED FROM

EXPERIMENTAL AND FINITE ELEMENT ANALYSIS
During ISF it is absolutely necessary to have a considerable amount of

knowledge about forming forces for the preservation of tooling and
machinery. Also, it is primitively important, especially when using a
robotic arm as a forming platform, because the end effector of a robotic
arm is not a stiff structure and due to forming forces it deflects. This causes
undesirable deviations in the tool path which affects the end results
inducing errors in the geometry of the achieved parts. One of the solutions
to this problem is by including a compensation for the deflection to be
expected in the tool path.
In the literature (Aerens et al., 2009), a study was done to establish
formulae allowing to predict the forces occurring during the SPIF process.
During experiments, forces were measured in steady state and truncated
cones were formed. A model was generated partially obtained through
experiments and regression techniques and partially through numerical
techniques using FEM simulations allowing to compute an approximated
value of axial force Fz as a function of tensile strength Rm of material, sheet
thickness t, wall angle ψ, tool diameter dt, and scallop height Δh.
Fz= 0.0716Rm t1.57 dt0.41 Δh0.09 ψCos(ψ),
here Fz is expressed in N, Rm in N/mm2, t in mm, dt in mm, Δh in mm, and

ψ in deg.
CONCLUSION
Forming limit analysis plays a major role in SPIF as it helps in

deciding or choosing a particular metal for sheet, type of tool and design of
the part. For the preservation of tooling and machinery, a good knowledge
of forming forces is essential. In addition, it is of utmost importance when
using a robotic arm as a forming platform. This is so because the end

effector of a robot is not a stiff structure and gets deflected under forming
forces. Using experimental results, a model is developed. With the help of
regression techniques and numerical techniques using FEM simulations an
approximated value of axial force Fz as a function of Tensile strength Rm of
material, sheet thickness t, wall angle ψ, tool diameter dt, and scallop
height Δh is computed. Using the results, a decision is to be made
regarding a servo controlled motor to give the base rotation on which blank
is clamped.
REFERENCES
Aerens R, Eyckens P, Bael A Van and Duflou JR. Force prediction for
single point incremental forming deduced from experimental and FEM
observations. 2009, Int J Adv Manuf Technol doi: 10.1007/s00170-
009-2160-2.
Bambach M, Hirt G, and Ames J, Modelling of optimization strategies in
the incremental CNC sheet metal forming process. In Materials
processing and design: modeling, simulation and applications, 2004.
Proceedings of the Eighth International Conference on Numerical
methods in industrial forming processes (Numiform 2004), AIP
Conference Proceedings, Vol. 712, Columbus, Ohio, USA, 13–17 June
2004, pp. 1969–1974 (American Institute of Physics, Melville, New
York).
Crina, Radu, Determination of the Maximum Forming Angle of some
Carbon Steel Metal Sheets, 2011, Journal of Engineering Studies and
Research – Volume 17 No. 3.
Ghosh AK and Hecker SS, Stretching limits in sheet metals: in-plane
versus out-of-plane deformation. 1974, Metall. Trans. A 5A (1974)
1607– 1616.
Goodwin GM, Application of strain analysis to sheet metal forming
problems. 1968, Metall. Ital. 60 (1968) 767–771.
Graf A and Hosford W, The influence of strain-path changes on forming

limit diagrams of Al 6111 T4, Int. J. Mech. Sci. 36 (1994) 897–910.
Hirt, G., Ames, J., Bambach, M., and Kopp, R. Forming strategies and
process modelling for CNC incremental sheet forming. 2004, CIRP
Ann. Mfg. Technol., 52(1), 203–206.
Keeler SP, Circular grid systems: A valuable aid for evaluating sheet
metal formability, 1968SAE Technical paper 680092 (1968) 633–640.
Khare Unmesh and Pandagale Martand, A Review of Fundamentals and
Advancement in Incremental Sheet Metal forming. 2014, International
Conference on Advances in Engineering & Technology (ICAET-2014).
IOSR Journal of Mechanical and Civil Engineering (IOSR-JMCE) e-
ISSN: 2278-1684, p-ISSN: 2320-334X PP 42-46.
Kim YH and Park JJ Effect of process parameters on formability in
incremental forming of sheet metal. 2002, J Mat Process Technol
130:42–46.
Martins, PAF, Kwiatkowski, L, Franzen, V, Tekkaya, AE and Kleiner, M
Single point incremental forming of polymers, CIRP Annals -
Manufacturing Technology 58 (2009) 229–232.
Paul SK Theoretical analysis of strain- and stress-based forming limit
diagrams.2013, J Strain Analysis, 48(3) 177–188.
Skjødt, M, Silva, MB, Martins, PAF, and Bay, N. (2010), Strategies and
limits in multi-stage single-point incremental forming. Journal of
Strain Analysis for Engineering Design, 45(1), 33-44. DOI:
10.1243/03093247JSA574.
Sowerby R and Duncan JL, Failure in sheet metal in biaxial tension. 1971,
Int. J. Mech. Sci. 13 (1971) 217–229.
Stachowicz F, Effects of microstructure on the mechanical properties and
limit strains in uniaxial and biaxial stretching. 1989, J. Mech. Work.
Technol. 19 (1989) 305–317.
Tisza, M, General overview of sheet incremental forming, 2012, JAMME,
Issue 1 Volume 55.
Yonan, S Alkas, Silva, MB, Martins, PAF and Tekkaya, AE, Plastic flow
and failure in single point incremental forming of PVC sheets, (2014),
eXPRESS Polymer Letters Vol.8, No.5 301–311.
ABOUT THE EDITOR
Sandip A. Kale is working as Associate Professor in Mechanical

Engineering Department at Trinity College of Engineering and Research,
Pune affiliated to Savitribai Phule University Pune (Previous Pune
University), India. Additionally, he is handling the responsibility of Dean,
Research and PhD Coordinator in the institute. He is also driving
Technology Research and Innovation Centre, Pune, India. He has
completed his PhD and Master’s Degree from Pune University. He has
more than eighteen years industrial, academic and research experience. He
has published more than fifty research articles in various conferences,
books and journals. He has worked as Guest Editor for publishing many
journal special issues. He has filed eleven patents. He has worked as
304 About the Editor
Conference Chair for five international conferences and one national

conference. He is working as Editorial Board member and reviewer for
many international journals.
INDEX
bending, 63, 176, 218, 219, 220, 226, 227,

A 237, 250
binary joints, 154
adjacency matrix, v, 153, 157, 158, 161, 162
binding energies, 179
algorithm, viii, 17, 18, 20, 23, 41, 42, 46,
bioceramics, vi, viii, 175, 176, 177, 182,
47, 59, 69, 75, 84, 100, 101, 116, 119,
239, 249, 255, 257
121, 126, 128, 131, 132, 133, 153, 163
biomaterials, viii, 175, 248, 249, 250, 252,
aluminum oxide, 188
253, 255, 257, 259, 260, 261, 262
amplitude, 111, 166
biomedical applications, 244, 249, 252
anisotropy, 291, 296, 298
biomolecules, 59, 63, 108, 112, 121, 122,
Arrhenius dependence, 48
124, 125, 126
atoms, 22, 23, 26, 30, 32, 33, 36, 37, 39, 45,
biopolymers, 133
47, 48, 50, 53, 79, 82, 109, 112, 178,
biotechnology, 178
179, 210, 228
Boltzmann constant, 20, 32, 65
automation, ix, 263, 264, 265, 266, 279, 281
Boltzmann distribution, 87, 94, 105
automobiles, 148, 292
Boolean algebra, 158, 164
axial force, vi, vii, ix, 1, 6, 14, 291, 292,
break-even point, 283, 285
300, 301
C
B
carbon nanotube, 177, 179, 183
barriers, 58, 92, 102, 103, 108, 109
carboxylic acids, 245
beam-column joint, vi, viii, 165, 166, 167,
casting, 198, 200, 202, 206, 269
168, 169, 171, 173, 174
ceramic, viii, 175, 186, 187, 188, 190, 191,
beams, 1, 166, 174, 226
192, 193, 198, 199, 205, 206, 207, 208,
bedding, 209
306 Index
211, 212, 214, 216, 219, 220, 221, 223, crystals, 17, 19, 20, 22, 23, 29, 30, 33, 34,
224, 225, 227, 229, 231, 232, 233, 234, 35, 40, 41, 42, 43, 45, 46, 47, 51, 52, 53,
235, 236, 237, 239, 240, 242, 248, 249, 54, 55, 115, 211, 215, 219, 247, 250
250, 251, 252, 254, 255, 256, 257, 258,
259, 260, 261, 262
D
ceramic bulk, vi, 175
ceramic materials, 186, 220, 240, 250, 251,
decomposition, 28, 30, 187, 205, 215, 216,
252
243
Chebyshev approximation theory, viii, 136
defect engineering, 18, 51, 52, 55
chute, 266, 267, 270, 274, 275, 279
defect formation, 20, 21, 40, 41, 42, 43, 44,
clay, 186, 187, 190, 191, 192, 193, 195,
46, 50, 51, 52, 55
198, 199, 200, 201, 202, 203, 204, 205,
defective structure, v, vii, 17, 18, 20, 21, 34,
206, 207, 214, 215, 216, 218, 224, 229,
35, 42
250, 253, 254, 256, 259, 261
defects, 19, 23, 35, 43, 50, 51, 52, 56, 219
CNC technology, 292
deformation, 4, 36, 37, 38, 39, 171, 194,
coal, 191, 220, 260
196, 197, 297, 301
coatings, 243, 252, 257
differential equations, 23, 35, 63
collet, 266, 267, 268, 269, 270, 276, 277,
diffusion, 17, 19, 22, 23, 26, 27, 36, 39, 40,
281
41, 42, 48, 49, 53, 56, 100, 115
computational algorithm, 17, 18, 23, 47
directional control valve, 264, 271
computer simulations, 69, 104
dislocation, 19, 20, 21, 22, 26, 31, 34, 35,
computer technology, 18, 47
36, 37, 38, 39, 40, 41, 43, 44, 46, 47, 48,
configuration, 6, 48, 50, 59, 64, 70, 80, 82,
51, 52, 54, 55
84, 86, 91, 93, 95, 105, 109, 110, 111,
displacement, viii, 2, 36, 37, 153, 159, 166,
128, 179
171, 210, 222
convergence, 73, 74, 98, 101, 102, 108
displacement isomorphism, viii, 153
cooling, 19, 20, 21, 22, 23, 30, 33, 34, 35,
distributed memory computer, 125
39, 40, 45, 46, 50, 54, 212, 223, 226, 233
distribution, 11, 20, 21, 22, 26, 30, 31, 32,
correlation function, 99, 105
33, 34, 35, 43, 44, 45, 64, 70, 72, 73, 76,
cost, ix, 263, 264, 265, 266, 281, 282, 283,
77, 81, 82, 84, 87, 89, 90, 94, 95, 100,
284, 285, 286, 287, 289, 292
104, 105, 180, 181, 198, 199, 204, 208,
cost reduction, 264, 265, 266
244, 245
Coulomb interaction, 47
distribution function, 20, 26, 30, 31
crystal growth, 19, 20, 32, 34, 39, 42, 43,
DNA, 57, 58
45, 52, 55
crystal structure, 42, 43, 210
crystalline, 33, 51, 53, 54, 209, 215, 228, E
248
crystallites, 198 ejector, 266, 268, 269, 270, 276, 277, 279,
crystallization, 21, 22, 23, 26, 33, 34, 35, 281
220, 243, 247, 251, 260 elastic deformation, 36, 40
Index 307
energy, 20, 38, 53, 58, 65, 69, 70, 72, 73, forming forces, 291, 300
74, 76, 77, 80, 81, 82, 83, 86, 87, 88, 89, forming limit diagrams, 291, 292, 297, 302
90, 91, 92, 93, 95, 96, 97, 99, 100, 101, four precision point synthesis, 138
102, 103, 104, 107, 108, 109, 113, 115, fracture toughness, 240, 249
118, 120, 122, 124, 125, 127, 128, 130, free energy, 48, 58, 96, 99, 100, 101, 102,
132, 133, 167, 178, 179, 189, 190, 220, 108, 109, 114, 115, 128, 130, 248
228 freedom, 1, 59, 60, 61, 63, 65, 66, 67, 68,
energy consumption, 220 71, 80, 81, 83, 92, 109, 131, 154, 156
energy density, 87, 90, 93 Freudenstein equation, 136, 138, 141
energy supply, 190 Freudenstein-Chebyshev approximation
engineering, vii, ix, 18, 41, 51, 52, 55, 165, method, 135
182, 190, 191, 250, 251, 252, 260, 289 friction, 168, 296, 297, 298
entropy, 47, 48, 49, 50, 53, 71, 76, 91, 94, fuel cell, 185
128 function generation, 135, 136, 137, 148,
epicyclic gear train, v, viii, 153, 154, 155, 149, 150, 151
156, 163, 164 functional representation, 154
expansion, 4, 93, 176, 208, 209, 210, 215,
216, 219, 229, 233, 234, 235, 238
G
F graphical representation, 154, 155

grown-in microdefects, 18, 19, 22, 26, 30,
finite element analysis, v, vii, 1, 2, 3, 8, 300 44, 46, 54, 55
firing, 176, 184, 186, 188, 190, 191, 192, growth, 19, 20, 21, 22, 23, 24, 25, 26, 27,
198, 201, 202, 205, 206, 208, 209, 210, 29, 30, 31, 32, 34, 35, 36, 37, 39, 40, 41,
214, 215, 217, 218, 219, 220, 221, 223, 42, 43, 44, 45, 46, 47, 52, 53, 54, 55, 185
226, 227, 229, 230, 231, 233, 235, 236, growth rate, 24, 43, 44, 45
238, 250, 253, 258
five precision point synthesis, 137, 140
H
flooding, viii, 58, 111, 112, 120
fluctuations, 58, 65, 110, 133
Hamiltonian, 64, 69, 84, 87, 99, 105, 109,
fluid, 63, 130, 196, 206, 221, 240, 294
111, 122
force, vii, ix, 1, 2, 3, 4, 5, 7, 9, 12, 14, 59,
hamming matrix, 158, 159, 161
60, 62, 64, 65, 67, 68, 71, 77, 78, 79, 89,
hamming number, 159
92, 112, 115, 127, 194, 200, 202, 291,
hamming procedure, 158
292, 300, 301
hamming string, 159, 160, 161
force transducers, 2, 4, 15
histogram, 69, 72, 73, 74, 75, 76, 87, 91, 96,
formability, 291, 292, 296, 297, 299, 302
97, 101, 117, 118, 119, 124, 130
formation, 17, 19, 20, 21, 22, 23, 26, 31, 33,
hybrid incremental forming, 294, 295
34, 36, 37, 39, 40, 41, 42, 43, 44, 45, 46,
48, 50, 52, 54, 55, 56, 129, 166, 209,
211, 215, 219, 249, 252, 257
308 Index
I M
incremental sheet forming, 292, 296, 302 acromolecular systems, 57, 58, 59, 109, 113
ions, 204, 243, 244, 245, 246 macromolecules, viii, 117, 120
iron, 207, 209 magnetic characteristics, 180
ISF using hydraulic fluid, 294, 295 magnitude, 1, 4, 13, 22, 36, 38, 63, 73, 77,
isomerization, 117 92, 107, 112
isomers, 159 manufacturing, vii, ix, 14, 178, 185, 190,
isomorphism, 153, 159, 163 191, 192, 220, 256, 264, 266, 269, 270,
isotope, 53 282, 283, 284, 288, 292
Markov chain, 116
materials, vii, viii, 18, 19, 20, 42, 51, 52, 53,
J
167, 175, 177, 180, 181, 182, 183, 184,
185, 186, 187, 189, 190, 192, 206, 208,
joints, 154, 155, 165, 166, 167, 174
212, 216, 218, 220, 229, 231, 236, 240,
247, 249, 250, 251, 252, 254, 255, 260,
K 261, 281
materials science, viii, 18, 51, 175, 249, 260
kinematics synthesis, 136 matrix, 3, 26, 36, 37, 110, 111, 140, 153,
kinetic equations, 23 157, 158, 159, 161, 162, 219, 221, 290
kinetic model, 31, 39, 46, 54 MD simulation, 57, 58, 59, 63, 64, 70, 71,
kinetics, 23, 26, 28, 32, 36, 42, 52, 108, 118, 72, 73, 75, 76, 77, 88, 89, 90, 91, 95, 99,
133 102, 103, 104, 106, 108, 110, 112, 113
MD technique, 58
mechanical properties, 176, 235, 247, 248,
L
254, 261, 302
landscape, 58, 100, 104, 108, 109, 118, 127, melting, 21, 48, 205, 208, 210, 216, 217,
128, 132, 133 219, 221, 227, 229
lattice parameters, 37 microdefects, 18, 19, 20, 22, 23, 26, 30, 35,
least-square method, 135 41, 43, 44, 46, 54, 55
life cycle cost, vi, ix, 263, 264, 265, 281, model, 1, 3, 6, 8, 17, 18, 19, 22, 23, 26, 31,
287, 288 32, 33, 34, 35, 36, 38, 39, 40, 41, 42, 46,
linkage adjacency matrix, 157, 161 47, 48, 49, 50, 51, 54, 55, 84, 108, 111,
liquids, 31, 117, 126, 128, 204 112, 113, 115, 117, 122, 123, 125, 128,
loader, 266, 267, 270, 275, 276, 281 129, 131, 132, 133, 135, 166, 167, 168,
low cost automation, vi, ix, 263, 264, 265, 169, 170, 171, 172, 173, 174, 264, 291,
266, 288, 289 300, 301
low temperatures, 50, 53, 70, 213 modulus, 5, 26, 38, 194, 201, 219, 240, 246
moisture content, 196, 198, 199, 200, 201,
203
Index 309
molecular dynamics (MD), v, viii, 57, 58, physical and mechanical properties, 261
59, 61, 63, 64, 65, 67, 68, 69, 70, 71, 72, physical properties, 31, 51, 204, 228, 242
73, 75, 76, 77, 79, 81, 83, 85, 87, 88, 89, physicochemical properties, 243
90, 91, 93, 95, 96, 97, 99, 101, 102, 103, plastic deformation, 292
104, 105, 106, 107, 108, 109, 110, 111, plasticity, 167, 187, 194, 195, 196, 197,
112, 113, 114, 115, 116, 117, 118, 119, 198, 203, 204, 205, 208, 215
120, 121, 122, 123, 124, 125, 126, 127, point defects, 19, 21, 47, 48, 50, 51, 53, 54
128, 129, 130, 131, 132, 133 polymerization, 242, 243, 249
molecular nanotechnology, 178 polymers, 113, 185, 302
Monte Carlo method, 120, 124 polypeptide, 125, 131
porosity, 176, 213, 221, 223, 225, 235, 237,
245, 250
N
precipitation, 19, 20, 22, 26, 31, 33, 34, 35,
36, 40, 42, 46, 47, 50, 51, 54, 209, 236
nano-Al2O3, 175, 183
precision points, viii, 135, 136, 145, 146,
nanoceramics, viii, 175, 184, 185, 187, 188,
148, 150
189, 190, 259, 261
probability, 70, 72, 73, 74, 77, 83, 84, 86,
nanomaterials, vi, viii, 175, 176, 177, 181,
87, 89, 90, 92, 93, 94, 95, 96, 102, 103,
182, 183, 184, 187, 249, 251
104, 105, 106, 117
nanometers, 177, 180, 184, 188
probability density function, 94, 95
nanoparticles, 175, 176, 180, 182, 184, 187,
probability distribution, 70, 72, 77, 84, 89,
188, 189, 250
93, 96, 102, 103, 104
nano-SiO2, 175, 183
production costs, 282, 283, 284
nanostructured materials, 180, 181
productivity, 263, 264, 265, 266, 281, 285,
nanostructures, 179, 182, 184, 188
286
nanotechnology, 175, 177, 178, 249, 251,
programmable logic controller, ix, 264, 279
253, 259, 260
programming, viii, 47, 51, 279, 281, 287
programming languages, 47
O properties of ceramics, 189, 261
protein(s), 57, 58, 68, 76, 78, 84, 85, 103,
optimization, 3, 15, 23, 92, 114, 120, 136, 106, 109, 112, 113, 114, 115, 119, 122,
137, 142, 143, 144, 145, 146, 148, 150, 123, 124, 126, 127, 128, 129, 130, 131,
301 132, 133
oxygen, 22, 23, 24, 26, 30, 31, 33, 34, 40, protein folding, 58, 76, 78, 85, 103, 122,
54, 216, 228 128, 130, 131, 133
pusher, 266, 267, 268, 270, 275, 276, 279,
281
P
peptide, 108, 115, 121, 123, 124, 126, 133 Q

phase transformation, 32
phase transitions, 32, 116 quality products, 266
310 Index
quartz, 192, 193, 207, 208, 209, 210, 211, 83, 87, 88, 89, 90, 91, 95, 96, 99, 100,
212, 214, 215, 216, 219, 248, 250, 254 101, 102, 103, 104, 105, 106, 108, 112,
115, 116, 118, 119, 121, 122, 123, 126,
127, 128, 129, 132, 133, 165, 167, 263,
R
264, 273, 274, 275, 276, 277, 278, 287,
288, 289, 301
random walk, 70, 73, 74, 75, 119, 132
single crystals, 20, 21, 22, 29, 31, 34, 39,
raw materials, 175, 190, 191, 193, 207, 218,
41, 43, 44, 46, 47, 48, 54, 55
220, 223, 224, 236, 249, 256, 260
single point incremental forming, 291, 292,
recombination, 21, 35, 36, 47, 48, 49, 50,
293, 294, 301, 302
53, 54
sintering, 176, 187, 188, 189, 220, 231, 247,
reduction of time, 266
248, 255, 258, 262
rheological and electromagnetic properties,
SiO2, 34, 175, 176, 182, 183, 185, 187, 209,
177
210, 213, 215, 218, 219, 220, 224, 230,
rotation graph, 155, 156, 157, 160, 163
236, 244, 247, 248, 252, 257, 258
rotational analysis, 159
sol-gel, 187, 188, 189, 242, 243, 244, 245,
rotational isomorphism, viii, 153, 163
248, 252, 255, 259, 260
SOL-GEL techniques and applications, 242
S solid state, viii, 18, 31, 34, 51, 55
special purpose machine, vi, ix, 263, 264,
seismic analysis, 166 266, 269, 284
semiconductor(s), 17, 19, 35, 40, 42, 51, 52, special purpose turning machine, 266, 269,
54, 55, 182, 185 286, 287
servo controlled, 292, 301 spindle collet, 267, 268
shear, 26, 37, 38, 166, 167, 169, 172, 174, strain gauges, vii, 1, 2, 3, 4, 5
194 strengthened beam-column joint, 166
sheet metal forming, vi, ix, 291, 295, 297,
299, 301
shrinkage, 176, 190, 198, 199, 200, 201, T
202, 205, 206, 208, 215, 217, 220, 238,
techniques, vii, 64, 69, 90, 125, 126, 137,
245, 246, 250, 253
187, 191, 242, 249, 287, 291, 295, 300,
sigmoidal function generation, v, 135, 137,
301
148
technology(ies), vii, 2, 18, 20, 41, 42, 51,
silica, 182, 208, 209, 210, 220, 233, 235,
52, 55, 178, 190, 252, 259, 265, 292
253, 259, 261
thermal expansion, 176, 208, 209, 210, 215,
silicon, vii, 17, 19, 20, 21, 22, 23, 24, 26,
219, 229, 233, 234
29, 30, 31, 33, 34, 37, 39, 40, 41, 42, 43,
transducer, vii, 1, 2, 4, 5, 7, 9, 10, 12, 14
44, 45, 46, 47, 50, 51, 52, 53, 54, 55, 56,
transfer vertex, v, viii, 153, 156, 160, 161,
228
163
simulation(s), v, vi, viii, ix, 9, 18, 57, 58,
tri-axial loadings, 1
59, 63, 69, 70, 71, 72, 73, 74, 75, 76, 78,
Index 311
water absorption, 176, 190, 217, 218, 220,

V 221, 223, 250
water vapor, 216
vacancies, 22, 23, 24, 26, 30, 39, 45, 48, 50
valve, 263, 269, 270, 271, 276
variables, 81, 86, 87, 101, 140 X
variations, 19, 35, 198
vector, 38, 62, 80, 82, 100, 109 X-ray diffraction (XRD), 228, 242, 247, 248
velocity, 63, 67
viscosity, 213, 227, 245
Y
W Yang-Mills, 32
Wales, 104, 118

Z
waste, 182, 187, 190, 205, 220, 223, 227,
233, 236, 237, 238, 239, 254, 255, 256,
ZnO, 230
258, 261

Desen Produto

Uploaded by

Document Information

Original Title

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

Desen Produto

Uploaded by

Copyright:

Available Formats

MECHANICAL ENGINEERING THEORY AND APPLICATIONS

Additional books and e-books in this series can be found on Nova’s

NOTICE TO THE READER

Library of Congress Cataloging-in-Publication Data

Published by Nova Science Publishers, Inc. † New York

Chapter 6 3-D Simulation Studies on Strengthening of

Design and manufacturing processes are the important aspects of all

process of solid state physics, material science, programming and the

The manufacturing industry is always focusing to reduce the

FINITE ELEMENT ANALYSIS OF THREE DOF

Ankur Jaiswal, H. P. Jawale* and Atul Sane

Strain Gauge Force transducers are the instruments deployed to

Keywords: force transducers, finite element analysis, strain gauges

The static and kinematic equations are developed through a

calibration and testing of three axis force Sensor is presented in

2. STRAIN GAUGE BASED FORCE SENSING

Force Transducers are devices which provide an observer the

The strain gauge cell is based on an elastic element to which a number

3.1 Converting of Cartesian Coordinates

The spherical coordinates of a point (r, θ, φ) can be obtained from its

Figure 3. Spherical Co-ordinate System.

Conversely, the Cartesian coordinates may be retrieved from the

x  r sin  sin  (6)

z  r sin  cos  (8)

4. FINITE ELEMENT ANALYSIS

4.1 Material Properties

Table 1. Material Properties of Structural Steel

Young’s Modulus 200 GPa

A ten node quadratic tetrahedron element C3D10 is used for meshing

4.3 Loading Conditions

The base of the transducer is made fixed support in order to replicate

(a) FT1 (b) FT2

(a) FT1 (b) FT2

Figure 5. Loading conditions on FT1 and FT2.

The stresses and strains in the members of the transducer were

Table 2. Stress values from ABAQUS for Maximum Principal Elastic

load (N) Maximum Principal Strain (*e^-5) for FT1

Figure 6. (a-f) Principal Strain distribution for TF1 and TF2.

Table 3. Stress values from ABAQUS for Maximum Principal Elastic

load (N) Maximum Principal Strain (*e^-5) for FT2

1. When load is applied in any one direction (uni-axial loading),

4. The difference between induced strain in FT1 and FT2 also

14 Force in X- direction (FT2)

Force in Y- direction (FT1)

Figure 7. Maximum Principal Strain V/s Applied Load.

Figure 8. Strain difference curve between TF1 and TF2.

Estimation of strains in a compact force transducer for application of

Bayo, Eduardo, and JR Stubbe. (1989), Six‐Axis Force Sensor Evaluation

Interference.” Paper presented at the Robotics and Automation, 1990.

THE BASIC PRINCIPLE OF CALCULATION

V. I. Talanin, I. E. Talanin, V. V. Zhdanova,

Keywords: grown-in microdefects, defective structure, model,

Physics, and in particular solid-state physics, is the basis for

Semiconductor silicon, due to the requirements imposed on it, is one of

secondary microdefects. The interrelation of these mathematical models

2. HIGH TEMPERATURE IMPURITY PRECIPITATION

Despite numerous experimental proofs of the existence of impurity

where 𝑘 is a Boltzmann constant; ∆𝐺 is a minimum energy that must be

2016a). There was an idea to introduce in the theoretical calculations the

1⁄𝑇 = 1⁄𝑇𝑚 + 𝐺(𝑟)𝑧⁄𝑇𝑚2

where 𝑧 is the distance from the crystallization front, 𝑇𝑚 is the melting

The solution of the second problem required to abandon the

residual impurities of oxygen and carbon is one to three orders of