Advances in Nanofibre Research Akbar K. Haghi Gennady Zaikov
iSmithers – A Smithers Group Company Shawbury, Shrewsbury...
64 downloads
782 Views
7MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Advances in Nanofibre Research Akbar K. Haghi Gennady Zaikov
iSmithers – A Smithers Group Company Shawbury, Shrewsbury, Shropshire, SY4 4NR, United Kingdom Telephone: +44 (0)1939 250383 Fax: +44 (0)1939 251118 http://www.ismithers.net
First Published in 2011 by
iSmithers Shawbury, Shrewsbury, Shropshire, SY4 4NR, UK
©2011, Smithers Rapra
All rights reserved. Except as permitted under current legislation no part of this publication may be photocopied, reproduced or distributed in any form or by any means or stored in a database or retrieval system, without the prior permission from the copyright holder.
A catalogue record for this book is available from the British Library.
Every effort has been made to contact copyright holders of any material reproduced within the text and the authors and publishers apologise if any have been overlooked.
ISBN: 978-1-84735-603-1 (Hardback) 978-1-84735-604-8 (Softback) 978-1-84735-605-5 (ebook) Cover image reproduced with permission from Juan P. Hinestroza, Cornell University. Source: http://nanotextiles.human.cornell.edu/
Typeset by Argil Services Indexed by Argil Services Printed and bound by Lightning Source
P
reface
Nanotechnology is revolutionising the world of materials. The research and development of nanofibres has gained much prominence in recent years due to the heightened awareness of its potential applications in the medical, engineering and defence fields. Among the most successful methods for producing nanofibres is the electrospinning process. Electrospinning introduces a new level of versatility and broader range of materials into the microfibre/nanofibre range. An old technology, electrospinning has been rediscovered, refined, and expanded into non-textile applications. Electrospinning has the unique ability to produce ultrathin fibres from a rich variety of materials that include polymers, inorganic or organic compounds and blends. With the enormous increase of research interest in electrospun nanofibres, there is a strong need for a comprehensive review of electrospinning in a systematic fashion. With the emergence of nanotechnology, researchers become more interested in studying the unique properties of nanoscale materials. Electrospinning, an electrostatic fibre fabrication technique has evinced more interest and attention in recent years due to its versatility and potential for applications in diverse fields. These notable applications include tissue engineering, biosensors, filtration, wound dressings, drug delivery, and enzyme immobilisation. The nanoscale fibres are generated by the application of a strong electric field on a polymer solution or melt. The non-woven nanofibrous mats produced by this technique mimic components of the extracellular matrix much more closely as compared with the conventional techniques. The sub-micron-range spun fibres produced by this process offer various advantages: high surface area-to-volume ratio, tunable porosity and the ability to manipulate nanofibre composition to obtain desired properties and functions. Over the years, >200 polymers have been electrospun for various applications, and the number is increasing gradually. Electrospinning is a highly versatile method to process solutions or melts (mainly of polymers) into continuous fibres with diameters ranging from a few micrometers to a few nanometers. This technique is applicable to virtually every soluble or fusible polymer. The polymers can be chemically modified and tailored with additives ranging from simple carbon-black particles to complex species such as enzymes, viruses, and bacteria. Electrospinning appears to be straightforward, but is an intricate process
iii
Advances in Nanofibre Research dependent upon a multitude of molecular, process, and technical parameters. The method provides access to entirely new materials which may have complex chemical structures. Electrospinning is not only a focus of intense academic investigation; the technique is being applied in many technological areas. This book presents some fascinating phenomena associated with the remarkable features of nanofibres in electrospinning processes and new progress in applications of electrospun nanofibres. It offers an overview of the structure–property relationships, synthesis and purification, and potential applications of electrospun nanofibres. The collection of topics in this book aims to reflect the diversity of recent advances in electrospun nanofibres with a broad perspective which may be useful for scientists as well as for graduate students and engineers. A.K. Haghi University of Guilan, Iran G.E. Zaikov Russian Academy of Sciences, Russia
iv
C
ontents
1
2
Electrospinning of Polymeric Nanofibres ..................................... 1 1.1
Introduction ...................................................................... 1
1.2
Processing Condition ......................................................... 2 1.2.1
Applied Voltage ...................................................... 2
1.2.2
Feed Rate ............................................................... 2
1.3
Theory and Modeling ........................................................ 4
1.4
Concluding Remarks ......................................................... 8
Polymeric Nanofibre Fabrication via Electrospinning Process .... 11 2.1
Introduction .................................................................... 11
2.2
Experimental ................................................................... 15
2.3
2.2.1
Solution Preparation and Electrospinning ............ 15
2.2.2
Choice of Parameters and Range .......................... 16
2.2.3
Experimental Design ............................................ 19
2.2.4
Response Surface Methodology ............................ 24
Results and Discussion .................................................... 26 2.3.1
2.3.2
Response Surfaces for Mean Fibre Diameter ........ 31 2.3.1.1
Solution Concentration.......................... 31
2.3.1.2
Spinning Distance .................................. 32
2.3.1.3
Applied Voltage ..................................... 33
2.3.1.4
Volume Flow Rate ................................. 34
Response Surfaces for Standard Deviation of Fibre Diameter ..................................................... 34 v
Advances in Nanofibre Research
2.4
3
2.3.2.2
Spinning Distance .................................. 36
2.3.2.3
Applied Voltage ..................................... 36
2.3.2.4
Volume Flow Rate ................................. 38
2.4.1
Mean Fibre Diameter ........................................... 39
2.4.2
Standard Deviation of Fibre Diameter .................. 39
Structure Formation of Polymeric Nanofibres in Electrospinning .......................................................................... 45 3.1
Introduction .................................................................... 45
3.2
Methodology ................................................................... 47 3.2.1
Simulation of Electrospun Webs ........................... 47
3.2.2
Fibre Diameter Measurement ............................... 48 3.2.2.1
Manual Method .................................... 48
3.2.2.2
Distance Transform ............................... 49
3.2.2.3
Direct Tracking...................................... 51
Real Webs Treatment ........................................... 52
3.3
Experimental ................................................................... 54
3.4
Results and Discussion .................................................... 54
3.5
Conclusion ...................................................................... 63
Optimisation of the Electrospinning Process .............................. 67 4.1
Introduction .................................................................... 67
4.2
Methodology ................................................................... 68 4.2.1
vi
Solution Concentration.......................... 34
Conclusion ...................................................................... 38
3.2.3
4
2.3.2.1
Measurement of Fibre Diameter ........................... 68 4.2.1.1
Manual Method .................................... 69
4.2.1.2
Distance Transform Method .................. 70
4.2.1.3
New Distance Transform Method ......... 74
Contents
5
6
4.2.2
Validation of the Methods .................................... 79
4.2.3
Thresholding ........................................................ 80
4.3
Experimental ................................................................... 82
4.4
Results and Discussion .................................................... 84
4.5
Conclusion ...................................................................... 91
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres ....................... 97 5.1
Introduction .................................................................... 97
5.2
Methodology ................................................................... 98 5.2.1
Sieving Methods ................................................. 101
5.2.2
Mercury Porosimetry ......................................... 101
5.2.3
Flow Porosimetry (Bubble Point Method) .......... 102
5.2.4
Image Analysis ................................................... 103 5.2.4.1
Real Webs ............................................ 104
5.2.4.2
Simulated Webs ................................... 105
5.3
Experimental ................................................................. 106
5.4
Results and Discussion .................................................. 108
5.5
Conclusion .................................................................... 118
Practical Hints on the Production of Electrospun Nanofibres from Regenerated Silk Fibroin ............................... 121 6.1
Introduction .................................................................. 121
6.2
Effect of Systematic Parameters on Electrospun Nanofibres ..................................................................... 122 6.2.1
Solution Properties ............................................. 122
6.2.2
Viscosity ............................................................. 122
6.2.3
Solution Concentration ...................................... 122
6.2.4
Molecular Weight ............................................... 123
vii
Advances in Nanofibre Research
6.3
6.4
6.5 7
6.2.5
Surface Tension .................................................. 123
6.2.6
Solution Conductivity ........................................ 123
6.2.7
Applied Voltage .................................................. 124
6.2.8
Feed Rate ........................................................... 124
Experimental ................................................................. 124 6.3.1
Electrospinning and Preparation of Nanofibrous Media ............................................ 124
6.3.2
Image Analysis using Image Processing Algorithms ......................................................... 125
Results and Discussion .................................................. 126 6.4.1
Diameter Distribution of Nanofibres .................. 126
6.4.2
Distribution of Nanofibre Orientation ............... 130
6.4.3
Porosity .............................................................. 130
Conclusions ................................................................... 130
Characterisation of Polymeric Electrospun Nanofibres ............ 133 7.1
Introduction .................................................................. 133 7.1.1
7.2
Electrospinning Setup ......................................... 135
Effect of Systematic Parameters on Electrospun Nanofibres ..................................................................... 139 7.2.1
Solution Properties ............................................. 139 7.2.1.1
Viscosity .............................................. 139
7.2.1.2. Solution Concentration........................ 140
7.2.2
viii
7.2.1.3
Molecular Weight ................................ 141
7.2.1.4
Surface Tension ................................... 141
7.2.1.5
Solution Conductivity .......................... 142
Processing Condition .......................................... 144 7.2.2.1
Applied Voltage ................................... 144
7.2.2.2
Feed Rate............................................. 144
Contents
8
9
7.3
Experimental ................................................................. 145
7.4
Result and Discussion .................................................... 147
7.5
Conclusion .................................................................... 150
Formation of Polymeric Electrospun Nanofibres ..................... 153 8.1
Overview ....................................................................... 153
8.2
Aim of the Project .......................................................... 153
8.3
Experimental ................................................................. 154
8.4
Results and Discussion .................................................. 155
8.5
Conclusion .................................................................... 163
Experimental Study on Electrospinning of Polymeric Nanofibres ............................................................................... 165 9.1
Introduction .................................................................. 165
9.2
Experimental ................................................................. 166
9.3
9.4
9.2.1
Materials ............................................................ 166
9.2.2
Sample Preparation ............................................ 167
9.2.3
Electrospinning .................................................. 167
9.2.4
Characterisation ................................................. 168
Results and Discussion .................................................. 169 9.3.1
Effect of Polyaniline Content .............................. 169
9.3.2
Effect of Electrospinning Temperature................ 172
9.3.3
Effect of Applied Voltage.................................... 177
Conclusions ................................................................... 178
Abbreviations .................................................................................... 181 Index ............................................................................................... 183
ix
Advances in Nanofibre Research
x
1
Electrospinning of Polymeric Nanofibres
1.1 Introduction Electrospinning [1, 2] is an economical and simple method used in the preparation of polymer fibres. The fibres prepared via this method typically have diameters much smaller than is possible to attain using standard mechanical fibre-spinning technologies [3]. Electrospinning of polymer solutions has gained much attention in the last few years as a cheap and straightforward method to produce nanofibres [4]. Electrospinning differs from the traditional wet/dry fibre-spinning in several ways, of which the most striking differences are the origin of the pulling force and the final fibre diameters. The mechanical pulling forces in the traditional industrial fibre-spinning processes lead to fibres in the micrometer range and are contrasted in electrospinning by electrical pulling forces that enable the production of nanofibres [5]. Depending on the solution properties, the throughput of single-jet electrospinning systems is ~10 ml/ min. This low fluid throughput may limit the industrial use of electrospinning. A stable cone-jet mode followed by the onset of the characteristic bending instability which eventually leads to great reductions in the jet diameter necessitate the low flow rate [6]. When the diameters of polymer fibre materials are shrunk from micrometers (e.g., 10–100 mm) to submicrons or nanometers, several amazing characteristics appear. These include a very large surface area-to-volume ratio (this ratio for a nanofibre can be as large as 103-times that of a microfibre), flexibility in surface functionalities, and superior mechanical performance (e.g., stiffness and tensile strength) compared with any other known form of the material. These outstanding properties make polymer nanofibres optimal candidates for many important applications [7]. These include filter media [8], composite materials [9], biomedical applications (tissue-engineering scaffolds) [10] bandages [11], drug-release systems [12]), protective clothing for the military [8], optoelectronic devices and semi-conductive materials [13] and biosensors/ chemosensors [14]. A schematic diagram to interpret electrospinning of polymer nanofibres is shown in Figure 1.1. There are basically three components to fulfill the process: a high-voltage supply, a capillary tube with a pipette or needle of small diameter, and a metal collecting screen [15–19].
1
Advances in Nanofibre Research
Syringe Metering pump
Connector 0
1
Electrical heater High voltage supply
25 ºC
Temperature controller
Figure 1.1 Electrospinning arrangement (schematic)
1.2 Processing Condition 1.2.1 Applied Voltage In general, an increase in the electrospinning current reflects an increase in the mass flow rate from the capillary tip to the grounded target when all other variables (conductivity, dielectric constant, and flow rate of solution to the capillary tip) are held constant [20, 21].
1.2.2 Feed Rate The morphological structure can be slightly altered by changing the solution flow rate (Figure 1.2). At a flow rate of 0.3 ml/h, a few big beads can be observed on the fibres. The flow rate can affect electrospinning process. A shift in the mass balance results in sustained (but unstable) jets and fibres with big beads [22].
2
Electrospinning of Polymeric Nanofibres
(a)
(b)
(c) Figure 1.2 Effect of flow rate of 7% polyvinyl alcohol water solution on fibre morphology (degree of hydrolysis = 98%, voltage = 8 kV, tip–target distance = 15 cm). Flow rate: (a) 0.1 ml/h; (b) 0.2 ml/h; and (c) 0.3 ml/h. Original magnification 10 k
3
Advances in Nanofibre Research
1.3 Theory and Modeling Typically, electrospinning has two stages. In the first, the polymer jet issues from a nozzle and thins steadily and smoothly downstream. In the second stage, the thin thread becomes unstable to non-axisymmetric instability and spirals violently in large loops. For the steady stretching in stage one, Spivak and Dzenis [23] published a simple model that assumes the electric field to be uniform and constant, unaffected by the charges carried by the jet. Hohman and co-workers [24, 25] developed a slender-body theory for electrospinning that couples jet stretching, charge transport, and the electric field. The model encounters difficulties, however, with the boundary condition at the nozzle. For stage two, bending instability has been carefully documented by two research teams: Reneker and co-workers [26, 27] and Shin and co-workers [28]. Each has proposed a theory for the instability. Hohman and co-workers [24] built an electrohydrodynamic instability theory and predicted that, under favorable conditions, nonaxisymmetric instability prevails over the familiar Rayleigh instability and varicose instability due to electric charges. The jet is governed by four steady-state equations representing the conservation of mass and electric charges, the linear momentum balance, and Coulomb’s law for the E field [25–32]. Mass conservation requires that
(1.1) where Q is a constant volume flow rate. Charge conservation may be expressed by
(1.2) where E is the z component of the electric field, K is the conductivity of the liquid, and I is the constant current in the jet. The momentum equation is formulated by Figure 1.3:
4
Electrospinning of Polymeric Nanofibres
Figure 1.3 Momentum balance on a short section of the jet
(1.3)
Where τzz is the axial viscous normal stress, p is the pressure, γ is the surface tension, e e and t t and t n are the tangential and normal tractions, respectively, on the surface of the jet due to electricity. The prime indicates a derivative with respect to z, and R´ is the slope of the jet surface. The ambient pressure has been set to zero. The electrostatic tractions are determined by the surface charge density and the electric field:
(1.4)
(1.5)
5
Advances in Nanofibre Research
where ε and ε are the dielectric constants of the jet and the ambient air, respectively, En and Et are the normal and tangential components of the electric field at the surface, and ║*║ indicates the jump of a quantity across the surface of the jet. We have used the jump conditions for En and Et: ║ ε En║= ε E - εEn=σ, ║Et║= E t Et = 0, and assumed that εEn<< ε E and Et ≈ E. The overbar indicates quantities in the surrounding air. The pressure p(z) is determined by the radial momentum balance. Applying the normal force balance at the jet surface leads to:
(1.6) Inserting Equations 1.4–1.6 into Equation 1.3 yields:
(1.7)
Fiber diameter (mm)
Figure 1.4 shows the relationship between mean fibre diameter and electric field with a concentration of 15% at spinning distances of 5, 7 and 10 cm.
T
25 ºC
250 200 150 100 50 0 Applied voltage (kv)
Figure 1.4 Mean fiber diameter of electrospun silk fibres at 10%, 12%, 14% and 25 ºC
6
Electrospinning of Polymeric Nanofibres The mean fibre diameter obtained at 2 kV/cm is larger than that obtained at other electric fields [30]. In general, the concentration has more effect on the fibre diameter than the electric field. The multiple regression analysis was carried out to evaluate the contribution of concentration and electric field on fibre diameter [30–32]. Sukigara and co-workers [31] used response surface methodology (RSM) analysis to the experimental results to develop a processing window which will produce nanoscale regenerated silk fibres by the electrospinning process. The steps in the procedure are described briefly as follows. 1. Identification of variables ζ1; ζ2; ζ3… for response η 2. Calculation of corresponding coded variables ًx1; x2; x3… by using the following equation.
(1.8)
where ζ Ai and ζ Bi refer to the high and low levels of the variables ζ I, respectively. 3. Determination of the empirical model by multiple regression analysis to generate theoretical responses (ŷ): The second-order model is widely used in RSM. The general equation for response h of the second-order model is given by:
(1.9) where k is the number of factors, xi are the coded variables and β are coefficients. 4. Calculation of the coefficients b to fit the experimental data as closely as possible. When k = 2; the empirical model from the general Equation 1.9 becomes:
(1.10)
7
Advances in Nanofibre Research
1.4 Concluding Remarks In recent years, nanotechnology has become a topic of great interest to scientists and engineers. It is now established as a prioritised research area in many countries. The reduction of size to the nanometre range brings an array of new possibilities in material properties, in particular with respect to achievable surface-to-volume ratios. Electrospinning of natural fibres is a novel process for producing superfine fibres by forcing a solution through a spinnerette with an electric field. A comprehensive review on this technique has been made in this contribution. Based on this review, many challenges exist in the electrospinning process of nanofibres, and several fundamental questions remain. The electrospinning technique provides an inexpensive and easy way to produce nanofibres on low basis weight, small fibre diameter, and small pore size. We hope that this chapter will pave the way toward a better understanding of the application of electrospinning of nanofibres.
References 1.
Y.Q. Wan, Q. Guo and N. Pan, International Journal of Nonlinear Sciences and Numerical Simulation, 2004, 5, 5.
2.
J.J. Feng, Journal of Non-Newtonian Fluid Mechanics, 2003, 116, 55.
3.
J. He, Y. Wan and J-Y. Yu, Polymer, 2005, 46, 2799.
4.
E. Zussman, A. Theron and AL. Yarin, Applied Physics Letters, 2003, 82, 73.
5.
D.H. Reneker, A.L. Yarin, H. Fong and S. Koombhongse, Journal of Applied Physics, 2000, 87, 4531.
6.
S.A. Theron, A.L. Yarin, E. Zussman and E. Kroll, Polymer, 2005, 46, 2889.
7.
Z-M. Huang, Y-Z. Zhang, M. Kotak and S. Ramakrishna, Composites Science and Technology, 2003, 63, 2223.
8.
H.L. Schreuder-Gibson, P. Gibson, K. Senecal, M. Sennett, J. Walker, W. Yeomans and D. Ziegler, Journal of Advanced Materials, 2002, 34, 3, 44.
9.
Z. Ma, M. Kotaki, R. Inai and S. Ramakrishna, Tissue Engineering, 2005, 11, 101.
10. Z. Ma, M. Kotaki, T. Yong, W. He and S. Ramakrishna, Biomaterials, 2005, 26, 2527.
8
Electrospinning of Polymeric Nanofibres 11. H.J. Jin, S. Fridrikh, G.C. Rutledge and D. Kaplan, Abstracts of Papers American Chemical Society, 2002, 224, 1/2, 408. 12. Y.K. Luu, K. Kim, B.S. Hsiao, B. Chu and M. Hadjiargyrou, Journal of Controlled Release, 2003, 89, 341. 13. K.J. Senecal, L. Samuelson, M. Sennett and G.H. Schreuder, inventors; US 0045547, 2001. 14. K. Sawicka, P. Goum and S Simon, Sensors and Actuators B, 2005, 108, 585. 15. K. Fujihara, M. Kotak and S. Ramakrishn, Biomaterials, 2005, 26, 4139. 16. X. Fang and D.H. Reneker, Journal of Macromolecular Science, Part B: Physics, 1997, 36, 169. 17. G.I. Taylor, Proceedings of the Royal Society Series A, 1969, 313, 453. 18. E-R. Kenawy, G.L. Bowlin, K. Mansfield, J. Layman, D.G. Simpson, E.H. Sanders and G.E. Wnek, Journal of Controlled Release, 2002, 81, 57 . 19. S.F. Fennessey and J.R. Farris, Polymer, 2004, 45, 4217. 20. E. Zussman, A. Theron and A.L. Yarin, Applied Physics Letters, 2003, 82, 973. 21. J.M Deitzel, J. Kleinmeyer, D. Harris and T.N. Beck, Polymer, 2001, 42, 261. 22. C.H. Zhang, X. Yuan, L. Wu, Y. Han and J. Sheng, European Polymer Journal, 2005, 41, 423. 23. A.F. Spivak and Y.A. Dzenis, Applied Physics Letters, 1998, 73, 3067. 24. M.M. Hohman, M. Shin, G. Rutledge and M.P. Brenner, Physics of Fluids, 2001, 13, 2201. 25. M.M. Hohman, M. Shin, G. Rutledge and M.P. Brenner, Physics of Fluids, 2001, 13, 2221. 26. D.H. Reneker, A.L. Yarin, H. Fong and S. Koombhongse, Journal of Applied Physics, 2000, 87, 4531. 27. A.L. Yarin, S. Koombhongse and D.H. Reneker, Journal of Applied Physics, 2001, 89, 47.
9
Advances in Nanofibre Research 28. Y.M. Shin, M.M. Hohman, M.P. Brenner and G.C. Rutledge, Applied Physics Letters, 2001, 78, 1149. 29. D.H. Reneker, A.L. Yarin, H. Fong and S. Koombhongse, Journal of Applied Physics, 2000, 87, 4531. 30. S. Sukigara, M. Gandhi, J. Ayutsede, M. Micklus and F. Ko, Polymer, 2003, 44, 5727. 31. S. Sukigara, M Gandhi, J. Ayutsede, M. Micklus and F. Ko, Polymer, 2004, 45, 3708. 32. K.E. Park, S.Y. Jung, S.J. Lee, B-M. Min and W.H. Park, International Journal of Biological Macromolecules, 2006, 38, 165.
10
2
Polymeric Nanofibre Fabrication via Electrospinning Process
2.1 Introduction Electrospinning is a novel and efficient method by which fibres with diameters in the nanometer range entitled ‘nanofibres’ can be achieved. In the electrospinning process, a strong electric field is applied on a droplet of polymer solution (or melt) held by its surface tension at the tip of a syringe needle (or a capillary tube). As a result, the pendent drop will become highly electrified and the induced charges are distributed over its surface. Increasing the intensity of the electric field ensures that the surface of the liquid drop will be distorted into a conical shape known as the ‘Taylor cone’ [1]. Once the strength of the electric field exceeds a threshold value, the repulsive electric force dominates the surface tension of the liquid, and a stable jet emerges from the cone tip. The charged jet is then accelerated toward the target and rapidly thins and dries as a result of elongation and solvent evaporation. As the jet diameter decreases, the surface charge density increases and the resulting high repulsive forces split the jet into smaller jets. This phenomenon may take place several times, leading to many small jets. Ultimately, solidification is carried out and fibres are deposited on the surface of the collector as a randomly oriented, non-woven mat [2–6]. Figure 2.1 shows a schematic illustration of electrospinning setup. Electrospun nanofibre mats have outstanding properties such as very small fibre diameters, large surface area per mass ratio [3], high porosity along with small pore sizes [7], flexibility, and superior mechanical properties [8]. They have found numerous applications in biomedicine (tissue engineering) [9–11], drug delivery [12], [13], and wound dressing [14, 15], protective clothing [7], filtration [16], reinforcement in composite materials [8, 17], as well as micro-electronics (battery [18], supercapacitors [19], transistors [20], sensors [21], and display devices [22]). The morphology of electrospun nanofibres (e.g., fibre diameter) are dependent upon many parameters which are mainly divided into three categories: solution properties (solution viscosity, solution concentration, polymer molecular weight, and surface tension), processing conditions (applied voltage, volume flow rate, spinning distance, and needle diameter), and ambient conditions (temperature, humidity, and atmosphere pressure) [23].
11
Advances in Nanofibre Research
Syringe Metering pump
Connector 0
1
Electrical heater High voltage supply
25 ºC
Temperature controller
Figure 2.1 Electrospinning setup (schematic)
As mentioned above, electrospun nanofibres have numerous applications, some of which have been commercialised. Most of these applications require nanofibres with desired properties, suggesting the importance of control of the process of manufacture. This may not be achieved without a comprehensive outlook of the process and quantitative study of the effects of governing parameters which makes control of the process possible. In addition, qualitative description of experimental observations is not adequate to derive general conclusions, and the equations governing the behaviour of the system must be found or appropriate empirical models presented. In the words of Ziabicki: ‘in the language of science ‘to explain’ means to put forward a quantitative model which is consistent with all the known data and capable of predicting new facts’ [24]. Employing a model to express the influence of electrospinning parameters will help us obtain a simple and systematic way for presenting the effects of variables, thereby enabling control of the process. Furthermore, it allows us to predict the results under a new combination of parameters. Hence, without conducting experiments, one can readily estimate features of the product under unknown conditions. That is, a model tells us to what extent the output of a system will change if one or more parameters increase or decrease. This is very helpful and leads to detailed understanding of the process and the effects of parameters. Despite the surge in attention in the electrospinning process, few investigations have addressed the quantitative study of the effects of the parameters, which has hindered
12
Polymeric Nanofibre Fabrication via Electrospinning Process control of the process. Changing the behaviour of materials at the nano-scale, the presence of an electric field, branching of the jet, and random orientation of fibres has made the analysis of the process so complex and difficult that to date there is no reliable theory that can describe the phenomenon. Furthermore, development of an empirical model has been impeded due to a lack of systematic and characterised experiments with appropriate designs. Adding to the difficulty is the number of parameters involved in the electrospinning process and the interactions between them. This has made it almost impossible to investigate the simultaneous effects of all variables. Fibre diameter is one of the most important structural features in electrospun nanofibre mats because it affects the characteristics of the final product (e.g., physical, mechanical and electrical properties). Podgorski and co-workers [25] indicated that filters made of fibres with smaller diameters have higher filtration efficiencies. This was also proved by the work of Qin and co-workers [16]. Ding and co-workers [26] reported that the sensitivity of sensors increase with decreasing mean fibre diameter due to the higher surface area. In a study by Kim and co-workers on designing polymer batteries consisting of electrospun poly(vinylidene fluoride) fibrous electrolyte [27], it was demonstrated that lower mean fibre diameter results in a higher uptake of electrolyte, thereby increasing the ionic conductivity of the mat. Moroni and co-workers [28] found that fibre diameters of electrospun poly(ethylene oxide terephthalate) and poly(butylene terephthalate) scaffolds influenced the seeding, attachment, and proliferation of cells. They also studied the release of dye incorporated in electrospun scaffolds and observed that, with increasing fibre diameter, cumulative release of the dye (methylene blue) decreased. Carbonisation and activation conditions as well as the structure and properties of the ultimate carbon fibres are also affected by the diameters of the precursor polyacrylonitrile (PAN) nanofibres [29]. Consequently, precise control of the electrospun fibre diameter is very important. Sukigara and co-workers [30] employed response surface methodology (RSM) to model the mean fibre diameter of electrospun-regenerated Bombyx mori silk with the electric field and concentration at two spinning distances. They applied a full factorial experimental design at three levels of each parameter (leading to nine treatments of factors) and used a quadratic polynomial to establish a relationship between mean fibre diameter and the variables. Increasing the concentration at a constant electric field resulted in an increase in mean fibre diameter. Different impacts for the electric field were observed depending on solution concentration. The trend of the effects of the two parameters on mean fibre diameter varied with changing the spinning distance, which suggests interaction and coupling between parameters. Gu and co-workers [31] and Gu and co-workers [32] also exploited the RSM for the quantitative study of PAN and poly D,L-lactide (PDLA) respectively. The only
13
Advances in Nanofibre Research difference observed in the procedure was the use of four levels of concentration in the case of PAN. They included the standard deviation of fibre diameter in their investigations by which they were able to provide additional information regarding the morphology of electrospun nanofibres and its variations under different conditions. Furthermore, they analysed the significance of the factors in the models to understand the level of influence of each parameter. In the case of PAN, voltage as well as its interaction with concentration had no significant effects on the mean and standard deviation of fibre diameter. Hence, they eliminated the terms corresponding to these factors, and thereby obtained simplified quadratic models according to which the mean and standard deviation of fibre diameter increased with polymer concentration. Conversely, voltage and its interaction with concentration were found to be significant in the case of PDLA. However, the effect of polymer concentration was more pronounced. Increasing the voltage at constant concentration favoured the formation of thinner fibres, which gained momentum with increasing concentration. Fibres with more uniform diameters (less standard deviation) were obtained at higher applied voltage or concentration. In the most recent investigation in this research area, Yördem and co-workers [33] utilised RSM to correlate the mean and coefficient of variation (CV) of electrospun PAN nanofibres to solution concentration, and applied a voltage at three spinning distances. They employed a face-centred central composite design along with a full factorial design at two levels, resulting in 13 treatments at each spinning distance. A cubic polynomial was then used to fit the data in each case. As with previous studies, fibre diameter was very sensitive to changes in solution concentration. The voltage effect was more significant at higher concentrations, demonstrating the interaction 2 between parameters. Despite reports of high R values, the presented models seemed to be inefficient and uncertain. Some terms in the models had very high p values (measure of statistical significance). For instance, in modeling the mean fibre diameter, a p value as high as 0.975 was calculated for the cubic concentration term at a spinning distance of 16 cm, where half of the terms had p values >0.8. This resulted 2 in low Rpred values which were not reported in their study. After calculations by our research team, these values were found to be almost zero in many cases, suggesting the poor predictive ability of their models. There are some interactions between electrospinning parameters. Previous studies however, have focused on the simultaneous effects of two variables. Therefore they could not thoroughly capture the interactions between the parameters. For instance, Sukigara and co-workers [30] and Yördem and co-workers [33] agreed that the spinning distance has a significant influence on fibre diameter, and that this effect varies when solution concentration and/or applied voltage is altered. However, they could not describe their findings in terms of quantitative relationships. Hence, the presented models are not comprehensive. In addition, in every study in which modeling of a
14
Polymeric Nanofibre Fabrication via Electrospinning Process process is targeted, the obtained models need to be evaluated with a set of test data which were not used in establishing the relationships. Otherwise, the effectiveness of the models will not be guaranteed and there will always be an uncertainty in the prediction of the models in new conditions. Hence, it is possible for a model to be very efficient in describing experimental data, but to present unsatisfactory prediction results. In none of the previous works, however, were the presented models evaluated with a series of test data. Therefore, their models may not generalise well to new data and their prediction ability is unclear. In this contribution, for the first time, the simultaneous effects of four electrospinning parameters (solution concentration, spinning distance, applied voltage, and volume flow rate) on mean and standard deviation of polyvinyl alcohol (PVA) fibre diameter were systematically investigated. PVA, the largest-volume synthetic water-soluble polymer produced in the world, is manufactured by the hydrolysis of polyvinyl acetate. The excellent chemical resistance and physical properties of PVA (along with non-toxicity and biodegradability) have led to its broad industrial applications in textile sizing, adhesive, paper-coating, fibres, and polymerisation stabilisers [34, 35]. Several patents reported a process for production of ultra-high tensile strength PVA fibres comparable with Kevlar® [36–38]. PVA has also found many applications in biomedicine due to its biocompatibility [39]. For instance, PVA hydrogels have been used in regenerating articular cartilages [40, 41], artificial pancreas [42], and drug-delivery systems [43, 44]. More recently, PVA nanofibres were electrospun and used as a protein delivery system [45], for retardation of enzyme release [45] and for wound dressing [46]. The objective of this contribution is to use RSM to establish quantitative relationships between electrospinning parameters and the mean and standard deviation of fibre diameter, as well as to evaluate the effectiveness of the empirical models with a set of test data.
2.2 Experimental 2.2.1 Solution Preparation and Electrospinning PVA with molecular weight of 72000 g/mol and degree of hydrolysis >98% was obtained from Merck and used as received. Distilled water as solvent was added to a predetermined amount of PVA powder to obtain 20 ml of solution with desired concentration. The solution was prepared at 80 °C and gently stirred for 30 minutes to expedite dissolution. After the PVA had completely dissolved, the solution was transferred to a 5 ml syringe and was ready to electrospin. The experiments were carried out on a horizontal electrospinning setup shown schematically in Figure 2.1.
15
Advances in Nanofibre Research The syringe containing PVA solution was placed on a syringe pump (new era NE-100) used to dispense the solution at a controlled rate. A high-voltage direct-current power supply (gamma high voltage ES-30) was used to generate the electric field needed for electrospinning. The positive electrode of the high-voltage supply was attached to the syringe needle via an alligator clip. The grounding electrode was connected to a flat collector wrapped with aluminium foil where electrospun nanofibres accumulated to form a nonwoven mat. Electrospinning was carried out at room temperature. Subsequently, the aluminium foil was removed from the collector. A small piece of mat was placed on the sample holder and gold sputter-coated (Bal-Tec). Thereafter, the morphology of electrospun PVA fibres was observed by an environmental scanning electron microscope (SEM) (Phillips XL-30) at 10000X magnification. For each specimen, the distribution of fibre diameter was determined from the SEM micrograph based on 100 measurements of random fibres. A typical SEM micrograph of an electrospun nanofibre mat and its corresponding diameter distribution are shown in Figure 2.2.
2.2.2 Choice of Parameters and Range As mentioned above, numerous variables can alter the electrospinning process. Hence, investigating all of them in the framework of one single research would be almost impossible. However, some of these parameters can be held constant during experimentation. For instance, carrying out the experiments in a controlled environmental condition would enable the ambient parameters (i.e., temperature, air pressure, and humidity) to be kept unchanged. Solution viscosity is affected by polymer molecular weight, solution concentration, and temperature. For a particular polymer (constant molecular weight) at a fixed temperature, solution concentration would be the only factor influencing viscosity. In this circumstance, the effect of viscosity could be determined by solution concentration. Therefore, there would be no need for viscosity to be considered as a separate parameter. In this regard, solution concentration (C), spinning distance (d), applied voltage (V), and volume flow rate (Q) were selected to be the most influential parameters in electrospinning of PVA nanofibres for the purpose of this study. The next step is to choose the region of interest (i.e., the ranges over which these factors are varied). Process knowledge, which is a combination of practical experience and theoretical understanding, is required to fulfill this step. The aim is to find an appropriate range for each parameter in which dry, bead-free, stable, and continuous fibres that do not break-up into droplets are obtained. This goal could be achieved by conducting a set of preliminary experiments while taking into consideration previous works, along with utilising the reported relationships.
16
Polymeric Nanofibre Fabrication via Electrospinning Process
(a)
0.025
Probability Density
0.02 0.015 0.01 0.005 0 100
150
200 250 300 Diameter (mm)
350
400
(b)
Figure 2.2 (a) a typical SEM micrograph of an electrospun nanofibre mat, and (b) its corresponding diameter distribution
17
Advances in Nanofibre Research The relationship between intrinsic viscosity ([η]) and molecular weight (M) is given by the well-known Mark–Houwink–Sakurada Equation as follows:
(2.1) where K and a are constants for a particular polymer–solvent pair at a given temperature [47]. For PVA with a molecular weight in the range 69000 g/mol <M<690000 g/mol in water at room temperature, K = 6.51 and a = 0.628 were found by Tacx and co-workers [48]. Using these constants in the equation, the intrinsic viscosity for PVA in this study (molecular weight of 72000 g/mol) was calculated to be [η] = 0.73. Entanglements of the polymer chain in a solution can be expressed in terms of the Berry number (B). This is a dimensionless parameter and defined as the product of intrinsic viscosity and polymer concentration (B = [η]C) [49]. At each molecular weight, there is a minimum concentration at which the polymer solution cannot be electrospun. Koski and co-workers [50] observed that B>5 is required to form stabilised fibrous structures in the electrospinning of PVA. Conversely, they reported the formation of flat fibres at B>9. Therefore, the appropriate range in this case could be found within 525 kV were rarely used. Afterwards, a series of experiments were carried out to obtain the desired voltage domain. At V<10 kV, the voltage was too low to spin fibres and 10 kV≤V<15 kV resulted in formation of fibres and droplets; in addition, electrospinning was impeded at high concentrations. In this regard, 15 kV≤V≤25 kV was selected to be the desired domain for applied voltage The use of 5–20 cm as the spinning distance has been reported in the literature. Short distances are suitable for highly evaporative solvents, whereas it results in wet conglutinated fibres for non-volatile solvents due to insufficient evaporation time. Water was used as a solvent for PVA in this study, so short spinning distances were
18
Polymeric Nanofibre Fabrication via Electrospinning Process not expected to be favourable for formation of dry fibres. This was subsequently proved by experimental observations, and 10 cm≤d≤20 cm was considered as the effective range for spinning distance. Few researchers have addressed the effect of volume flow rate. Therefore in this case, attention was focused on experimental observations. At Q<0.2 ml/h, in most cases (especially at high polymer concentrations), fibre formation was hindered due to insufficient supply of solution to the tip of the syringe needle. In contrast, excessive feed of solution at Q>0.4 ml/h incurred formation of droplets along with fibres. As a result, 0.2 ml/h≤Q≤0.4 ml/h was chosen as the favourable range of flow rate in this study.
2.2.3 Experimental Design Experiments often involve several factors and the objective of the experimenter is to determine how these factors affect the output (response) of the system. How many observations are required for the purpose of a research and how to obtain the most information about a system are important questions. The aim of the experimental design is to provide reasonable and scientific answers to such questions. That is, experimental design is a sequential step taken to ensure that data will be obtained in the most efficient form for the problem being considered, and that the analyses will lead to valid statistical inferences [51, 52]. Let us consider a process in which several factors affect a response of the system. In this case, a conventional strategy of experimentation which is extensively used in practice is the one-factor-at-a-time approach. The major disadvantage of this design is that it fails to consider possible interactions between factors—the failure of one factor to produce the same effect on the response at different levels of another factor. As mentioned above, interactions exist between electrospinning parameters that make this approach an inappropriate choice for the present work. The correct strategy to dealing with several factors is to use a full factorial design in which factors are all varied together. Therefore all possible combinations of the levels of the factors are investigated. The advantages of a full factorial design are that it is very efficient, makes the most use of the experimental data, and takes into account the interactions between factors [51, 52]. At least two points are needed to draw a line; at least three points are required for a quadratic curve. Hence, three levels were selected for each parameter in this study so that it would be possible to use quadratic models. These levels chosen were equally spaced. A full factorial experimental design with four factors (solution concentration, spinning distance, applied voltage, and flow rate) each at three levels (34 design) was
19
Advances in Nanofibre Research employed, resulting in 81 treatment combinations (Figure 2.3). The coded variables corresponding to low, intermediate and high levels of each factor were –1, 0, and 1, respectively. The coded variables (xj) were calculated using Equation 2.2 from natural variables (ξi). Indices 1 to 4 represent solution concentration, spinning distance, applied voltage, and flow rate, respectively. In addition to experimental data, 15 treatments inside the design space were selected as test data and used for evaluation of the models. The natural and coded variables for experimental data (numbers 1–81) as well as test data (numbers 82–96) are listed in Table 2.1:
(2.2)
C 0
-1
1
1 d 0 -1 -1
0 V
1 -1
0
1 Q
Figure 2.3 The 34 full factorial experimental design used in this study
20
Polymeric Nanofibre Fabrication via Electrospinning Process
Table 2.1 Natural and coded variables for experimental and test data along with corresponding responses Number
Natural variables C (%) d (cm) V (kV) Q (ml/h)
Coded variables x1 x2 x3 x4
Responses MFD StdFD (nm) (nm)
1
8
10
15
0.2
–1
–1
–1
–1
232.62
26.60
2
8
10
15
0.3
–1
–1
–1
0
235.50
24.52
3
8
10
15
0.4
–1
–1
–1
1
252.02
25.89
4
8
10
20
0.2
–1
–1
0
–1
236.84
37.30
5
8
10
20
0.3
–1
–1
0
0
232.08
30.22
6
8
10
20
0.4
–1
–1
0
0
249.21
34.49
7
8
10
25
0.2
–1
–1
1
–1
196.05
34.76
8
8
10
25
0.3
–1
–1
1
0
201.38
35.15
9
8
10
25
0.4
–1
–1
1
1
215.00
39.00
10
8
10
15
0.2
–1
0
–1
–1
221.10
38.88
11
8
10
15
0.3
–1
0
–1
1
238.63
20.17
12
8
10
15
0.4
–1
0
–1
1
242.32
21.99
13
8
10
20
0.2
–1
0
0
–1
219.76
36.19
14
8
10
20
0.3
–1
0
0
0
228.56
28.29
15
8
10
20
0.4
–1
0
0
1
242.01
28.30
16
8
10
25
0.2
–1
0
1
–1
202.62
33.22
17
8
10
25
0.3
–1
0
1
0
208.21
37.14
18
8
10
25
0.4
–1
0
1
1
213.66
34.84
19
8
10
15
0.2
–1
1
–1
–1
196.63
30.69
20
8
10
15
0.3
–1
1
–1
0
197.73
24.55
21
8
10
15
0.4
–1
1
–1
1
206.28
22.11
22
8
10
20
0.2
–1
1
0
–1
206.69
31.56
23
8
10
20
0.3
–1
1
0
–1
224.38
27.41
24
8
10
20
0.4
–1
1
0
0
242.06
26.51
25
8
10
25
0.2
–1
1
1
–1
205.25
40.32
26
8
10
25
0.3
–1
1
1
0
215.70
30.54
27
8
10
25
0.4
–1
1
1
1
231.34
32.40
28
8
10
15
0.2
0
–1
–1
–1
269.91
30.35
29
8
10
15
0.3
0
–1
–1
0
270.05
28.88
21
Advances in Nanofibre Research
22
30
8
10
15
0.4
0
–1
–1
1
291.99
33.98
31
8
10
20
0.2
0
–1
0
–1
256.11
38.54
32
8
10
20
0.3
0
–1
0
0
264.86
35.70
33
8
10
20
0.4
0
–1
0
1
278.34
49.13
34
8
10
25
0.2
0
–1
1
–1
228.21
42.33
35
8
10
25
0.3
0
–1
1
0
239.28
40.30
36
8
10
25
0.4
0
–1
1
1
238.74
46.57
37
8
10
15
0.2
0
0
–1
–1
263.67
34.16
38
8
10
15
0.3
0
0
–1
0
269.29
31.54
39
8
10
15
0.4
0
0
–1
1
277.71
29.40
40
8
10
20
0.2
0
0
0
–1
284.20
38.18
41
8
10
20
0.3
0
0
0
0
281.82
36.27
42
8
10
20
0.4
0
0
0
1
282.39
42.07
43
8
10
25
0.2
0
0
1
–1
249.42
40.79
44
8
10
25
0.3
0
0
1
0
278.22
46.15
45
8
10
25
0.4
0
0
1
1
286.96
51.16
46
8
10
15
0.2
0
1
–1
–1
239.45
27.98
47
8
10
15
0.3
0
1
–1
0
244.04
27.43
48
8
10
15
0.4
0
1
–1
1
251.58
27.26
49
8
10
20
0.2
0
1
0
–1
285.67
35.62
50
8
10
20
0.3
0
1
0
0
273.05
30.74
51
8
10
20
0.4
0
1
0
1
280.62
34.66
52
8
10
25
0.2
0
1
1
–1
278.10
40.79
53
8
10
25
0.3
0
1
1
0
280.95
44.58
54
8
10
25
0.4
0
1
1
1
306.28
44.04
55
8
10
15
0.2
1
–1
–1
–1
286.23
27.12
56
8
10
15
0.3
1
–1
–1
0
295.60
32.91
57
8
10
15
0.4
1
–1
–1
1
293.41
40.48
58
8
10
20
0.2
1
–1
0
–1
271.20
34.86
59
8
10
20
0.3
1
–1
0
0
291.89
42.78
60
8
10
20
0.4
1
–1
0
1
295.93
49.43
61
8
10
25
0.2
1
–1
1
–1
234.13
39.31
62
8
10
25
0.3
1
–1
1
0
247.65
48.60
63
8
10
25
0.4
1
–1
1
1
247.13
59.02
Polymeric Nanofibre Fabrication via Electrospinning Process
64
8
10
15
0.2
1
0
–1
–1
271.93
33.05
65
8
10
15
0.3
1
0
–1
0
297.65
26.75
66
8
10
15
0.4
1
0
–1
1
296.79
39.84
67
8
10
20
0.2
1
0
0
–1
297.94
38.82
68
8
10
20
0.3
1
0
0
0
310.06
36.84
69
8
10
20
0.4
1
0
0
1
312.15
41.69
70
8
10
25
0.2
1
0
1
–1
272.24
39.55
71
8
10
25
0.3
1
0
1
0
282.04
42.35
72
8
10
25
0.4
1
0
1
1
288.00
51.72
73
8
10
15
0.2
1
1
–1
–1
259.63
34.63
74
8
10
15
0.3
1
1
–1
0
278.40
25.35
75
8
10
15
0.4
1
1
–1
1
279.25
27.25
76
8
10
20
0.2
1
1
0
–1
307.42
42.25
77
8
10
20
0.3
1
1
0
0
327.77
35.71
78
8
10
20
0.4
1
1
0
1
337.88
45.16
79
8
10
25
0.2
1
1
1
-1
321.78
46.21
80
8
10
25
0.3
1
1
1
0
334.54
40.68
81
8
10
25
0.4
1
1
1
1
342.45
47.94
82
9
20
15
0.3
–0.5
1
–1
0
216.53
24.25
83
10
12.5
15
0.3
0
0.5
–1
0
259.61
25.67
84
10
20
22.5
0.3
0
1
0.5
0
300.27
35.71
85
10
20
15
0.25
0
1
–1
–0.5
235.04
29.64
86
9
12.5
15
0.3
–0.5
–0.5
–1
0
247.57
26.65
87
9
20
22.5
0.3
–0.5
1
0.5
0
247.16
31.12
88
9
20
15
0.25
–0.5
–1
–1
–0.5
212.82
30.26
89
10
12.5
22.5
0.3
0
–0.5
0.5
0
263.70
45.06
90
10
12.5
15
0.25
0
–0.5
–1
–0.5
258.26
26.16
91
10
20
22.8
0.25
0
1
0.5
–0.5
272.03
36.28
92
9
12.5
22.5
0.3
–0.5
–0.5
0.5
0
235.75
33.16
93
9
12.5
15
0.25
–0.5
–0.5
–1
–0.5
244.43
24.87
94
9
20
22.5
0.25
–0.5
1
0.5
–0.5
252.50
36.01
95
10
12.5
22.5
0.25
0
–0.5
0.5
–0.5
260.71
42.25
96
9
12.5
22.5
0.25
–0.5
–0.5
0.5
–0.5
231.97
32.86
23
Advances in Nanofibre Research
2.2.4 Response Surface Methodology Thus far, the mechanism of some scientific phenomena has been understood well, and models depicting the physical behaviour of the system have been drawn in the form of mathematical relationships. However, there are numerous processes which have not been sufficiently understood to permit a theoretical approach. RSM is a combination of mathematical and statistical techniques useful for empirical modeling and analyses of such systems. The application of RSM is in situations where several input variables potentially influence some performance measure or quality characteristic of the process (often called ‘responses’). The relationship between the response (y) and k input variables (ξ1,ξ2,...,ξk) can be expressed in terms of mathematical notations as follows:
(2.3) where the true response function f is unknown. It is often convenient to use coded variables (x1,x2,..,xk) instead of natural (input) variables. The response function will then be:
(2.4) The form of the true response function f is unknown, so it must be approximated. Therefore, the successful use of RSM is critically dependent upon the choice of appropriate function to approximate f. Low-order polynomials are widely used as approximating functions. First-order (linear) models cannot capture the interaction between parameters which is a form of curvature in the true response function. A second-order (quadratic) model will probably perform well in these circumstances. In general, the quadratic model is in the form of:
(2.5) where ε is the error term in the model. The use of polynomials of higher order is also possible but infrequent. The βs are a set of unknown coefficients needed to be
24
Polymeric Nanofibre Fabrication via Electrospinning Process estimated. To do that, the first step is to make some observations on the system being studied. The model in Equation 2.5 may now be written in matrix notations as:
y = Xâ + å
(2.6)
where y is the vector of observations, X is the matrix of levels of the variables, β is the vector of unknown coefficients, and ε is the vector of random errors. Afterwards, the method of least squares (which minimises the sum of squares of errors) is employed to find the estimators of the coefficients ( ∠) through:
∠= ( X ′X) −1 X ′y
(2.7)
The fitted model will then be written as:
yˆ = Xâˆ
(2.8)
Finally, response surfaces or contour plots are depicted to help visualise the relationship between the response and the variables, and to see the influence of the parameters [53, 54]. As you might notice, there is a close connection between RSM and linear regression analysis [55]. In this study, RSM was employed to establish empirical relationships between four electrospinning parameters (solution concentration, spinning distance, applied voltage, and flow rate) and two responses (mean fibre diameter and standard deviation of fibre diameter). Coded variables were used to build the models. The choice of three levels for each factor in the experimental design allowed us to take advantage of quadratic models. Afterwards, the significance of terms in each model was investigated by testing hypotheses on individual coefficients, and simpler yet more efficient models were obtained by eliminating statistically unimportant terms. Finally, the validity of the models was evaluated using the 15 test data. The analyses were carried out using statistical software Minitab 15.
25
Advances in Nanofibre Research
2.3 Results and Discussion After the unknown coefficients (βs) were estimated by the least squares method, the quadratic models for the mean fibre diameter (MFD) and standard deviation of fibre diameter (StdFD) in terms of coded variables are written as:
(2.9)
(2.10)
In the next step, two very important hypothesis-testing procedures were carried out to measure the usefulness of the models presented here. First, the test for significance of the model was done to determine if there is a subset of variables which contributes significantly in representing the response variations. The appropriate hypotheses are: H 0 : β1 = β 2 = = β k H1 : β j ≠ 0
for at least one j
(2.11)
The F statistics of this test along with the p values for both models are shown in Table 2.2:
Table 2.2 Summary of the results from statistical analysis of the models
26
F
P
R2
2 Radj
2 Rpred
MFD
106.02
0.000
95.74%
94.84%
93.48%
StfFD
42.05
0.000
89.92%
87.78%
84.83%
Polymeric Nanofibre Fabrication via Electrospinning Process The p values of the models are very small (almost zero). Therefore, it is concluded that the null hypothesis is rejected in both cases, suggesting that there are some significant 2 2 terms in each model. Also included in Table 2.2 are the values of R 2 , Radj , and Rpred . These are a measure of the amount of response variation which is explained by the variables, and will always increase when a new term is added to the model, regardless 2 of whether the inclusion of the additional term is statistically significant or not. Radj 2 is the R adjusted for the number of terms in the model. Therefore it will increase only if the new terms improve the model and decreases if unnecessary terms are 2 added. Rpred implies how well the model predicts the response for new observations, 2 whereas R 2 and Radj indicate how well the model fits the experimental data. The R 2 values demonstrate that 95.74% of MFD and 89.92% of StdFD are explained by the 2 variables. The Radj values are 94.84% and 87.78% for MFD and StdFD respectively, 2 which account for the number of terms in the models. Values of R 2 and Radj indicate 2 that the models fit the data very well. The slight difference between the values of R 2 and Radj suggests that there might be some insignificant terms in the models. The 2 2 Rpred values are so close to the values of R 2 and Radj that the models do not appear to ‘overfit’ and have very good predictive ability. The second testing hypothesis carried out in this study was the test on individual coefficients. This would be useful in determining the value of the variables in the models. The hypotheses for testing the significance of any individual coefficient are:
H0 : β j = 0 H1 : β j ≠ 0
(2.12)
The model might be more effective with inclusion or perhaps exclusion of one or more variables. By means of this test, we could evaluate the value of each term in the model and eliminate the statistically insignificant terms, thereby obtaining more efficient models. The results of this test for the models of MFD and StdFD are summarised in Tables 2.3 and 2.4, respectively.
27
Advances in Nanofibre Research
Table 2.3 The test on individual coefficients for the model of mean fibre diameter Term
Coefficient
T
P
Constant
282.031
102.565
0.000
C
34.953
31.136
0.000
d
5.622
5.0008
0.000
V
–2.113
–1.882
0.064
Q
9.013
8.028
0.000
C2
–11.613
–5.973
0.000
d2
–4.304
–2.214
0.030
V2
–15.500
–7.972
0.000
Q2
–0.414
–0.213
0.832
Cd
12.517
9.104
0.000
CV
4.020
2.924
0.005
CQ
–0.162
–0.118
0.906
dV
20.643
15.015
0.000
dQ
0.741
0.539
0.592
VQ
0.877
0.638
0.526
T = Statistic is a measure of the difference between an observed statistic and its hypothesised population value in units of standard error.
2 As depicted, the terms Q , , , and in the model of MFD and d 2 , , and in the model of StdFD have very high p values. Therefore they do not contribute significantly in representing the variation of the corresponding response. Eliminating these terms will enhance the efficiency of the models. Recalculating the unknown coefficients, the new models are then given by:
(2.13)
28
Polymeric Nanofibre Fabrication via Electrospinning Process
(2.14)
in terms of coded variables and:
(2.15)
(2.16)
in terms of natural (uncoded) variables. The results of the test for significance as well 2 2 as R 2 , Radj , and Rpred for the new models are given in Table 2.5. It is obvious that the p values for the new models are close to zero, indicating the existence of some significant terms in each model. Comparing the results of this table with Table 2.2, the F statistic increased for the new models, indicating improvement of the models 2 after eliminating the insignificant terms. Despite the slight decrease in R , the values 2 2 2 of Radj and Rpred increased a great deal for the new models. As mentioned above, R will always increase with the number of terms in the model. Therefore, the smaller values of R 2 were expected for the new models due to the fewer terms. However, this does not necessarily mean that the pervious models were more efficient. Looking at the 2 tables, Radj (which provides a more useful tool for comparing the explanatory power of models with a different number of terms) increased after eliminating the unnecessary variables. Hence, the new models can better explain the experimental data. Due to 2 higher values of Rpred obtained, the new models also have higher prediction ability. That is, eliminating the insignificant terms, simpler models were obtained which not only better explain the experimental data, but also are more powerful in predicting new conditions.
29
Advances in Nanofibre Research
Table 2.4 The test on individual coefficients for the model deviation of fibre diameter Term
Coefficient
T
P
Constant
36.1574
39.381
0.000
C
4.5788
12.216
0.000
D
–1.5536
–4.145
0.000
V
6.4012
17.078
0.000
Q
1.1531
3.076
0.003
C2
–2.2937
–3.533
0.001
d2
–0.1115
–0.172
0.864
V2
–1.1891
–1.832
0.072
Q2
3.0980
4.772
0.000
Cd
-0.2088
–0.455
0.651
CV
1.0010
2.180
0.033
CQ
2.7978
6.095
0.000
dV
0.1649
0.359
0.721
dQ
–2.4876
–5.419
0.000
VQ
1.5182
3.307
0.002
The test for individual coefficients was done again for the new models. This time (as anticipated), no terms had a higher p value than expected, which needs to be eliminated. Here is another advantage of removing unimportant terms. The values of the T statistic increased for the terms already in the models, implying that their effects on the response became stronger.
Table 2.5 Summary of the results from statistical analysis of the models after eliminating the insignificant terms F
30
R2
P
2 Radj
2 Rpred
MFD
155.56
0.000
95.69%
95.08%
94.18%
StfFD
55.61
0.000
89.86%
88.25%
86.02%
Polymeric Nanofibre Fabrication via Electrospinning Process Now that the relationships have been developed, the test data were used to investigate the prediction ability of the models. Root mean square errors (RMSE) between the calculated responses (Ci) and real responses (Ri) were determined using Equation 2.1 for experimental data as well as test data for the evaluation of MFD and StdFD models (Table 2.6). The models present acceptable RMSE values for test data. This indicates the ability of the models to generalise well the experimental data for predicting new conditions. Although the values of RMSE for the test data are slightly higher than experimental data, these small discrepancies were expected because it is almost impossible for an empirical model to express the test data as well as experimental data, and higher errors are often obtained when new data are presented to the models. Hence, the results imply the acceptable prediction ability of the models.
n
RMSE =
∑ (C i =1
i
− Ri ) 2
n
(2.17)
Table 2.6 RMSE values of the models for the experimental and test data Experimental data
Test data
MFD
7.489
10.647
StdFD
2.493
2.890
2.3.1 Response Surfaces for Mean Fibre Diameter
2.3.1.1 Solution Concentration Increasing polymer concentration will result in greater entanglements in the polymer chain. This causes the viscoelastic force to increase, enabling the charged jet to withstand a larger electrostatic stretching force, leading to a larger diameter of fibres. A monotonic increase in MFD with concentration was observed in this study (Figure 2.4a–c), which concurs with the previous observations [23, 29, 56–58]. The concentration effect was more pronounced at further spinning distances (Figure 2.4a). 31
Advances in Nanofibre Research This could be attributed to the twofold effect of distance (see below). At low concentrations, there is more solvent in the solution, and a longer distance provides more time not only to stretch the jet in the electric field but also to evaporate the solvent, thereby encouraging formation of thinner fibres. At higher concentrations, however, there are extensive entanglements of polymer chains, resulting in higher viscoelastic forces which tend to resist the electrostatic stretching force. Conversely, V increasing the spinning distance will reduce the electric field strength ( E = ), causing d the electrostatic force to decrease. As a result, increasing MFD with concentration gains more momentum at longer spinning distances. Higher applied voltages also accelerate the concentration impact on MFD (Figure 2.4b). This may be ascribed to the twofold effect of voltage explained below. At higher voltages, where the electric field is the strong and dominant factor, increasing polymer concentration tends to encourage the effect of voltage on mass flow of the polymer. Hence, more solution acn be removed from the tip of the needle, resulting in a further increase in MFD. A combined effect between solution concentration and volume flow rate was not observed (Figure 2.4c). Therefore, concentration had interactions with spinning distance and applied voltage, which had been suggested by the existence of terms Cd and CV in the model of MFD. Recall that the term CQ was statistically insignificant and therefore had been removed from the model of MFD.
2.3.1.2 Spinning Distance The effect of spinning distance on the diameter of electrospun fibre is twofold. Varying the distance has a direct influence on the jet flight time as well as electric field strength. Longer spinning distance will provide more time for the jet to stretch in the electric field before it is deposited on the collector. Furthermore, solvents will have more time to evaporate. Hence, the fibre diameter will be prone to decrease. Conversely, increasing the spinning distance will decrease the electric field strength (E =
V ), resulting in less acceleration and hence stretching of the jet, which leads d
to formation of thicker fibres. The balance between these two effects will determine the final fibre diameter. Increase in fibre diameter [57, 59, 60] as well as decrease in fibre diameter [29] with increasing spinning distance has been reported. There were also some cases in which spinning distance did not have a significant influence on fibre diameter [56, 61–63]. The impact of spinning distance on MFD is illustrated in Figures 2.4a, d and e. As depicted in these figures, the effect of spinning distance is not always identical. As mentioned above, there will be more chain entanglements at higher concentrations, resulting in an increase in viscoelastic force. Furthermore, the longer the distance, the lower is the electric field strength. Hence, the electrostatic
32
Polymeric Nanofibre Fabrication via Electrospinning Process stretching force, which has now become weaker, will be dominated more readily by the viscoelastic force. As a result, the increasing effect of spinning distance on fibre diameter will be assisted, rendering a higher MFD (Figure 2.4a). The effect of spinning distance will alter at different applied voltages (Figure 2.4d). At low voltages, longer spinning distance brought about formation of thinner fibres whereas, at high voltages, the effect of spinning distance was totally reversed and fibres with thicker diameters were obtained at longer distances. It is supposed that, at low voltages, where the electric field is weak, stretching time becomes the dominant factor. Hence, a longer spinning distance (which gives more time to the jet to stretch and thin and for to solvent to evaporate) will result in fibres with smaller diameters. At high voltages, however, the electric field strength is high and dominant. Therefore, increasing the distance (which reduces the electric field) causes an increase in fibre diameter. The function of spinning distance was observed to be independent from volume flow rate for MFD (Figure 2.4e). The interaction of spinning distance with solution concentration and applied voltage is demonstrated in Figures 2.4a and d, and proved the existence of terms Cd and dV in the model of MFD.
2.3.1.3 Applied Voltage Applied voltage has two major effects on fibre diameter. First, increasing the applied voltage will increase the electric field strength, and a larger electrostatic stretching force causes the jet to accelerate more in the electric field, thereby favouring formation of thinner fibres. Second, charge transport is carried out only by the flow of the polymer in the electrospinning process [64] and increasing the voltage would introduce more surface charges on the jet. Hence, the mass flow rate from the needle tip to the collector will increase, so the solution will be drawn more quickly from the tip of the needle, causing the fibre diameter to increase. A combination of these two effects will determine the final fibre diameter. Hence, increasing applied voltage may decrease [65–67], increase [56, 57, 60] or may not change [23, 29, 61, 68] fibre diameter. Figures 2.4b, d and f show the effect of applied voltage on MFD. Increasing the voltage, MFD underwent an increase followed by a decrease. According to the explanation given, at low voltages (where the electric field strength is low), the effect of the mass of solution could be dominant. Therefore, fibre diameter increases when the applied voltage rises. However, as the voltage exceeds a limit, the electric field will be sufficiently high to be a determining factor. Hence, fibre diameter decreases as voltage increases. The effect of voltage on MFD was influenced by solution concentration to some extent (Figure 2.4b). At high concentrations, the increase in fibre diameter with voltage was more pronounced. This could be because the effect of the mass of the solution will be more important for solutions of higher concentrations. Spinning distance dramatically influenced the way voltage affects fibre diameter (Figure 2.4d). At a short distance, the electric field is high and the dominant factor. Therefore, 33
Advances in Nanofibre Research increasing applied voltage (which strengthens the electric field) resulted in a decrease in fibre diameter. At long distances (where the electric field was low), the effect of the mass of the solution would be the determining factor according to which fibre diameter increased with applied voltage. The effect of applied voltage on MFD was found to be independent from volume flow rate. Looking at the figures, it is apparent that there was: a huge interaction between applied voltage and spinning distance; a slight interaction between applied voltage and solution concentration; no interaction between applied voltage and volume flow rate. This is in agreement with the presence of CV and dV and absence of VQ in the model of MFD.
2.3.1.4 Volume Flow Rate It has been suggested that a minimum value for solution flow rate is required to form the drop of polymer at the tip of the needle for maintaining a stable Taylor cone [69]. Hence, flow rate could affect the morphology of electrospun nanofibres (e.g., fibre diameter). If the flow rate is increased, more solution is delivered to the tip of the needle, enabling the jet to carry the solution away faster. This could bring about an increase in the jet diameter, thereby favouring formation of thicker fibres. In this study, MFD slightly increased with volume flow rate (Figures 2.4c, e and f), which is in agreement with previous research [29, 69–71]. Flow rate was also found to influence MFD independent of solution concentration, applied voltage, and spinning distance as suggested earlier by the absence of CQ, dQ, and VQ in the model of MFD.
2.3.2 Response Surfaces for Standard Deviation of Fibre Diameter
2.3.2.1 Solution Concentration As depicted in Figures 2.5a, b and c, StdFD increased with concentration, which concurs with previous observations [23, 29, 31, 56, 61, 72, 73, 74]. In electrospinning, the elongation flow of the jet causes the coiled macromolecules in the solution to transform into an oriented entangled network. Increasing the polymer concentration, the macromolecular chain entanglements increase, making it more difficult for the jet to stretch and split. This could result in less uniform fibres (higher StdFD). Concentration affected StdFD regardless of spinning distance (Figure 2.5a), suggesting that there was no interaction between these two parameters (absence of Cd in the model of StdFd). At low applied voltages, the formation of more uniform fibres with decrease in concentration was facilitated. In agreement with existence of the term CV in the model of StdFd, solution concentration was found to have a slight interaction
34
Polymeric Nanofibre Fabrication via Electrospinning Process with applied voltage (Figure 2.5b). The curvature of the surface in (Figure 2.5c) suggested that there was a noticeable interaction between concentration and flow rate. This is in agreement with the presence of the term CQ in the model of StdFD.
• Design points above predicted value • Design points below predicted value
• Design points above predicted value • Design points below predicted value
Mean Fiber Diameter (mm)
•
300 270 240 210
•
225
• X1
A C X2 B d Actual Factors C V 20 00 D Q 0 30
•
•
•
20 0 17 5
15 0 Distance (cm) 12 5
10
9
10 0 8
•
290
200
•
•
22 5 20 0 Voltage (kV) 17 5
Concentration (%)
(a)
225 X1 A C X2 C V Actual Factors B d 15 0 D Q 0 30
260 230
325
•
25 0
12
11
•
320
325
Mean Fiber Diameter (mm)
•
330
9
15 0 8
12
11
10
Concentration (%)
(b) • Design points above predicted value • Design points below predicted value • Design points above predicted value • Design points below predicted value
Mean Fiber Diameter (mm)
•
290 260 230
•
•
200
•
•
•
0 40
225 X1 A C X2 D Q Actual Factors B d 15 0 C V 20 0
•
280 260 240
•
•
325
•
225
•
•
9
0 20 8
10
11
22 5 20 0 Voltage (kV) 17 5
12
Concentration (%)
(c)
12 5
15 0 10 0
300
325
• •
• •
265
•
•
•
240 0 40 0 35
0 30 Flow Rate (ml/h) 0 25
(e)
Distance (cm)
• Design points above predicted value • Design points below predicted value
0 20 10 0
12 5
15 0
17 5
Distance (cm)
225 X1 B d X2 D Q Actual Factors A C 10 C V 20 0
20 0
Mean Fiber Diameter (mm)
Mean Fiber Diameter (mm)
300
270
20 0
17 5
15 0
X1 B d X2 C V Actual Factors A C 10 D Q 0 30
(d) • Design points above predicted value • Design points below predicted value
285
•
220 25 0
0 35
0 30 Flow Rate (ml/h) 0 25
325
300 Mean Fiber Diameter (mm)
•
320
•
285
• 270
•
265 240
•
325
• •
•
225
•
0 40 0 35 0 30 Flow Rate (ml/h) 0 25
0 20 15 0
17 5
20 0
22 5
X1 C V X2 D Q Actual Factors A C 10 B d 15 0
25 0
Distance (cm)
(f)
Figure 2.4 Response surfaces for mean fibre diameter in terms of: (a) solution concentration and spinning distance; (b) solution concentration and applied voltage; (c) solution concentration and flow rate; (d) spinning distance and applied voltage; (e) spinning distance and flow rate; and (f) applied voltage and flow rate 35
Advances in Nanofibre Research
2.3.2.2 Spinning Distance More uniform fibres (lower StdFD) were obtained by increasing the spinning distance (Figures 2.5a, d and e). At a longer spinning distance, more time is given to the jet to fly from the tip of the needle to the collector and for the solvent to evaporate. Therefore, the processes of stretching the jet and evaporating the solvent will be carried out more gently, resulting in more uniform fibres. Our finding is consistent with the trend observed by Zhao and co-workers [74]. Spinning distance influenced StdFD regardless of solution concentration and applied voltage (Figures 2.5a and d), meaning that no interaction exists between these variables as could be inferred from the model of StdFD. However, the interaction of spinning distance with volume flow rate was noticeable (Figure 2.5e). The presence of dQ in the model of StdFD proves this observation. The effect of spinning distance was more pronounced at higher flow rates. This could be because more solution is delivered to the tip of the needle at higher flow rates, so the threads will need more time to dry. If the distance is sufficiently high to provide sufficient time, uniform fibres will be formed. Decreasing the distance will mean there is less time for the solvent to evaporate, thereby favouring the production of non-uniform fibres (high StdFD).
2.3.2.3 Applied Voltage StdFD was found to increase with applied voltage (Figures 2.5b, d and f), as observed in other works [56, 57, 60, 73]. Increasing the applied voltage causes the effect of the electric field on the charged jet to increase. Hence, the flight speed of the jet increases, shortening the time that the jet travels towards the collector. As a result, less time is given to the jet to stretch and thin and for the solvent to evaporate. This may result in formation of less uniform fibres (higher StdFD). The effect of applied voltage on StdFD was influenced by solution concentration (Figure 2.5b). This implies the interaction of voltage with concentration, which was addressed above by the presence of the corresponding term in the model of StdFD. At low concentrations, the formation of uniform fibres was facilitated by decreasing the applied voltage. An interaction was not observed between the applied voltage and spinning distance (Figure 2.5d), as suggested by the absence of the term dV in the model of StdFD. Figure 2.5f shows a slight interaction of voltage with flow rate, which concurs with the existence of VQ in the model of StdFD.
36
Polymeric Nanofibre Fabrication via Electrospinning Process • Design points above predicted value • Design points below predicted value • Design points above predicted value • Design points below predicted value
Std Fiber Diameter (mm)
40
• •
35 30
• 25
•
•
Distance (cm)
•
•
15 0
12 5
45 25
X1 A C X2 B d Actual Factors C V 20 00 D Q 0 30
45
•
45
25
•
35
•
25
10
9
10 0 8
12
11
• 22.5
20.0
Voltage (kV)
Concentration (%)
(a)
17.5
15.0 8
• Design points above predicted value • Design points below predicted value
•
37 50 31 25 25 00
•
•
•
•
•
0 40 0 35 Flow Rate (ml/h)
0 30
0 25
9
0 20 8
11
10
Std Fiber Diameter (mm)
Std Fiber Diameter (mm)
25 X1 A C X2 D Q Actual Factors B d 15 00 C V 20 0
43 75 37 50
•
•
50 00
45
•
43 75
•
•
31 25
•
25 00
22 5
Concentration (%)
•
•
25 0
12
45 25
•
Voltage (kV)
(c)
20 0
17 5
15 0 10 00
12 5
17 5
15 0
X1 A C X2 C V Actual Factors B d 15 00 D Q 0 30
20
Distance (cm)
(d) • Design points above predicted value • Design points below predicted value
• Design points above predicted value • Design points below predicted value
•
•
45 40 35
•
•
30
•
0 40 0 35 Flow Rate (ml/h)
0 30
0 25
0 20 10 0
12 5
15 0
17 5
45
55 0
25
47 5
X1 B d X2 D Q Actual Factors A C 10 C V 20 0
20 0
Distance (cm)
Std Fiber Diameter (mm)
Std Fiber Diameter (mm)
Concentration (%)
(b)
50 00
(e)
X1 A: C X2 C: V Actual Factors B: d 15.00 D: Q 0.30
12
11
10
9
• Design points above predicted value • Design points below predicted value
50
•
•
•
25
25.0
20 0 17 5
55 Std Fiber Diameter (mm)
45
• •
40 0 32 5 25 0
•
•
45
•
25
•
•• •
0 40 0 35 Flow Rate (ml/h)
0 30
0 25
0 20 15 0
17 5
20 0
22 5
X1 C V X2 D Q Actual Factors A C 10 B d 15 0
25 0
Voltage (kV)
(f)
Figure 2.5 Response surfaces for standard deviation of fibre diameter in terms of: (a) solution concentration and spinning distance; (b) solution concentration and applied voltage; (c) solution concentration and flow rate; (d) spinning distance and applied voltage; (e) spinning distance and flow rate; and (f) applied voltage and flow rate
37
Advances in Nanofibre Research
2.3.2.4 Volume Flow Rate As demonstrated in Figures 2.5c, e and f, increasing the flow rate meant that the uniformity of fibres increased (StdFD decreased), reached an optimum value and then decreased (StdFD increased). When the flow rate is low, the amount of solution fed to the tip of the needle is not sufficient, whereas an excess amount of solution is delivered to the tip of the needle at high flow rates. Therefore, unstable jets are formed in the two extremes, resulting in the production of non-uniform fibres. Solution concentration, applied voltage, and spinning distance were found to influence the impact of flow rate on StdFD. This indicated the interaction of flow rate with the other variables, as demonstrated by the terms CQ, dQ, and VQ in the model of StdFD. Increasing the solution concentration favoured the formation of non-uniform fibres at high flow rates (Figure 2.5c) which may be the result of the greater difficulty of solution removal due to the increased content of polymeric material in the solution. The effect of flow rate on StdFD was more pronounced as the spinning distance decreased (Figure 2.5e). The shorter the distance, the less time provided to the jet to thin and dry. Therefore, at high flow rates (at which more solution is delivered to the tip of the needle), the given time may not be sufficient, resulting in formation of less uniform fibres. A high applied voltage encouraged the increase in StdFD at fast flow rates (Figure 2.5f).
2.4 Conclusion For the first time, the simultaneous effects of four processing variables (solution concentration, applied voltage, spinning distance, and volume flow rate) on MFD and StdFD were investigated quantitatively and qualitatively. The appropriate range of parameters in which dry, bead-free, and continuous fibres that did not break-up into droplets were selected by reference to the literature along with conducting preliminary experiments. A full factorial experimental design at three levels of each factor (34 design) was carried out and 15 treatments inside the design space were selected as the test set for evaluating the prediction ability of the models. Nanofibres PVA were then prepared for experimental and test sets through the electrospinning method. After that, MFD and StdFD were determined from SEM micrographs of each sample. RSM was used to establish quadratic models for MFD and StdFD. The test for significance of the coefficients demonstrated that the terms Q2, CQ, dQ, and VQ in the model of MFD and d2, Cd, and dV in the model of StdFD were not of important value in representing the responses. Eliminating these terms, simpler yet more efficient models were obtained which not only explained the experimental data in a better manner, but also had more prediction ability. Afterwards, to show the generalisation ability of the models for predicting new conditions, the test set
38
Polymeric Nanofibre Fabrication via Electrospinning Process was used. Low RMSE of the test set for MFD and StdFD were obtained, indicating the good prediction ability of the models. Finally, to qualitatively study the effects of variables on MFD and StdFD, response surface plots were generated using these relationships .
2.4.1 Mean Fibre Diameter • Increasing solution concentration meant that MFD increased drastically. The effect of concentration was more pronounced at longer spinning distance and also at higher applied voltage • The effect of spinning distance on MFD changed depending on solution concentration and applied voltage. At low applied voltages, MFD decreased as the spinning distance became longer, whereas higher MFD resulted with lengthening the spinning distance when the applied voltage was high. Increasing the solution concentration tended to assist the formation of thicker fibres at a longer spinning distance • If the applied voltage was increased, MFD was observed to first increase and then decrease. High solution concentrations (partly) and long spinning distances (largely) favoured the increase of MFD with applied voltage • MFD slightly increased with flow rate. The impact of flow rate on MFD was not related to the other variables
2.4.2 Standard Deviation of Fibre Diameter • The higher the solution concentration, the less uniform were the fibres (higher StdFD) formed. Low applied voltages facilitated the formation of more uniform fibres (lower StdFD) by decreasing the concentration. The increase of StdFD with concentration gained momentum at high flow rates • A longer spinning distance resulted in more uniform fibres (lower StdFD). The effect of spinning distance was more pronounced at higher flow rates • Increasing the applied voltage increased StdFD. Low concentrations facilitated the formation of uniform fibres (high StdFD) by decreasing the applied voltage • Flow rate was found to have a significant impact on the uniformity of fibres (StdFD). As flow rate increased, StdFD decreased and then increased. Higher solution concentration, higher applied voltage, and shorter spinning distance encouraged the formation of non-uniform fibres (high StdFD) at fast flow rates
39
Advances in Nanofibre Research
References 1.
G.I. Taylor in the Proceedings of the Royal Society of London, 1969, 313, 1515, 453.
2.
J. Doshi and D.H. Reneker, Journal of Electrostatics, 1995, 35, 151.
3.
H. Fong and D.H. Reneker in Structure Formation in Polymeric Fibres, Hanser, Cincinnati, OH, USA, 2001, p.225.
4.
D. Li and Y. Xia, Advanced Materials, 2004, 16, 14, 1151.
5.
R. Derch, A. Greiner and J.H. Wendorff in Polymer Nanofibres, Dekker Encyclopedia of Nanoscience and Nanotechnology, Eds., J. A. Schwarz, C. I. Contescu and K. Putyera, CRC, New York, NY, USA, 2004, p.182.
6.
A.K. Haghi and M. Akbari, Physica Status Solidi A, 2007, 204, 1830.
7.
P.W. Gibson, H.L. Schreuder-Gibson and D. Rivin, AIChE Journal, 1999, 45, 1, 190.
8.
Z.M. Huang, Y.Z. Zhang, M. Kotaki and S. Ramakrishna, Composites Science and Technology, 2003, 63, 2223.
9.
M. Li, M.J. Mondrinos, M.R. Gandhi, F.K. Ko, A.S. Weiss and P.I. Lelkes, Biomaterials, 2005, 26, 5999.
10. E.D. Boland, B.D. Coleman, C.P. Barnes, D.G. Simpson, G.E. Wnek and G.L. Bowlin, Acta Biomaterialia, 2005, 1, 115. 11. J. Lannutti, D. Reneker, T. Ma, D. Tomasko and D. Farson, Materials Science and Engineering C, 2007, 27, 504. 12. J. Zeng, L. Yang, Q. Liang, X. Zhang, H. Guan, C. Xu, X. Chen and X. Jing, Journal of Controlled Release, 2005, 105, 43. 13. E.R. Kenawy, G.L. Bowlin, K. Mansfield, J. Layman, D.G. Simpson, E.H. Sanders and G.E. Wnek, Journal of Controlled Release, 2002, 81, 57. 14. M-S. Khil, D-I. Cha, H-Y. Kim, I-S. Kim and N. Bhattarai, Journal of Biomedical Materials Research, Part B: Applied Biomaterials, 2003, 67, 675. 15. B.M. Min, G. Lee, S.H. Kim, Y.S. Nam, T.S. Lee and W.H. Park, Biomaterials, 2004, 25, 1289.
40
Polymeric Nanofibre Fabrication via Electrospinning Process 16. X.H. Qin and S.Y. Wang, Journal of Applied Polymer Science, 2006, 102, 1285. 17. J.S. Kim and D.H. Reneker, Polymer Engineering and Science, 1999, 39, 5, 849. 18. S.W. Lee, S.W. Choi, S.M. Jo, B.D. Chin, D.Y. Kim and K.Y. Lee, Journal of Power Sources, 2006, 163, 41. 19. C. Kim, Journal of Power Sources, 2005, 142, 382. 20. N.J. Pinto, A.T. Johnson, A.G. MacDiarmid, C.H. Mueller, N. Theofylaktos, D.C. Robinson and F.A. Miranda, Applied Physics Letters, 2003, 83, 20, 4244. 21. D. Aussawasathien, J-H. Dong and L. Dai, Synthetic Metals, 2005, 54, 37. 22. S-Y. Jang, V. Seshadri, M-S. Khil, A. Kumar, M. Marquez, P.T. Mather and G.A. Sotzing, Advanced Materials, 2005, 17, 2177. 23. S-H. Tan, R. Inai, M. Kotaki and R. Ramakrishna, Polymer, 2005, 46, 6128. 24. A. Ziabicki in Fundamentals of Fibre Formation: The Science of Fibre Spinning and Drawing, Wiley, New York, NY, USA, 1976, p.76. 25. A. Podgóski, A. Bałazy and L. Grado , Chemical Engineering Science, 2006, 61, 6804. 26. B. Ding, M. Yamazaki and S. Shiratori, Sensors and Actuators B, 2005, 106, 477. 27. J.R. Kim, S.W. Choi, S.M. Jo, W.S. Lee and B.C. Kim, Electrochimica Acta, 2004, 50, 69. 28. L. Moroni, R. Licht, J. de Boer, J.R. de Wijn and C.A. van Blitterswijk, Biomaterials, 2006, 27, 4911. 29. T. Wang and S. Kumar, Journal of Applied Polymer Science, 2006, 102, 1023. 30. S. Sukigara, M. Gandhi, J. Ayutsede, M. Micklus and F. Ko, Polymer, 2004, 45, 3701. 31. S.Y. Gu, J. Ren and G.J. Vancso, European Polymer Journal, 2005, 41, 2559.
41
Advances in Nanofibre Research 32. S.Y. Gu and J. Ren, Macromolecular Materials and Engineering, 2005, 290, 1097. 33. O.S. Yördem, M. Papila and Y.Z. Mencelo lu, Materials and Design, 2008, 29, 34. 34. I. Sakurada in Polyvinyl Alcohol Fibres, CRC, New York, NY, USA, 1985, p.167. 35. F.L. Marten, Encyclopedia of Polymer Science and Technology, Ed., H. F. Mark, 3 Edition, Volume 8, Wiley, New York, NY, USA, 2004, p.154. 36. Y.D. Kwon, S. Kavesh and D.C. Prevorsek, inventors; Allied Corporation, assignee; US 4440711, 1984. 37. S. Kavesh and D.C. Prevorsek, inventors; Allied Corporation, assignee; US 4551296, 1985. 38. H. Tanaka, M. Suzuki and F. Uedo, inventors; Toray Industries, Inc., assignee; US 4603083, 1986. 39. G. Paradossi, F. Cavalieri, E. Chiessim, C. Spagnoli and M.K. Cowman, Journal of Materials Science: Materials in Medicine, 2003, 14, 687. 40. G. Zheng-Qiu, X. Jiu-Mei and Z. Xiang-Hong, Bio-Medical Materials and Engineering, 1998, 8, 75. 41. M. Oka, K. Ushio, P. Kumar, K. Ikeuchi, S.H. Hyon, T. Nakamura and H. Fujita, Journal of Engineering in Medicine, 2000, 214, 59. 42. K. Burczak, E. Gamian and A. Kochman, Biomaterials, 1996, 17, 2351. 43. J.K. Li, N. Wang and X.S. Wu, Journal of Controlled Release, 1998, 56, 117. 44. A.S. Hoffman, Advanced Drug Delivery Reviews, 2002, 43, 3. 45. J. Zeng, A. Aigner, F. Czubayko, T. Kissel, J.H. Wendorff and A. Greiner, Biomacromolecules, 2005, 6, 1484. 46. K.H. Hong, Polymer Engineering and Science, 2007, 47, 43. 47. L.H. Sperling in Introduction to Physical Polymer Science, 4th Edition, Wiley, New Jersey, NJ, USA, 2006, p.76.
42
Polymeric Nanofibre Fabrication via Electrospinning Process 48. J.C.J.F. Tacx, H.M. Schoffeleers, A.G.M. Brands and L. Teuwen, Polymer, 2000, 41, 947. 49. F.K. Ko in Nanomaterials Handbook, Ed., Y. Gogotsi, CRC Press, Boca Raton, FL, USA, 2006, Chapter 19. 50. A. Koski, K. Yim and S. Shivkumar, Materials Letters, 2004, 58, 493. 51. D.C. Montgomery in Design and Analysis of Experiments, 5th Edition, Wiley, New York, NY, USA, 1997, p.95. 52. A. Dean and D. Voss in Design and Analysis of Experiments, Springer, New York, NY, USA, 1999, p.129. 53. G.E.P. Box and N.R. Draper in Response Surfaces, Mixtures, and Ridge Analyses, Wiley, New Jersey, NJ, USA, 2007, p.94. 54. K.M. Carley, N.Y. Kamneva and J. Reminga, CASOS Technical Report, CMU-ISRI-04-136, 2004. 55. S. Weisberg in Applied Linear Regression, 3rd Edition, Wiley, New Jersey, NJ, USA, 2005, p.69. 56. C. Zhang, X. Yuan, L. Wu, Y. Han and J. Sheng, European Polymer Journal, 2005, 41, 423. 57. Q. Li, Z. Jia, Y. Yang, L. Wang and Z. Guan in the Proceedings of IEEE International Conference on Solid Dielectrics, Winchester, UK, 2007, 8–13th July. 58. C. Mit-uppatham, M. Nithitanakul and P. Supaphol, Macromolecular Chemistry and Physics, 2004, 205, 2327. 59. T. Jarusuwannapoom, W. Hongrojjanawiwat, S. Jitjaicham, L. Wannatong, M. Nithitanakul, C. Pattamaprom, P. Koombhongse, R. Rangkupan and P. Supaphol, European Polymer Journal, 2005, 41, 409. 60. S.C. Baker, N. Atkin, P.A. Gunning, N. Granville, K. Wilson, D. Wilson and J. Southgate, Biomaterials, 2006, 27, 3136. 61. S. Sukigara, M. Gandhi, J. Ayutsede, M. Micklus and F. Ko, Polymer, 2003, 44, 5721.
43
Advances in Nanofibre Research 62. X. Yuan, Y. Zhang, C. Dong and J. Sheng, Polymer International, 2004, 53, 1704. 63. C.S. Ki, D.H. Baek, K.D. Gang, K.H. Lee, I.C. Um and Y.H. Park, Polymer, 2005, 46, 5094. 64. J.M. Deitzel, J. Kleinmeyer, D. Harris and N.C. Beck Tan, Polymer, 2001, 42, 261. 65. C.J. Buchko, L.C. Chen, Y. Shen and D.C. Martin, Polymer, 1999, 40, 7397. 66. J.S. Lee, K.H. Choi, H.D. Ghim, S.S. Kim, D.H. Chun, H.Y. Kim and W.S. Lyoo1, Journal of Applied Polymer Science, 2004, 93, 1638. 67. S.F. Fennessey and R.J. Farris, Polymer, 2004, 45, 4217. 68. S. Kidoaki, I.K. Kwon and T. Matsuda, Biomaterials, 2005, 26, 37. 69. X. Zong, K. Kim, D. Fang, S. Ran, B.S. Hsiao and B. Chu, Polymer, 2002, 43, 4403. 70. D. Li and Y. Xia, Nano Letters, 2003, 3, 4, 555. 71. W-Z. Jin, H-W. Duan, Y-J. Zhang and F-F. Li in the Proceedings of the 1st IEEE International Conference on Nano/Micro Engineered and Molecular Systems, Zhuhai, China, 2006, p.42. 72. Y.J. Ryu, H.Y. Kim, K.H. Lee, H.C. Park and D.R. Lee, European Polymer Journal, 2003, 39, 1883. 73. X.M. Mo, C.Y. Xu, M. Kotaki and S. Ramakrishna, Biomaterials, 2004, 25, 1883. 74. S. Zhao, X. Wu, L. Wang and Y. Huang, Journal of Applied Polymer Science, 2004, 91, 242.
44
3
Structure Formation of Polymeric Nanofibres in Electrospinning
3.1 Introduction Conventional fibre spinning (e.g., melt, dry and wet spinning) produces fibres with diameters in the micrometres range. In recent years, electrospinning has gained much attention as a useful method to prepare fibres with diameters in the nanometer range [1–4]. These ultra-fine fibres are classified as ‘nanofibres’. The unique combination of high specific surface area [3], extremely small pore size [5], flexibility ad superior mechanical performance [6] makes nanofibres a preferred material form for many applications. Proposed uses of nanofibres include tissue engineering [7–9], drug delivery [10, 11], wound dressing [12, 13], protective clothing [5], filtration [14], reinforcement [6, 15], electronic applications [16–19] and aerospace-based applications [20]. In the electrospinning process, a polymer solution held by its surface tension at the end of a capillary tube is subjected to an electric field. Charge is induced on the liquid surface by an electric field. Mutual charge repulsion causes a force directly opposite to the surface tension. As the intensity of the electric field is increased, the hemispherical surface of the solution at the tip of the capillary tube elongates to form a conical shape known as the ‘Taylor cone’. When the electric field reaches a critical value at which the repulsive electric force overcomes the surface tension force, a charged jet of the solution is ejected from the tip of the cone. This jet is charged, so its trajectory can be controlled by an electric field. As the jet travels in air, the solvent evaporates, leaving behind a charged polymer fibre which lays itself randomly on a collecting metal screen. Thus, continuous fibres are laid to form a nonwoven fabric [2–4]. Figure 3.1 illustrates the electrospinning setup. Analysing the electrospun non-woven webs yield results and information which helps researchers to improve the quality and prediction of the overall performance of the electrospun webs. Some of the reasons for characterisation probably are process control, process development, product control or quality control. Physical and mechanical properties of non-woven textiles are dependent upon the material properties of the component fibre (e.g., melting temperature and glass transition temperature) as well as its structural characteristics (e.g., fibre orientation [21, 22],
45
Advances in Nanofibre Research fibre diameter [23], pore size [24], uniformity [25] and other structural features [26]). The properties of electrospun non-woven webs are also determined by the aforementioned features. However, in this case, fibre diameter is more important than the others. The diameter of the fibres produced by electrospinning varies depending on variables related to the process and material. Understanding how fibre diameter and its distribution are affected by the electrospinning parameters is essential to produce webs with desired properties. Many researchers have reported the effects of processing variables on electrospun fibre diameter. These have been reported despite a lack of standard technique to measure the fibre diameter and analyse its distribution. This explains the importance of the study of the fibre diameter of electrospun webs. Recently, image analysis has been used to identify fibres and to measure structural characteristics in non-woven mats [21–26].
Syringe Metering pump
Connector 0
1
Electrical heater High voltage supply
25 ºC
Temperature controller
Figure 3.1 Electrospinning setup
The objective of this chapter is to use image analysis for measuring the diameter of electrospun fibres. Two methods are presented; distance transform and direct tracking. The methods are compared with the conventionally used manual method and tested with some samples with known characteristics generated by a simulation algorithm.
46
Structure Formation of Polymeric Nanofibres in Electrospinning
3.2 Methodology 3.2.1 Simulation of Electrospun Webs Reliable evaluation of the accuracy of the developed methods requires samples with known characteristics. It is almost impossible to obtain real electrospun webs with specific characteristics through experimentation, and a method that precisely measures fibre diameter to compare the results with is not available. Hence, the method will not be well-evaluated using only real webs. To that end, a simulation algorithm has been employed for generating non-wovens with known characteristics. The physical characteristics of simulated images are known exactly, so one can employ them to test the usefulness of algorithms used in characterising diameter and other structural features. Simulation algorithms were first proposed by Abdel-Ghani and co-workers [27] and Pourdeyhimi and co-workers [21] for creation of nonwovens with continuous and discontinuous fibres using straight or curved lines. The most important component of simulation is the way in which lines or curves are generated. Abdel-Ghani and co-workers [27] presented three methods for generating a random network of lines: • Surface randomness (‘S-randomness) • Mean free path (‘µ-randomness’) • Internal randomness (‘I-randomness’) Under the first scheme, the position of intersection of the line with the image boundary is chosen and a line with a specified slope drawn from this point. Under the third scheme, a point in the image plane is chosen at random, then a slope is selected from an appropriate distribution, and a line is drawn to pass through the point with the corresponding slope [27]. However, neither of the two procedures described above is appropriate for simulation of non-wovens of continuous fibres because they both produce biased arrays. The aim of the simulation is to obtain unbiased arrays which are spatially homogeneous. It was revealed by Pourdeyhimi and co-workers [21] that the best way to simulate non-wovens of continuous fibres is through the second method. For continuous fibres, it is assumed that the lines are infinitely long so that, in the image plane, all lines intersect the boundaries. Under this scheme (Figure 3.2), a line with a specified thickness is defined by the perpendicular distance d from a fixed reference point O located in the centre of the image and the angular position of the perpendicular α. Distance d is limited to the diagonal of the image [27].
47
Advances in Nanofibre Research
a d
O
Figure 3.2 Procedure for µ-randomness
Several variables are allowed to be controlled during the simulation: • Web density can be controlled using the line density (which is the number of lines to be generated in the image) • Angular density is useful for generating fibrous structures with specific orientation distribution. The orientation may be sampled from a normal or a uniform random distribution • Distance from the reference point normally varies between zero and the diagonal of the image. It is restricted by the boundary of the image and is sampled from a uniform random distribution • Line thickness (fibre diameter) is sampled from a normal distribution. The mean diameter and its standard deviation are needed • Image size can also be chosen as required
3.2.2 Fibre Diameter Measurement The first step in determining fibre diameter is to produce a high-quality image of the web (called a micrograph) at a suitable magnification using electron microscopy techniques. The methods for measuring electrospun fibre diameter are described in the following sections.
3.2.2.1 Manual Method The conventional method of measuring the fibre diameter of electrospun webs is to analyse the micrograph manually. The manual analysis usually involves determining
48
Structure Formation of Polymeric Nanofibres in Electrospinning the length of a pixel of the image (setting the scale), identifying the edges of the fibres in the image and counting the number of pixels between two edges of the fibre (the measurements are made perpendicular to the direction of fibre-axis), converting the number of pixels to nanometres using the scale and recording the result. Typically 100 measurements are carried out (Figure 3.3). However, this process is tedious and time-consuming (especially for many samples). Furthermore, it cannot be used as an online method for quality control because an operator is needed for carrying out the measurements. Thus, developing automated techniques which eliminate the use of an operator and which can be employed as online quality control is of great importance.
Figure 3.3 Manual method
3.2.2.2 Distance Transform The distance transform of a binary image is the distance from every pixel to the nearest non-zero-valued pixel. The centre of an object in the distance-transformed image will have the highest value and lie exactly over the object’s ‘skeleton’. The skeleton of the object can be obtained by ‘skeletonisation’ or ‘thinning’. The algorithm removes pixels on the boundaries of objects but does not allow objects to break apart. This reduces a thick object to its corresponding object with one-pixel width. Skeletonisation or thinning often produces short spurs which can be cleaned up automatically with a ‘pruning’ procedure [28].
49
Advances in Nanofibre Research The algorithm for determining fibre diameter uses a binary input image and creates its skeleton and distance-transformed image. The skeleton acts as a guide for tracking the distance-transformed image by recording the intensities to compute the diameter at all points along the skeleton. This method was proposed by Pourdeyhimi and coworkers [23]. Figure 3.4 shows a simple simulated image comprising five fibres with diameters of 10, 13, 16, 19 and 21 pixels, together with its skeleton and distance map (including the histogram of the fibre diameter obtained by this method).
(a)
(b)
250
Distance Transform
Frequency
200 150 100 50 0
(c)
0
5
10 15 20 Diameter (pixel)
25
30
(d)
Figure 3.4 (a) A simple simulated image; (b) skeleton of (a); (c) distance map of (a) after pruning; and (d) histogram of fibre diameter distribution obtained by the distance-transform method
50
Structure Formation of Polymeric Nanofibres in Electrospinning
3.2.2.3 Direct Tracking The direct tracking method uses a binary image as an input data to determine fibre diameter based on information acquired from two scans: first a horizontal and then a vertical scan. In the horizontal scan, the algorithm searches for the first white pixel adjacent to a black pixel. Pixels are counted until reaching the first black pixel. The second scan is then started from the mid-point of the horizontal scan and pixels are counted until the first black pixel is encountered. The direction changes if the black pixel is not found. Having the number of horizontal and vertical scans, the number of pixels in a perpendicular direction (which is the fibre diameter) can be measured from a geometrical relationship (Figure 3.5).
Figure 3.5 Diameter measurement based on two scans in the direct tracking method
In electrospun non-woven webs, nanofibres cross each other at intersection points. This brings about the possibility for some untrue measurements of fibre diameter in these regions. To circumvent this problem, ‘fibre identification’ is employed. First, black regions are labeled and a couple of regions between which a fibre exists is selected. In the next step, the two selected regions are connected by carrying out a ‘dilation’ operation with a sufficiently large ‘structuring element’. Dilation is an operation that grows or thickens objects in a binary image by adding pixels to the boundaries of objects. The specific manner and extent of this thickening is controlled by the size and shape of the structuring element used [28, 29]. In the subsequent process, an ‘erosion’ operation with the same structuring element is done and the fibre is recognised. Erosion shrinks or thins objects in a binary image by removing pixels on object boundaries. As in dilation, the manner and extent of shrinking is controlled by a structuring element [28, 29]. Then, to enhance the processing speed, the image
51
Advances in Nanofibre Research is cropped to the size of selected regions. Afterwards, fibre diameter is measured according to the previously explained algorithm. This trend is continued until all of the fibres are analysed. Finally, the data in pixels may be converted to nanometres and the histogram of the distribution of fibre diameters plotted. Figure 3.6 shows a labeled simulated image and the histogram of fibre diameter obtained by this method.
300
Direct Tracking
Frequency
250 200 150 100 50 0
(a)
0
5
10 15 20 Diameter (pixel)
25
30
(b)
Figure 3.6 a) A simple simulated image which is labeled, and b) a histogram of the distribution of fibre diameter obtained by direct tracking
3.2.3 Real Webs Treatment The distance-transform and direct tracking algorithms for measuring fibre diameter require binary images as inputs. Hence, the micrographs first have to be converted to black and white. This can be carried out by ‘thresholding’ (also known as ‘segmentation’), which produces a binary image from a greyscale (intensity) image [28, 29]. This is a critical step because the segmentation affects the result. Before the segmentation, an ‘intensity adjustment’ operation and a two-dimensional ‘median’ filter are applied to enhance the contrast of the image and remove noise. In the simplest thresholding technique, called ‘global thresholding’, the image is partitioned using a single constant threshold. A simple way to choose a threshold is by trial and error. Then each pixel is labeled as ‘object’ or ‘background’ depending on whether the grey level of that pixel is greater or less than the value of threshold, respectively [28, 29].
52
Structure Formation of Polymeric Nanofibres in Electrospinning The main problem of global thresholding is its possible failure in the presence of nonuniform illumination or local grey-level unevenness. An alternative to circumvent this problem is to use ‘local thresholding’ instead. In this approach, the original image is divided to sub-images and different thresholds are used for segmentation [28, 29]. Another variant of this approach (which has been used in this contribution) consists of estimating the background illumination using a ‘morphological opening’ operation, subtracting the obtained background from the original image, and applying a global thresholding to produce the binary version of the image. The morphological opening is a sequential application of an erosion operation followed by a dilation operation (i.e., opening = erosion + dilation) using the same structuring element [28, 29]. Otsu’s method could be employed to automatically compute the appropriate threshold [30]. This method chooses the threshold to maximise the interclass variance and minimise the intraclass variance of the black and white pixels. As shown in Figure 3.7, global thresholding resulted in some broken fibre segments. This problem was solved using local thresholding.
(a)
(b)
(c)
Figure 3.7 a) A real web; b) global thresholding; and c) local thresholding
53
Advances in Nanofibre Research
3.3 Experimental Electrospun non-woven webs used as real webs in image analysis were obtained from electrospinning of polyvinyl alcohol (PVA) with an average molecular weight of 72000 g/mol (Merck) at different processing parameters. The micrographs of the webs were obtained using a Philips (XL-30) environmental scanning electron microscope (SEM) under 10000× magnification after being gold-coated.
3.4 Results and Discussion Two sets-each comprising five simulated images generated by the µ-randomness procedure were used as samples with known characteristics to demonstrate the validity of the techniques. The first set had random orientation with increasing constant diameters; the second was also randomly oriented but with a varying diameter sampled from normal distributions with a mean of 15 pixels and standard deviations ranging from 2 pixels to 10 pixels. Table 3.1 and Table 3.2 show the structural features of these simulated images which are shown in Figure 3.8 and Figure 3.9.
Table 3.1 Structural characteristics of images in the first set Image number
Angular range
Line density
Line thickness
C1
0–360
30
5
C2
0–360
30
10
C3
0–360
30
15
C4
0–360
30
20
C5
0–360
30
25
Table 3.2 Structural characteristics of images in the second set Image number Angular range
54
Line density
Line thickness Mean
Standard deviation
V1
0–360
30
15
2
V2
0–360
30
15
4
V3
0–360
30
15
6
V4
0–360
30
15
8
V5
0–360
30
15
10
Structure Formation of Polymeric Nanofibres in Electrospinning
(C1)
(C2)
(C3)
(C4)
(C5)
Figure 3.8 Simulated images with constant diameter
55
Advances in Nanofibre Research
(V1)
(V2)
(V3)
(V4)
(V5)
Figure 3.9 Simulated images with varying diameter
The mean and standard deviation of nanofibre diameter for the first and second set of simulated images obtained by different methods are shown in Table 3.3 and Table 3.4, respectively. Figure 3.10 and Figure 3.11 also show histograms of the distribution of fibre diameter for simulated images for the first and second set, respectively.
56
Structure Formation of Polymeric Nanofibres in Electrospinning 0.7 0.6
0.35
0.5 0.4 0.3 0.2
0.3 0.25 0.2 0.15 0.1
0.1 0
Distance Transform Direct Tracking Distance Transform Direct Tracking
0.4 Probability Density
Probability Density
0.45
Distance Transform Direct Tracking Distance Transform Direct Tracking
0.05 0
2
4
6 8 10 Diameter (pixel)
12
0 0
14
5
10 15 Diameter (pixel)
(C1)
Distance Transform Direct Tracking Distance Transform Direct Tracking
0.16 0.14 0.12 0.1 0.08 0.06
0.3
Distance Transform Direct Tracking Distance Transform Direct Tracking
0.25 Probability Density
0.2
0.04
0.2 0.15 0.1 0.05
0.02 0
5
10
15 20 25 Diameter (pixel)
30
35
40
0
10
0
20 30 Diameter (pixel)
(C3)
40
50
(C4)
0.16
Distance Transform Direct Tracking Distance Transform Direct Tracking
0.14 Probability Density
Probability Density
25
(C2)
0.18
0
20
0.12 0.1 0.08 0.06 0.04 0.02 0 0
10
20
30 40 50 Diameter (pixel)
60
70
(C5)
Figure 3.10 Histograms for simulated images with constant diameter
57
Advances in Nanofibre Research
Table 3.3 Values of mean and standard deviation for series 1 Simulation
Distance transform Direct tracking
C1
C2
C3
C4
C5
Mean
5
10
15
20
25
Standard deviation
0
0
0
0
0
Mean
5.486
10.450
16.573
23.016
30.063
Standard deviation
1.089
2.300
5.137
6.913
10.205
Mean
5.625
11.313
17.589
22.864
29.469
Standard deviation
1.113
2.370
4.492
5.655
7.241
Table 3.4 Values of mean and standard deviation for series 2 Simulation
Distance transform Direct tracking
V1
V2
V3
V4
V5
Mean
15.247
15.350
15.243
15.367
16.628
Standard deviation
1.998
4.466
5.766
8.129
9.799
Mean
16.517
16.593
17.135
17.865
19.394
Standard deviation
5.350
6.165
7.597
9.553
11.961
Mean
16.075
15.803
16.252
16.770
18.756
Standard deviation
2.606
5.007
6.129
9.319
10.251
In the first set, for simulated images with line thickness of 5 pixels and 10 pixels, distance transform presents closer values of mean and standard deviation of fibre diameter to the fibre diameter of the simulated picture. For the line thickness of 15 pixels, the standard deviation of diameter obtained from the direct tracking method is closer to the data artificially made by simulation. However, in this case, distance transform measured the average diameter more accurately. For the simulated webs with a line thickness >15 pixels in the first set, the direct tracking method resulted in better estimation of the mean and standard deviation of fibre diameter. This is because as the lines get thicker, there is a higher possibility of branching during skeletonisation (or thinning) and these branches remain even after pruning. Although these branches are small, their orientation is typically normal to the fibre axis, thus widening the distribution obtained by the distance transform method. For fibres with small diameters, however, these branches are lower in number and more accurate measurements are made by distance transform. 58
Structure Formation of Polymeric Nanofibres in Electrospinning
Simulation Distance Transform Direct Tracking Simulation Distance Transform Direct Tracking
Probability Density
0.18 0.15 0.12 0.09 0.06
0.14
0.1 0.08 0.06 0.04 0.02
0.03 0
Simulation Distance Transform Direct Tracking Simulation Distance Transform Direct Tracking
0.12 Probability Density
0.21
5
0
10
15 20 25 Diameter (pixel)
30
35
0
40
0
5
10
(V1)
0.08 0.06 0.04
0.06
35
40
Simulation Distance Transform Direct Tracking Simulation Distance Transform Direct Tracking
0.05 Probability Density
Simulation Distance Transform Direct Tracking Simulation Distance Transform Direct Tracking
0.1
0.02
0.04 0.03 0.02 0.01
0
5
10
15 20 25 30 Diameter (pixel)
35
40
45
0
0
5
10 15
(V3)
20 25 30 35 40 45 Diameter (pixel)
50 55
(V4)
0.08
Simulation Distance Transform Direct Tracking Simulation Distance Transform Direct Tracking
0.07 Probability Density
Probability Density
30
(V2)
0.12
0
15 20 25 Diameter (pixel)
0.06 0.05 0.04 0.03 0.02 0.01 0
0
10
20
30 40 50 Diameter (pixel)
60
70
(V5)
Figure 3.11 Histograms for simulated images with varying diameter
59
Advances in Nanofibre Research Furthermore, in the distance transform method, the value of the centre of the object in the distance map is related to fibre diameter only for one single fibre. At intersections where two or more fibres cross each other, it is associated to more than one fibre and is no longer related to fibre diameter. Both of the distance-transformed images and the skeleton are broken at intersections. The problem becomes more serious as fibres get thicker and for points where more fibres cross each other. Hence, the method fails in measuring fibre diameter at intersections. This causes an overestimation of fibre diameter. In the direct tracking method, the image is divided into parts where single fibres exist, so the effect of intersections which causes inaccurate measurement of fibre diameter is eliminated. Therefore, there will be a better estimation for fibre diameter. In the second set, regardless of the line thickness in the simulation, for all simulated webs, direct tracking resulted in better measurement of the mean and standard deviation of fibre diameter. Note that the mean and standard deviation of diameter for the simulated images with varied diameter are slightly different to those set as the simulation. There are several reasons for the deviation of the computed results using direct tracking method and true gathered results. The differences observed can be attributed to the failure of the technique to correctly distinguish between multiple fibres being joined together and a single fibre. Also, a one-pixel error occurs in the selection of the midpoint pixel (as a starting point for the second scan) if the number of pixels in the first scan is even. Furthermore, fibre segments must be of minimum lengths so that the diameter may be measured. For dense webs or dense regions in a web, the process of fibre identification creates some artifacts other than fibres which result in untrue measurements. Further advancements in this field could improve the process of fibre identification and try to circumvent the other problems mentioned. The applicability of the techniques was also tested using five real webs obtained from the electrospinning of PVA. SEM micrographs of the webs (Figure 3.12) were first thresholded for diameter measurement. The fibre-diameter distributions were determined for each image using distance-transform and direct tracking methods and the results compared with those obtained by the manual method. Table 3.5 shows the results for real webs in terms of pixel and nanometres. Histograms for real webs are given in Figure 3.13. For the real webs, the mean and standard deviation of fibre diameter for direct tracking were closer to those of the manual method, which is in accordance with the trends observed for the simulated images.
60
Structure Formation of Polymeric Nanofibres in Electrospinning
(R1)
(R2)
(R3)
(R4)
(R5)
Figure 3.12 Micrographs of electrospun webs
61
Advances in Nanofibre Research 0.1
0.07
0.08
0.06 0.05 0.04 0.03
0.07 0.06 0.05 0.04 0.03
0.02
0.02
0.01
0.01
0
0
100
200
300 400 500 600 Fiber Diameter (nm)
Manual Distance Transform Direct Tracking Manual Distance Transform Direct Tracking
0.09 Probability Density
0.08 Probability Density
0.1
Manual Distance Transform Direct Tracking Manual Distance Transform Direct Tracking
0.09
0
700 800
0
100
200
(R1)
0.12
Manual Distance Transform Direct Tracking Manual Distance Transform Direct Tracking
0.14 Probability Density
Probability Density
0.16
Manual Distance Transform Direct Tracking Manual Distance Transform Direct Tracking
0.14
0.1 0.08 0.06 0.04 0.02
0.12 0.1 0.08 0.06 0.04 0.02
0
100
200 300 400 Fiber Diameter (nm)
500
600
0
0
100
200 300 400 500 Fiber Diameter (nm)
(R3)
(R4)
0.16
Manual Distance Transform Direct Tracking Manual Distance Transform Direct Tracking
Probability Density
0.14 0.12 0.1 0.08 0.06 0.04 0.02 0
0
100
200 300 400 500 Fiber Diameter (nm)
600
700
(R5)
Figure 3.13 Histograms for real webs
62
700 800
(R2)
0.16
0
300 400 500 600 Fiber Diameter (nm)
600
700
Structure Formation of Polymeric Nanofibres in Electrospinning
Table 3.5 Values of mean and standard deviation for real webs Manual
Distance transform
Direct tracking
R1
R2
R3
pixel
24.358
24.633
18.583
18.827 17.437
nm
318.67
322.27
243.11
246.31 228.12
Standard deviation
pixel
3.193
3.179
2.163
1.984
2.230
nm
41.77
41.59
28.30
25.96
29.18
Mean
pixel
27.250
27.870
20.028
23.079 20.345
nm
356.49
364.61
262.01
301.94 266.17
Standard deviation
pixel
8.125
7.462
4.906
7.005
6.207
nm
106.30
97.62
64.18
91.64
81.21
Mean
pixel
27.195
27.606
20.638
21.913 20.145
nm
355.78
361.15
269.99
286.68 263.55
pixel
4.123
5.409
4.148
4.214
3.800
nm
53.94
70.77
54.27
55.14
49.72
Mean
Standard deviation
R4
R5
In addition to the reasons mentioned above, the small discrepancies between these results may also be attributed to the different number of measurements utilised in each technique. Distance transform and direct tracking measures >1000 diameters. In the manual method, however, the number of measurements is limited to mostly 100 due to the time-consuming nature of the procedure.
3.5 Conclusion Fibre diameter is the most important structural characteristic in electrospun nonwoven webs. The typical way of measuring electrospun fibre diameter is through the manual method. This is a tedious, time-consuming and operator-based method that cannot be used as an automated technique for quality control. The use of image analysis was investigated in this chapter for determining fibre diameter and development of an automated method called direct tracking. This is a new technique so its accuracy needs to be evaluated using samples with known characteristics. To that end, the µ-randomness procedure was used to simulate electrospun non-woven webs. Based on this scheme, two sets of simulated images (each containing 5 webs) were generated. The first set had random orientation with increasing constant diameter.
63
Advances in Nanofibre Research For the second set, the diameter values were sampled from normal distributions with a mean of 15 pixels and a standard deviation ranging from 2 pixels to 10 pixels. The results obtained from our method were compared with those data acquired by the distance-transform method. For all the simulated webs with varying diameter and for those with constant diameter >15, the direct tracking method resulted in a value of mean and standard deviation closer to the simulation. However, for the simulated webs with smaller constant diameter, distance transform measured the mean and standard deviation of fibre diameter more accurately. These results suggest that the direct tracking method is an accurate, direct measurement technique because it extracts the fibre diameter for the samples by tracking fixed segments of the fibre and eliminating the effect of intersections. The general applicability of the method using real webs was also demonstrated using five real electrospun non-woven webs obtained by the electrospinning of PVA. The methods needed binary images as inputs, so the images initially had to be segmented. A local thresholding method together with Otsu’s method was employed to automatically compute the appropriate threshold. The results obtained for real webs confirm the trends suggested by simulated images. The mean and standard deviation values obtained by direct tracking were significantly closer to the manual method compared with those obtained by distance transform. This suggested that direct tracking could, in general, perform better but distance transform may produce more accurate results in webs with very low fibre diameter. The results show that the use of image analysis to determine the fibre diameter in electrospun non-woven webs has been successful.
References 1.
A.K. Haghi and M. Akbari, Physica Status Solidi A, 2007, 204, 1830.
2.
J. Doshi and D.H. Reneker, Journal of Electrostatics, 1995, 35, 151.
3.
H. Fong and D.H. Reneker in Structure Formation in Polymeric Fibres, Ed., D.R. Salem, Hanser, Cincinnati, OH, USA, 2001, p.225.
4.
T. Subbiah, G.S. Bhat, R.W. Tock, S. Parameswaran and S.S. Ramkumar, Journal of Applied Polymer Science, 2005, 96, 557.
5.
P.W. Gibson, H.L. Schreuder-Gibson and D. Rivin, AIChE Journal, 1999, 45, 190.
6.
Z-M. Huang, Y-Z. Zhang, M. Kotaki and S. Ramakrishna, Composites Science Technology, 2003, 63, 2223.
64
Structure Formation of Polymeric Nanofibres in Electrospinning 7.
M. Li, M.J. Mondrinos, M.R. Gandhi, F.K. Ko, A.S. Weiss and P.I. Lelkes, Biomaterials, 2005, 26, 5999.
8.
E.D. Boland, B.D. Coleman, C.P. Barnes, D.G. Simpson, G.E. Wnek and G.L. Bowlin, Acta Biomaterialia (italics), 2005, 1, 115.
9.
J. Lannutti, D. Reneker, T. Ma, D. Tomasko and D. Farson, Materials Science Engineering: C, 2007, 27, 504.
10. J. Zeng, L. Yang, Q. Liang, X. Zhang, H. Guan, X. Xu, X. Chen and X. Jing, Journal of Controlled Release, 2005, 105, 43. 11. E.R. Kenawy, G.L. Bowlin, K. Mansfield, J. Layman, D.G. Simpson, E.H. Sanders and G.E. Wnek, Journal of Controlled Release, 2002, 81, 57. 12. M. Khil, D. Cha, H. Kim, I. Lim and N. Bhattarai, Journal of Biomedical Materials Research, Part B: Applied Biomaterials, 2003, 67, 675. 13. B.M. Min, G. Lee, S.H. Kim, Y.S. Nam, T.S. Lee and W.H. Rark, Biomaterials, 2004, 25, 1289. 14. X.H. Qin and S.Y. Wang, Journal of Applied Polymer Science, 2006, 102, 1285. 15. J.S. Kim and D.H. Reneker, Polymer Engineering & Science, 1999, 39, 849. 16. A.G. MacDiarmid, W.E. Jones, I.D. Norris, J. Gao, A.T. Johnson, N.J. Pinto, J. Hone, B. Han, F.K. Ko, H. Okuzaki and M. Llaguno, Synthetic Metals, 2001, 119, 27. 17. N.J. Pinto, A.T. Johnson, A.G. MacDiarmid, C.H. Mueller, N. Theofylaktos, D.C. Robinson and F.A. Miranda, Applied Physics Letters, 2003, 83, 4244. 18. D. Aussawasathien, J-H. Dong and L. Dai, Synthetic Metals, 2005, 154, 37. 19. S.W. Lee, S.W. Choi, S.M. Jo, B.D. Chin, D.Y. Kim and K.Y. Lee, Journal of Power Sources, 2006, 163, 41. 20. G. Zhang, W. Kataphinan, R. Teye-Mensah, P. Katta, L. Khatri, E.A. Evans, G.G. Chase, R.D. Ramsier and D.H. Reneker, Materials Science and Engineering: B, 2005, 116, 353. 21. B. Pourdeyhimi, R. Ramanathan and R. Dent, Textile Research Journal, 1996, 66, 713.
65
Advances in Nanofibre Research 22. B. Pourdeyhimi, R. Dent, A. Jerbi, S. Tanaka and A. Deshpande, Textile Research Journal, 1999, 69, 185. 23. B. Pourdeyhimi and R. Dent, Textile Research Journal, 1999, 69, 233. 24. A.H. Aydilek, S.H. Oguz and T.B. Edil, Journal of Computing in Civil Engineering, 2002, 16, 280. 25. R. Chhabra, International Nonwoven Journal, Spring 2003, 43. 26. B. Xu and Y.L. Ting, Textile Research Journal, 1995, 65, 41. 27. M.S. Abdel-Ghani and G.A. Davies, Chemical Engineering Science, 1985, 40, 117. 28. R.C. Gonzalez and R.E. Woods in Digital Image Processing, 2nd Edition, Prentice Hall, New Jersey, NJ, USA, 2001, p.28. 29. M. Petrou and P. Bosdogianni in Image Processing: the Fundamentals, John Wiley and Sons, Chichester, UK, 1999. 30. N. Otsu, IEEE Transactions on Systems Man and Cybernetics, Part C, 1979, 9, 62.
66
4
Optimisation of the Electrospinning Process
4.1 Introduction In general, fibres with a diameter of ~100 nm are classified as nanofibres. What makes nanofibres of great interest is their extremely small size. Compared with conventional fibres, nanofibres have higher surface area-to-volume ratios and smaller pore size. They can be used in a wide variety of applications. To date, the most successful method of producing nanofibres is through electrospinning. The electrospinning process uses a high voltage to create an electric field between a droplet of polymer solution at the tip of a needle and a collector plate. When the electrostatic force overcomes the surface tension of the drop, a charged, continuous jet of polymer solution is ejected. As the solution moves away from the needle and towards the collector, the solvent evaporates and the jet rapidly thins and dries. On the surface of the collector, a non-woven web of randomly oriented solid nanofibres is deposited [1–6]. Figure 4.1 illustrates the electrospinning setup. The properties of electrospun nanofibre webs are dependent upon not only the nature of the component fibres but also on their structural characteristics. In the last few years, image analysis methods have been developed to identify fibres and measure non-woven characteristics such as fibre orientation [7-14], fibre diameter [15, 16], pore size [17], [18], uniformity [19] and other structural features [13, 20]. However, these are new techniques and their accuracy and limitations have not been verified, so samples with known characteristics are required to evaluate the accuracy of the methods: these can be produced by simulation schemes [7, 21]. Fibre diameter is the most important structural characteristic in electrospun nanofibre webs. Despite the importance, there is no successful method for determining fibre diameter and a few works have been conducted to develop a method for measuring it. Furthermore, large-scale production of nanofibres requires unique online quality control. Hence, developing an accurate and automated technique for measurement of fibre diameter is crucial. In a method proposed by Pourdyhimi and co-workers [15], image analysis has been used to measure fibre diameter in non-woven textiles. Nevertheless, the method has problems at the intersections of fibres, making it
67
Advances in Nanofibre Research inefficient for measuring electrospun nanofibre diameter. In this contribution, an attempt has been made to circumvent the problems associated with this method to develop a reliable, efficient and automated method for measuring nanofibre diameter in electrospun webs.
Syringe Metering pump
Connector 0
1
Electrical heater High voltage supply
25 ºC
Temperature controller
Figure 4.1 Electrospinning setup
4.2 Methodology 4.2.1 Measurement of Fibre Diameter Understanding how fibre diameter and its distribution are affected by the electrospinning variables is essential to produce nanofibres with desired properties. The extremely small fibre size and random production of nanofibres make measurement of their diameter very difficult. Most commercially available measurement equipment cannot work with nanofibres [16]. To measure fibre diameter, images of the webs are required. These images (called ‘micrographs’) are usually obtained by scanning electron microscopy (SEM), transmission electron microscopy (TEM) or atomic force
68
Optimisation of the Electrospinning Process microscopy (AFM). Dealing with fibre diameter requires high-quality images with appropriate magnifications. The methods for measuring fibre diameter are presented below.
4.2.1.1 Manual Method Routine measurement of fibre diameter and its distribution are carried out by a manual method using micrographs obtained from SEM. First the length of a pixel in the image is determined (i.e., the scale is set). Then a fibre is selected and the pixels between two edges of the fibre perpendicular to the fibre axis are counted. The number of the pixels is then converted to nanometres using the scale and the resulting diameter recorded. This procedure is repeated for other selections until any fibre is processed. Typically 100 diameters are measured (Figure 4.2). Finally the histogram of the distribution of fibre diameters is plotted.
Figure 4.2 Manual method
This process is very time-consuming, and operator consistency and fatigue can reduce the accuracy. Identifying the edges of the fibres needs attention and the measurements are not made exactly perpendicular to the fibre axis. Furthermore, it is an operator-based method, so it cannot be used as an online method for quality control. Automating the measurement of fibre diameter eliminates the use of an operator and is a natural solution to this problem.
69
Advances in Nanofibre Research
4.2.1.2 Distance Transform Method The ‘skeleton’ of an object in a binary image, which provides helpful information about the shape of the object, is defined as the corresponding object with one-pixel width. There are two approaches for assessing the object’s skeleton: ‘skeletonisation’ and ‘thinning’. In the former, using a medial-axis transformation, the centre points of the object which are equidistant from the two closest points of the object’s boundary are attained and set to as the skeleton [22]. In thinning, the pixels on the boundary of the object are removed without allowing it to break apart, thereby shrinking a thick object to a centrally located object with one-pixel width. In thinning operations, the following conditions must be satisfied: • An object must not break into pieces • The endpoints must not be removed so that the object does not become shorter • An object must not be deleted [23] ‘Skeletonisation’ and ‘thinning’ result in line-like structures with one-pixel in thickness preserving the topology of the object. However, the skeleton obtained by skeletonisation is often different to that obtained by thinning and has more branches. Figure 4.3a shows a binary image to which skeletonisation and thinning is applied and the resultant skeletons are depicted in Figure 4.3b and Figure 4.3c, respectively. These operations often produce short ‘spurs’ (also called ‘parasitic components’) which may be cleaned up further by a post-processing procedure called ‘pruning’ (Figure 4.3d). Identifying and removing the spurs iteratively, this procedure is an essential complement to skeletonisation and thinning [24]. The distance transform is an operation which is applied to a binary image consisting of 1s and 0s corresponding to objects and background respectively. it results in a greyscale image often called a ‘distance map’ (known also as a ‘distance-transformed image’). For each pixel in the binary image, the corresponding pixel in the distance map has a value equal to the minimum distance between that pixel and the closest object pixel, i.e., the distance from that pixel to the nearest non-zero-valued pixel [23, 24]. There are several sorts of distance transform according to which distance metric is being used to measure the distance between the pixels. Three common distance metrics used in this approach are: ‘city block’, ‘chessboard’ and ‘Euclidean’. The city-block distance gives the length of a path between the pixels according to a four-connected neighbourhood (moving only in horizontal and vertical directions). The city-block distance between (x1,y1) and (x2,y2) is given by:
(4.1)
70
Optimisation of the Electrospinning Process
(a)
(b)
(c)
(d)
Figure 4.3 Obtaining the skeleton of a binary image: a) a binary image; b) skeletonisation; c) thinning; and d) resulting skeleton after pruning
In contrast, the chessboard-distance metric measures the path between the pixels based on an eight-connected neighbourhood (diagonal move is also allowed) as if a King moves in a game of chess. This metric is given by:
(4.2) With the city-block metric, distances in the direction of diagonals are longer, resulting in diamond-shaped structures. If a chessboard metric is used, square-shaped structures are obtained [23]. Even though it could be used in certain applications, a Euclidean metric is more practical and relevant because it is the only one that preserves the isotropy of the continuous space (Figure 4.4) [25–27]. The Euclidean distance (which is the straight-line distance between two pixels) is defined as:
(4.3)
71
Advances in Nanofibre Research
(a)
(b)
(c)
(d)
Figure 4.4 Distance map of a binary image: a) a small binary image and its distance map obtained by b) city block; c); chessboard; and d) Euclidean metric
The centre of an object in a distance-transformed image has the highest value which coincides with the axis of the object. Interestingly, the skeleton of an object will lie exactly over the maximum of the distance transform for that object [15]. This fact is clearly demonstrated in Figure 4.5. Note that the z-position of the skeleton in Figure 4.5d is quite arbitrary, just to show the coincidence. Serving as the basic component of the methods, this remarkable feature will be utilised later for determining the distribution of nanofibre diameters in electrospun webs. The algorithm for determining fibre diameter uses the skeleton and distance map of binary input images. Our images consist of light fibres on a dark background, so they first need to be complemented. In the complement of a binary image, zeros become ones and ones become zeros; black and white are reversed [24]. Thus, fibres become black and the background white. The complemented image is used to create a distancetransformed image. Then the skeleton of the objects is created from the input binary image by skeletonisation or thinning. Fibre diameter is then determined using the distance-transformed image and the skeleton. The latter acts as a guide for tracking the distance map, and distances at all points along the skeleton (which coincide with the centre of the objects in the distance-transformed image) are recorded to compute fibre diameters. Finally, the recorded results are doubled and fibre diameters obtained. The values (in pixels) may be converted further to nanometres and the histogram
72
Optimisation of the Electrospinning Process of fibre diameter distribution plotted. This method was proposed by Pourdeyhimi and co-workers [15]. Figure 4.6 shows a simple simulated image together with its skeleton and distance map (including the histogram of the fibre diameter obtained by this method).
(a)
(b)
(c)
Distance
15 10 5
0 100
75
50
Height
25
0 0
25
50
75
100
Width
(d)
Figure 4.5 The skeleton lies exactly over the centre of a distance map: a) a simple 100 × 100 binary image; b) Euclidean distance map; c) skeleton obtained by thinning after pruning; and d) 3D plot showing the coincidence of the skeleton and the centre of the object in a distance map
73
Advances in Nanofibre Research
(a)
(b)
350 300
Frequency
250 200 150 100 50 0
(c)
5
10
15 20 25 Diameter (Pixel)
30
35
(d)
Figure 4.6 Distance transform method: a); a simple simulated image; b) skeleton of (a); c) distance map of (a) after pruning; and d) histogram of distribution of fibre diameters
4.2.1.3 New Distance Transform Method The problem of the distance transform method is that skeletons are often broken at intersection points. Furthermore, two or more fibres cross each other at the intersections, so the value of the centre of the object in the distance-transformed image does not coincide with the fibre diameter because it does not correspond to a single fibre. As depicted in Figure 4.7a, the intersections in the distance map are brighter if a single fibre is present. This demonstrates that higher values than expected were
74
Optimisation of the Electrospinning Process returned at these points. Figure 4.7b shows the broken skeleton at intersections. This problem becomes more pronounced as fibres get thicker and for points where more fibres cross each other. Hence, the distance transform method fails in measuring fibre diameter at intersections.
(a)
(b)
Figure 4.7 Failure of the distance transform method at intersection points: a) distance map of the image shown in Figure 4.6a, and b) broken skeleton obtained from thinning of Figure 4.6a (area around an intersection has been magnified for more clarity)
We modified the distance transform method so that the problems associated with the intersections are solved. Furthermore, in the method proposed by Pourdeyhimi and co-workers [15], a city-block distance transform was used which, as mentioned above, is not a realistic metric because it does not preserve the isotropy. To provide more rational results, we used a Euclidean distance metric. The method uses a binary image as an input. Hence, the distance-transformed image and its skeleton are created. To solve the problem of the intersections, these points are identified and deleted from the skeleton. First, to find the intersection points, a ‘sliding neighbourhood operation’ is employed. This is an operation that is applied to one pixel at a time; the value of that pixel in the output image is determined by implementation of a given function to the values of the corresponding input pixel’s neighbourhood (Figure 4.8). The neighbourhood about a pixel, which is usually called the ‘centre point’, is a square or rectangular region centred at that pixel. The operation consists of five steps:
75
Advances in Nanofibre Research • Defining a centre point and a neighbourhood block • Starting from the first (normally top left) pixel in the image • Carrying out an operation (a function given) that involves only the pixels in the defined block • Finding the pixel in the output corresponding to the centre pixel in the block and setting the result of the operation as the response at that pixel • Repeating steps 3 to 4 for each pixel in the input image [24]
f(x)
Figure 4.8 Sliding neighbourhood operation with a 3-by-3 neighbourhood block
At an intersection point, two or more fibres meet each other. It could therefore be defined as a location where a white pixel in the skeleton has more than two neighbouring pixels, each leading a branch. Hence, carrying out a sliding neighbourhood operation on the skeleton with a 3-by-3 sliding block and summation as the function (which is applied over all pixels in the block), the intersections could be identified as the points having values >3 (Figure 4.9; intersections are shown by arrows). After the intersection points have been located, the next step is to find the width of each one. This is carried out using the distance map of the binary input image by
76
Optimisation of the Electrospinning Process finding the pixel corresponding to that intersection point. The value of the distance map at the pixel is then considered to be the width of that intersection. After that, the pixels in the skeleton which lie inside a square with that width around the intersection point are cleaned. This procedure is replicated until each intersection is identified and cleaned. Figure 4.10a exhibits the skeleton of the simple simulated image shown in Figure 4.6a after deleting the intersection points followed by a pruning procedure.
Figure 4.9 Identifying intersection points using a sliding neighbourhood operation with a 3-by-3 neighbourhood block
Finally, the resultant skeleton (of which the intersections are deleted) is used as a guide for tracking the distance-transformed image. Fibre diameters are obtained by recording the intensities to at all points along the skeleton (white pixels in Figure 4.10a show the skeleton) and doubling the results. The distance map of the image in Figure 4.6a is also shown in Figure 4.10b for better understanding of the procedure. Setting the length of a pixel in the image, the values may then be converted to nanometres and the histogram of the distribution of fibre diameters plotted. Figure 4.10c demonstrates the histogram of fibre diameter (in term of pixels) obtained by this method. The procedure for determining fibre diameter via this approach is summarised in Figure 4.11. The method is efficient, reliable, accurate, rapid and can be used as an online method for quality control.
77
Advances in Nanofibre Research
(a)
(b)
200
Frequency
150
100
50
0
5
10
15 20 Diameter (Pixel)
25
30
(c)
Figure 4.10 New distance transform method: a) the skeleton of the simple simulated image shown in Figure 4.6a after deleting the intersection points; b) the distance map; and c) histogram of the distribution of fibre diameters
78
Optimisation of the Electrospinning Process
Figure 4.11 The new distance transform method
4.2.2 Validation of the Methods To validate the methods for determining fibre diameter, test samples with known characteristics are required. It is almost impossible to obtain real electrospun webs with specific characteristics through the experiment, and a method that precisely measures fibre diameter to compare the results with is lacking. Hence, the method will not be well evaluated using real webs only. To that end, a simulation algorithm has been employed for generating test samples (which are binary images that resemble electrospun webs) with known characteristics. A geometric model has been considered here to simulate electrospun fibrewebs. There are three widely used methods for generating a random network of lines. These are called ‘S-randomness’, ‘µ-randomness’ (suitable for generating a web of continuous filaments) and ‘I-randomness’ (suitable for generating a web of staple fibres). These methods have been described in elaborations by Abdel-Ghani and Davies [21] and Pourdeyhimi and co-workers [7]. The physical characteristics of simulations are known exactly, so one can employ them to test the usefulness of
79
Advances in Nanofibre Research algorithms used in characterising diameter and other structural features [7]. In this chapter, the µ-randomness procedure has been used for generating simulated images with known characteristics. Under this scheme, a line with a specified thickness is defined by the perpendicular distance d from a fixed reference point O located in the centre of the image and the angular position of the perpendicular α. Distance d is limited to the diagonal of the image. Figure 4.12 demonstrates this procedure.
a d
O
(a)
(b)
Figure 4.12 µ-randomness: a) schematic view of the procedure, and b) a typical simulated image generated using this approach
One of the most important features of simulation is that it allows several structural characteristics to be taken into consideration with the simulation parameters. These parameters are: web density (controlled as line density), angular density (sampled from a normal or random distribution), distance from the reference point (sampled from a random distribution), line thickness (sampled from a normal distribution) and image size.
4.2.3 Thresholding Determination of fibre diameter by the use of image analysis requires initial segmentation of the micrographs to produce binary images. This is a critical step because the segmentation affects the results dramatically. The typical way of producing a binary image from a greyscale image is by ‘global thresholding’, where a single constant threshold is applied to segment the image. All pixels up to and equal to the threshold belong to the object and the remaining belong to the background. One
80
Optimisation of the Electrospinning Process simple way to choose the threshold is by picking different thresholds until one is found that produces a good result as judged by the observer. Global thresholding, however, is very sensitive to inhomogeneities in the grey-level distributions of object and background pixels [22–24]. Figure 4.13a illustrates a typical micrograph obtained from electron microscopy. As shown in Figure 4.13b, global thresholding resulted in some broken fibre segments. To eliminate the effect of inhomogeneities, a ‘local thresholding scheme’ could be used. In this approach, the image is divided into subimages where the inhomogeneities are negligible. Then, optimal thresholds are found for each subimage [22–24]. A common practice in this case is to use morphological ‘opening’ to compensate for non-uniform background illumination. The morphological opening is a sequential application of an ‘erosion’ operation followed by a ‘dilation’ operation (i.e., opening = erosion + dilation) using the same ‘structuring element’. Dilation is an operation that grows or thickens objects in a binary image by adding pixels to the boundaries of objects. Erosion shrinks or thins objects in a binary image by removing pixels on object boundaries. The specific manner and extent of the thickening or thinning is controlled by the size and shape of the structuring element, which is a matrix consisting of 0 s and 1 s having any arbitrary shape and size. Opening the image produces an estimate of the background provided a sufficiently large structuring element is used so that it does not fit entirely within the objects (Figure 4.13c). Subtracting the opened image from the original image (which is called the ‘top-hat transformation’ results in an image with a reasonably even background (Figure 4.13d) [22–24]. Now that the background is homogeneous and the edges of the objects are clearer, a global thresholding could be applied to provide the binary image. It could be shown that this process is equivalent to segment the image with locally varying thresholds [24]. To automatically select the appropriate threshold, Otsu’s method [28] is employed. This method is a simple but efficient technique in which the optimal threshold is chosen automatically by the discriminant criterion, thereby maximising the interclass variance and minimising the intraclass variance of the black and white pixels. Figure 4.13d depicts the binary image obtained using this approach. It is apparent that the problem associated with global thresholding has been solved. Note that the process is extremely sensitive to noise contained in the image; before the segmentation, a procedure to clean the noise and enhance the contrast of the image is necessary.
81
Advances in Nanofibre Research
(a)
(b)
(c)
(d)
Figure 4.13 Thresholding: a) a typical micrograph of electrospun web; b) global thresholding; c) top-hat transformation; and d) local thresholding
4.3 Experimental Electrospun nanofibre webs used as real webs for image analysis were obtained from the electrospinning of polyvinyl alcohol (PVA) with an average molecular weight of 72000 g/mol (Merck) at different processing parameters. The micrographs of the webs were obtained using Philips (XL-30) environmental SEM under of 10,000× magnification after being gold-coated. Figure 4.14 shows the micrographs of the electrospun webs which were used as real webs.
82
Optimisation of the Electrospinning Process
(R1)
(R2)
(R3)
(R4)
(R5)
Figure 4.14 Micrographs of the electrospun webs
83
Advances in Nanofibre Research
4.4 Results and Discussion We evaluated the accuracy of the fibre diameter determined by the two methods using two sets of test samples each composed of five simulated images. The firstset images had constant diameters which increased from 5 pixels to 25 pixels for different samples. The second-set images had varying diameters sampled from normal distributions with a mean of 15 pixels and standard deviation of 2–10 pixels. For both cases, the line density was set to 30 and the angular density sampled from a random distribution in the range 0–360°. The simulation parameters for the two sets are presented in Table 4.1 and Table 4.2. Figure 4.15 and Figure 4.16 show the simulated images in the two sets.
Table 4.1 Structural characteristics of images in the first set Image number
Angular range
Line density
Line thickness
C1
0–360
30
5
C2
0–360
30
10
C3
0–360
30
15
C4
0–360
30
20
C5
0-360
30
25
Table 4.2 Structural characteristics of images in the second set Image number Angular range
84
Line density
Line thickness Mean
Std
V1
0–360
30
15
2
V2
0–360
30
15
4
V3
0–360
30
15
6
V4
0–360
30
15
8
V5
0–360
30
15
10
Optimisation of the Electrospinning Process
(C1)
(C2)
(C3)
(C4)
(C5)
Figure 4.15 Simulated images with constant diameters
85
Advances in Nanofibre Research
(V1)
(V2)
(V3)
(V4)
(V5)
Figure 4.16 Simulated images with varying diameters
86
Optimisation of the Electrospinning Process The results for the simulated images are given in Table 4.3 and Table 4.4. Figure 4.17 and Figure 4.18 show the distribution of fibre diameter for the two sets of simulated images obtained by the methods. A normal distribution was also fitted on the histogram in each case.
Table 4.3 Values for mean and standard deviation for series 1 C1
C2
C3
C4
C5
Mean
5
10
15
20
25
Std
0
0
0
0
0
Distance transform
Mean
5.486
10.450
16.573
23.016
30.063
Std
1.089
2.300
5.137
6.913
10.205
New distance transform
Mean
5.366
9.917
15.106
20.013
24.645
Std
0.747
1.053
1.707
1.765
2.837
Simulation
Table 4.4 Values for mean and standard deviation for series 2 V1
V2
V3
V4
V5
Mean
15.247
15.350
15.243
15.367
16.628
Std
1.998
4.466
5.766
8.129
9.799
Distance transform
Mean
16.517
16.593
17.135
17.865
19.394
Std
5.350
6.165
7.597
9.553
11.961
New distance transform
Mean
14.876
15.020
14.812
14.651
15.546
Std
2.403
4.797
6.047
7.851
9.942
Simulation
87
Advances in Nanofibre Research 0.7 0.6 0.5 0.4 0.3 0.2
0.35 0.3 0.25 0.2 0.15 0.1
0.1 0
Distance Transform New Distance Transform Distance Transform New Distance Transform
0.4 Probability Density
Probability Density
0.45
Distance Transform New Distance Transform Distance Transform New Distance Transform
0.05 0
2
4 6 8 Diameter (pixel)
10
0
12
0
5
10 15 Diameter (pixel)
(C1)
Distance Transform New Distance Transform Distance Transform New Distance Transform
0.35 Probability Density
Probability Density
0.4
Distance Transform New Distance Transform Distance Transform New Distance Transform
0.2 0.15 0.1 0.05
0.3 0.25 0.2 0.15 0.1 0.05
0
5
10
15 20 Diameter (pixel)
25
30
35
0
0
10
20 30 Diameter (pixel)
(C3)
Distance Transform New Distance Transform Distance Transform New Distance Transform
0.2 Probability Density
40
(C4)
0.25
0.15 0.1 0.05 0
0
10
20
30 40 Diameter (pixel)
50
60
70
(C5)
Figure 4.17 Histograms for simulated images with constant diameters
88
25
(C2)
0.25
0
20
50
Optimisation of the Electrospinning Process 0.22
0.16 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0
Manual Distance Transform New Distance Transform Manual Distance Transform New Distance Transform
0.14 Probability Density
0.18 Probability Density
0.16
Manual Distance Transform New Distance Transform Manual Distance Transform New Distance Transform
0.2
0.12 0.1 0.08 0.06 0.04 0.02
0
5
10
15 20 25 Diameter (pixel)
30
35
0
40
5
0
10
15 20 25 Diameter (pixel)
(V1)
0.06 0.05 0.04 0.03
0.07 0.06 0.05 0.04 0.03
0.02
0.02
0.01
0.01 5
10
40
15 20 25 Diameter (pixel)
30
35
Manual Distance Transform New Distance Transform Manual Distance Transform New Distance Transform
0.08 Probability Density
0.07
40
0
10
0
20 30 Diameter (pixel)
(V3)
40
50
(V4)
0.07
Manual Distance Transform New Distance Transform Manual Distance Transform New Distance Transform
0.06 Probability Density
Probability Density
0.09
Manual Distance Transform New Distance Transform Manual Distance Transform New Distance Transform
0.08
0
35
(V2)
0.09
0
30
0.05 0.04 0.03 0.02 0.01 0
0
10
20 30 40 Diameter (pixel)
50
60
(V5)
Figure 4.18 Histograms for simulated images with varying diameters
89
Advances in Nanofibre Research From Table 4.3 and Table 4.4 it is apparent that for all the simulated samples, the results obtained by the new method were significantly better than for the old one. For both sets of simulated images, the mean and standard deviation of fibre diameter obtained by the new distance transform were very close to those of the simulation. It is noteworthy that the true mean and standard deviation of the diameter in samples with varying line thickness differed slightly from those used as simulation parameters (see Table 4.2). As described above, the distance transform method fails in the measurement of the fibre diameter at intersection points. The intersections result in overestimation of fibre diameter. These points were deleted in our developed method, so the effects of intersections which cause imprecise measurement of fibre diameter were eliminated. Therefore, the fibre diameter was determined more accurately. Neither method can distinguish multiple fibres being joined together. This can happen in simulation by laying one line over the other. In real webs, fibre bundling sometimes happens (often in high-density webs comprising many fibres). There is not any black pixel (associated with the background) between joined fibres, so they are segmented as a single fibre in the step of thresholding. There may also be up to half-a-pixel error in both directions, i.e., up to 1 pixel error in measuring fibre diameter. The error may be more significant if the fibres are thicker. The slight difference between the diameters obtained by our method and the simulation can be attributed to the 1 pixel measurement error, bundling of fibres, some remaining parts of branches after pruning, and slight variations of skeleton adjacent to intersections which were not deleted. The last two problems could be readily solved by further pruning for the former and increasing the area to be deleted for each intersection for the latter. Authors preferred to leave it unchanged because it causes a decrease in the number of diameter measurements. It can be shown that the errors due to these problems do not play an important part in variation of the diameter because they are in the range of measurement error. Furthermore, some parts of the image are deleted due to the presence of intersections and not counted in diameter measurement; this can be another reason for the variation observed. In most cases (except where thick fibres are present), the difference between our method and simulation was within the 1 pixel measurement error. That is, the effects of other errors are negligible. In addition, there were five real webs for testing the applicability of the methods for real samples. The results for the real webs obtained by two methods together with the manual method are presented in Table 4.5. Figure 4.19 shows the diameter
90
Optimisation of the Electrospinning Process distribution of these samples in term of nanometres. The curved line over each histogram corresponds to its fitted normal distribution. The results for the real webs are in complete agreement with the trends observed by the simulation. The mean and standard deviation of fibre diameter obtained by our method are very to those obtained by the manual method. In addition to the reasons mentioned above, the differences can also be attributed to the different number of measurements. For each sample, our method measured >2000 fibre diameters whereas the operator could measure only 100 fibre diameters in the manual method because of the time-consuming nature of this work. Despite all of these facts, the differences here are also within the 1 pixel measurement error, which suggests that other errors are less significant.
4.5 Conclusion Fibre diameter is an important structural characteristic in electrospun webs. Understanding how it is affected by processing variables is essential for producing nanofibres with desired properties. Electrospun fibre diameter is often measured by the manual method. This is a time-consuming and operator-based technique that cannot be used for online quality control. An image analysis-based method called distance transform was reported in the literature to be an automated technique for measurement of fibre diameters in non-woven textiles. Despite the usefulness, the method fails in measuring the diameter at intersections because the skeleton and distance map are broken at these points, so the centre of the object in the distancetransformed image no longer coincides with the fibre diameter. We developed a novel method in which the intersections are identified and deleted from the skeleton, thus solving the associated problem. These techniques have been validated by applying the methods to test images with known characteristics generated using the µ-randomness procedure. The results show the effectiveness of our method for diameter measurement. For the entire simulated images, a new algorithm resulted in the values for the mean and standard deviation of fibre diameter being closer to the simulation. In addition, five electrospun webs of PVA were used to test the general applicability of the methods for real webs. Due to the need of binary input images, local thresholding was applied to the micrographs of the webs obtained from SEM. Otsu’s method was used to automatically determine the appropriate threshold. The results for the real webs show that the mean and standard deviation of fibre diameter obtained by the new algorithm were in perfect agreement with the manual method. That is, our attempt to develop a method for measuring fibre diameter has been successful.
91
Advances in Nanofibre Research
0.01
0.007
0.009
0.006 0.005 0.004 0.003
0.008 0.007 0.006 0.005 0.004 0.003
0.002
0.002
0.001
0.001 0
0
0
100
200
300 400 500 600 Fiber Diameter (nm)
700
Manual Distance Transform New Distance Transform Manual Distance Transform New Distance Transform
0.01
Probability Density
0.008
Probability Density
0.011
Manual Distance Transform New Distance Transform Manual Distance Transform New Distance Transform
0.009
800
0
100
200
300 400 500 600 Fiber Diameter (nm)
(R1)
0.12
Manual Distance Transform New Distance Transform Manual Distance Transform New Distance Transform
0.14 Probability Density
Probability Density
0.16
Manual Distance Transform New Distance Transform Manual Distance Transform New Distance Transform
0.14
0.1 0.08 0.06 0.04 0.02
0.12 0.1 0.08 0.06 0.04 0.02
0
100
200 300 400 Fiber Diameter (nm)
500
600
0
100
0
200 300 400 500 Fiber Diameter (nm)
(R3)
(R4) 0.16
Manual Distance Transform New Distance Transform Manual Distance Transform New Distance Transform
Probability Density
0.14 0.12 0.1 0.08 0.06 0.04 0.02 0
0
100
200 300 400 Fiber Diameter (nm)
500
600
(R5)
Figure 4.19 Histograms for real webs
92
800
(R2)
0.16
0
700
600
700
Optimisation of the Electrospinning Process
Table 4.5 Values for mean and standard deviation for real webs Manual
Mean Std
Distance transform
Mean Std
New distance transform
Mean Std
R1
R2
R3
R4
R5
Pixel
24.358
24.633
18.583
18.827
17.437
nm
318.67
322.27
243.11
246.31
228.12
Pixel
3.193
3.179
2.163
1.984
2.230
nm
41.77
41.59
28.30
25.96
29.18
Pixel
27.250
27.870
20.028
23.079
20.345
nm
356.49
364.61
262.01
301.94
266.17
Pixel
8.125
7.462
4.906
7.005
6.207
nm
106.30
97.62
64.18
91.64
81.21
Pixel
24.741
25.512
18.621
20.100
18.299
nm
323.681
333.767
243.610
262.954
239.395
Pixel
3.854
3.961
2.826
2.903
2.795
nm
50.417
51.821
36.976
37.980
36.571
References 1.
A.K. Haghi and M. Akbari, Physica Status Solidi A, 2007, 204, 1830.
2.
J. Doshi and D.H. Reneker, Journal of Electrostatics, 1995, 35, 151.
3.
H. Fong and D.H. Reneker in Structure Formation in Polymeric Fibres, Ed., D.R. Salem, Hanser, Cincinnati, OH, USA, 2001, p.225.
4.
D. Li and Y. Xia, Advanced Materials, 2004, 16, 1151.
5.
Z-M. Huang, Y-Z. Zhang, M. Kotaki and S. Ramakrishna, Composites Science & Technology, 2003, 63, 2223.
6.
T. Subbiah, G.S. Bhat, R.W. Tock, S. Parameswaran and S.S. Ramkumar, Journal of Applied Polymer Science, 2005, 96, 557.
7.
B. Pourdeyhimi, R. Ramanathan and R. Dent, Textile Research Journal, 1996, 66, 713
8.
B. Pourdeyhimi, R. Ramanathan and R. Dent, Textile Research Journal, 1996, 66, 747.
93
Advances in Nanofibre Research 9.
B. Pourdeyhimi, R. Dent and H. Davis, Textile Research Journal, 1997, 67, 143.
10. B. Pourdeyhimi and R. Dent, Textile Research Journal, 1997, 67, 181. 11. B. Pourdeyhimi, R. Dent, A. Jerbi, S. Tanaka and A. Deshpande, Textile Research Journal, 1999, 69, 185. 12. B. Pourdeyhimi and H.S. Kim, Textile Research Journal, 2002, 72, 803. 13. B. Xu and Y.L. Ting, Textile Research Journal, 1995, 65, 41. 14. I. Krucinska and S. Krucinski, Textile Research Journal, 1999, 69, 363. 15. B. Pourdeyhimi and R. Dent, Textile Research Journal, 1999, 69, 233. 16. D.M. Luzhansky in the Proceedings of the International Nonwovens Technical Conference, Baltimore, MD, USA, 2003, 15–18 September. 17. H.S. Kim and B. Pourdeyhimi, International Nonwovens Journal, Winter 2000, p.15. 18. A.H. Aydilek, S.H. Oguz and T.B. Edil, Journal of Computing in Civil Engineering, 2002, 16, 280. 19. R. Chhabra, International Nonwovens Journal, Spring 2003, p.43. 20. E. Ghassemieh, H.K. Versteeg and M. Acar, International Nonwovens Journal, Summer 2001, p.26. 21. M.S. Abdel-Ghani and G.A. Davies, Chemical Engineering & Science, 1985, 40, 117. 22. W.K. Pratt, Digital Image Processing, 3rd Edition, John Wiley and Sons, New York, NY, USA, 2001, p.58. 23. B. Jähne, Digital Image Processing, 5th Edition, Springer, Germany, 2002, p.92. 24. R.C. Gonzalez and R.E. Woods, Digital Image Processing, 2nd Edition, Prentice Hall, New Jersey, NJ, USA, 2001, p.102. 25. H. Breu, J. Gil, D. Kirkpatrick and M. Werman, IEEE Transactions on Pattern Analysis and Machine Intelligence, 1995, 17, 529.
94
Optimisation of the Electrospinning Process 26. N. Sudha, S. Nandi, P.K. Bora and K. Sridharan in the Proceedings of IEEE Region 10 International Conference on Global Connectivity in Energy, Computer, Communication and Control, Delhi, India, 1998, p.49. 27. Q-Z. Ye in the Proceedings of the 9th International Conference on Pattern Recognition, Rome, Italy, 1988, p.495. 28. N. Otsu, IEEE Transactions on Systems Man and Cybernetics, Part C, 1979, 9, 62.
95
Advances in Nanofibre Research
96
5
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres
5.1 Introduction Over the recent decades, fabrication of polymer nanofibres used in many biomedical applications such as tissue engineering, drug delivery, wound dressing, enzyme immobilisation and so on has been extensively studied. The nanofibre fabrications have unique characteristics such as very large surface area, ease of functionalisation for various purposes and superior mechanical properties. The electrospinning with simple process is an important technique which can be used for the production of polymer nanofibres with diameter in the range from several micrometers down to ten of nanometers. In electrospinning, the charged jets of a polymer solution which are collected on a target are created by using an electrostatic force. Many parameters can influence the quality of fibres including the solution properties (polymer concentration, solvent volatility and solution conductivity), governing variables (flow rate, voltage and distance between tip-to-collector), and ambient parameters (humidity, solution temperature and air velocity in the electrospinning chamber) [2-6]. Figure 5.1 illustrates the electrospinning setup. Material properties such as melting temperature and glass transition temperature as well as structural characteristics of nanofibre webs (e.g., distribution of: (i) fibre diameter, (ii) pore size; and (iii) fibre orientation) determine the physical and mechanical properties of the webs. The surface of electrospun fibres is important when considering end-use applications. For example, the ability to introduce porous surface features of known size is required if nanoparticles need to be deposited on the surface of the fibre or if drug molecules are to be incorporated for controlled release, as tissue-scaffold materials, and for acting as a ‘cradle’ for enzymes [7]. The filtration performance of nanofibres is also strongly related to their pore structure parameters, i.e., percent open area (POA) and pore-opening size distribution (PSD). Hence, the control of the pore of electrospun webs is of prime importance for nanofibres that are being produced for these purposes. There is no literature available about the pore size and its distribution in electrospun fibres. In this chapter, the pore size and its distribution were measured using an image analysis technique.
97
Advances in Nanofibre Research
Syringe Metering pump
Connector 0
1
Electrical heater High voltage supply
25 ºC
Temperature controller
Figure 5.1 Electrospinning setup
Current methods for determining PSD are mostly indirect and contain inherent disadvantages. Recent technological advancements in image analysis offer great potential for a more accurate and direct way of determining the PSD of electrospun webs. Overall, the image analysis method provides a unique and accurate method that can measure pore-opening sizes in electrospun nanofibre webs.
5.2 Methodology The porosity, eV, is defined as the percentage of the volume of the voids, Vv, to the total volume (voids plus constituent material), Vt, and is given by
(5.1)
98
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres Similarly, the POA, eA, that is defined as the percentage of the open area, Ao, to the total area At, is given by
(5.2) Usually, porosity is determined for materials with a three-dimensional structure (e.g., relatively thick non-woven fabrics). Nevertheless, for two-dimensional textiles such as woven fabrics and relatively thin non-wovens, it is often assumed that porosity and POA are equal [8]. The size of an individual opening can be defined as the surface area of the opening, although it is mostly indicated with a diameter called ‘equivalent opening size’ (EOS). EOS is not a single value because each opening may differ. The commonly used term in this case is the diameter, Oi, corresponding with the equivalent circular area, Ai, of the opening.
(5.3) This diameter is greater than the side dimension of a square opening. A spherical particle with that diameter will never pass through the opening (Figure 5.2a) and may therefore not be considered as an equivalent dimension or equivalent diameter. This will be possible only if the diameter corresponds with the side of the square area (Figure 5.2b). However, not all openings are squares, yet the equivalent square area of openings is used to determine their equivalent dimension because this simplified assumption results in one single opening size from the open area. It is the diameter of a spherical particle that can pass the equivalent square opening, hence the equivalent opening or pore size, Oi, results from:
(5.4)
99
Advances in Nanofibre Research
(a)
(b)
Figure 5.2 Equivalent opening size, Oi, based on (a) equivalent area, and (b) equivalent size
From the EOS, PSD and an equivalent diameter for which a certain percentage of the opening have a smaller diameter (Ox, pore opening size that x percent of pores are smaller than that size) may be measured. PSD curves can be used to determine the uniformity coefficient, Cu, of the investigated materials. The uniformity coefficient is a measure for the uniformity of the openings and is given by:
(5.5) The ratio equals 1 for uniform openings and increases with decreasing uniformity of the openings [8]. Pore characteristics are one of the main tools for evaluating the performance of any non-woven fabric and electrospun webs. Understanding the link between processing parameters and pore structure parameters allows for better control over the properties of electrospun fibres. Therefore, there is a need for the design of nanofibres to meet specific application needs. Various techniques may be used to evaluate the pore characteristics of porous materials, including sieving techniques (dry, wet and hydrodynamic sieving), mercury porosimetry and flow porosimetry (‘bubble point method’) [9, 10]. As one selects a suitable technique for characterisation, the associated virtues and pitfalls of each technique should be examined. The most attractive option is a single technique which is non-destructive yet capable of providing a comprehensive set of data [11].
100
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres
5.2.1 Sieving Methods In dry sieving, glass bead fractions (from finer to coarser) are sieved through the porous material. In theory, most of the glass beads from the first glass bead fraction should pass. As larger and larger glass bead fractions are sieved, more and more glass beads should become trapped within and on top of the material. The number of pores of a certain size should be reflected by the percentage of glass beads passing through the porous material during each glass bead fraction sieved; however, electrostatic effects between glass beads and between glass beads and the material can affect the results. Glass beads may stick to fibres, making the pores effectively smaller, and they may also agglomerate to form one large glass bead that is too large to pass through the any of the pores. Glass beads may also break from hitting each other and the sides of the container, resulting in smaller particles that can pass through smaller openings. In hydrodynamic sieving, a glass bead mixture is sieved through a porous material under alternating water flow conditions. The use of glass-bead mixtures leads to results that reflect the original glass-bead mixture used. Therefore, this method is useful only for evaluating large pore openings (e.g., O95). Another problem occurs if particles of many sizes interact, which probably results in particle blocking and bridge formation. This is a particular problem in hydrodynamic sieving because the larger glass bead particles settle first when water is drained during the test. If this occurs, fine glass beads which are smaller than the pores are prevented from passing through by the coarser particles. In wet sieving, a glass-bead mixture is sieved through a porous material aided by a water spray. The same basic mechanisms that occur when using the hydrodynamic sieving method also take place when using the wet sieving method. Bridge formation is not as pronounced in the wet sieving method as in the hydrodynamic sieving method; however, particle blocking and agglomeration of glass beads are more pronounced [9, 10]. The sieving tests are very time-consuming. In general, 2 hours are required to carry out a test. The sieving tests are far from providing a complete PSD curve because the accuracy of the tests for pore sizes <90 μm is questionable [12, 13].
5.2.2 Mercury Porosimetry Mercury porosimetry is a well known method which is often used to study porous materials. This technique is based on the fact that mercury, as a non-wetting liquid, does not intrude into pore spaces except under application of sufficient pressure.
101
Advances in Nanofibre Research Therefore, a relationship can be found between the size of pores and the pressure applied. In this method, a porous material is completely surrounded by mercury and pressure is applied to force the mercury into pores. As mercury pressure increases the large pores are filled with mercury first. Pore sizes are calculated as the mercury pressure increases. At higher pressures, mercury intrudes into the fine pores and, when the pressure reaches a maximum, total open-pore volume and porosity are calculated. Mercury porosimetry thus gives a PSD based on total pore volume but gives no information regarding the number of pores of a porous material. Pore sizes ranging from 0.0018 μm to 400 μm can be studied using mercury porosimetry. Pore sizes <0.0018 μm are not intruded with mercury, and this is a source of error for porosity and PSD calculations. Furthermore, mercury porosimetry does not account for closed pores because mercury does not intrude into them. Due to the application of high pressures, sample collapse and compression is possible, hence it is not suitable for fragile compressible materials such as nanofibre sheets. Other concerns include the fact that it is assumed that the pores are cylindrical, which is not the case in reality. After the mercury intrusion test, sample decontamination at specialised facilities is required because the highly toxic mercury is trapped within the pores. Therefore this dangerous and destructive test can be undertaken only in well-equipped laboratories [7, 9, 10].
5.2.3 Flow Porosimetry (Bubble Point Method) Flow porosimetry is based on the principle that a porous material will allow a fluid to pass only if the pressure applied exceeds the capillary attraction of the fluid in the largest pore. In this test, the specimen is saturated with a liquid and continuous air flow is used to remove liquid from the pores. At a critical pressure, the first bubble will come through the largest pore in the wetted specimen. As the pressure increases, the pores are emptied of liquid in order from largest to smallest and the flow rate is measured. PSD, the number of pores, and porosity can be derived once the flow rate and the applied pressure are known. Flow porosimetry can be used to measure pore sizes within the range 0.013–500 μm. The air passes through only the pores, so the characteristics of these pores are measured while those of closed pores and blind pores are omitted. It should be noted that during the spinning process, on many occasions, 100% total flow is not reached. This is due to pore wick evaporation from the pores when the flow rate is too high. Extreme care is required to ensure that the air flow does not disrupt the pore structure of the specimen. The flow porosimetry method is also based on the assumption that
102
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres the pores are cylindrical, which is not the case in reality. Finding a liquid with low surface tension which could cover all the pores, has no interaction with the material and does not cause swelling in the material is not easy [7, 9, 10].
5.2.4 Image Analysis It seemed to be advantageous to use image analysis techniques for pore measurement because of its convenience in detecting individual pores in a non-woven image. Image analysis has been used to measure the pore characteristics of woven [12] and nonwoven geotextiles [13]. In the former, successive erosion operations with increasing size of structuring element were used to count the pore openings larger than a given structuring element. The main purpose of the erosion was to simulate the conditions in the sieving methods. In this method, the voids are first connected to the border of the image ((these are not complete pores and should not be taken into account during measurement)). Undertaking opening and then closing operations after pore measurement causes the pore sizes and shapes to deviate from the real ones. The method is suitable for measuring the pore sizes of woven geotextiles with fairly uniform pore sizes and shapes, but is not appropriate for electrospun nanofibre webs of different pore sizes. In the latter case, cross-sectional images of non-woven geotextile were used to calculate the pore structure parameters. A slicing algorithm based on a series of morphological operations for determining the mean fibre thickness and the optimal position of the uniform slicing grid was developed. After recognition of the fibres and pores in the slice, the distribution of pore-opening sizes of the cross-sectional image may be determined. The method is useful for measuring pore characteristics of relatively thick non-wovens but cannot be applied to electrospun nanofibre webs due to the extremely small size. Therefore, there is a need for developing an algorithm suitable for measuring the pore structure parameters in electrospun webs. In response to this need, a new image analysis-based method has been developed which is presented below. In this method, a binary image of the web is used as an input. Firstly, voids connected to the image border are identified and cleared using morphological reconstruction [14, 15] where the mask image is the input image and the marker image is zero everywhere except along the border. Total area (which is the number of pixels in the image) is measured. Then the pores are labeled and each considered to be an object. Here the number of pores may be obtained. In the next step, the number of pixels of each object as the area of that object is measured. Having the area of pores, the porosity and EOS with respect to each pore may be calculated. The data in pixels
103
Advances in Nanofibre Research may then be converted to nanometres. Finally, the PSD curve is plotted and O50, O95 and Cu determined.
5.2.4.1 Real Webs To measure the pore characteristics of electrospun nanofibres using image analysis, images of the webs are required. These images (called ‘micrographs’) are usually obtained by scanning electron microscopy (SEM), transmission electron microscopy (TEM) or atomic force microscopy (AFM). The images must be of high-quality and taken under appropriate magnifications. The image analysis method for measuring pore characteristics requires initial segmentation of the micrographs to produce binary images. This is a critical step because the segmentation affects the results dramatically. The typical way of producing a binary image from a greyscale image is by ‘global thresholding’ [14, 15] in which a single constant threshold is applied to segment the image. All pixels up to and equal to the threshold belong to the object and the remaining belong to the background. One simple way to choose the threshold is by picking different thresholds until one is found that produces a good result as judged by the observer. Global thresholding is very sensitive to inhomogeneities in the grey-level distributions of object and background pixels. To eliminate the effect of inhomogeneities, a local thresholding scheme [14, 15] could be used. In this approach, the image is divided into sub-images where the inhomogeneities are negligible. Optimal thresholds are then found for each sub-image. A common practice in this case (which is used in this contribution) is to pre-process the image to compensate for the illumination problems and then apply a global thresholding to the pre-processed image. It can be shown that this process is equivalent to segmenting the image with locally varying thresholds. To automatically select the appropriate thresholds, Otsu’s method [16] is employed. This method chooses the threshold to minimise intraclass variance of the black and white pixels. As shown in Figure 5.3, global thresholding resulted in some broken fibre segments. This problem was solved using local thresholding. The process is extremely sensitive to noise contained in the image so, before segmentation, a procedure to clean the noise and enhance the contrast of the image is necessary.
104
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres
(a)
(b)
(c)
Figure 5.3 a) A real web; b) global thresholding; and c) local thresholding
5.2.4.2 Simulated Webs In is known that the pore characteristics of non-woven webs are influenced by web properties and so are those of electrospun webs. There are no reliable models available for predicting these characteristics as a function of web properties [17]. To explore the effects of some parameters on pore characteristics of electrospun nanofibres, simulated webs are generated. These webs are images simulated by straight lines. There are three widely used methods for generating random network of lines. These are called ‘S-randomness’, ‘μ-randomness’ (suitable for generating a web of continuous filaments) and ‘I-randomness’ (suitable for generating a web of staple fibres). These methods have been described in detail by Abdel-Ghani and co-workers [18] and Pourdeyhimi and co-workers [19]. In this chapter, the μ-randomness procedure for generating simulated images was used. Under this scheme, a line with a specified thickness is
105
Advances in Nanofibre Research defined by the perpendicular distance d from a fixed reference point O located in the centre of the image and the angular position of the perpendicular α. Distance d is limited to the diagonal of the image. Figure 5.4 demonstrates this procedure.
a d
O
Figure 5.4 Procedure for μ-randomness
One of the most important features of simulation is that it allows several structural characteristics to be taken into consideration with the simulation parameters. These parameters are: web density (controlled as line density), angular density (sampled from a normal or random distribution), distance from the reference point (sampled from a random distribution), line thickness (sampled from a normal distribution) and image size.
5.3 Experimental Nanofibre webs were obtained from the electrospinning of polyvinyl alcohol (PVA) with an average molecular weight of 72000 g/mol (Merck) at different processing parameters for attaining different pore characteristics. Table 5.1 summarises the electrospinning parameters used for preparing the webs. The micrographs of the webs were obtained using Philips (XL-30) environmental SEM under of 10,000× magnification after being gold-coated. Figure 5.5 shows the micrographs of the electrospun webs.
106
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres
(1)
(2)
(3)
(4)
(5)
Figure 5.5 Micrographs of electrospun
107
Advances in Nanofibre Research
Table 5.1 Electrospinning parameters used for preparing nanofibre webs Number
Concentration (%)
Spinning distance (Cm)
Voltage (KV) Flow rate (ml/h)
1
8
15
20
0.4
2
12
20
15
0.2
3
8
15
20
0.2
4
8
10
15
0.3
5
10
10
15
0.2
5.4 Results and Discussion Due to previously mentioned reasons, sieving methods and mercury porosimetry are not applicable for measuring pore structure parameters at the nano-scale. The only method which seems to be practical is flow porosimetry. However, in this contribution, the nanofibres were made of PVA, so finding an appropriate liquid for the test to be done is almost impossible because of the solubility of PVA in organic and inorganic liquids. As an alternative, image analysis was employed to measure pore structure parameters in electrospun nanofibre webs. PSD curves of the webs determined using the image analysis method are shown in Figure 5.6. The pore characteristics of the webs (O50, O95, Cu, number of pores, porosity) measured by this method are presented in Table 5.2. It is seen that when the porosity is reduced, O50 and O95 decrease. Cu also decreases with respect to porosity, i.e., the uniformity of the pores increases. The number of pores has an increasing trend with decreasing the porosity. The image analysis method presents valuable and comprehensive information regarding pore structure parameters in nanofibre webs. This information may be exploited in preparing the webs with the required pore characteristics to use in filtration, biomedical applications, nanoparticle deposition and other purposes. The advantages of the method are listed below: • The method can be used to measure pore structure parameters in any nanofibre web with any pore features, and is applicable even if other methods may not be employed
108
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres • It is very fast. It takes less <1 s for an image to be analysed using a 3-GHz processor • The method is direct and very simple. Pore characteristics are measured from the area of the pores (which is defined as the number of pixels of the pores) • There is no systematic error in the measurement (such as assuming pores to be cylindrical in mercury and flow porosimetry, and the errors associated with the sieving methods mentioned above). Once the segmentation is successful, pore sizes will be measured accurately. The quality of images affects the segmentation procedure. High-quality images reduce the possibility of poor segmentation and enhance the accuracy of the results • It gives a complete PSD curve • There is no cost involved in the method and minimal technical equipments are needed (SEM for obtaining the micrographs of the samples and a computer for analysis) • It can be used as an online quality-control technique for large-scale production • The results obtained by image analysis are reproducible • It is not a destructive method. A very small amount of sample is required for measurement
Table 5.2 Pore characteristics of electrospun webs Number
O20
O95
Cu
Pore number
Porosity
Pixel
nm
Pixel
nm
1
39.28
513.9
94.56
1237.1
8.43
31
48.64
2
27.87
364.7
87.66
1146.8
5.92
38
34.57
3
26.94
352.5
64.01
837.4
3.73
64
26.71
4
22.09
289.0
60.75
794.8
3.68
73
24.45
5
19.26
252.0
44.03
576.1
2.73
69
15.74
109
Advances in Nanofibre Research
100 90
No.1 No.2 No.3 No.4 No.5
O95
Cumulative Percent
80 70 60 50
O50
40 30 20 10 0 0
250
500
750 1000 1250 Pore Size (nm)
1500
1750
2000
Figure 5.6 PSD curves of electrospun webs
In an attempt to establish the effects of some structural properties on the pore characteristics of electrospun nanofibres, two sets of simulated images with varying properties were generated. The simulated images reveal the degree to which the diameter and density of fibres affect the pore structure parameters. The first set contained images with the same density varying in fibre diameter and images with the same fibre diameter varying in density. Each image had a constant diameter. The second set contained images with the same density and mean fibre diameter whereas the standard deviation of fibre diameter varied. The details are given in Table 5.3 and Table 5.4. Typical images are shown in Figure 5.7 and Figure 5.8.
110
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres
Table 5.3 Structural characteristics of images in the first set Number
Angular range
Line density
Line thickness
1
0–360
20
5
2
0–360
30
5
3
0–360
40
5
4
0–360
20
10
5
0–360
30
10
6
0–360
40
10
7
0–360
20
20
8
0–360
30
20
9
0–360
40
20
Table 5.4 Structural characteristics of images in the second set Number
Angular range
Line density
Line thickness Mean
Std
1
0–360
30
15
0
2
0–360
30
15
4
3
0–360
30
15
8
4
0–360
30
15
10
Pore structure parameters of the simulated webs were measured using the image analysis method. Table 5.5 summarises the pore characteristics of the simulated images in the first set. For the webs with the same density, increasing fibre diameter resulted in a decrease in O95, number of pores, and porosity. Assuming the web density to be constant, increasing fibre diameter means that the ratio of area of fibres to total area (i.e., the proportion of white pixels to total pixels in the image) increases, thereby reducing the porosity. It could be imagined that, as the fibres get thicker, small pores are covered with the fibres, lowering the number of pores. An increase in fibre diameter at a given web density results in smaller pores, hence O95 decreases. No particular trends were observed for O50 and Cu. In the case of O50, it is because the effect of fibre diameter is more significant on larger pores whereas O50 is related to mostly small pores. Also, there seem to be other parameters (e.g., fibre arrangement) which influence O50 more significantly rather than fibre diameter. In Equation 5.5, O10 is in 111
Advances in Nanofibre Research the denominator of the fraction, so Cu is very sensitive to variation in O10. O10 tends to vary a lot and almost regardless of fibre diameter (because it is related to very small pores). Hence, other factors (e.g., the way fibres arrange) are more dominant, and Cu varies regardless of fibre diameter.
Table 5.5 Pore characteristics of the first set of simulated images Number
O50
O95
Cu
Pore number
Porosity
1
27.18
100.13
38.38
84
79.91
2
15.52
67.31
22.20
182
71.78
3
13.78
52.32
18.71
308
69.89
4
36.65
94.31
43.71
67
66.10
5
17.89
61.64
22.67
144
53.67
6
12.41
51.60
16.70
245
47.87
7
24.49
86.90
33.11
58
41.05
8
16.31
56.07
21.66
108
32.53
9
13.11
45.38
17.75
126
22.01
112
(1)
(2)
(3)
(4)
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres
(5)
(6)
(7)
(8)
(9)
Figure 5.7 Simulated images of the first set
Figure 5.9 and Figure 5.10 show the PSD curves of the simulated images in the first set. As the web density increases, the effects of fibre diameter are less pronounced because the PSD curves of the webs become closer to each other.
113
Advances in Nanofibre Research
(1)
(2)
(3)
(4)
Figure 5.8 Simulated images of the second set
For the webs with the same fibre diameter, increasing the density resulted in a decrease in O50, O95, Cu and porosity, whereas the number of pores increased with density. For the same fibre diameter, the total number of fibres and indeed the total number of crossovers increases as web density increases, suggesting a greater number of pores. It is quite trivial but, at a given fibre diameter, the ratio of area of fibres to total area increases as the webs get denser, thus lowering the porosity. Increasing the web density leads to a greater number of crossovers. Therefore, large pores are split into several smaller pores. As a result, O50 and O95 decrease. Furthermore, this fracture of the pores results in less variation in the pore size. Hence, uniformity increases (i.e., Cu decreases).
114
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres 100 90
No.1 No.2 No.3
O95
Cumulative Percent
80 70 60 50
O50
40 30 20 10 0 0
20
40
60 80 Pore Size (pixel)
100
120
140
(a) 100 90
No.4 No.5 No.6
O95
Cumulative Percent
80 70 60 50
O50
40 30 20 10 0 0
20
40
60 80 100 Pore Size (pixel)
120
140
160
(b) 100 90
No.7 No.8 No.9
O95
Cumulative Percent
80 70 60 50
O50
40 30 20 10 0 0
20
40
60 80 Pore Size (pixel)
100
120
140
(c)
Figure 5.9 PSD curves of the first set of simulated images; effect of density, images with a diameter of a) 5 pixels; b) 10 pixels; and c) 20 pixels
115
Advances in Nanofibre Research 100 90
No.1 No.4 No.7
O95
Cumulative Percent
80 70 60 50
O50
40 30 20 10 0 0
20
40
60 80 100 Pore Size (pixel)
120
140
160
(a) 100 90
No.2 No.5 No.8
O95
Cumulative Percent
80 70 60 50
O50
40 30 20 10 0 0
20
40
60 80 Pore Size (pixel)
100
120
140
(b) 100 90
No.3 No.6 No.9
O95
Cumulative Percent
80 70 60 50
O50
40 30 20 10 0 0
20
40 60 Pore Size (pixel)
80
100
(c)
Figure 5.10 PSD curves of the first set of simulated images; effect of fibre diameter, images with a density of a) 20 lines; b) 30 lines; and c) 40 lines
116
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres Table 5.6 summarises the pore characteristics of the simulated images in the second set. No significant effects for variation of fibre diameter on pore characteristics were observed. This suggested that average fibre diameter is the determining factor, not the variation of diameter. Figure 5.11 shows the PSD curves of the simulated images in the second set. Holding web density and average fibre diameter constant, the ratio of area of fibres to total area remains the same or fluctuates mostly due to fibre arrangement and regardless of the variation in fibre diameter. As a result, porosity is not related to the variation in fibre diameter. No trends in O50 and O95 with respect to variation in fibre diameter were observed. This could be attributed to different pore sizes with respect to how thin and thick fibres arrange. Changes in the number of pores also seem to be independent of variation of fibre diameter. It could also be attributed to fibre arrangement.
100 90
No.1 No.2 No.3 No.4
O95
Cumulative Percent
80 70 60 50
O50
40 30 20 10 0 0
20
40
60 80 Pore Size (pixel)
100
120
140
Figure 5.11 PSD curves of the second set of simulated images: the effect of variation in fibre diameter
117
Advances in Nanofibre Research
Table 5.6 Pore characteristics of the second set of simulated images Number
O50
O95
Cu
Pore number
Porosity
1
14.18
5.56
18.79
133
35.73
2
13.38
61.66
20.15
136
41.89
3
18.14
59.35
22.07
121
41.03
4
15.59
62.71
20.20
112
37.77
5.5 Conclusion The evaluation of electrospun nanofibre pore structure parameters is necessary because it facilitates improvement of the design process and the eventual applications of nanofibres. Various techniques have been developed to assess pore characteristics in porous materials. However, most of these methods are indirect, have inherent problems, and are not applicable for measuring the pore structure parameters of electrospun webs. In this chapter, an image analysis-based method has been developed to respond to this need. The method is simple, comprehensive, rapid and directly measures pore structure parameters. The effects of web density, fibre diameter and its variation on pore characteristics of the webs were also explored using some simulated images. As fibre diameter increased, O95, number of pores and porosity decreased. No particular trends were observed for O50 and Cu. Increasing the density resulted in a decrease in O50, O95, Cu and porosity, whereas the number of pores increased with density. The effects of variation of fibre diameter on pore characteristics were not significant.
References 1.
A.K. Haghi and M. Akbari, Physica Status Solidi A, 2007, 204, 1830.
2.
D.H. Reneker and I. Chun, Nanotechnology, 1996, 7, 216.
3.
H. Fong and D.H. Reneker in Structure Formation in Polymeric Fibres, Ed., D.R. Salem, Hanser, Cincinnati, OH, USA, 2001, p.225.
4.
T. Subbiah, G.S. Bhat, R.W. Tock, S. Parameswaran and S.S. Ramkumar, Journal of Applied Polymer Science, 2005, 96, 557.
118
Practical Hints on the Processing Parameters and Geometric Properties of Electrospun Nanofibres 5.
A. Frenot and I.S. Chronakis, Current Opinion in Colloid and Interface Science, 2003, 8, 64.
6.
D. Li and Y. Xia, Advanced Materials, 2004, 16, 1151.
7.
C.L. Casper, J.S. Stephens, N.G. Tassi, D.B. Chase and J.F. Rabolt, Macromolecules, 2004, 37, 573.
8.
W. Dierickx, Geotext Geomembranes, 1999, 17, 231.
9.
S.K. Bhatia and J.L. Smith, Geosynthetics International, 1996, 3, 155.
10. S.K. Bhatia and J.L. Smith, Geosynthetics International, 1996, 3, 301. 11. S.T. Ho and D.W. Hutmacher, Biomaterials, 2006, 27, 1362. 12. A.H. Aydilek and T.B. Edil, Geotechnical Testing Journal, 2004, 27, 1. 13. A.H. Aydilek, S.H. Oguz and T.B. Edil, Journal of Computing in Civil Engineering, 2002, 16, 280. 14. R.C. Gonzalez and R.E. Woods in Digital Image Processing, 2nd Edition, Prentice Hall, New Jersey, NJ, USA, 2001. 15. B. Jähne in Digital Image Processing, 5th Edition, Springer, Germany, 2002, p.125. 16. N. Otsu, IEEE Transactions on Systems Man and Cybernetics, Part C, 1979, 9, 62. 17. H.S. Kim and B. Pourdeyhimi, International Nonwovens Journal, Winter 2000, p.15. 18. M.S. Abdel-Ghani and G.A. Davies, Chemical Engineering & Science, 1985, 40, 117. 19. B. Pourdeyhimi, R. Ramanathan and R. Dent, Textile Research Journal, 1996, 66, 713.
119
Advances in Nanofibre Research
120
6
Practical Hints on the Production of Electrospun Nanofibres from Regenerated Silk Fibroin
6.1 Introduction Elecrospinning is a process that produces continuous polymer fibres with diameters in the submicron range. Electrospinning has emerged as a specialised processing technique for the formation of submicron fibres (typically between 100 nm and 1 µm in diameter), with high specific surface areas. Due to their high specific surface area, high porosity, and small pore size, these unique fibres have been suggested to be excellent candidates for use in filtration [1, 2]. In the non-woven industry, one of the fastest growing segments is in filtration applications. Traditionally wet-laid, melt blown and spun non-woven articles containing micron-size fibres are most popular for these applications because of their low cost, easy processing and good filtration efficiency. Their applications in filtration can be divided into two major areas: air filtration and liquid filtration [3]. Air and water are the bulk transportation media for the transmission of particulate contaminants. The contaminants during air filtration are a complex mixture of particles. Most of them are usually <1000 µm in diameter; chemical and biological aerosols are frequently in the range 1–10 µm. Particulate materials may carry some gaseous contaminants. In water filtration, removal of particulates and biological contaminants is an important step. Currently, the filtration industry is looking for energy-efficient high-performance filters for filtration of particles <0.3 µm and absorbed toxic gases [4]. Nanofibrous media have low basis weight, high permeability, and small pore size that make them appropriate for a wide range of filtration applications. In addition, the nanofibre membrane offers unique properties such as high specific surface area (1–35 m2/g depending on fibre diameter), good interconnectivity of pores, and the potential to incorporate active chemistry or functionality at the nanoscale [4, 5]. The structural characteristics of nanofibrous filtering media such as layer thickness, fibre diameter, nanofibre orientation, representative pore size, and porosity dictate the filter properties and fibre quality [4].
121
Advances in Nanofibre Research The properties of nanofibrous media will be dependent upon their structural characteristics as well as the nature of the component fibres. Understanding and determining these characteristics is therefore desirable. In this work, we tried to identify the orientation distribution function (ODF) of nanofibres in nanofilters, the distribution of fibre thickness, and the porosity of nanofibrous media using image processing algorithms. Fourier methods are useful for extracting orientation information by transforming an intensity image into a frequency image in which a higher rate of change in greyscale intensity is reflected in higher amplitudes [6–7].
6.2 Effect of Systematic Parameters on Electrospun Nanofibres It has been found that the morphology (e.g., fibre diameter and fibre uniformity) of electrospun nanofibres is dependent upon many processing parameters. These parameters can be divided into three main groups: (a) solution properties; (b) processing conditions; and (c) ambient conditions. Each of the parameters has been found to affect the morphology of electrospun fibres.
6.2.1 Solution Properties Parameters such as the viscosity, solution concentration, and molecular weight of a solution, as well as its electrical conductivity, elasticity and surface tension, have important effects on the morphology of nanofibres.
6.2.2 Viscosity The viscosity range of different nanofibre solutions which can be electrospun is different. One of the most significant parameters influencing fibre diameter is solution viscosity. A higher viscosity results in a large fibre diameter. Beads and beaded fibres are less likely to be formed from more viscous solutions. The diameter of the beads become bigger and the average distance between beads on the fibres longer as viscosity increases.
6.2.3 Solution Concentration In the electrospinning process, for fibre formation to occur, a minimum solution concentration is required. As the solution concentration increases, a mixture of beads 122
Practical Hints on the Production of Electrospun Nanofibres from Regenerated Silk Fibroin and fibres is obtained. The shape of the beads changes from spherical to spindle-like if the solution concentration varies from low to high. The fibre diameter increases with increasing solution concentration because of the higher viscosity resistance. Nevertheless, at higher concentrations, the viscoelastic force (which usually resists rapid changes in fibre shape) may result in uniform fibre formation. However, it is impossible to electrospin if the solution concentration or the corresponding viscosity becomes too high due to the difficulty in formation of a liquid jet.
6.2.4 Molecular Weight Molecular weight also has a significant effect on the rheological and electrical properties (e.g., viscosity, surface tension, conductivity and dielectric strength). It has been observed that too low-molecular-weight solutions tend to form beads rather than fibres, and that high-molecular-weight nanofibre solutions give fibres with larger average diameters.
6.2.5 Surface Tension The surface tension of a liquid is defined as the force acting at right angles to any line of unit length on the liquid surface. By reducing the surface tension of a nanofibre solution, fibres could be obtained without beads. The surface tension seems more likely to be a function of solvent compositions, but is negligibly dependent upon the solution concentration. Different solvents may contribute different surface tensions. However, a lower surface tension of a solvent will not necessarily be more suitable for electrospinning. In general, the surface tension determines the upper and lower boundaries of the ‘electrospinning window’ if all other variables are held constant. The formation of droplets, beads and fibres can be driven by the surface tension of a solution and a lower surface tension of the spinning solution helps electrospinning to occur at a lower electric field.
6.2.6 Solution Conductivity There is a significant drop in the diameter of electrospun nanofibres if the electrical conductivity of the solution increases. Beads may also be observed due to low conductivity of the solution, which results in insufficient elongation of a jet by electrical forces to produce uniform fibres. In general, electrospun nanofibres with the smallest fibre diameter can be obtained with the highest electrical conductivity. This suggests that the drop in size of the fibres is due to increased electrical conductivity.
123
Advances in Nanofibre Research
6.2.7 Applied Voltage In electrospinning, the electric current due to the ionic conduction of charge in the nanofibre solution is usually assumed to be sufficiently small to be negligible. The only mechanism of charge transport is the flow of solution from the tip to the target. Thus, an increase in the electrospinning current generally reflects an increase in the mass flow rate from the capillary tip to the grounded target when all other variables (conductivity, dielectric constant, and flow rate of solution to the capillary tip) are held constant. Increasing the applied voltage (i.e., increasing the electric field strength) will increase the electrostatic repulsive force on the fluid jet, which favours the formation of thinner fibres. Conversely, the solution will be removed from the capillary tip more quickly as the jet is ejected from the Taylor cone. This results in an increase in fibre diameter.
6.2.8 Feed Rate The morphological structure can be slightly changed by changing the solution flow rate. If the flow rate exceeds a critical value, the delivery rate of the solution jet to the capillary tip exceeds the rate at which the solution is removed from the tip by the electric forces. This shift in the mass balance results in a sustained but unstable jet and fibres with big beads.
6.3 Experimental 6.3.1 Electrospinning and Preparation of Nanofibrous Media Silk fibre wastes were degummed in aqueous 0.5% (w/w) NaHCO3 and rinsed with water to extract sericin and obtain silk fibroin (SF). The degummed silk was then dissolved in ternary CaCl2/CH3CH2OH/H2O (molar ratio 1:2:8) at 70 °C for 6 hours and then dialysed with cellulose tubular membrane (pore size, 25 Å; Aldrich) for 3 days. Dialysed SF was lyophilized because SF became sponge-like. A 8% w/w and 12% w/w of SF solution in formic acid was obtained for producing silk nanofibrous filter media. A 8% w/w and 13% w/w polyacrylonitrile (PAN) solution for electrospinning was prepared by dissolving a pre-determined quantity of PAN (MW 150,000; Polyacryle Company) in n,n-dimethyl formamide. The electrospinning apparatus (Figure 6.1) consisted of a 5.0 ml syringe, a highvoltage power supply (able to produce 0–30 kV), syringe pump and a rotating collector (stainless-steel drum) with a diameter of 6.75 cm and a length of 13 cm. The 124
Practical Hints on the Production of Electrospun Nanofibres from Regenerated Silk Fibroin electrospinning parameters for silk were: voltage, 15 kV; needle distance, 7 cm; and collector drum speed, 100 rpm. For PAN they were: voltage, 12 kV; needle distance, 10 cm; and collector drum speed, 100 rpm.
polymer solution syringe
pump Taylor cone
straight jet
needle unstable region
high voltage
collector drum step motor ground Figure 6.1 Electrospinning setup (schematic)
6.3.2 Image Analysis using Image Processing Algorithms The morphologies of nanofibres were observed by scanning electron microscopy (SEM). Nanofibrous mats were imaged using a Philips XL-30 scanning electron microscope. SEM images were converted to greyscale. Fast Fourier transform (FFT) was carried out on all greyscale images. Figure 6.2 and Figure 6.3 illustrate application
125
Advances in Nanofibre Research of Fourier transform on a sample image with known orientation angles 0o, 20o and 90o for the porosity analysis. SEM micrographs were converted to a binary format and then used (the picture pixels had only two values: 0 and 255).
Figure 6.2 Sample image with known orientation angles, 0o, 20o and 90o
6.4 Results and Discussion 6.4.1 Diameter Distribution of Nanofibres The diameter distribution of nanofibres and their average values were extracted using the ImageJ programme (http://rsb.Info.ni h.gov/ij/). Figure 6.4 shows the nanofibrous media obtained from solutions of 8% w/w and 12% w/w silk/formic acid. At a concentration of 12% w/w, the average diameter of fibres was much larger than that of fibres spun at 8% w/w. The distribution of fibre diameters at 8% w/w and 12% w/w concentrations is shown on the right of Figure 6.4. The fibre distribution becomes broader with increasing concentration. Figure 6.5 shows the same results for PAN nanomats.
126
Practical Hints on the Production of Electrospun Nanofibres from Regenerated Silk Fibroin 1200 1000
APS
800 600 400 200 0 0
20
(a)
40
60
80 100 120 theta (degree)
140
160
180
(b)
Figure 6.3 Fast Fourier transform (FFT) and angular power spectrum of a sample
25
Frequency percent
20
Fiber Diameter Distribution average: 339 ± 20 (nm)
15 10 5 0
30
Frequency percent
25
100 200 300 400 500 600 700 800 900 Fiber Diameter (nm) Fiber Diameter Distribution average: 1213 ± 20 (nm)
20 15 10 5 0
200 400 600 800 1000 1200 1400 1600
Fiber Diameter (nm)
Figure 6.4 Distribution of fibre diameters and morphology of silk nanofibres at concentrations a) 8% w/w and b) 12% w/w at a constant tip-to-collector distance of 7 cm, applied voltage of 15 kV, and collector speed of 100 rpm
127
Advances in Nanofibre Research
30
Frequency percent
25 20
Fiber Diameter Distribution
average: 700 ± 20 (nm)
15 10 5 0
Frequency percent
18 16 14 12
200 400 600 800 10001200 1400 1600 Fiber Diameter (nm)
Fiber Diameter Distribution average: 895 ± 20 (nm)
10 8 6 4 2 0
400 500 600 700 800
900
1000 1100 1200 1300
Fiber Diameter (nm)
Figure 6.5 Distribution of fibre diameters and morphology of PAN nanofibres at concentrations a) 8% w/w and b) 13% w/w at a constant tip-to-collector distance of 10 cm, applied voltage of 12 kV, and collector speed of 100 rpm
128
Practical Hints on the Production of Electrospun Nanofibres from Regenerated Silk Fibroin 22
x 104
2 18 16 APS
14
12 1 08 06 04 0
20
40
60
80 100 120 140 160 180 theta (degree)
15000
APS
1000
5000
0 0
20
40
60
80 100 120 140 160 180 theta (degree)
2000 1800 1600 APS
1400 1200 1000 800 600 0
2
20
40
60
80 100 120 140 160 180 theta (degree)
40
60
80 100 120 140 160 180 theta (degree)
x 104
18 16 14 APS
12 1
08 06 04 02 0
20
Figure 6.6 FFT and angular power spectrum of a) PAN 8% w/w (Figure 6.6a); b) PAN 13% w/w (Figure 6.6b); c) silk 8% w/w (Figure 6.5a); and d) silk 13% w/w (Figure 5b)
129
Advances in Nanofibre Research
6.4.2 Distribution of Nanofibre Orientation The results for the FFT and orientation distribution of nanofibres (angular power spectrum histogram) is shown in Figure 6.6. The FFT could detect the approximate angular orientation of fibres. Although the collector speed was constant for all samples of nanomat, nanofibres from low-concentration solutions were more uniform than nanofibres from highconcentration solutions.
6.4.3 Porosity Figure 6.7 is a sample with known porosity for checking the accuracy of the MatLab programme written for calculating porosity. The porosity of samples is depicted in Table 6.1. As shown in Table 6.1 the porosity of nanofibres electrospun from highconcentration solutions was more than those electrospun from low-concentration solutions. It is clear that by increasing solution concentration the fibre diameter increases. Subsequently, by increasing the diameter, the surface-to-volume ratio of fibres decreases and the pore size between fibres become larger. This is assuming that fibre numbers and subsequently pore numbers do not change significantly, and that the pore size dominates the other parameters.
Figure 6.7 Sample image with known porosity (white pixels are fibres; porosity = 0.75)
6.5 Conclusions The porosity of nanofilters and nanofibre diameter and its statistical parameters (average and distribution) can be calculated by analysing SEM images.
130
Practical Hints on the Production of Electrospun Nanofibres from Regenerated Silk Fibroin The results indicated that increasing solution concentration leads to larger fibre diameter and broader diameter distribution in silk nanofibres and PAN nanofibres. Image analysis of porosity illustrated that, in nanofibrous media with larger fibre diameter, there is more porosity and many more empty spaces than nanomats with finer nanofibres. It is clear that Fourier methods can provide good approximations of the ODF and can be a useful tool for nanofibrous characterisation.
Table 6.1 Porosity of electrospun nanomats Figure number
Figure size (pixels)
Fibre pixels (white)
Porosity
8
472 × 472
167088
0.25
5a (silk 8% w/w)
1056 ×732
428792
0.446
5b (silk 12% w/w)
2202 × 1686
2051467
0.492
6a (PAN 8% w/w)
2203 × 1640
1886830
0.432
6b (PAN 13% w/w)
2200 × 737
1539663
0.572
References 1.
M. Ziabari, V. Mottaghitalab and A.K. Haghi, Brazilian Journal of Chemical Engineering, 2009, 26, 1, 53.
2.
M. Ziabari, V. Mottaghitalab and S.T. McGovern, Chinese Physics Letters, 2008, 25, 8, 3071.
3.
M. Ziabari, V. Mottaghitalab, S.T. McGovern and A.K. Haghi, Nanoscale Research Letters, 2007, 2, 297.
4.
M. Ziabari, V. Mottaghitalab and A.K. Haghi, Korean Journal of Chemical Engineering, 2008, 25, 4, 919.
5.
M. Ziabari, V. Mottaghitalab and A.K. Haghi, Korean Journal of Chemical Engineering, 2008, 25, 4, 923.
131
Advances in Nanofibre Research 6.
M. Ziabari, V. Mottaghitalab and A.K. Haghi, Korean Journal of Chemical Engineering, 2008, 25, 4, 905.
7.
A.K. Haghi and M. Akbari, Physica Status Solidi A, 2007, 204, 6, 1830.
132
7
Characterisation of Polymeric Electrospun Nanofibres
7.1 Introduction In recent years, nanotechnology has become a topic of great interest to scientists and engineers, it is now established as a prioritised research area in many countries. The reduction of the size to the nanometre range brings an array of new possibilities in terms of material properties, in particular with respect to achievable surface-tovolume ratios. Electrospinning of nanofibres is a novel process for producing superfine fibres by forcing a solution through a spinnerette with an electric field. An emerging technology of manufacturing of thin natural fibres is based on the principle of electrospinning. In conventional fibre spinning, the mechanical force is applied to the end of a jet. In the electrospinning process, the electric body force acts on elements of the charged fluid. Electrospinning has emerged as a specialised processing technique for the formation of sub-micron fibres (typically between 100 nm and 1 µm in diameter), with high specific surface areas. Due to their high specific surface area, high porosity, and small pore size, these unique fibres have been suggested for a wide range of applications. Electrospinning of nanofibres offers unique capabilities for producing novel natural nanofibres and fabrics with controllable pore structure. About 4–9% of cotton fibre is lost in textile mills in ‘opening and cleaning’. This involves mechanically separating compressed clumps of fibres for removal of trapped debris. Another 1% is lost in ‘drawing’ and ‘roving-pulling’ lengths of fibre into longer and longer segments, which are then twisted together for strength. An average of 20% is lost during ‘combing’ and ‘yarn production’. Typically, waste cotton is used in relatively low-value products such as cotton balls, yarn, and cotton batting. A new process for electrospinning waste cotton that employs a less harmful solvent has been developed. Electrospinning is an economical and simple method used in the preparation of polymer fibres. The fibres prepared by this method typically have diameters much smaller than is possible to attain using standard mechanical fibre-spinning technologies [1]. Electrospinning has gained much attention in the last few years as a cheap and
133
Advances in Nanofibre Research straightforward method to produce nanofibres. Electrospinning differs from the traditional wet/dry fibre spinning in a number of ways, of which the most striking differences are the origin of the pulling force and the final fibre diameters. The mechanical pulling forces in the traditional industrial fibre spinning processes lead to fibres in the micrometer range and are contrasted in electrospinning by electrical pulling forces that enable the production of nanofibres. Depending on the solution properties, the throughput of single-jet electrospinning systems is ~10 ml/min. This low fluid throughput may limit the industrial use of electrospinning. A stable cone-jet mode followed by the onset of the characteristic bending instability (which eventually leads to a great reduction in the jet diameter) necessitate the low flow rate [2]. If the diameters of cellulose fibre materials are shrunk from micrometers (e.g., 10–100 mm) to submicrons or nanometres, several amazing characteristics appear. These include a very large surface area-to-volume ratio (this ratio for a nanofibre can be 103-times that of a microfibre), flexibility in surface functionalities, and superior mechanical performance (e.g., stiffness and tensile strength) compared with any other form of the material. Such outstanding properties make polymer nanofibres optimal candidates for many important applications [3]. These include filter media, composite materials, biomedical applications (tissue-engineering scaffolds, bandages, drug-release systems), protective clothing for the military, optoelectronic devices, semi-conductive materials, and biosensors/chemosensors [4]. Another biomedical application of electrospun fibres currently receiving much attention is drug-delivery devices. Researchers have monitored the release profile of several drugs from various biodegradable electrospun membranes. Another application for electrospun fibres is porous membranes for filtration devices. Due to the interconnected network structure that electrospun fibres form, they exhibit good tensile properties, low air permeability, and good aerosol protection capabilities. Moreover, by controlling the fibre diameter, electrospun fibres can also be produced over a wide range of porosities. Research has also focused on the influence of the charging effects of electrospun non-woven mats on their filtration efficiency. The filtration properties were found to slightly depend upon the surface charge of the membrane, but the fibre diameter was found to have the strongest influence on aerosol penetration. Electrospun fibres are also being utilised for several other applications, including nanocomposites. Figure 7.1 compares the dimensions of nanofibres, microfibres and conventional fibres. When the diameters of polymer fibre materials are shrunk from micrometres (e.g., 10–100 µm) to submicrons or nanometres (e.g., 10 × 10–3 µm to 100 × 10–3 µm), several amazing characteristics appear. These include a very large surface area-to-volume ratio, flexibility in surface functionalities, and superior mechanical performance compared with other forms of material. These outstanding properties make the polymer nanofibres optimal candidates for many important applications [4].
134
Characterisation of Polymeric Electrospun Nanofibres
Nano Fiber: <1 µm Micro Fiber: 10-50 µm Ordinary Fiber: 50-200 µm
Figure 7.1 Classification of fibres by diameter
7.1.1 Electrospinning Setup The electrospinning setup of nanofibres is shown in Figure 7.2.
135
Advances in Nanofibre Research
Polymer Solution High Voltage power supply
Collector Plate
Figure 7.2 Electrospinning set up
Syringe Metering pump
Connector 0
1
Electrical heater High voltage supply
25 ºC
Temperature controller
Figure 7.3 Electrospinning process
136
Characterisation of Polymeric Electrospun Nanofibres There are basically three components to fulfill the process: a high-voltage supplier, a capillary tube with a pipette or needle of small diameter, and a metal collecting screen. In the electrospinning process (Figure 7.3), a high-voltage is used to create an electrically charged jet of polymer solution or melt to emerge out of the pipette. Before reaching the collecting screen, the solution jet evaporates or solidifies, and is collected as an interconnected web of small fibres [5]. One electrode is placed into the spinning solution/melt or needle, and the other attached to the collector. In most cases, the collector is simply grounded. The electric field is subjected to the end of the capillary tube that contains the solution fluid held by its surface tension. This induces a charge on the surface of the liquid. Mutual charge repulsion and contraction of the surface charges to the counter electrode cause a force directly opposite to the surface tension [6]. As the intensity of the electric field is increased, the hemispherical surface of the fluid at the tip of the capillary tube elongates to form a conical shape known as the ‘Taylor cone’ [7]. Further increase in the electric field results in a critical value with which the repulsive electrostatic force overcomes the surface tension and the charged jet of fluid is ejected from the tip of the Taylor cone [8]. The jet exhibits bending instabilities due to repulsive forces between the charges carried with the jet. The jet extends through spiraling loops. As the loops increase in diameter, the jet grows longer and thinner until it solidifies or collects on the target [9]. Schematic diagrams to show the electrospinning of polymer nanofibres are shown in Figure 7.4 and Figure 7.5. The basic three components to fulfill the process are the same as described above. In the electrospinning process a high voltage is used to create an electrically charged jet of polymer solution or melt out of the pipette. Before reaching the collecting screen or drum, the solution jet evaporates or solidifies, and is collected as an interconnected web of small fibres. When an electric field is applied between a needle capillary end and a collector, surface charge is induced on a polymer fluid, deforming a spherical pendant droplet to a conical shape. As the electric field surpasses a threshold value where electrostatic repulsion force of surface charges overcome surface tension, the charged fluid jet is ejected from the tip of the Taylor cone and the charge density on the jet interacts with the external field to produce instability. It has been found that the morphology such as fibre diameter and the uniformity of the electrospun polymer fibres are dependent upon many processing parameters. These parameters can be divided into three groups (Table 7.1). Under certain conditions, not only uniform fibres but also bead-like fibres can be produced by electrospinning. Although the parameters of the electrospinning process have been analysed in each type of polymer, this information has been inadequate to support the electrospinning of ultra-fine nanometre-scale polymer fibres. A more systematic parametric study is required.
137
Advances in Nanofibre Research PPSN + nanofiber
High voltage
Syringe pump
Traversing plate
Rotating drum
Figure 7.4 The electrospinning setup by the drum collector (schematic)
Polymer solution
Pt Electrode Jet
Syringe Metering Pump
Glass Pipette Taylor cone
High Voltage Supply
Target (Rotating or Stationary)
Figure 7.5 The electrospinning setup by the screen collector (schematic)
138
Characterisation of Polymeric Electrospun Nanofibres
Table 7.1 Processing parameters in electrospinning Solution properties
Viscosity Polymer concentration Molecular weight of polymer Electrical conductivity Elasticity Surface tension
Processing conditions
Applied voltage Distance from needle to collector Volume feed rate Needle diameter
Ambient conditions
Temperature Humidity Atmospheric pressure
7.2 Effect of Systematic Parameters on Electrospun Nanofibres The three parameters mentioned above can be divided into three main groups: (a) solution properties; (b) processing conditions; and (c) ambient conditions. Each of the parameters has been found to affect the morphology of electrospun fibres.
7.2.1 Solution Properties Parameters such as the viscosity, concentration, and molecular weight of a solution, as well as electrical conductivity, elasticity and surface tension, have important effects on the morphology of nanofibres.
7.2.1.1 Viscosity The viscosity range of different nanofibre solutions which can be spun is different. One of the most significant parameters influencing fibre diameter is solution viscosity.
139
Advances in Nanofibre Research A higher viscosity results in a large fibre diameter. Figure 7.6 shows the representative images of beads formed in electrospun nanofibres. Beads and beaded fibres are less likely to be formed frome more viscous solutions. The diameter of the beads becomes bigger and the average distance between beads on the fibres longer as viscosity increases.
Figure 7.6 Electron micrograph of bead formation in electrospun nanofibres
7.2.1.2. Solution Concentration In the electrospinning process, for fibre formation to occur, a minimum solution concentration is required. As the solution concentration increases, a mixture of beads and fibres is obtained (Figure 7.7). The shape of the beads changes from spherical to spindle-like when the solution concentration varies from low to high. The fibre diameter increases with increasing solution concentration because of the higher viscosity resistance. Nevertheless, at higher concentration, viscoelastic forces which usually resist rapid changes in fibre shape may result in the formation of uniform fibres. However, it is impossible to electrospin if the solution concentration or the corresponding viscosity become too high due to the difficulty in formation of a liquid jet.
140
Characterisation of Polymeric Electrospun Nanofibres
Figure 7.7 Electron micrograph of formation of beads and fibres in electrospun nanofibres
7.2.1.3 Molecular Weight Molecular weight also has a significant effect on the rheological and electrical properties such as viscosity, surface tension, conductivity and dielectric strength. It has been reported that too low-molecular-weight solutions tend to form beads rather than fibres and that high-molecular-weight nanofibre solutions give fibres with larger average diameter (Figure 7.8).
7.2.1.4 Surface Tension The surface tension of a liquid is often defined as the force acting at right angles to any line of unit length on the liquid surface. However, this definition is misleading because there is no elastic skin or tangential force at the surface of a pure liquid. It is more satisfactory to define surface tension and surface free energy as the work required to increase the area of a surface isothermally and reversibly by a unit amount. As a consequence of surface tension, there is a balancing pressure difference across any curved surface, the pressure being greater on the concave side. By reducing surface tension of a nanofibre solution, fibres could be obtained without beads
141
Advances in Nanofibre Research (Figures 7.9–7.10). This might be correct in some sense, but should be applied with caution. The surface tension seems more likely to be a function of solvent compositions, but is negligibly dependent upon the solution concentration. Different solvents may contribute different surface tensions. However, not necessarily a lower surface tension of a solvent will always be more suitable for electrospinning. In general, surface tension determines the upper and lower boundaries of the ‘electrospinning window’ if all other variables are held constant. The formation of droplets, beads and fibres can be driven by the surface tension of a solution and a lower surface tension of the spinning solution helps electrospinning to occur at lower electric fields.
Figure 7.8 Electron micrograph of formation of variable diameters in electrospun nanofibres
7.2.1.5 Solution Conductivity There is a significant drop in the diameter of electrospun nanofibres if the electrical conductivity of the solution increases. Beads may also be observed due to low conductivity of the solution, which results in insufficient elongation of a jet by electrical force to produce uniform fibres. In general, electrospun nanofibres with the smallest fibre diameter can be obtained with the highest electrical conductivity. This means that the drop in the size of the fibres is due to the increased electrical conductivity. 142
Characterisation of Polymeric Electrospun Nanofibres
Figure 7.9 Electron micrograph of electrospun nanofibres without bead formation
Figure 7.10 Electron micrograph of electrospun nanofibres with minor beads formation
143
Advances in Nanofibre Research
7.2.2 Processing Condition
7.2.2.1 Applied Voltage In the case of electrospinning, the electric current due to the ionic conduction of charge in the nanofibre solution is usually assumed to be small enough to be negligible. The only mechanism of charge transport is the flow of solution from the tip to the target. Thus, an increase in the electrospinning current generally reflects an increase in the mass flow rate from the capillary tip to the grounded target when all other variables (conductivity, dielectric constant, and flow rate of solution to the capillary tip) are held constant. With the increase of the electrical potential the resulting nanofibres became rougher. It is sometimes reported that a diameter of electrospun fibres is not significantly affected by an applied voltage. This voltage effect is particularly diminished if the solution concentration is low. The applied voltage may affect some factors such as the mass of solution fed out from a tip of a needle, the elongation level of a jet by an electrical force, and the morphology of a jet (single or multiple jets). A balance among these factors may determine a final diameter of electrospun fibres. Beaded fibres may be found to be electrospun with too high a level of applied voltage. Although voltage effects show different tendencies, the voltage generally does not have a significant role in controlling fibre morphology. Nevertheless, increasing the applied voltage (i.e., increasing the electric field strength) will increase the electrostatic repulsive force on the fluid jet, which favors the formation of thinner fibres. Conversely, the solution will be removed from the capillary tip more quickly as the jet is ejected from the Taylor cone. This results in the increase in fibre diameter.
7.2.2.2 Feed Rate The morphological structure can be slightly changed by changing the solution flow rate (Figure 7.11). At a flow rate of 0.3 ml/h, a few big beads were observed on the fibres. When the flow rate exceeded a critical value, the delivery rate of the solution jet to the capillary tip exceeded the rate at which the solution was removed from the tip by the electric forces. This shift in the mass-balance resulted in a sustained but unstable jet and fibres with big beads. The electrical conductivity of the solution was found to be a dominant parameter to control the morphology of electrospun nanofibres [9]. In a low-molecular-weight
144
Characterisation of Polymeric Electrospun Nanofibres liquid, when a high electrical force is applied, formation of droplets can occur. A theory proposed by Rayleigh explained this phenomenon. As evaporation of a droplet occurs, the droplet decreases in size. Therefore, the charge density of its surface is increased. This increase in charge density due to Coulomb repulsion overcomes the surface tension of the droplet and causes the droplet to split into smaller droplets. However, in the case of a solution with a high molecular weight, the emerging jet does not break up into droplets, but is stabilised and forms a string of beads connected by a fibre. As the concentration is increased, a string of connected beads is seen, and with further increase there is reduced bead formation until only smooth fibres are formed. Also, sometimes spindle-like beads can form due to the extension caused by electrostatic stress. The changing of fibre morphology can probably be attributed to a competition between surface tension and viscosity. As concentration is increased, the viscosity of the solution also increases. The surface tension attempts to reduce surface area per unit mass, thereby causing the formation of beads/spheres. Viscoelastic forces resist the formation of beads and allow for the formation of smooth fibres. Therefore, formation of beads at lower solution concentration (low viscosity) occurs if surface tension has a greater influence than the viscoelastic force. However, bead formation can be reduced and finally eliminated at higher solution concentration if viscoelastic forces have a greater influence in comparison with surface tension. However, if the concentration is too high, high viscosity and rapid evaporation of solvent makes the extension of jet more difficult, thicker and non-uniform fibres will be formed. A suitable level of processing parameters must be optimised to electrospin solutions into nanofibres with desired morphology, and the parameter levels are dependent upon the properties of the solution and solvents used in each of the electrospinning processes. Understanding how each of the processing parameters affect the morphology of the electrospun nanofibres is essential. All the parameters can be divided into two main groups, parameters that affect: (i) the mass of solution fed out from the tip of the needle, and (ii) an electrical force during electrospinning. Solution concentration, applied voltage and volume feed rate are usually considered to affect the mass. Increased solution concentration and feed rate tend to bring more mass into the jet. A high applied voltage reflects the force needed to pull a solution out from the needle, so a higher applied voltage causes more solution to come out. Conversely, the electrical conductivity and applied voltage of the solution affect the charge density, thus an electrical force acts to elongate a jet during electrospinning.
7.3 Experimental Polyacrylonitrile (PAN) fibre (Dolan) was obtained from Hoechst. N-methyl-2pyrolidon (NMP) was purchased from Riedel-de Haën. The polyaniline used was synthesised in our laboratory. Polyaniline (PANi) was synthesised by the oxidative 145
Advances in Nanofibre Research polymerisation of aniline in acidic media. Three millilitres of distilled aniline was dissolved in 150 ml of 1 N HCl and kept at 0 °C, 7.325 g of (NH4)2S2O8 was dissolved in 35 ml of 1 N HCl and added dropwise under constant stirring to the aniline/HCl solution over 20 minutes. The resulting dark-green solution was maintained under constant stirring for 4 hours. It was then filtered and washed with methanol, then with water. It was dried before being added to 150 mL of 1 N (NH4)OH solution. After an additional 4 hours, the solution was filtered and a deep-blue emeraldine base form of polyaniline obtained. It was dried and crushed into fine powder and then passed trough 100 mesh.
Figure 7.11 Electron micrograph of electrospun nanofibres when flow rate exceeded the critical value
Polymer solutions were prepared by first dissolving an exact amount of PANi in NMP. The PANi was slowly added to the solvent with constant stirring at room temperature. This solution was then allowed to stir for 1 hour in a sealed container. The PAN/NMP solution was prepared separately and added dropwise to the wellstirred PANi solution. The blend solution was agitated with a mechanical stirrer for an additional 1 hour. By mixing different solution ratios (0/100, 50/50, 60/40, 75/25) of 5% PANi solution and 20% PAN solution, various polymer blend solutions were prepared with the
146
Characterisation of Polymeric Electrospun Nanofibres concentration of polyaniline ranging from 5 wt% to 42 wt%. The fibre diameter and polymer morphology of the electrospun polyaniline/polyacrylonitrile NMP solution were determined using an optical microscope (Nikon Microphot-FXA). A small section of the non-woven mat was placed on the glass slide and placed on the microscope sample holder. A scanning electron microscope (SEM) Philips XL-30 was used to take photographs to do a more precise characterisation. A small section of the web was placed on the SEM sample holder and coated with gold (BAL-TEC SCD 005 sputter coater).
7.4 Result and Discussion In our first experiment, we tried to ascertain if electrospinning of PANi pure solution results in web formation. Without the addition of PAN to PANi dissolved in NMP, no web formation occurred because the concentration and viscosity of the solution was not high enough to form a stable drop at the end of the needle, and only a few dispersed drops were formed on the collector. Adding more polyaniline did not increase the solution viscosity and resulted in gelation of the solution. Based on these results, the blend solution was electrospun. The initial results showed that fine fibres are formed at room temperature. In these blends, the concentration of PANi ranged from 5 wt% to 42 wt%. The potential difference between the needle tip and the electrode was 20 Kv. Optical microscope photomicrographs (Figure 7.12) showed that the fibres are formed but they are entangled with each other, and that bead forming is observed. To get more uniform webs, we tried to obtain webs at the gap between two metal stripes which were placed on the collector plate. Webs became more uniform but entanglement and beads were observed (Figure 7.13). For examining the web formation of PAN, it was electrospun from NMP at different concentrations (10% and 15%). Webs were formed between 17 V and 20 V. PAN webs showed excellent web-forming behaviour and the resulting webs were uniform (Figure 7.14). Optical microscope micrographs confirmed that fibres were formed. Higher concentrations of PANi were used, and the resulting webs were examined by SEM to study their diameter and morphology more precisely. SEM photomicrographs revealed that the diameter of fibres in non-woven mats ranged from a minimum of 160 nm to a maximum of 560 nm, with an average fibre diameter of 358 nm. It was noticed that, by increasing the amount of PAN that fibre formation was enhanced and more uniform fibres were obtained. Also, the variation in fibre diameter was smaller. It could be related to the intrinsic fibre-forming behaviour of PAN because
147
Advances in Nanofibre Research it is used widely as the base material for producing fibres and yarns. By increasing the PANi ratio, the amount of beading increased and fibres became entangled with each other before reaching the collector; a uniform web was formed in some parts. It seemed that the fibres were wet and that this was the cause of fibres sticking together. In 42% PANi, web formation was not seen and instead we had an entangled bulk with some polymer drops. With increasing PANi ratio the fibre diameter decreased (Table 7.2):
Table 7.2 Fibre diameters in different PANi ratios PANi percent (blending ratio)
0% (0/100)
20% (50/50)
27% (60/40)
Fibre diameter
445 nm
372 nm
292 nm
By increasing the temperature of the electrospinning environment to 75 oC to let the solvent evaporate more rapidly, the problem of twisted fibres was overcome and more uniform webs and finer nanofibres were formed, but the beading problem remained. More research is in progress to enhance the web characterisation and decrease the fibre diameter to real nanometre size.
Figure 7.12 Optical microscope micrograph of 16% PANi blend solution; beads can be clearly observed
148
Characterisation of Polymeric Electrospun Nanofibres
Figure 7.13 Optical microscope micrograph of 16% PANi blend solution caught in air
Figure 7.14 SEM photomicrograph of pure PAN
149
Advances in Nanofibre Research
7.5 Conclusion Nanofibres of pure PAN dissolved in NMP were prepared, but pure PANi/NMP solution did not show web formation. By adding PAN, fibre formation was observed. Different PANi/PAN blends were electrospun, and the average diameter of nanofibres was 385 nm. By increasing the amount of PANi, the resulting nanofibre diameter decreased, and the webs became more irregular and non-uniform. The electrospinning technique provides an inexpensive and easy way to produce nanofibres of low basis weight, small fibre diameter and pore size. It is hoped that this chapter will pave the way toward a better understanding of the application of electrospinning of nanofibres. There are three categories of variables that influence the electrospun fibre diameter: (1) polymer solution variables; (2) process variables; and (3) environmental variables. Examples of solution variables are viscosity or polymer concentration, solvent volatility, conductivity, and surface tension. Process variables consist of electric field strength, fluid flow rate, and distance between electrodes. Low-molecular-weight fluids form beads or droplets in the presence of an electric field, whereas high-molecular-weight fluids generate fibres. However, an intermediate process is the occurrence of the ‘beads on a string’ (Figure 7.15 and Figure 7.16) morphology. In many instances, bead formation is also observed in addition to fibre growth. This morphology is a result of capillary break-up of the spinning jet caused by surface tension. Solution conductivity is another polymer solution property that greatly influences electrospun fibre diameter. The addition of salts to polymer solutions has been shown to increase the resulting net charge density of the electrospinning jet. The surface tension of the polymer solution also influences the resulting fibre morphology because large surface tensions promote the formation of polymer droplets. The surface tension of the fluid must be overcome by the electrical voltage for emission of an electrified jet from the syringe. Process variables also control the morphology of fibres during the electrospinning process. In general, fibre diameter is insensitive to process conditions when compared with varying the polymer solution properties. However, extensive work has been done on the influence of voltage, flow rate, and working distance on electrospun fibre morphology. The distance between the electrodes or the working distance influences the electrospinning process. In general, as the working distance decreases, the time for the flight of the path for the fluid jet decreases. Moreover, temperature is a convoluted variable when attempting to discern its influence on electrospun fibre formation. Increasing the solution temperature causes: (1) a change in chain conformation in solution; (2) a decrease in solution viscosity; and (3) an increase in rate of solvent evaporation. Thus, quantifying the effect of temperature on electrospinning proves difficult because all of the above can influence fibre morphology. Humidity has also been shown to control the surface morphology of electrospun fibres. 150
Characterisation of Polymeric Electrospun Nanofibres
Figure 7.15 Formation of ‘beads on a string’
Figure 7.16 Formation of minor‘beads on a string’
151
Advances in Nanofibre Research
References 1.
Y. Wan, Q. Guo and N. Pan, International Journal of Nonlinear Sciences and Numerical Simulation, 2004, 5, 5.
2.
J. He, Y. Wan and J. Yu, International Journal of Nonlinear Sciences and Numerical Simulation, 2004, 5, 3, 243.
3.
J. He, Y.Q. Wan and J.Y. Yu, International Journal of Nonlinear Sciences and Numerical Simulation, 2004, 5, 3, 253.
4.
J. He and Y.Q. Wan, Polymer, 2004, 45, 19, 6731.
5.
X-H. Qin, Y.Q. Wan and J.H. He, Polymer, 2004, 45, 18, 6409.
6.
S. Therona, E. Zussmana and A.L. Yarin, Polymer, 2004, 45, 2017.
7.
M. Demir, I .Yilgor, E. Yilgor and B. Erman, Polymer, 2002, 43, 3303.
8.
A. Ganan-Calvo, Journal of Aerosol Science, 1999, 30, 7, 863.
9.
J. Feng, Journal of Non-Newtonian Fluid Mechanics, 2003, 116, 55.
152
8
Formation of Polymeric Electrospun Nanofibres
8.1 Overview In the second half of the twentieth century, the use of polymers in daily life has grown tremendously. Polymers are used in different forms and for a wide range of applications. Noticeable among these are the synthetic and regenerated polymers that have found applications not only in the textile and apparel sector, but also in numerous industrial usages such as tyre cords, reinforcing and structural agents, barrier films, food and packaging industry, and automotive parts. The process of making fibres from polymers generally involves spinning, wherein the polymer is extruded through a spinnerete to form fibres under suitable shear rates and temperatures. This conventional fibre formation process is generally followed by drawing, which involves the plastic strands to increase the strength and modulus depending on whether the polymer is in the molten state or in solution; the process is likewise termed ‘meltspinning’ or ‘solution spinning’ respectively. Typical average diameters obtained by these convectional spinning methods are ≥10 μm [1]. Electrospinning is a process by which a polymer solution is charged to high voltage to produce submicron-scaled fibres [2, 3]. At a voltage sufficient to overcome surface tension forces, fine jets of polymer solution shoot out toward a grounded collector. The jet is stranded and elongates before it reaches the target, dries and is collected as an interconnected web of small fibres with typical diameter of several hundreds of nanometres. Developed more than 50 years ago, electrospinning is far from a new technique, but received an extensively renewed research interest in recent years [4].
8.2 Aim of the Project It has been found that the morphology of nanofibres is dependent upon many processing parameters. These parameters can be divided into three main groups (Table 8.1). The effects of ‘polymer concentration’, ‘applied voltage’ and ‘distance from needle to collector’ on the structure of polyacrylonitrile (PAN) nanofibres are the particular interest of this work.
153
Advances in Nanofibre Research
Table 8.1 Processing parameters in electrospinning Solution properties
Viscosity Polymer concentration* Molecular weight of polymer Molecular weight distribution Electrical conductivity Elasticity Topology (branched, linear) of the polymer Surface tension
Processing conditions
Applied voltage* Distance from needle to collector* Motion of target screen Volume feed rate Needle diameter
Ambient conditions
Temperature Humidity Atmospheric pressure
* These parameters were studied in this experiment
8.3 Experimental Our electrospinning apparatus consisted of a syringe pump, a 0–50 kV direct-current power supply, an ammeter, and various take-up devices, including metal screens, belts and other targets in an enclosed Faraday cage. PAN fibres (thickness = 16 dtex; length = 150 mm; bright) were used to prepare solutions. This polymer was dissolved in dimethylformamide (DMF) solvent at 40–50 ºC and at different concentrations.
154
Formation of Polymeric Electrospun Nanofibres
8.4 Results and Discussion At concentrations of 8 wt% (using 10 kV and 12 kV), formation of beaded fibres was observed. Using ≥14 kV, nanofibres without beads were formed (Figure 8.1). In Figure 8.1c the ‘fracture points’ (indicated by arrows) are due to the jet resistance to deformation during the whipping process.
(a)
(b)
(c)
(d)
Figure 8.1 Scanning electron microscopy (SEM) images of nanofibres produced from 8 wt% PAN/DMF: tip-target distance 10 cm, feed rate 0.5 ml/h, voltages used: (a) 10 kV; (b) 12 kV; (c) 14 kV; and (d) 16 kV
155
Advances in Nanofibre Research At this concentration (15 wt%), with increasing the applied voltage from 10 kV to 16 kV and holding the tip-target distance at 10 cm, nanofibres could not completely dry. At 10 kV and 14 kV, PAN webs with almost uniform diameter were obtained, but broader distributions in the diameter of nanofibres resulted at 16 kV. Fracture points were obtained at 16 kV. No bead defect was present under these conditions (Figure 8.2).
(a)
(b)
(c)
(d)
Figure 8.2 SEM images of nanofibres produced from 15 wt% PAN/DMF: tiptarget distance 15 cm, feed rate 0.5 ml/h, voltage: (a) 10 kV; (b) 12 kV; (c) 14 kV; and (d) 16 kV
156
Formation of Polymeric Electrospun Nanofibres At a concentration of 15 wt% and at a tip-target distance of 15 cm in comparison with 10 cm, the time for the solvent evaporate increased and, as a result, dry solid fibres were collected at the target. The distribution of diameters was higher at ≥12 kV, and fracture points and beaded fibres were formed at 14 kV (Figure 8.3).
(a)
(b)
(c)
Figure 8.3 SEM images of beaded nanofibres produced from: (a) 8 wt%; (b) 10 wt%; and (c) 15 wt%, PAN/DMF. Tip-target distance 10 cm, feed rate 0.5 ml/h, voltage 14 kV
Drastic morphological changes were found when the concentration of the polymer solution was changed. That is, the concentration or the corresponding viscosity was
157
Advances in Nanofibre Research
diameter (nm)
one of the most effective variables to control fibre morphology. To investigate the effect of concentration on fibre diameter, experiments were carried out at a concentration of 8, 10, 15 wt% of PAN/DMF, with a constant feeding rate of 0.5 ml/h and a tip-target distance of 10 cm. The applied voltages were varied from 10 kV to 16 kV. The effect of concentration on nanofibre diameter in these voltages is shown in Figures 8.4–8.7.
300 280 260 240 220 200 180 160 140 120 100
applied voltage = 10kv
6
8 10 12 14 concentration (wt%)
16
Figure 8.4 Effect of concentration on nanofibre diameter for applied voltage: 10 kV, feed rate: 0.5 ml/h and tip-target distance 10 cm
The results displayed graphically in Figure 8.4 show that, as the solution concentration increased, the average charge per mass ratio decreased, whereas the mean fibre diameter increased. Other considerations are the viscoelastic properties of the solutions, the solvent evaporation rate, and the details of the whipping motion. These figures indicated a uniform increasing trend in diameter of PAN webs with increasing concentration at 10 kV and 14 kV. A considerable amount of very fine nanofibres with diameters between 153 nm and 230 nm were produced when the concentration was 8 wt%.
158
diameter (nm)
Formation of Polymeric Electrospun Nanofibres
350 330 310 290 270 250 230 210 190 170 150
applied voltage = 12kv
6
8
10
12
14
16
concentration (wt%) Figure 8.5 Effect of concentration on nanofibre diameter in applied voltage: 12 kV, feed rate: 0.5 ml/h, tip-target distance: 10 cm
In Figure 8.5, it was observed that nanofibre diameter decreased with increasing concentration from 8 wt% to 10 wt% that originated from increasing the amount of strand or elongation of the jet due to increasing the electrostatic forces at 10 cm and 12 kV. As shown in Figures 8.6 and 8.7, there was a decrease in the average diameter of nanofibres with increasing concentration from 10 wt% to 15 wt%. In this case, at a tip target distance of 10 cm, the electric field strength is increased, so it will increase the electrostatic repulsive force on the fluid jet. Conversely, the solution will be removed from the needle tip more quickly, which favours the formation of thinner fibres. To investigate the effect of applied voltage on fibre diameter, experiments were carried out at voltages 10, 12, 14, and 16 kV with concentrations 8 wt% to 15 wt%, a flow rate of 0.5 ml/h, and a tip-target distance of 10 cm (Figures 8.8–8.11).
159
Advances in Nanofibre Research
310 diameter (nm)
280 250 220 190
applied voltage = 14kv
160 130 100
6
8 10 12 14 16 concentration (wt%)
diameter (nm)
Figure 8.6 Effect of concentration on nanofibre diameter in applied voltage: 14 kV, feed rate: 0.5 ml/h, tip-target distance: 10 cm
430 400 370 340 310 280 250 220 190 160 130 100
applied voltage = 16 kv
6
12 10 8 14 concentration (wt%)
16
Figure 8.7 Effect of concentration on nanofibre diameter in applied voltage: 16 kV, feed rate: 0.5 ml/h, tip-target distance: 10 cm
160
Formation of Polymeric Electrospun Nanofibres
240
diameter (nm)
220 200 180 concentration 8%
160 140 120 100
8
10
12
14
16
18
applied voltage (kv) Figure 8.8 Effect of applied voltage on nanofibre diameter at concentration: 8 wt%, feed rate: 0.5 ml/h, tip-target distance: 10 cm
450
diameter (nm)
420 390 360 330 300
concentration 10%
270 240 210 180 150 8
16 10 12 14 applied voltage (kv)
18
Figure 8.9 Effect of applied voltage on nanofibre diameter at concentration: 10 wt%, feed rate: 0.5 ml/h, tip-target distance: 10 cm 161
Advances in Nanofibre Research
350 330 310 diameter (nm)
290 270
applied voltage = 15 wt%, distance = 10 cm
250 230 210 190 170 150
8
10 12 14 16 applied voltage (kv)
18
diameter (nm)
Figure 8.10 Effect of applied voltage on nanofibre diameter at concentration: 15 wt%, feed rate: 0.5 ml/h, tip-target distance: 10 cm
350 330 310 290 270 250 230 210 190 170 150
concentration = 15 wt%, distance = 15 cm
8
10 12 14 16 applied voltage (kv)
18
Figure 8.11 Effect of applied voltage on nanofibre diameter at concentration: 15 wt%, feed rate: 0.5 ml/h, tip-target distance: 15 cm 162
Formation of Polymeric Electrospun Nanofibres This increasing/decreasing or opposing trend in diameter in the figures can be explained. The diameter of the jet reached a minimum after an initial increase in field strength and then became much larger with increasing electric field. With the increasing in applied voltage, the electric field strength also increased, so it increased the electrostatic repulsive force on the jet.
8.5 Conclusion 1) At 8 wt% and 14 kV, fibres without any beads were formed and the structure of the nanofibres was not completely dry. 2) At 10 wt%, uniform webs without any beads were formed. 3) At 15 wt% at 15 cm tip-target distance in comparison with 10 cm, the time for the solvent to evaporate increased, with increased distance between the needle and collector. 4) Higher voltage needed to provide a strong electric field and formation of thinner fibres. 5) It was observed that the diameter of the electrospun fibres was not dramatically changed with variation of applied voltage. 6) With increasing concentration, the fibre diameter was observed to increase in all voltages except at 12 kV due to increasing the electrostatic forces.
References 1.
S-H. Tan and R. Inai, Polymer, 2005, 46, 6128.
2.
J.M. Deitzel, Polymer, 2001, 42, 8163.
3.
S. Zarkoob, Polymer, 2004, 45, 3973.
4.
S. Atheron, Polymer, 2004, 45, 2017.
163
Advances in Nanofibre Research
164
9
Experimental Study on Electrospinning of Polymeric Nanofibres
9.1 Introduction Polymers that exhibit high electrical conductivity have been synthesised in the last few decades [1]. These electrically conductive polymers have increasing numbers of applications in different areas of microelectronics and chemical analyses. Traditional formulations for the calculation of a metal resistor can be written in the form of:
(9.1) where R is the resistance of a conductor, A is the section area, L its length and k is the resistance coefficient. Actually, Equation 9.1 is valid only for metal conductors if there are plenty of electrons in the conductor. He [2] modified Equation 9.1 based on allometric scaling law to accurately describe polymer conduction. Among all conducting polymers, polyaniline has been of particular interest because of its environmental stability, controllable electrical conductivity and interesting redox properties associated with the chain nitrogen. Polyaniline also exhibits solutionor counter-ion-induced processability. Furthermore, the electrical properties of polyaniline can be substantially improved through secondary doping. The excellent processability, together with the presence of several intrinsic redox states, has enhanced the potential applications of aniline-based polymers for use in devices [3]. Polyaniline can be used in diodes, field effect transistors, and different types of sensors (e.g., biosensors, gas sensors, humidity sensors) [4–8]. The electrical conductivity of conductive polymers may increase if the polymers are in the nanoscale [9]. Systems at the nanoscale may possess entirely new physical and chemical characteristics For example, higher electrical conductivity arises if the size of a wire is reduced below certain critical thickness (nanoscale). Using these properties raises the potential of electrospinning to make fibres at the nano-level with unusual properties that is impossible at the level of the visible world [9]. Using the
165
Advances in Nanofibre Research nanofibre structure of conducting polymers, due to the high surface-to-volume ratio of nanofibres, the performance of conducting polymers can be enhanced. It has been reported that optical sensors based on electrospun nanofibres showed sensitivity up to three orders of magnitude higher than that obtained from thin-film sensors for the detection of nitro compounds, ferric ions, and mercury ions [10–12]. The higher sensitivities reported for these electrospun nanofibres can be attributed to their high ratios of surface area-to-volume. Electrospinning is one of the processes that can be used for producing nanofibres and nanoporous materials [13, 14]. It uses a high voltage electric field to produce high surface area submicron fibres. Electrospinning is a straightforward method for producing continuous polymeric fibres with diameters in the nanometre range [13]. Several studies have been done to produce nanofibres of polyaniline. Reneker and Chun reported that polyaniline fibres could be successfully electrospun from sulfuric acid into a coagulation bath. Similar work was done by MacDiarmid and co-workers in which the average diameter was reported to be ~139 nm [15, 16]. Later on, work was done on electrospinning polyaniline/polyethylene oxide (PEO) blend. PEO was added to assist in fibre formation. Norris and co-workers reported conducting ultrafine fibres with diameters <2 µm using an electrospinning process [17]. Leon reported electrospinning of polyaniline/polystyrene with diameters <100 nm [18]. Fabrication of polyaniline-based nanofibres with diameters <30 nm has also been reported [19]. In the present work, we used a polyaniline/polyacrylonitrile blend to form a non-woven mat. Polyaniline exists in many intrinsic redox states. The half-oxidised emeraldine base is the most stable and widely investigated state in the polyaniline family that can be dissolved in N-methyl-2-pyrolidone (NMP). Polyaniline emeraldine base/ polyacrilonitrile blend solution in NMP was prepared and electrospun with different blending ratios. Fibre diameter, fibre morphology and electrical conductivity of the mats were analysed and discussed.
9.2 Experimental 9.2.1 Materials Commercial polyacrylonitrile (PAN) polymer containing 6% methylacrylate with a molecular weight of 100,000 was supplied by the Polyacryl Iran Company. N-methyl2-pyrolidone (NMP) was from Riedel-de Haën. Aniline was from Merck and was vacuum-distilled before use. The polyaniline (PANi) used was synthesised in our laboratory.
166
Experimental Study on Electrospinning of Polymeric Nanofibres
9.2.2 Sample Preparation Polyaniline was synthesised by the oxidative polymerisation of aniline in acidic media. Three millilitres of distilled aniline was dissolved in 150 ml of 1 N HCl and kept at 0–5 oC. A total of 7.325 g of (NH4)2S2O8 was dissolved in 35 ml of 1 N HCl and added dropwise under constant stirring to the aniline/HCl solution over 20 minutes. The resulting dark-green solution was maintained under constant stirring for 4 hours. The prepared suspension was dialysed in a cellulose tubular membrane (Dialysis Tubing D9527, molecular cutoff = 12,400; Sigma–Aldrich) against distilled water for 48 hours. Then it was filtered and washed with water and methanol. The synthesised polyaniline was added to 150 mL of 1 N (NH4)OH solution. After an additional 4 hours, the solution was filtered and a deep-blue emeraldine base form of polyaniline obtained. The synthesised polyaniline was dried and crushed into fine powder and then passed through 100 mesh. The intrinsic viscosity of the synthesised polyaniline dissolved in sulfuric acid (98%) was 1.18 dl/g at 25 oC. PANi solution with a concentration of 5% (w/w) was prepared by dissolving an exact amount of PANi in NMP. PANi was slowly added to the NMP with constant stirring at room temperature. This solution was then allowed to stir for 1 hour in a sealed container. A 20% (w/w) solution of PAN in NMP was prepared separately and added dropwise to the well-stirred PANi solution. The blend solution was allowed to agitate with a mechanical stirrer for an additional 1 hour. Various polymer blends with PANi content ranging from 10 wt% to 30 wt% were prepared by mixing different amounts of 5% PANi solution and 20% PAN solution. The total concentration of the blend solutions were kept as 12.5%.
9.2.3 Electrospinning Polymeric nanofibres can be made using the electrospinning process, which has been described in the literature and in patents [20–21]. Electrospinning uses a high electric field to draw a polymer solution from the tip of a capillary towards a collector. A voltage is applied to the polymer solution that causes a jet of the solution to be drawn towards a grounded collector. The fine jets dry to form polymeric fibres, which can be collected as a web. Our electrospinning equipment used a variable high-voltage power supply from Gamma High Voltage Research. The applied voltage can be varied from 1 kV to 30 kV. A 5 ml syringe was used and a positive potential applied to the polymer blend solution by attaching the electrode directly to the outside of the hypodermic needle (internal diameter, 0.3 mm). The collector screen was a 20 × 20 cm aluminium foil placed 10
167
Advances in Nanofibre Research cm horizontally from the tip of the needle. An electrode of opposite polarity was attached to the collector. A metering syringe pump from New Era Pump Systems Incorporated was used. It was responsible for supplying polymer solution at a constant rate of 20 µl/min. Electrospinning was done in a temperature-controlled chamber. The temperature of the electrospinning environment was adjusted to 25, 50 and 75 °C (Figure 9.1). A factorial experiment was designed to investigate and identify the effects of parameters on fibre diameter and morphology (Table 9.1).
Syringe Metering pump
Connector 0
1
Electrical heater High voltage supply
25 ºC
Temperature controller
Figure 9.1 Electrospinning setup
9.2.4 Characterisation Shear viscosities of the fluids were measured at a shear rate of 500 s-1 and 22 °C using a Brookfield viscometer (DVII+). Fibre formation and morphology of the electrospun PANi/PAN fibres were determined using a scanning electron microscope (SEM) Philips XL-30A. A small section of the prepared samples was placed on a SEM sample holder and then coated with gold by a BAL-TEC SCD 005 sputter coater. The
168
Experimental Study on Electrospinning of Polymeric Nanofibres diameter of electrospun fibres was measured with image analyser software (manual microstructure distance measurement). For each experiment, average fibre diameter and distribution were determined from ~100 measurements of the random fibres. The electrical conductivity of the electrospun mats was measured by the standard four- probe method after doping with HCl vapor.
Table 9.1 Shear viscosity of PANi/PAN blend solutions at 22 oC and shear rate of 500 s–1 and average diameter of electrospun nanofibres PANi/PAN blend ratio percentage (w/w)
Shear viscosity (Pa.s)
Average nanofibre diameter (nm)
100/0 (5% solution)
0.159
No fibre
30/70
0.413
164
20/80
0.569
425
10/90
0.782
602
0/100
1.416
652
9.3 Results and Discussion Published literature has shown that, in the electrospinning process, the system configuration and operation conditions differ vastly from one material to another, depending on the material and choice of solvent. The physical and chemical parameters of polymer solutions such as viscosity, electrical conductivity, surface tension and air temperature can determinedly affect the formability and morphology of electrospun fibres. In the following sections, the effects of some electrospinning parameters on the fibre formation and morphology of PANi/PAN blend solutions are discussed and the optimal condition for obtaining PANi/PAN fibres examined.
9.3.1 Effect of Polyaniline Content We could not obtain the fibres from the pure PANi solution because a stable drop at the end of the needle was not maintained. Figure 9.2 shows the SEM micrographs of PANi nanoparticles electrospun from pure PANi solution. As seen in Figure 9.2,
169
Advances in Nanofibre Research most of PANi particles have a round shape, whereas a fibrous structure is not observed. The major complication in the electrospinning of PANi is the poor solubility of PANi. At low polymer concentration, the solution does not contain sufficient material to produce stable solid fibres. With increasing polymer concentration, the number of insoluble PANi particles in the solution increases rapidly, resulting in an un-spinnable solution. Therefore, we prepared PANi/PAN blend solutions with different PANi content using NMP as solvent. At PANi content >30% regardless of electrospinning conditions, drops were formed instead of fibres. Experiments were carried out when the PANi weight percent was varied from 10% to 30%. The applied voltage was 20 to 30 kV and the chamber temperature was held at 25, 50 and 75 °C. Figure 9.2 shows the SEM micrographs and the surface morphology of obtained fibres at 25 °C and 25 kV. In a solution containing 30% PANi, the fibrous structure was not completely stabilised and a bead-on-a-string structure with non-uniform morphology was obtained. The fibres between the beads had a circular cross-section, with a diameter typically between 60 nm and 460 nm and mean fibre diameter of 164 nm. As the PANi content decreased to <20%, a fibrous structure was stabilised. At 20% PANi content, the mean diameter of fibres increased to 425 nm with some beads on the fibres. At 10% PANi content, continuous fibres without beads resulted regardless of the electric field with a mean fibre diameter of 602 nm at 25 kV. Smooth and uniform fibres with an average diameter of 652 nm were electrospun from PAN solution at the same electrospinning condition. These results reveal that as the PANi contents in the blends increase up to 30%, the average diameter of blend nanofibre gradually decreases from 602 nm to 164 nm, and its distribution becomes significantly broader with a higher standard deviation (Figure 9.2). It is also observed that fibres with non-uniform morphology are electrospun at 25 °C. Figure 9.3 shows SEM photomicrographs of electrospun PANi/PAN blend fibres at 50 °C at various blend ratios. This figure shows that fibres with uniform morphology without remarkable beads are formed regardless of PANi content.
170
Experimental Study on Electrospinning of Polymeric Nanofibres
Figure 9.2 SEM micrographs of electrospun fibres at an applied voltage of 25 kV and temperature of 25 oC with a constant spinning distance of 10 cm
It is also observed that at 50 °C the average diameter of electrospun fibres decreases from 194 nm at 10% PANi content to 124 nm at 30% PANi content. Similar to the results obtained at 25 °C, fibre formation from pure PANi solution and blends containing >30% PANi was not possible. In electrospinning, the coiled polymer chains in the solution are transformed by the elongational flow of the jet into oriented entangled networks. Experimental observations in electrospinning confirm that for fibre formation to occur, a minimum chain entanglement is required [22]. Below this critical chain entanglement, application of voltage results in beads and droplets due to jet instability. The gradual increase in fibre diameter with content of PAN in the blends 171
Advances in Nanofibre Research may be explained by the increase in solution viscosity due to a higher viscosity of PAN solution. Shear viscosity values of the PANi/PAN blends are shown in Table 9.2. It is clearly seen that the shear viscosity of the solutions decrease with PANi content in the blends. Therefore, as the concentration of PAN in the blend is increased, the solution viscosity and resulted polymer chain entanglements increase significantly. During electrospinning, the stable jet ejected from Taylor’s cone [13] is subjected to tensile stresses and may undergo significant elongational flow. The nature of this elongational flow may determine the degree of stretching of the jet. The characteristics of this elongational flow can be determined by the elasticity and viscosity of the solution. The results show that the viscosity of the PAN solution is higher than that of the PANi solution. Hence, the viscosity of the blend solution decreases with an increase in PANi content. Therefore, jet stretching during the electrospinning is more effective at higher PANi content. As a result, the fibre diameters decrease with increasing PANi content in the blends. Conversely, at high PANi content, an insufficiently deformable entangled network of polymer chain exists, and the ejected jet reaches the collector before the solvent fully evaporates. Therefore, at low solution viscosity, ejected jet breaks into droplets and a mixture of beads and fibres is obtained. This explains the formation of droplets and beads at high PANi content. The effect of electrospinning temperature is discussed in the following section. He and coworkers [23] showed that the diameters of electrospun nanofibres are greatly affected by solution viscosity, and solution viscosity has an allometric relationship with its concentration. Our results shows that the electrospun nanofibres diameters (d) of PANi/PAN blends has a relationship with PANi content in the form of:
(9.2)
9.3.2 Effect of Electrospinning Temperature Studies on electrospinning show [13, 24–26] that many parameters may influence the transformation of polymer solution into nanofibres. Some of these parameters include: (1) solution-related properties such as viscosity and surface tension; (2) process variables such as electric potential at the capillary tip; and (3) ambient parameters such as air temperature in the electrospinning chamber. To study the effect of electrospinning temperature on the morphology and texture of electrospun PANi/PAN nanofibres, a solution containing 20% PANi was electrospun at 25, 50 and 75 °C. SEM micrographs of electrospun fibres at 20 kV are shown in Figure 9.4. Interestingly, electrospinning of the solution shows bead-free fibre morphology at 50 and 75 °C, whereas fibres with large beads are observed at 25 °C, especially at high PANi contents (Figure 9.4). The electrospun sample at 25 °C shows fibres with several beads and non-uniform surface morphology. 172
Experimental Study on Electrospinning of Polymeric Nanofibres
Figure 9.3 SEM micrographs of electrospun nanofibres at an applied voltage of 25 kV and temperature of 50 oC with a constant spinning distance of 10 cm
Table 9.2 Shear viscosity of PANi/PAN blend solutions at 22 oC and shear rate of 500 s–1 and average diameter of electrospun nanofibres PANi/PAN blend ratio percentage (w/w)
Shear viscosity (Pa.s)
Average nanofibre diameter (nm)
100/0 (5% solution)
0.159
No fibre
30/70
0.413
164
20/80
0.569
425
10/90
0.782
602
0/100
1.416
652
173
Advances in Nanofibre Research With an increase in electrospinning temperature, fibre morphology gradually changes from a mixture of beads and fibres through uniform fibres. As shown in Figure 9.4, at 50 °C, continuous fibres with uniform morphology were obtained, whereas increasing the electrospinning temperature to 75 °C caused bead-free but fragile and cracked fibres. Diameter measurement of electrospun fibres at 25 °C showed a size range of ~400–700 nm (with 480 nm being the most frequently occurring). These data were within the same range of the reported size for electrospun PANi/PEO nanofibres [18]. With increasing the electrospinning temperature to 50 °C, fibre diameter was decreased to a range of ~110–290 nm (with 170 nm the most occurring frequency). At 75 °C, fibres dimensions were 70–170 nm (with 110 nm the most occurring frequency). It was obvious that that diameter of electrospun fibres decreased with increasing electrospinning temperature. The distributions of fibre diameters electrospun at 25, 50 and 75 °C are shown in Figure 9.5. At 25 °C, a broad distribution of fibre diameters was obtained, whereas a narrow distribution of fibre diameters was observed at 50 and 75 °C. Several factors with PANi/PAN blends may explain the effects of electrospinning temperature and PANi content on morphology of the electrospun fibres. Nanofibres result from evaporation of solvent from polymer solution jets, so the fibre diameters will be dependent upon the jet sizes, elongation of the jet and evaporation rate of the solvent [24]. At constant PANi content, as the electrospinning temperature is increased, the rate of solvent evaporation from the ejected jet increases significantly. In the case of electrospinning at 25 °C, due to the high boiling point of NMP (~202 °C), fibres with relatively high solvent content travel during electrospinning and reach the collector. Therefore, the collected fibres have irregular morphology due to contraction of the fibres during the electrospinning and on the collector. At higher electrospinning temperature, the rate of solvent evaporation from the ejected jet increases significantly and a skin is formed on the surface of the jet, which results in collection of dry fibre with smooth surface. The presence of a thin, mechanically distinct polymer skin on the liquid jet during electrospinning has been discussed by Koombhongse and co-workers [27]. Conversely, a higher electrospinning temperature results in a higher degree of stretching and more uniform elongation of the ejected jet due to higher mobility and lower viscosity of the solution. Therefore, fibres with smaller diameters and a narrower distribution of diameters will be electrospun at a higher electrospinning temperature.
174
Experimental Study on Electrospinning of Polymeric Nanofibres
Figure 9.4 SEM micrographs of electrospun nanofibres at an applied voltage of 20 kV and PANi content of 20% with a constant spinning distance of 10 cm
175
Frequency
Advances in Nanofibre Research
18 16 14 12 10 8 6 4 2 0 0
40
0
42
0
44
0
46
e 0 0 0 0 0 0 0 0 0 48 50 52 54 56 58 60 62 64 Mor Diameter Range (nm)
(a) 60
Frequency
50 40 30 20 10 0
50
70
90
0 0 0 0 0 0 0 0 0 ore 0 11 13 15 17 19 21 23 25 27 29 M Diameter Range (nm)
(b) 60
Frequency
50 40 30 20 10 0
50
70
90
110
130
150
170
190
210 More
Diameter Range (nm)
(c)
Figure 9.5 Distribution of fibre diameter electrospun at PANi content of 20%, applied voltage of 20 kV, spinning distance of 10 cm and electrospinning temperature of (a) 25 ºC; (b) 50 ºC; and (c) 75 ºC 176
Experimental Study on Electrospinning of Polymeric Nanofibres
9.3.3 Effect of Applied Voltage
60 40
80
12 0 26 0 20 0 M or e
20 0
0 40
Occurrence
To study the effects of applied voltage, the blend solutions were electrospun at various applied voltages and temperatures. From the results shown in Figure 9.6, it is obvious that the diameter of electrospun PANi/PAN fibres at 50 °C decreased as the applied voltage increased. Similar results were observed for electrospun fibres at 25 and 70 °C (results not shown). The same results were found by Fenessey and co-workers [28], Ding and co-workers [29] and others [30–31]. However, the results are contradictory with the results obtained by Renker and co-workers [32] and Gu and co-workers [33], who found insignificant changes of the diameter of electrospun fibres over the range of applied voltage. This inconsistency may be due to a difference in experimental conditions. The flow rate of solution in our experiments was maintained constant with the help of a syringe pump, whereas in the experiments of Gu and co-workers [33] the solution was brought down automatically by the electrostatic force and hydrodynamic force of the fluid. Therefore, the flow rate in their experiments was not constant. The diameter of fibres is a combination of the results of flow rate and electrostatic force due to applied voltage. Increasing the applied voltage at constant flow rate increases the electrostatic force and creates smaller-diameter fibres. However, if the increasing of the applied voltage draws more solution out of the capillary, the fibre diameter would increase with increasing applied voltage, as reported by Demir and co-workers [34]. In some reports [33], a combination of increasing of the applied voltage and flow rate meant that the fibre diameter did not significantly change with the applied voltage.
Fiber diameter (nm) Figure 9.6 Average fibre diameter of electrospun fibres at various applied voltages and PANi content at temperature of 50 oC and electrospinning distance of 10 cm
177
Advances in Nanofibre Research
9.4 Conclusions The electrospinning of PANi/PAN blend in NMP was investigated and fibres with diameter ranging from 60 nm to 600 nm obtained depending on electrospinning conditions. The morphology of fibres was investigated at various blend ratios and electrospinning temperature. At 30% PANi content and 25 °C, fibres with an average diameter of 164 nm were formed with beads (droplets of polymer over the woven mat) and non-uniform morphology. At this condition, solution viscosity and chain entanglements may not be sufficient, resulting in spraying of large droplets connected with very thin fibres. The average values of fibre diameters decreased with PANi content in the solutions, but PANi/PAN solution containing >30% PANi did not form a stable jet regardless of applied voltage and electrospinning temperature. For pure PANi solution, because the viscosity is too low to obtain stable drops and jets, we could not get the fibres. It was found that at 25 °C, fibre morphology was changed to beaded fibres when PANi content was >20%. With increasing electrospinning temperature, the morphology changed from beaded fibres to a uniform fibrous structure, and the fibre diameter decreased from 500 nm to 100 nm when the electrospinning temperature changed from 25 to 75 °C. The mean fibre diameter is the smallest and the fibre diameter distribution is narrowest for electrospun fibres at 75 °C. However, some cracks are observed on the surface of the electrospun fibres. There was a slight decrease in average fibre diameter with increasing applied voltage. It is concluded that the optimal condition for nanoscale and uniform PANi/PAN fibre formation is 20% PANi content and 50 °C electrospinning temperature regardless of the applied voltage.
References 1.
N. Hall, Chemical Communications, 2003, 1, 1.
2.
J.H. He, Polymer, 2004, 45, 9067.
3.
E.T. Kang, K.G. Neoha and K.L. Tan, Progress in Polymer Science, 1998, 23, 277.
4.
S.A. Carter, M. Angelopoulos, S. Karg, P.J. Brock and J.C. Scott, Applied Physics Letters, 1997, 70, 2067.
5.
C.T. Kuo, R.L. Huang and S.Z. Weng, Synthetic Metals, 1997, 88, 101.
6.
B. Adhikari and S. Majumdar, Progress in Polymer Science, 2004, 29, 699.
7.
M. Trojanowicz, Microchimica Acta, 2003, 143, 75.
178
Experimental Study on Electrospinning of Polymeric Nanofibres 8.
E. Bakker, Analytical Chemistry, 2004, 76, 3285.
9.
S.H. Lee, B.C. Ku, X. Wang, L.A. Samuelson and J. Kumar, Material Research Society Symposium Proceedings, 2002, 708, 403.
10. J.H. He, Y.Q. Wan and L. Xu, Chaos, Solitons & Fractals, 2007, 33, 26. 11. X.Y. Wang, S.H. Lee, C. Drew, K.J. Senecal, J. Kumar and L.A. Samuelson, Material Research Society Symposium Proceedings, 2002, 708, 397. 12. X.Y. Wang, C. Drew, S.H. Lee, K.J. Senecal, J. Kumar and L.A. Samuelson, Nano Letters, 2004, 2, 1273. 13. Z.M. Huang, Y.Z. Zhang, M. Kotaki and S. Ramakrishna, Composite Science and Technology, 2003, 63, 2223. 14. L. Xu, J.H. He and Y. Liu, International Journal of Nonlinear Science and Numerical Simulation, 2007, 8, 2, 199. 15. D.H. Reneker and I. Chun, Nanotechnology, 1996, 7, 216. 16. A.G. MacDiarmid, W.E. Jones, Jr., I.D. Norris, J. Gao, A.T. Johnson, Jr., N.J. Pinto, J. Hone, B. Han, F.K. Ko, H. Okuzaki and M. Llaguno, Synthetic Metals, 2001, 119, 27. 17. I.D. Norris, M.M. Shaker, F.K. Ko and A.G. MacDiarmid, Synthetic Metals, 2000, 114, 109. 18. M.J. Díaz-de León in the Proceedings of the National Conference on Undergraduate Research, University of Kentucky, Lexington, KY, USA, 2001, 15-17 March. 19. Y. Zhou, M. Freitag, J. Hone, C. Staii, A.T. Johnson, N.J. Pinto and A.G. MacDiarmid, Applied Physical Letters, 2003, 83, 18. 20. A. Formhals, inventors; Richard Schreiber-Gastell, assignee; US 1975504, 1934. 21. W.E. Teo and S. Ramakrishna, Nanotechnology, 2006, 17, R89. 22. S.L. Shenoy, W.D. Bates, H.L. Frisch and G.E. Wnek, Polymer, 2005, 46, 3372. 23. J.H. He, Y.Q. Wan and J.Y. Yu, Fibres and Polymers, 2008, 9, 2, 140.
179
Advances in Nanofibre Research 24. J.M. Deitzel, J. Kleinmeyer, D. Harris and N.C. Beck Tan, Polymer, 2001, 42, 261. 25. S.A. Theron, E. Zussman and A.L. Yarin, Polymer, 2004, 45, 2017. 26. S.H. Tan, R. Inai, M. Kotaki and S. Ramakrishna, Polymer, 2005, 46, 6128. 27. S. Koombhongse, W. Liu and D. Denker, Journal of Polymer Science, Part B: Polymer Physics Edition, 2001, 39, 2598. 28. S.F. Fennessey and R.J. Farris, Polymer, 2004, 45, 4217. 29. B. Ding, H.Y. Kim, S.C. Lee, C.L. Shao, D.R. Lee, S.L. Park, G-B. Kwag and K-J. Choi, Journal of Polymer Science, Part B: Polymer Physics Edition, 2002, 40, 13, 1261. 30. X.M. Mo, C.Y. Xu, M. Kotaki and S. Ramakirishna, Biomaterials, 2004, 25, 1883. 31. D.S. Katti, K.W. Robinson, F.K. Ko and C.T. Laurencin, Journal of Biomedical Materials Research Part B, 2004, 70, 286. 32. D.H. Reneker and I. Chun, Nanotechnology, 1996, 7, 216. 33. S.Y. Gu, J. Ren and G.J. Vancso, European Polymer Journal, 2005, 41, 2559. 34. M.M. Demir, I. Yilgor, E. Yilgor and B. Erman, Polymer, 2002, 43, 3303.
180
A
bbreviations
AFM
Atomic force miscroscopy
CV
Current of variation
DMF
Dimethylformamide
EOS
Equivalent opening size
FFT
Fast Fourier transform
MFD
Mean fibre diameter
NMP
N-methyl-2-pyrolidone
ODF
Orientation distribution function
PAN
Polyacrylonitrile
PANi
Polyaniline
PDLA
Poly D,L-lactide
PEO
Polyethylene oxide
POA
Percent open area
PSD
Pore-opening size distribution
PVA
Polyvinyl alcohol
RSM
Response surface methodology
SEM
Scanning electron microscopy
SF
Silk fibroin
181
Advances in Nanofibre Research StdFD
Standard deviation of fibre diameter
TEM
Transmission electron microscopy
182
I
ndex
34 Full factorial experimental design 19-20 3-by-3 Neighbourhood block 76-77 3-GHz Processor 109
A Adhesive 15 Air filtration 121 Angular Density 48, 80, 84, 106 Orientation 130 Power spectrum 129 Power spectrum histogram 130 Applied voltage 16, 18-20, 25, 33-39, 124, 127, 139, 144-145, 153, 156, 159-162, 167, 171, 173, 175-177 Atomic force microscopy 69
B BAL-TEC SCD 005 sputter coater 147, 168 Bead formation 140, 143, 150 Bead-on-a-string structure 150-151, 170 Beads 122-127, 141-142, 145, 150, 163, 174 Berry number 18 Biocompatibility 15 Bridge formation 101 Brookfield viscometer 168 Bubble Point Method 100
C Carbon fibres 13 Carbonisation 13 Cellulose tubular membrane 167 Centre point 75 Chessboard-distance 71 City block 72 183
Advances in Nanofibre Research Distance transform 75 Metric 71 Coded variables 25-26 Coefficient of variation 14, 32 Combing 133 Contour plots 25 Conventional fibres 134 Cotton fibre 133 Coulomb repulsion 145 Coulomb’s Law for the E field 4
D Degree of hydrolysis 15 Degummed silk 124 Dilation operation 51, 81 Dimethylformamide 154 Direct tracking 46, 51-52, 58, 60, 63-64 Algorithms 52 Distance from needle to collector 153 Distance map 70, 72-74, 78 Distance Transform Method 46, 49, 50, 52, 58, 60, 63-64, 70, 74-75, 91, 93 Distance-transformed image 50, 70, 72, 74-75, 77 Doping 165 Drawing 133 Drum collector 138 Dry fibres 19 Dry sieving 101
E Electrohydrodynamic instability theory 4 Electrospinning Equipment 167 Jet 150 Parameters 14-15, 19, 25, 46, 106, 108, 125 Waste cotton 133 Window 123, 142 Fibre formation 150 Fibres 32, 134, 150, 163, 169, 171, 174, 177-178 Fibrewebs 79 Nanofibre 14, 16, 34, 104, 105, 118, 122, 123, 139, 140, 141, 142, 143, 144, 146, 166, 172, 173, 175
184
Index Electrospun Nanofibre mat 11, 17, 131 Nanofibre webs 67, 82, 98 Non-woven webs 45-46, 51, 63-64 PANi/PAN fibres 168 Poly (vinylidene fluoride) fibrous electrolyte 13 Polyaniline/polyacrylonitrile NMP 147 Polymer fibres 137 PVA fibres 16 Scaffolds 13 Web 45-48, 61, 68, 72, 79, 82-83, 91, 97, 100, 103, 105, 109, 110, 118 Electrospun-regenerated Bombyx mori silk 13 Electrostatic force 97 Electrostatic stretching force 32-33 Empirical Model 7, 12, 15, 31 Empirical modeling 24 Equivalent opening size 99-100 Erosion operation 51, 53, 81 Error term 24 Euclidean distance 71 Map 73 Metric 71-72, 75
F Fabrication 97 Face-centred central composite design 14 Faraday cage 154 Fast Fourier Transform 125-127, 129 Feed rate 2, 124, 144, 158, 160-162 Fibre Beaded 18, 38, 137, 140, 144, 155, 157 Diameter measurement 48 Formation 19, 153, 166, 168-169 Identification 51 Micro 134-135 Morphology 3, 144-145, 150, 158, 166, 174, 178 Bead-free 172 Spinning 45, 133 Spun 126 Sub-micron 133, 137, 166 Uniform 36, 39 First-order (linear) models 24
185
Advances in Nanofibre Research Flow porosimetry (Bubble Point Method) 100, 102, 108-109 Flow rate 19-20, 25, 35, 37 Four-probe Method 169 Fracture points 155
G Glass transition temperature 97 Glass-bead mixtures 101 Global thresholding 52-53, 80-82, 104-105
H Half-a-pixel error 90 High-voltage direct-current power supply 16 Hydrodynamic sieving 101 Hydrolysis 15 Hypotheses 27 Hypothesis-testing procedures 26
I Image analysis 67, 80, 91, 97-98, 103-104, 108-109, 111, 118, 125, 131, 169 Image processing algorithms 125 Image size 48, 80, 106 ImageJ programme 126 Intensity adjustment operation 52 Internal randomness 47, 79, 105 Intrinsic viscosity 18
J Jet stretching 172
L Least squares method 26 Line density 84 Line thickness 48, 80, 106 Linear regression analysis 25 Liquid filtration 121 Local thresholding 53, 82, 104-105 Low-order polynomials 24
M Manual Method 48-49, 60, 69, 90-91, 93 Mark-Houwink-Sakurada Equation 18
186
Index MatLab programme 130 Mean fibre diameter 6-7, 13-14, 25-28, 28, 31-34, 35, 36, 38-39 Mean free path (μ-randomness) 47-48, 54, 63, 79-80, 91, 105-106 Mechanical fibre-spinning technologies 1 Medial-axis transformation 70 Melting temperature 97 Melt-spinning 153 Mercury porosimetry 100-101, 108 Modeling 14 Momentum Equation 4 Morphological opening 81 Morphological opening operation 53 Morphology 11, 14, 16, 34, 122, 127-128, 137, 139, 144-145, 147, 150, 153, 168-169, 172, 174
N Nanocomposites 134 Nanofibre Beaded 157 Diameter 56 Orientation 130 Polyaniline-based 166 Webs 106, 108 Electrospun 130 Nanofibres PVA 38 Nanofibrous mats 125 Nanofibrous media 121-122, 124, 126 Nanotechnology 8, 133 Natural (uncoded) variables 29 Natural fibres 8, 133 New Distance Transform Method 74, 78-79, 93 N-methyl-2-pyrolidone 145-147, 150, 166-167, 170, 174, 178 Non-uniform Fibres 36, 38-39 Morphology 170 Surface morphology 172 Non-woven webs 105 Null hypothesis 27
O One-factor-at-a-time approach 19 One-pixel error 60
187
Advances in Nanofibre Research Online quality-control technique 109 Opening and cleaning 133 Operator-based method 69 Optical microscope (Nikon Microphot-FXA) 147 Orientation distribution function 122, 131 Otsu’s Method 53, 64, 81, 91, 104 Oxidative polymerisation 167
P Polyacrylonitrile (PAN) 14, 124-125, 129, 147, 149, 150, 171-172 Fibres 154 Nanofibres 13, 128, 131, 153 Nanomats 126 Solution 146 Webs 158 PAN/DMF 158 Polyaniline 145-147, 149, 165-167, 170, 172, 175-178 Ratio 148 Solution 169, 171 PANi/NMP solution 150 PANi/PAN Blend solutions 170, 173 Blends 172, 174, 178 Fibres 169, 177 PANi/PEO nanofibres 174 Paper-coating, fibres 15 Parasitic components 70 Percent open area 97, 99 Philips (XL-30) environmental scanning electron microscope 54, 82, 106, 125 Poly D, L-Lactide 13-14 Polyacrylonitrile polymer 166 Polyaniline/polyethylene oxide blend 166 Polyaniline-based nanofibres 166 Polymer Aniline-based 165 Concentration 14, 18, 32, 34, 153 Conduction 165-166 Fibres 1, 121, 133 Morphology 147 Nanofibres 97, 137 Solution 11, 45, 67, 97, 137, 146, 150, 153, 157, 167-169, 172, 174 Polymerisation 146
188
Index Stabilisers 15 Polynomials 24 Polyvinyl alcohol 18, 54, 60, 82, 106, 108 Concentration 18 Fibre diameter 15 Hydrogels 15 Nanofibres 16, 38 Solution 16 Pore-opening size distribution 97-98, 100-102, 104, 108-110, 113, 115-117 Porosity 98-99, 103, 108, 114, 118, 131, 133 Porosity analysis 126 Pruning 49-50, 58, 71, 73-74, 77, 90
Q Quadratic Model 24-26
R Radial momentum balance 6 Rayleigh instability 4 Real responses 31 Real Web 62-63, 82, 91-93, 104-105 Fibre bundling 90 Treatment 52 Resistance coefficient 165 Response surface methodology analysis 7, 13, 15, 24-25, 38 Root mean square errors 31, 39 Roving-pulling 133
S Scanning electron microscope 16-17, 38, 48, 60, 68-69, 81, 91, 104, 109, 125126, 130, 147, 155, 168-171, 173, 175 Screen collector 138 Second-order (quadratic) Model 24 Segmentation 52 Sensors 165 Sieving Methods 100, 103, 108-109 Silk Fibre wastes 124 Fibroin 124 Nanofibres 127, 131 Simulated Images 85-86, 88-90, 112-116, 118
189
Advances in Nanofibre Research Webs 105 Simulation 48 Algorithms 47 Single-jet electrospinning systems 1, 134 Skeletonisation 49, 58, 70-72 Slender-body Theory 4 Slicing algorithm 103 Sliding neighbourhood operation 75-77 Spheres 145 Spinning distance 14, 16, 18-20, 25, 31-35, 37-39, 36 Spinning process 102 Spurs 70 Square opening 99 Stabilised bead-free fibres 18 Standard deviation of fibre diameter 25-28, 31, 34, 36, 38-39 Standard mechanical fibre-spinning technologies 133 Statistical software minitab 15 25 Surface morphology 150, 170 Surface randomness 47, 79, 105 Surface tension 45 Swelling 103
T Taylor’s cone 11, 34, 45, 137, 144, 172 Textile sizing 15 Thinning 49, 70-73, 75 Thresholding 52, 64, 80, 82, 90 Tip-target distance 159-163 Tissue engineering 45 Tissue-scaffold materials 97 Top-hat transformation 81-82 Transmission electron microscopy 68, 104
V Varicose instability 4 Viscoelastic forces 32, 145 Viscosity 16, 122-123, 139-141, 145, 147, 150, 157, 167, 169, 172-174 Voids 98, 103 Volume feed rate 139, 145 Volume flow rate 16, 34, 38
190
Index
W Web density 48, 80, 106, 111, 113, 117-118 Wet sieving method 101 Wet/dry fibre spinning 1, 134 Whipping 155 Motion 158
191
Advances in Nanofibre Research
192